Nucleolin interacts with influenza A nucleoprotein and contributes to viral ribonucleoprotein complexes nuclear trafficking and efficient influenza viral replication

Influenza viruses replicate their single-stranded RNA genomes in the nucleus of infected cells and these replicated genomes (vRNPs) are then exported from the nucleus to the cytoplasm and plasma membrane before budding. To achieve this export, influenza viruses hijack the host cell export machinery. However, the complete mechanisms underlying this hijacking remain not fully understood. We have previously shown that influenza viruses induce a marked alteration of the nucleus during the time-course of infection and notably in the nucleolar compartment. In this study, we discovered that a major nucleolar component, called nucleolin, is required for an efficient export of vRNPs and viral replication. We have notably shown that nucleolin interacts with the viral nucleoprotein (NP) that mainly constitutes vRNPs. Our results suggest that this interaction could allow vRNPs to “catch” the host cell export machinery, a necessary step for viral replication.

Influenza viruses replicate their single-stranded RNA genomes in the nucleus of infected cells and these replicated genomes (vRNPs) are then exported from the nucleus to the cytoplasm and plasma membrane before budding. To achieve this export, influenza viruses hijack the host cell export machinery. However, the complete mechanisms underlying this hijacking remain not fully understood. We have previously shown that influenza viruses induce a marked alteration of the nucleus during the time-course of infection and notably in the nucleolar compartment. In this study, we discovered that a major nucleolar component, called nucleolin, is required for an efficient export of vRNPs and viral replication. We have notably shown that nucleolin interacts with the viral nucleoprotein (NP) that mainly constitutes vRNPs. Our results suggest that this interaction could allow vRNPs to "catch" the host cell export machinery, a necessary step for viral replication.
Influenza A viruses (IAV) are responsible for respiratory diseases that affect millions of people worldwide in recurrent annual epidemics, sometimes fatally. The ability of this major human pathogen to mutate and lead to the emergence of new pandemic subtypes, constitutes a constant serious threat to human health 1 . The genome of IAV consists of eight single-stranded negative RNA segments encoding for up to 17 distinct proteins 2 . These RNA segments are encapsidated by the viral nucleoprotein (NP) in association with proteins of the viral polymerase complex (PA, PB1 and PB2). The resulting viral ribonucleoproteins (vRNPs) constitute autonomous functional units for viral transcription and replication 1,3 .
In contrast with most RNA viruses, IAV have a nuclear infectious cycle and thus have direct access to the host nuclear machineries, including the transcriptosome or the spliceosome, to successfully achieve their viral replication [2][3][4] . Numerous cellular components have been identified as functional interacting partners of IAV proteins and some are subject to different extents of viral "retasking" 3,[5][6][7][8][9] . In view of their nuclear replication, IAV require the efficient export of their vRNPs from the nucleus towards the cytoplasm and budding regions of the cell membrane for a productive infection 10 . Extensive investigation of the nuclear export of vRNPs has provided a model in which the viral matrix protein (M1) binds to vRNPs and the viral nuclear export protein (NEP/NS2) acts as a bridge between M1 and the cellular export receptor Crm1 11 . Alternatively, NP has been shown to bind to both M1 and Crm1 thereby allowing its exportation 12 . Nevertheless, deciphering the precise functional interplay between

Results
Nucleolin redistributes with IAV ribonucleoprotein complexes to the nuclear periphery of infected cells. Considering the strongly altered host nucleolar ultrastructure and composition during IAV infection, we analyzed the subcellular localization of three major constitutive nucleolar proteins, nucleolin, B23 and fibrillarin, during the time course of influenza infection. We used immuno-fluorescence confocal microscopy in human lung epithelial A549 cells, mock-infected or infected by influenza A/Moscow/10/99 (H3N2) virus, at 7 and 10 hours post infection (hpi). In addition to nucleolin, B23 or fibrillarin immunostaining, we also used NP as an indicator of viral cycle advancement, as previously carried out 3,[17][18][19][20] . As expected in mock-infected cells, the three nucleolar proteins mainly localized in nucleoli (Fig. 1, panels a to h and Supplementary Figure 1B,C). During infection, localization patterns of B23 and fibrillarin remained unmodified up to 7 hpi (data not shown) compared to mock-infected cells, as previously reported 7,27 . In contrast, when NS1 is notably accumulated within nucleolus at this time (Supplementary Figure 1A), nucleolin no longer preferentially localized in the subnuclear compartment and instead displayed a broad relocalization throughout the nucleus (Fig. 1, panels i to p; and Supplementary Figure 1A), in accordance with previous study 28 . Later in infection (10 hpi), as illustrated by the mainly cytoplasmic localization of NP, nucleolin mostly redistributed at the nuclear periphery of more than 95% of the infected cells (Fig. 1, panels q to x). At this time most of the fibrillarin accumulated in a large nuclear spot (Supplementary Figure 1B), probably linked to the ultrastructural remodeling of the nucleolar compartment 3,17-20 , whereas B23 remained associated (Supplementary Figure 1C).
