SEARCH

SEARCH BY CITATION

Keywords:

  • cholesterol;
  • cyclodextrin;
  • haemagglutinin;
  • microdomain;
  • polarised sorting;
  • ribonucleoprotein

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References

Influenza virus acquires a lipid raft-containing envelope by budding from the apical surface of epithelial cells. Polarised budding involves specific sorting of the viral membrane proteins, but little is known about trafficking of the internal virion components. We show that during the later stages of virus infection, influenza nucleoprotein (NP) and polymerase (the protein components of genomic ribonucleoproteins) localised to apical but not lateral or basolateral membranes, even in cell types where haemagglutinin was found on all external membranes. Other cytosolic components of the virion either distributed throughout the cytoplasm (NEP/NS2) or did not localise solely to the apical plasma membrane in all cell types (M1). NP localised specifically to the apical surface even when expressed alone, indicating intrinsic targeting. A similar proportion of NP associated with membrane fractions in flotation assays from virus-infected and plasmid-transfected cells. Detergent-resistant flotation at 4 °C suggested that these membranes were lipid raft microdomains. Confirming this, cholesterol depletion rendered NP detergent-soluble and furthermore, resulted in its partial redistribution throughout the cell. We conclude that NP is independently targeted to the apical plasma membrane through a mechanism involving lipid rafts and propose that this helps determine the polarity of influenza virus budding.

Low pathogenicity influenza virus infections are associated with superficial infection of the respiratory epithelium in mammals. In avian species, the infection is similarly limited but as a gut infection (1). One major determinant of this tropism concerns the availability of particular host proteases necessary for post-translational cleavage of the viral haemagglutinin (HA) (1). However, another potential determinant of superficial infection is the fact that virus budding from infected epithelial cells is polar, as progeny virions are only released from the apical surface of the cell (2–4). This polarity has long been thought to be in major part determined by specific sorting of the viral membrane proteins to the apical plasma membrane (PM) as each of the HA, neuraminidase (NA) and M2 proteins are independently targeted there (4–8). The HA and NA glycoproteins also associate with lipid rafts (9,10) and several studies indicate that virus particles are assembled at and incorporate lipid rafts (11–14). Although raft association has been implicated in apical transport (15), the sequence determinants of both HA and NA responsible for the two functions are subtly different (10,12,14,16).

Assembly of virus particles also requires that the internal components of the virus be delivered to the cytosolic surface of the apical PM. For influenza virus, this includes the matrix (M1) protein, the minor virion component NS2/NEP and the virus genome. The last component consists of eight segments of single-strand, negative polarity RNA that are always found as ribonucleoproteins (RNPs) with four viral proteins: a trimeric RNA-dependent RNA polymerase and stoichiometric quantities of the single-strand RNA-binding NP (17). Trafficking of RNPs is particularly complex as they are first assembled in the nucleus of infected cells and subsequently exported to the cytoplasm in the second half of the infectious cycle (reviewed in [17,18]). This temporally regulated RNP trafficking is reflected in the distribution of their major protein component, NP. At early times (≤ 5 h postinfection; p.i.) it localises in the nucleus, but after that it is mainly found in the cytoplasm (17). Nuclear export of RNPs involves the sequential binding of M1, NS2/NEP and cellular Crm1/exportin 1 to the RNPs (19–23). Once in the cytoplasm, the internal virion components must be assembled into budding virus particles. Influenza virus morphogenesis is imperfectly understood but a widely accepted hypothesis involves the M1 protein acting as an adapter molecule between the membrane and the RNPs, through its ability to interact simultaneously with lipid membranes, the cytoplasmic tails of the viral glycoproteins and RNPs (reviewed in [24]). The binding of NS2/NEP to M1-RNP complexes is thought to explain the small amount of NS2/NEP that becomes incorporated into virus particles (25,26). With this model for virion assembly, directed transport of the cytoplasmic virion components is not required for polarised virus budding because the cytoplasmic tails of the viral glycoproteins that M1 binds to are only present at the apical surface of the plasma membrane. However, recent work has shown that destruction of the apical targeting signal in HA does not lead to a loss of polarity in viral budding (27,28). Similarly, polarised virus budding still occurs in viruses lacking an intact NA gene (29). Possibly this reflects functional redundancy between HA, NA and M2 for targeting virus assembly to the apical surface, as has been demonstrated for the process of particle formation itself (30). Alternatively or additionally, it is possible that elements of the internal virion components independently localise to the apical surface.

Accordingly, we set out to test the hypothesis that RNP particles are targeted specifically to the apical plasma membrane. In support of this hypothesis, we found that late in infection in a variety of cell lines, NP was not distributed throughout the cytoplasm but instead accumulated underneath the apical or active plasma membrane surface. Significantly, this occurred even in cell types where HA and M1 showed no polarity in their distribution or when NP was expressed by itself. Detergent extraction and membrane flotation experiments indicated a cholesterol-dependent association of NP with lipid rafts in the presence or absence of other viral components. Furthermore, cholesterol depletion of infected cells resulted in the reduction of NP accumulation at the apical plasma membrane. We propose that RNPs are targeted to the apical surface independently of other viral components through an NP-dependent interaction with lipid rafts.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References

Intracellular localization of influenza virus NP

Budding of influenza virus requires the assembly of all virion components at the apical cell surface. Although the viral membrane proteins contain targeting signals that direct them to the apical plasma membrane, it is unclear how the internal virion components reach this surface at late time points during infection. With respect to RNPs, an immunoelectron microscopy study showed a general distribution of NP throughout the cytoplasm and a low accumulation of it at the apical plasma membrane (31). In contrast, a recent study employed confocal microscopy to show NP apparently localised specifically at the apical plasma membrane (27). As an initial experiment therefore, we examined the distribution of NP and HA in virus A/PR/8/34 (PR8)-infected cells at late time points, using Madin–Darby canine kidney (MDCK) cells as a model polarised epithelial cell, baby hamster kidney 21 (BHK) cells as a source of nonpolarised fibroblasts, and human embyronic kidney cells (293-T) as a cell type with an intermediate epithelioid phenotype. All three cell types are fully permissive for influenza virus replication and produce infectious progeny (e.g. 13,22,32).

Immunofluorescent staining of HA on infected MDCK cells fixed at 8 h p.i. showed the expected pattern of distribution. In z-axis cross-sections taken through the depth of the cells, HA was present only on apical and not on basolateral surfaces (Figure 1a, panel iii). Single optical sections taken in the xy-plane through small clumps of cells showed predominant labelling of the exterior (i.e. apical) surface of the cell clumps, with weaker labelling of internal lateral membranes (Figure 1a, panel iii). This latter localization pattern might be due to some missorting of HA to lateral membranes, or possibly to visualization of HA at areas of apical membrane above tight junctions. The staining pattern observed for NP was similar to that of HA in that, in xy-sections, the protein was found predominantly at the exterior of the cell clump, while z-sections showed its association with the apical surface (Figure 1a, panel i). In contrast to HA, no NP was seen to be associated with lateral membranes. Cells similarly infected and double-stained for NP and DNA (to delineate the nucleus) confirmed that the majority of NP had exited the nucleus and, furthermore, clearly showed that NP did not localise to all areas of the cytoplasm but instead was specifically found in areas adjacent to the apical plasma membrane (Figure 1a, panel iv). This pattern was maintained when cells were grown to confluency on porous filters to promote full differentiation of apical and basolateral surfaces (Figure 1d, panel i). Identical results were obtained with the human polarised epithelial CACO-2 cell line, either as confluent monolayers on permeable filters or as subconfluent cells on solid supports (data not shown). Examination of infected 293-T cells at 8 h p.i. showed a very similar localization pattern for NP in which although the protein had largely exited the nucleus, it showed a marked accumulation at apical cell surfaces rather than being distributed throughout the cytoplasm, either in subconfluent (Figure 1b, panels i and iv) or confluent cells (Figure 1d, panel ii). However, in 293-T cells, HA did not show a marked preference for the apical membrane but instead was found on all surfaces of the PM (Figure 1b, panel iii). Similarly, when infected BHK fibroblasts were tested, HA distributed all over the PM (Figure 1c, panel iii). In many cells, NP was distributed diffusely throughout the cytoplasm (data not shown). However, in approximately 50% of cells, NP associated preferentially to areas of cytoplasm adjacent to the plasma membrane (Figure 1c, panel i). Furthermore, this juxta-membrane positioning of NP was polar, with little NP associated with the basal plasma membrane (Figure 1c, panel i). Thus at late times postinfection, NP localises specifically to regions of the cytoplasm adjacent to apical (or active) membrane surfaces in several different cell types. Furthermore, in confirmation of the results of Mora et al. (27), this behaviour of NP does not obligatorily correlate with HA localization. This apical localization pattern is unlikely to be a staining artifact resulting from incomplete permeabilization of the cells because it was also seen with methanol permeabilised cells and with the histological technique of applying antibody to cut sections of fixed cells (data not shown).

image

Figure 1. Cellular distribution of NP and HA at late times postinfection. Virus-infected MDCK (a, d, i), 293-T (b, d, ii) or BHK (c) cells were fixed at 8 h p.i. and double-stained for NP and HA, NP and DNA (with propidium iodide) as labelled, or (d), NP (green) and LAP-2 (red). Cells in (d) were grown to confluency on (i) porous supports until tight junctions had formed as assessed by measurement of the trans-epithelial electrical resistance or (ii) glass coverslips. Single optical sections in the xy plane (upper panels) and xz planes (lower panels) are shown for individual or merged fluorophores as labelled. Scale bar = 5 μm.

Since NP localises to the apical plasma membrane of cells apparently independently of HA, we went on to examine the localization of other internal virion components known to be involved in RNP trafficking. Experiments were carried out on subconfluent clumps of cells because this permitted visualization of apical targeting in the xy plane, where the microscope has greater resolving power. In addition, it permitted direct comparison between infection and transfection (see later) as the latter technique requires subconfluent cells for reasonable efficiency. In MDCK cells at 8 h p.i., the M1 polypeptide was found primarily at apical plasma membrane surfaces in a very similar localization pattern to that of NP and the two antigens showed strong colocalization (Figure 2a). Similar results were obtained when infected 293-T cells were examined except that a slightly larger proportion of M1 was detectable in the nucleus (data not shown). However, in BHK fibroblasts, M1 showed little tendency to accumulate at the plasma membrane and was generally found throughout the cytoplasm even in cells where NP had localised preferentially to the upper surface of the cell membrane (Figure 2b). When MDCK cells were stained for NS2/NEP the protein was found throughout both nucleus and cytoplasm, showing no preference for any membrane surface or any polarity in z-axis projections (Figure 2c, panel iii). This behaviour was in contrast to NP, which as before localised preferentially at the apical plasma membrane, where it colocalised with a minor fraction of NS2/NEP (Figure 2c, panels i and ii). A similar pattern was found in infected 293-T cells (Figure 2d) and BHK cells (data not shown) in which NS2/NEP localised throughout the cell only showing limited colocalization with NP present at the apical plasma membrane. Therefore the marked accumulation of NP at the apical surface is not matched by similar specific trafficking of the NS2/NEP polypeptide, or, in BHK cells, the M1 protein.