Altogether, these results reveal the differential modification of the subnuclear localization of the three abundant constitutive nucleolar proteins by H3N2 infection, and suggest a specific, earlier and dynamic redistribution of nucleolin during the time course of infection.
We then investigated the impact of infection on nucleolin localization by a different human (H1N1 pdm2009) and several avian (H5N1, H5N2 and H7N7) IAV strains. Despite their specific host-range restriction 56 , all viruses efficiently infected A549 cells with comparable kinetics, as we previously reported 3,17 . We performed immunolabeling at 24 hpi, a time corresponding to late stage of infection for all these viruses (Supplementary Figure 2A). As observed for H3N2, each infection led to a similar redistribution of endogenous nucleolin towards the nuclear periphery of infected cells (Supplementary Figure 2A). Interestingly, the observed subnuclear localization pattern of nucleolin in IAV infected cells clearly evoked the previously described polarized distribution of trafficking vRNPs that formed a cap at the apex of the nuclear envelope prior to their export towards the cytoplasm 57,58 . To Immunofluorescence staining of nucleolin (green) and B23 (red) in mock-infected cells (panels a to h) or NP (red) in A549 cells infected with H3N2 virus at a MOI of 1 (panels i to x) was performed at different times, as indicated. Nuclei were counterstained with DAPI (blue, panels c,g,k,o,s,w). Merged fluorescent signals are presented in panels d,h,i,p,t and x. Cell details are enlarged (inset). Scale bar = 10 μ m. note, similar NP-nucleolin colocalization and subnuclear localization patterns were also observed using another specific monoclonal antibody raised against NP (MAb 3/1, kind gift of Dr Webster) that was reported to recognize IAV vRNP complexes 59 (Supplementary Figure 2B).
To further investigate these correlated observations, we studied the localization of endogenous nucleolin by 3D reconstruction of confocal images. Our observations confirmed an asymmetric localization pattern of endogenous nucleolin within H3N2 infected cells (10 hpi) (Fig. 2). Single optical sections through the nucleus (Fig. 2, panels A1 to A3) and Z-axis reconstruction revealed a polarized co-localization of nucleolin with NP towards the nuclear apex, as revealed by the asymmetric merged fluorescence staining in the z1-z2-axis plane (Fig. 2). Altogether, these results suggest a putative involvement of nucleolin in the nuclear trafficking of vRNPs during IAV infection.

Endogenous nucleolin interacts with transiently expressed NP.
To further explore this hypothesis, we first performed a GST pull-down assay to examine whether endogenous nucleolin could interact with nucleoprotein, the main component of vRNPs. A549 cells were transfected with a plasmid encoding NP fused to GST (GST-NP, H3N2), and cell lysates were pulled down using glutathione sepharose beads at 48 h post transfection. Complexes were analyzed by western blot with anti-GST and anti-nucleolin antibodies. As controls, cell lysates Figure 2. Influenza virus induces a nuclear polarized redistribution of nucleolin. A549 cells were infected with H3N2 (MOI 1) and fixed at 8 hpi. Immunofluorescence staining of NP (red) and nucleolin (green) were performed. Single optical sections through the top (A1), the midline (A2) and the bottom (A3) of nuclei (xy) and z-axis reconstruction (z1 and z2, respectively) are represented. Merged fluorescent signals are presented, as indicated. Scale bar = 10 μ M. JAcoP plugin for imageJ was used to highlight the colocalization of the two stainings by thresholding the signal above the background 80 . For easier visualization, colocalized pixels are colorized in white (right panels).
Scientific RepoRts | 6:29006 | DOI: 10.1038/srep29006 obtained after transfection of non-fused GST, GST-NS1 or GST-NEP were also examined as controls. As shown in Fig. 3A, endogenous nucleolin was pulled down from GST-NP lysates (Fig. 3A, lane 2). In comparison, no interaction with nucleolin was characterized with cell lysate obtained after transfection of non-fused GST protein (Fig. 3A, lane 1). In addition, a very weak interaction with nucleolin was characterized with GST-NS1 (Fig. 3A, lane 3). This partial discrepancy with a previous study 29 could be explained by the different strain origin of the NS1 tested (A/WSN/33), especially considering other reports showing strikingly different properties among NS1 of various origins 28 . Another explanation would be our experimental conditions being based on transient expression in A549 cells and not on the use of recombinant proteins expressed from bacteria 29 . On the other hand, our use of GST-NEP protein expression as a negative control did confirm no interaction with nucleolin (Fig. 3A, lane 4). Additionally, this interaction between endogenous nucleolin and transiently expressed NP was also confirmed by immunoprecipitation assay using a specific rabbit polyclonal anti-nucleolin antibody 43 (data not shown).

Endogenous nucleolin is a putative interacting partner of IAV ribonucleoproteins in infected cells.
To determine whether NP association with nucleolin also occurs in a context of infection, we performed . Immunoprecipitations (IP) were performed on the whole cell extract with polyclonal anti-nucleolin or control IgG antibodies, as indicated. The presence of nucleolin and NP in immunopurified complexes was checked by western blot analysis. Whole cell extracts (Input) were used as control. (C) Cells were infected during 8 or 24 hours with H3N2, as before. Immunoprecipitations were performed with the corresponding whole cell extract proteins, using polyclonal anti-nucleolin or control IgG antibody. RNAs associated with purified complexes were extracted and M vRNAs were quantified by specific RT-qPCR (M vRNAs copies number/mL). Total RNAs were extracted from Mock-or H3N2-infected cells and used as controls (Input). * * * P-value < 0.001. . Intranuclear localization of NP was evaluated by counting cells harboring nuclear and/or cytoplasmic labeling in a total of 100 cells per condition, on a total of three distinct fields (lower panel). (C,D) A549 cells were transfected with si-Nucleolin or si-Control and infected with H3N2 virus, as in A and submitted to nucleocytoplasmic fractionation. Equal amounts of proteins from each fraction were analyzed by western blot (C). NP and PB1 proteins were detected using specific antibodies. H3 and β -tubulin were detected as loading and fractionation controls. (D) Total RNAs were extracted from cytoplasmic and nuclear fractions and vRNAs were quantified by specific RT-qPCR (M vRNAs copies number/mL), * P-value < 0.05. (E) A549 cells were immunoprecipitation assays with extracts prepared from A549 cells infected with H3N2 virus, at 8 h and 24 hpi. Extracts from mock-infected cells were used as negative controls. As expected, western blot confirmed the immunoprecipitation of nucleolin specifically by the rabbit polyclonal anti-nucleolin antibody compared to IgG control (Fig. 3B, lanes 2, 5 and 8 versus 3, 6 and 9). The immunoprecipitated complexes from infected cells contained NP, the amount of which increased from 8 hpi to 24 hpi (Fig. 3B, lane 5 versus 8). Otherwise, RNAse A treatment did not abolished immunoprecipitation of NP, suggesting nucleolin could interact with both vRNP and non-vRNP forms of NP (Supplementary Figure 3). Alternatively, we investigated the presence of vRNA in immunoprecipitated complexes in similar experimental conditions (Fig. 3C). To this end, we subjected cell lysates of infected cells (8 and 24 hpi) to anti-nucleolin immunoprecipitation before extracting the putative associated vRNAs for quantification by specific RT-qPCR (Fig. 3C), described in our previous study 60 . As expected, total cell lysates at 24 hpi contained more vRNA copies as compared to those at 8 hpi; total cell lysates from mock-infected cells contained no detectable vRNA. Accordingly, we detected vRNAs in complex containing NP at 8 hpi and mostly at 24 hpi, whereas no vRNA in control (IgG) (Fig. 3C, right panel).