image

Figure 2. Cellular distribution of NP, M1 and NEP at late times postinfection. Virus infected MDCK (a c), 293-T (d) or BHK (b) cells were fixed at 8 h p.i. and double-stained for NP and M1 or NP and NEP as labelled. Single optical sections in the xy plane (upper panels) and xz planes (lower panels) are shown for individual or merged fluorophores as labelled. Scale bar = 5 μm.

Nevertheless, a proportion of the M1 and NS2/NEP polypeptides were found to colocalise with NP at the apical PM (Figure 2). Potentially this could reflect a ternary complex which, once assembled, contains the necessary signals for apical targeting. Alternatively, it is possible that NP itself contains a signal that directs its specific accumulation at this surface independently of other viral polypeptides. To test this latter hypothesis, we examined the intracellular localization of exogenously expressed NP. When expressed in the absence of other viral proteins, NP shuttles between nucleus and cytoplasm and depending on expression levels and the cellular environment can appear resident in either compartment when examined by immunofluorescence (22,33,34). Accordingly, 293-T cells were transfected with a high dose of a plasmid (pCDNA-NP) containing the NP gene under the control of an RNA polII promoter and incubated for 48 h before fixation and immunofluorescent analysis. As expected (22,34) under these conditions of high-level NP expression the majority of transfected cells contained cytoplasmic NP. The high transfection efficiency obtainable with 293-T cells resulted in the formation of many contiguous clumps of NP-expressing cells. Similar to the pattern observed in virus-infected cells, NP in these transfected cells localised predominantly to areas of cytoplasm adjacent to the edges of the cell sheet (Figure 3a). Z-axis reconstructions confirmed that this did indeed result from preferential localization of NP to the apical surface of the cells (Figure 3a). The poor transfection efficiency of MDCK cells meant that large clumps of NP-expressing cells could not be obtained. Nevertheless, when individual or pairs of NP expressing cells were examined, the protein showed preferential accumulation at the apical surface (data not shown). When a similar experiment was carried out in BHK fibroblasts, again the majority of transfected cells contained cytoplasmic NP. In many cases, this NP localised diffusely throughout the cytoplasm (data not shown). However, in approximately half the cells, single optical sections taken through the body of the cell showed marked peripheral accumulation of NP with clear regions of unstained cytoplasm between them and the nucleus (Figure 3b, panels i and ii). Z-axis reconstructions generated from several such optical planes of focus confirmed that this resulted from close association of NP with the upper but not basal membrane surfaces of the cell (Figure 3b, panel iii). Overall, therefore, NP localises to the apical cell periphery in the absence of other influenza virus proteins.

image

Figure 3. Cellular distribution of NP in transfected and infected cells.a) 293-T or b) BHK cells were transfected with plasmid pCDNA-NP and subsequently fixed and stained for NP (green) and (a) DNA with propidium iodide or (b) Nup 62 (red). c) Virus infected MDCK cells were fixed at 8 h p.i. and double- stained for NP and the PB2 subunit of the viral polymerase. Single optical sections in the xy plane (large panels) and xz planes (small panels) are shown for individual or merged fluorophores as labelled. Scale bar = 5 μm.

Although NP is generally considered to be a useful marker for RNP localization (17,18), the finding that NP can traffick to the apical PM in the absence of other viral components raised the possibility that the NP seen there in virus-infected cells might not necessarily be in the form of RNPs. To test whether the apical NP staining did represent RNPs, we stained infected MDCK cells at 8 h p.i. for the PB2 component of the viral RNA-dependent RNA polymerase. The majority of PB2 remained in the nucleus (consistent with the existence of a pool of non-RNP-associated polymerase [35]), but significant antibody reactivity was also visible at the apical PM (Figure 3c, panel iii). Furthermore, this subpopulation of PB2 colocalised with NP also present at the membrane (Figure 3c, panels i, ii). Identical results were obtained when infected MDCK cells were double-stained for PB1 and NP and when the same experiments were performed in 293-T cells (data not shown). This suggests that the NP seen at the apical plasma membrane late in virus infection is at least in part in the form of RNP particles.

Membrane association of NP

Immunofluorescence analysis showed the specific accumulation of NP at the apical plasma membrane both during virus infection and when expressed alone. To provide a biochemical test of the association of NP with the plasma membrane, these findings were further investigated by sucrose gradient flotation analysis of cellular membrane fractions. Infected 293-T cells were separated into nuclear and cytoplasmic fractions at 8 h p.i. Cytoplasmic membrane-associated proteins were then separated by sucrose gradient flotation assay and analysed by Western blotting for viral polypeptides. The proportion of membrane-associated protein (defined here as the top four fractions) was quantified by densitometry of the Western blots. These analyses revealed the presence of HA, M1 and NP in low buoyant density membrane-associated fractions (Figure 4a,b). Densitometric quantification of the amount of HA present in the top four and bottom six fractions showed that around half of the total protein was present in the form of low buoyant density material (Figure 4e). This was in agreement with earlier studies, which also observed the presence of this protein in both nonbuoyant and membrane associated fractions by flotation analysis (36,37). Likewise, a substantial fraction of M1 (50%) was found in membrane-associated fractions, although in a slightly lower proportion to that reported previously (∼ 70%) (37,38). Most NP protein remained at the bottom of the gradients, but a small but reproducible proportion (14 ± 0%) was found in membrane-associated fractions (Figure 4a,e). This finding suggests the association of some NP with membranes and is in agreement with earlier studies that also detected NP in low buoyant density fractions after flotation assay of virus-infected cells (36,37). Virus specific polypeptides were not detected when extracts from mock-infected cells were subjected to similar analysis (data not shown). This observed association of NP with membranes might be due its interaction with any of the influenza virus membrane-associated proteins, such as HA, NA, or M1. Therefore, to determine whether NP associates with membranes in the absence of other influenza virus proteins, 293-T cells were transfected or mock-transfected with plasmid pCDNA-NP and analysed by flotation assay at 48 h post transfection. No NP was detected in gradient fractions from untransfected cells (Figure 4d). Analysis of transfected cells, however, again showed the presence of NP in membrane-associated fractions (Figure 4c). Although the majority of NP was found at the bottom of the sucrose gradients, a similar proportion to that found in infected cells (12 ± 4%) floated to the top fractions (Figures 4c,e), indicating that NP is able to associate with membranes even in the absence of other viral proteins.

image

Figure 4. Membrane flotation analysis of infected and transfected cells. Postnuclear supernatants prepared from 293-T cells (a b) infected with virus for 8 h (c) transfected with plasmid pCDNA-NP or (d) mock transfected were separated by sucrose gradient flotation and fractions analysed by SDS-PAGE and Western blotting with anti-PR8 virus (a, b) or anti-RNP serum (c, d). Arrows mark the indicated polypeptides. e) The proportions of HA and NP from infected or transfected (tNP) cells present in low density fractions (top four) from untreated cells (– TX), cells treated with TX-100 (+ TX) or TX-100 and the indicated concentrations (mm) of MBCD were determined by densitometry of the Western blots and plotted as a percentage of the total protein. The mean and range of duplicate experiments is shown except for the + TX samples, where the mean and standard error from five experiments is shown.

Several studies have shown that the influenza virus glycoproteins associate with lipid raft domains in the plasma membrane and that these domains are incorporated into budding virus particles (9–14,16,39). We therefore tested whether NP also associated with detergent-resistant membrane domains. First, 293-T cells were infected with virus and at 8 h p.i., treated with the nonionic detergent Triton X-100 (TX-100) on ice. Proteins associated with TX-100-insoluble lipid rafts were isolated by sucrose gradient centrifugation and analysed by Western blotting as before. The proportion of each protein present in low buoyant density fractions (again defined as the top four fractions) was quantified by densitometry (Figure 4e). HA and M1 were recovered in both low buoyant density (raft) and nonraft fractions (Figure 5a, panels i and iii), consistent with previously published results (11–14,37). When the proportion of each polypeptide that partitioned into low buoyant density fractions was quantified, one third of the HA was found to be raft associated (Figure 4e). This is comparable to but slightly lower than previously published values (43–49% [12,13]). In the case of M1, 13 ± 5% (n = 5) was found to be raft associated. This is substantially lower than the 60% value found by Zhang et al. (12). However, the study of Ali et al. (37) showed that the fraction of M1 associated with detergent-insoluble lipid domains during flotation analysis is sensitive to small changes in TX-100 concentration, providing a possible methodological explanation for our findings. When NP was examined, the majority of it remained at the bottom of the gradient, but a reproducible and meaningful proportion (on average, 16%) floated with the detergent-insoluble membrane fractions similarly to HA and M1 (Figure 5a, panel ii; quantification data in Figure 4e). However, this behaviour was not a universal feature of influenza virus polypeptides as the NS1 polypeptide showed no tendency to associate with low buoyant density material (Figure 5a, panel iv). Thus, consistent with previous observations (12,14,37) we found that some NP associates with lipid rafts in the context of virus infection. We therefore went on to test whether this was an intrinsic property of the polypeptide that did not require other viral components. 293-T cells were transfected with plasmid pCDNA-NP, treated on ice with TX-100 at 48 h post transfection, and analysed as above. As with previous flotation experiments, a high proportion of NP remained in the high density fractions at the bottom of the gradient, but the same amount as in infected cells (16%) floated to the low buoyant density fractions (Figure 5b, panel ii; quantification data in Figure 4e). This could not be ascribed to a failure of the lipid raft extraction procedure as the cellular clathrin heavy chain protein, a membrane-associated polypeptide that does not interact with lipid rafts (40,41), remained wholly in the detergent-soluble fractions (Figure 5b, panel i). In addition, flotation of NP was abolished if cells were detergent extracted at room temperature (data not shown), a condition expected to solubilise lipid rafts (42). This suggests that NP associates with lipid rafts in the absence of other viral polypeptides.

image

Figure 5. Lipid raft flotation analysis of infected and tran- sfected cells. Infected (a) or pCDNA-NP transfected (b) cells were lysed on ice with TX-100, separated by sucrose gradient flotation and the resulting fractions analysed by SDS-PAGE and Western blotting with anti-sera to the indicated poly-peptides. Where indicated, the cells were first depleted of cholesterol by treatment with MBCD as described in the text.

Effect of cholesterol depletion on NP trafficking

Cholesterol is a known structural component of lipid rafts. If it is removed, disorganization of these microdomains and dissociation of proteins bound to the rafts ensues (9,43,44). Therefore, the association of NP with lipid rafts was evaluated after treatment with cholesterol depleting agents. For this, 293-T cells were grown in the presence of mevastatin and mevalonic acid lactone (to inhibit the cholesterol biosynthesis pathway) and treated with either low or high doses of the cholesterol-extracting agent methyl-β-cyclodextrin (MBCD) for long or short times, respectively (45). In addition, the cells were either infected with virus or transfected with plasmid pCDNA-NP before lipid rafts and associated proteins were isolated by gradient centrifugation, analysed by Western blotting and quantified by densitometry as described above. When cells were treated with 3.5 mm MBCD from 1 h p.i and harvested at 8 h p.i. the proportion of HA associated with low buoyant density material decreased by approximately two-thirds (Figure 4e). Similarly, when cells were treated with 30 mm MBCD for 40 min prior to harvest, less than 10% of the HA remained raft associated (Figure 4e). Metabolic labelling experiments showed that the drug treatments did not significantly affect synthesis of the NP, M1/NS1 and NEP polypeptides, although HA synthesis was reduced (Figure 6a). Western blotting experiments confirmed normal accumulation of NP (Figure 6b) and M1 after cholesterol depletion (data not shown). However, when the cells were fractionated into soluble cytoplasmic and insoluble/nuclear fractions, the amount of soluble NP was significantly reduced by both drug treatment regimens (Figure 6b, compare lanes 2,4 and 6). Furthermore, when this soluble NP was subjected to flotation analysis, a much lower proportion (2% with both high and low MBCD concentrations) was associated with low buoyant density fractions (Figure 5a, panels v,vi; quantification data in Figure 4e) compared to that from untreated cells. When the parallel experiment was carried out on cells transfected with an NP-expressing plasmid, a similar decrease in the levels of soluble NP was seen (data not shown). Again, the proportion of low buoyant density NP in the soluble fraction decreased to 3% in cells treated with 3.5 mm MBCD for 7 h prior to harvesting (Figures 4e and 5b, panel iii), and to undetectable levels in cells treated with 30 mm MBCD for 40 min (Figures 4e and 5b, panel iv). Thus NP shows a cholesterol-dependent association with detergent-resistant membrane fractions that does not require the presence of other influenza virus proteins.

image

Figure 6. Analysis of viral protein synthesis in infected cells with and without cholesterol depletion.a) Lysates from 293-T cells infected with PR8 or mock (M)-infected and pulse labelled with 35S-methionine for 2-h periods ending at the indicated times p.i. were separated by SDS-PAGE and detected by autoradiography. Arrows denote abundant viral polypeptides. b) Aliquots of infected 293-T cells taken at 8 h p.i. and treated with TX-100 and the indicated concentrations of MBCD were analysed by SDS-PAGE and Western blotting with anti-NP either before (T) or after subcellular fractionation into soluble cytoplasmic (S) fractions.

In addition to the effect of cholesterol depletion on the solubility of raft-associated proteins, previous studies have shown that extraction of cholesterol causes missorting of apically targeted membrane proteins, including the influenza virus HA, possibly due to the disruption of lipid rafts (45). We therefore determined the effect of cholesterol-depleting agents on the cellular distribution of NP and HA. 293-T cells were grown in the presence of mevastatin and mevalonic acid lactone, infected with influenza virus and treated with MBCD as above. Untreated cells were used as a control. Cells were fixed at 8 h p.i., double-stained for NP and HA, and analysed by confocal microscopy. In the absence of drug treatment, HA localised to all plasma membrane surfaces, while NP was found concentrated at the apical surface (Figure 7a, panel i). Treatment with MBCD from 1 h p.i. resulted in the failure of HA to reach the plasma membrane in detectable amounts, instead localising to foci near the nuclei (Figure 7a, panel ii), presumably representing its retention in the Golgi apparatus (45). Furthermore, NP was no longer found only at the apical plasma membrane. Although it still showed a bias towards the apical sides of cells, its overall staining pattern was less polarised and more diffuse, including areas of cytoplasm not directly opposed to the apical membrane (Figure 7 a, panel ii, arrows). Extraction of cells with a higher dose of MBCD for 40 min prior to cell fixation had less effect on HA distribution with some cell surface staining still visible (Figure 7 a, panel iii). However, the tight association of NP with the apical plasma membrane was lost, with the protein again showing an increased tendency to localise throughout the cell, including areas of cytoplasm away from the apical cell surface (Figure 7a, panel iii, arrows). To provide a semiquantitative measure of NP distribution in treated and untreated cells, the fluorescence intensity across the width of several similarly sized clumps of cells was measured and the average plotted. The profile derived from untreated cells confirmed the predominant localization of NP towards the apical surface, with around 10-fold higher fluorescence intensities found at the exterior of the cell sheet than the interior (Figure 8). However, after cholesterol depletion (with either high or low concentrations of MBCD), markedly higher fluorescence intensities were observed in the interior of the cell sheet, and although a bias towards peripheral staining was still evident, the difference in average intensity between interior and exterior was now only around twofold (Figure 8). In part, the more diffuse localization of NP also appeared to result from increased nuclear staining. To test this, cells were double-stained for NP and nuclear lamin-associated polypeptide 2 (LAP-2) in order to visualise the nuclear envelope. At 7 h 20 min p.i., in cells treated only with mevastatin and mevalonic acid lactone, the nuclei were largely devoid of NP staining (Figure 7b, panel i). The identical pattern was seen in infected cells not treated with any drug (data not shown). However, after treatment with 3.5 mm MBCD for the preceding 7 h, a higher proportion of NP remained inside the nuclei at 8 h p.i. (Figure 7b, panel ii). Similarly, when cells were treated with 30 mm MBCD for 40 min at 7 h 20 min p.i., a substantial proportion of NP was found within the nuclei (Figure 7b, panel iii). Thus cholesterol depletion decreases the association of RNPs with the apical PM at late times postinfection and also results in their nuclear retention or possibly re-import.

image

Figure 7. Effect of cholesterol depletion on NP distribution in infected cells. PR8-infected 293-T cells were treated with MBCD as indicated, fixed and stained for NP and HA (a) or LAP-2 (b) as labelled. Single optical sections in the xy plane (upper panels) and xz planes (lower panels) are shown for individual or merged fluorophores as labelled. Arrows indicate areas of cytoplasmic NP staining not immediately adjacent to the apical PM. Scale bar = 5 μm.

image

Figure 8. Semiquantitative analysis of NP localization in infected cells. The average fluorescence intensity in the xy plane across four similarly sized clumps of cells (including those shown in Figure 7) treated with the indicated concentrations of MBCD is plotted.

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References

A widely accepted model for influenza virus assembly proposes a cascade of interactions between the cytoplasmic tails of the virus membrane proteins, the matrix/M1 protein and RNPs (24). For many years, this model also plausibly explained the polarised budding seen in epithelial cells, because specific sorting of the three viral membrane proteins (3,5–8) meant that their cytoplasmic tails were only available to interact with M1 and RNPs at the apical PM. However, more recent analyses suggest that other factors may influence the direction of virus release. M2 seems an unlikely determinant because it is not incorporated into lipid rafts (12) and there is no evidence for functional interactions with its cytoplasmic tail (46). Major disruption of the NA gene does not prevent polarised virus budding (29). Similarly, it has recently been shown that mutations in HA that perturb its apical targeting do not lead to promiscuous budding or prevent the accumulation of NP underneath the apical PM (27,28). Consistent with and extending these data, we found that NP accumulated only at apical membrane surfaces late in infection, even in poorly polarised cell types where wild-type HA localised to lateral, basolateral and apical membranes. Furthermore, it did so even in fibroblast cells where M1 localised diffusely through the cytoplasm. Similar observations have been made with influenza A/WSN/33 infected mouse L929 fibroblasts, showing NP preponderantly at the PM but M1 distributed throughout the cytoplasm (33). This seemingly independent trafficking of NP to the apical PM might reflect functional redundancy in the ability of HA and NA to direct accumulation of the internal virion components to the correct membrane surface. However, NP also accumulated at apical but not lateral and basolateral membranes when expressed in the absence of other influenza virus proteins, strongly suggesting that, like HA, NA and M2, it contains a targeting mechanism in its primary sequence. Consistent with this, flotation analysis showed that NP associated with membranes when expressed by itself. Although most of the cytoplasmic NP remained at the bottom of flotation gradients, around one sixth of the material reproducibly formed a secondary peak of low buoyant density material. Furthermore, this interaction was resistant to treatment with TX-100 on ice but sensitive to cholesterol extraction, strongly suggesting that NP associates with lipid raft microdomains independently of HA, NA and M1. The fraction of NP which associated with membranes (with or without detergent extraction) did not increase in the context of virus infection. This is in contrast to the behaviour of the M1 protein, whose membrane association and detergent insolubility have been suggested in several studies to be decreased by exogenous expression (36,37,47). However, the flotation profile of NP from infected cells closely recapitulated previous observations (12,14,36,37) in that rather than showing a distinct separation into high and low buoyant density populations, it trailed throughout the gradient, with the majority of the protein remaining towards the bottom. This may reflect a difference in density between exogenously expressed NP and authentic RNPs. Indeed, there is evidence that the presence of M1 alters the maturation and sedimentation properties of RNPs (48).

The finding that NP migrates specifically to the apical PM in the absence of other viral proteins can be most plausibly explained by it interacting with a cellular component resident there. Although the membrane flotation assays do not distinguish between the various types of cell membrane, our working hypothesis is that the low buoyant density exhibited by NP results from its association with a lipid raft-associated component of the apical PM. However, with both transfection and infection, the relatively low proportion of NP defined biochemically as membrane-bound contrasts with the immunofluorescence data, suggesting a tight association with the apical PM. This may well reflect a relatively weak membrane interaction of a nonintegral membrane protein that is partially disrupted by the fractionation procedures used to analyse membrane flotation. A similar discrepancy between biochemical and fluorescence analysis was seen when the effects of cholesterol depletion were analysed. Although MBCD treatment caused a profound redistribution of NP within the cell (as assessed biochemically and by immunofluorescence), with a far greater proportion of it remaining in the nuclear/insoluble fraction, only a partial redistribution of the remaining cytoplasmic protein away from the apical PM was seen by cell staining. In contrast, almost complete abrogation of NP flotation was seen. This might reflect the lower sensitivity of the flotation assay. Alternatively or additionally, the retention of a degree of apical targeting (as judged by immunofluorescence) even after cholesterol extraction could result from either incomplete removal of the sterol or the existence of more than one mechanism to direct NP to the apical PM. In favour of incomplete cholesterol extraction, the conditions used here did not totally prevent detergent-resistant flotation of HA or the cellular protein decay accelerating factor (DAF) (data not shown), both commonly used markers for lipid rafts. Previous work has shown that very high concentrations of MBCD (> 50 mm) can be required to solubilise DAF fully (49).

The identity of the cellular component(s) that NP interacts with at the apical PM is not known, but of the several cellular proteins NP is known to bind (17), actin provides an interesting candidate. The β-isoform of actin is predominantly found associated with active areas of PM (50–54). In the cell types used here, β-actin distribution therefore matches that seen for cytoplasmic NP; primarily underneath the apical PM in cell types that form sheets, or under the upper/active membrane surface of fibroblasts. Cytoplasmic RNPs are known to associate with the actin cytoskeleton (47) and NP binds directly to F-actin in vitro (34). Furthermore, transfected NP shows stronger colocalization with β-actin than stress fibres (34) and, in infected cells, the distribution of NP at the apical PM alters after treatment with inhibitors of actin treadmilling (13,47). However, the relative insensitivity of influenza virus budding to such drug treatment (13,55) would argue against the hypothesis that RNPs are targeted to the apical PM solely via interactions with cortical actin. Nevertheless, the hypothesis deserves further investigation.

Disruption of the actin cytoskeleton in infected polarised epithelial cells leads to the redistribution of HA, M1 and NP around clumps of cortical β-actin (13). Since many lines of evidence indicate a linkage between lipid rafts and the actin cytoskeleton (56–59), influenza virus budding incorporates lipid rafts (11,12) and continues unabated (albeit as nonfilamentous particles only) after cytoskeletal disruption (13,55), this suggests that RNPs accumulate at lipid rafts in the apical PM. This is consistent with the finding here that cholesterol extraction (expected to disperse lipid rafts) decreases the accumulation of NP at the apical PM. However, the finding that cholesterol depletion also affects the nucleo-cytoplasmic trafficking of RNPs was less expected. Both short- (40 min) and long-term (7 h) treatment of infected cells with MBCD led to increased nuclear residency of NP. In the case where cells were treated from 1 h p.i., this could have resulted from a block to RNP nuclear export (although M1 and NEP/NS2 synthesis was unaffected). However, as very similar results were obtained when cells were treated with MBCD for a short period of time relatively late in infection, after the majority of RNPs had already been exported, it seems likely that cholesterol depletion leads to re-import of the RNPs. Heterokaryon fusion experiments show that while exogenously expressed NP shuttles between nucleus and cytoplasm, this does not occur for RNPs in infected cells (33). This difference in trafficking behaviour seems to arise at least in part from the presence of the M1 protein in infection (60,61), suggesting the possibility that cholesterol depletion affects the interaction of M1 with RNPs. Removal of membrane cholesterol also decreased the amount of cytosolic transfected NP (data not shown), suggesting that dispersing lipid rafts may affect the trafficking of NP even in the absence of M1. However, as transfection of NP leads to varying expression levels of the protein in individual cells, there is always a background of cells with nuclear NP (22), which rendered the difference between MBCD treated and untreated cells less clear cut. At present, therefore, we can not say for certain that cholesterol depletion affects the intracellular trafficking of plasmid-expressed NP in exactly the same way as in virus-infected cells.

In conclusion, we show that NP trafficks specifically to the apical PM independently of the other viral structural proteins but at least in part dependently upon a direct or indirect interaction with lipid rafts. We propose that this plays a role in determining the polarity of virus budding. We hypothesise that under normal circumstances, specific targeting of NP helps promote efficient viral budding by increasing the concentration of RNPs at the PM microdomains utilised by the virus as budding platforms. Under artificial laboratory conditions where mutant viruses are analysed, or incompletely polarised cell types are infected, this polarised trafficking of NP helps maintain directional budding even when the correct targeting of the viral integral membrane proteins breaks down.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References

Plasmids, cells, drugs and antisera

Plasmid pCDNA-NP containing a cDNA copy of influenza virus A/PR/8/34 strain segment 5 under the control of the human cytomegalovirus immediate early promoter was constructed by excising the NP coding sequence from plasmid pKT5 (34) by digestion with the restriction enzymes NcoI (subsequently endfilled with Klenow fragment DNA polymerase) and XbaI and inserting the DNA fragment into plasmid pCDNA-3 (Invitrogen, Renfrew, UK) digested with Eco RV and XbaI.

Polarised epithelial MDCK cells (62) with a transepithelial electrical resistance of 800 ω/cm2 at confluency, epithelioid cell line 293-T (derivative of human embryonic kidney 293 cells which express the SV40 large T antigen [63]), and fibroblastic cell line BHK-21 (64) were used. MDCK and 293-T cells were cultured in Glasgow minimum essential medium (GMEM, Sigma Co., Surrey, UK) supplemented with 2 mm glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin and 10% (v/v) foetal calf serum (FCS). BHK cells were cultured in BHK-21 GMEM (Gibco BRL, Gaithersburg, MD) supplemented as above except that newborn calf serum (NCS) was used and 5% (v/v) tryptose phosphate broth was also added. Cells were seeded onto 13 mm glass coverslips for immunofluorescence or 6 cm dishes for biochemical analysis and were grown at 37 °C in a humidified atmosphere containing 5% CO2. Metabolic labelling of proteins was carried out using 35S-methionine as previously described (22). Mevastatin (10 mm stock solution dissolved in dimethyl sulfoxide), mevalonic acid lactone and MBCD (both kept as a 0.2 m stock in water) were all purchased from Sigma. To deplete 293-T cells of cholesterol, they were grown for 48 h in serum-free medium in the presence of 10 μm mevastatin and 250 μm mevalonic acid lactone (to inhibit cholesterol biosynthesis) before treatment with MBCD (45).

Polyclonal rabbit antisera to influenza virus A/FPV/34 RNP and influenza virus PR8 NS1, NS2/NEP and whole virus particles have been previously described (22,65,66). Rabbit polyclonal antisera to influenza virus PR8 PB2 was prepared by immunising rabbits with the fusion protein MBP-PB2-C containing the C-terminal 180 amino acids of PB2 (67). A mouse monoclonal antibody reactive against influenza virus PR8 NP (clone AA5H) was purchased from Serotec (Oxford, UK), while goat polyclonal antiserum against M1 was purchased from Biogenesis (Poole, UK). A mouse monoclonal antibody reactive against influenza virus HA (clone H28-E23) was kindly provided by Dr Jonathon Yewdell. Mouse monoclonal antisera to the cellular proteins LAP-2 and Nup62 were purchased from BD Biosciences (San Diego, CA), and goat polyclonal antiserum against clathrin heavy chain from Santa Cruz Biotechnology (Santa Cruz, CA). Cellular DNA was stained with propidium iodide (Sigma).

Virus infection, transfection and immunofluorescence

Influenza virus PR8 was grown in embryonated eggs and titred by plaque assay on MDCK cells according to standard procedures (13,22). For analytical purposes, cells were infected at a multiplicity of infection of 10 in serum-free medium for 1 h before being overlaid with complete medium. Cells were transfected with plasmid (0.3 μg per 24-well or 10 μg per 6-cm dish) using a cationic lipid mixture (Lipofectin; Gibco-BRL) essentially according to the manufacturer's instructions. For immunofluorescence, cells were fixed with 4% formaldehyde in PBS for 20 min at room temperature, washed with PBS containing 2% NCS (PBS/NCS) and permeabilised with 0.2% TX-100 in PBS for 5 min. After further washing with PBS/NCS, cells were probed with primary antibodies diluted in PBS/NCS for 1 h, followed by secondary fluorophore-conjugated antibodies for 30 min with PBS/NCS washes in between and after. Coverslips were mounted in Citifluor (Agar Scientific, Stansted, UK) and examined by confocal microscopy using a Leica TCS-SP instrument (13,22). For semiquantitative analysis, images of similarly sized clumps of cells were scaled to be identical widths and fluorescence intensities across a slice profiled using the program Image J ( (http://www.rsb.info.nih.gov/nih-image/).

Membrane flotation analyses

To analyse membrane-associated proteins, postnuclear supernatants of 293-T cells were prepared by washing monolayers on 6-cm dishes once with ice-cold phosphate-buffered saline (PBS), before scraping them into 0.5 mL ice-cold hypotonic lysis buffer (10 mm Tris pH 7.5, 10 mm KCl, 5 mm MgCl2). Cells were collected by centrifugation at 2000 × g for 5 min, resuspended in 0.5 mL of hypotonic lysis buffer, and incubated on ice for 10 min before disruption by 20 passages through a 25-gauge hypodermic needle. Unbroken cells and nuclei were pelleted by centrifugation at 1000 × g for 5 min at 4 °C, and the resulting supernatant was subjected to flotation analysis as described by Sanderson et al. (68) with the following modifications. Postnuclear supernatants (0.4 mL) were dispersed into 2 mL of 72% (w/w) sucrose in low-salt buffer (LSB; 50 mm Tris HCl pH 7.5, 25 mm KCl, 5 mm MgCl2) and transferred into 6 mL centrifuge tubes at 4 °C. Samples were sequentially overlaid with 2.5 mL of 55% and 0.6 mL of 10% sucrose in LSB, then centrifuged at 170 000 × g for 8 h at 1 °C. After centrifugation, 0.6 mL fractions were collected from the bottom of the gradient using a Haake-Buchler Auto Densiflow IIC gradient harvester (Buchler Instruments, Saddle Brook NJ). Protein was concentrated from the fractions by precipitation with methanol-chloroform (69) before further analysis by chemoluminescent Western blotting. Blots were quantified by densitometric analysis of exposed X-ray films using the program image j/NIH Image. For analysis of lipid raft microdomain-associated proteins, a modification of the method described by Vincent et al. (70) was used. Cells were washed with ice-cold PBS, incubated with 0.5 mL of 0.5% (w/v) TX-100 in LSB for 20 min and removed from plates by scraping. 0.4 mL aliquots of the lysates were dispersed into 2 mL of 55% (w/w) sucrose in LSB and transferred into 6 mL centrifuge tubes. Samples were sequentially overlaid with 2 mL of 30% and 1 mL of 2.5% (w/w) sucrose in LSB. All steps were carried out in a 4 °C room with solutions and dishes maintained on ice where possible. Gradients were then centrifuged at 152 000 × g for 16 h at 1 °C before fractions were harvested and processed as described above.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References

We thank Dr Konrad Bishop for constructing plasmid pCDNA-NP and Drs Amanda Stuart, Paul Luzio and Debra Elton for helpful discussion. This work was supported by grants from the Royal Society, Wellcome Trust (no. 073126) and BBSRC (no. S18874) to P.D. M.J.A. is supported by the Gulbenkian PhD program in Biomedicine and the Fundacao para a Ciencia e Tecnologia.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgements
  7. References