Altogether, these results demonstrate that NP and vRNA are co-immunoprecipitated with endogenous nucleolin in IAV infected cells. In accordance with previous results of immunostaining experiments and NP interaction assays, our results suggest a putative association of endogenous nucleolin with IAV ribonucleoproteins.

Silencing of endogenous nucleolin expression impairs nucleocytoplasmic trafficking of IAV ribonucleoprotein complexes.
We investigated functional relevance of such putative association by studying the contribution of nucleolin to the nuclear trafficking of vRNPs. We first assessed, as an indicator of infection progress, the subcellular distribution pattern of NP in conditions of endogenous nucleolin silencing. After validating silencing efficiency in non-infected A549 cells (Supplementary Figure 4), we transfected A549 cells with siRNA control (si-Control) or a pool of siRNAs targeting nucleolin (si-Nucleolin) before their subsequent infection at 48 h with H3N2 virus (MOI of 1). As expected, the si-Nucleolin treated cells displayed very low levels of nucleolin staining at 8 and 24 hpi, as compared to si-Control (Fig. 4A, panels e and m). Interestingly, the silencing of nucleolin expression appeared to only slightly decrease the number of NP positive cells (Fig. 4A, panels b versus f, j versus n and Fig. 4B, upper panel), suggesting a limited impact on cell infectivity. In contrast, the si-nucleolin treatment strongly affected the subcellular localization of NP during infection (Fig. 4A, panels b versus f, j versus n) in such a way that 1.9 to 3 times more cells harbored a stringent nuclear pattern only for NP in this condition, compared to the si-Control condition at 8 and 24 hpi, respectively. At both time intervals post-infection, the viral cycle appeared more advanced in the presence of endogenous nucleolin as shown by a more pronounced cytoplasmic staining for NP in infected cells (Fig. 4B, lower panel).
In a parallel experiment, we assessed by western blot the levels of two protein components of vRNP (NP and PB1) in subcellular compartments (Fig. 4C). A549 cells, previously transfected with si-Control or si-Nucleolin, were infected with H3N2 virus (MOI 1) for 24 h and harvested for nucleocytoplasmic fractionation. As a control, we studied the expression of histone H3 and β -tubulin. As expected, endogenous nucleolin had been silenced in the different sub-cellular fractions (Fig. 4C, compare lanes 1, 3 and 5 with lanes 2, 4 and 6, respectively). Moreover, the two experimental si-RNA conditions showed no difference in terms of expression levels of NP and PB1 viral proteins in whole cell extracts, suggesting no effect on viral protein expression (Fig. 4C, lanes 5 and 6). In contrast, we noted a strong increase of NP and PB1 in the nuclear fractions of si-Nucleolin-treated cells as compared to those of si-Control (Fig. 4C, lanes 3 and 4), while the cytoplasm fractions displayed similar protein ratios between the two si-RNA conditions (Fig. 4C, lanes 1 and 2). We completed this analysis by performing total RNA extraction from the cytoplasmic and nuclear fractions, followed by a quantification of viral genome copy numbers by RT-qPCR (Fig. 4D). Results revealed no difference in cytoplasmic fractions between si-Nucleolin and si-Control (Fig. 4D, left panel). In contrast, the nuclear fraction of si-Nucleolin treated cells contained a significative higher level of vRNA copies than that of the control (Fig. 4D, right panel, p-value < 0.05). In accordance with these results, the silencing of nucleolin also significantly reduced the yield of progeny virus produced at 24 and 30 hpi, as compared to si-Control conditions. Quantification of vRNA copies number/ml in cell supernatants revealed a marked decrease in viral progeny yield that could reach up to 95.5% at 30 hpi, (MOI 0.2, Fig. 4E) or up to 63% at 30 hpi (MOI 2, Fig. 4E). This result was also confirmed by quantification of infectious titers (TCID50/mL) in supernatants, with a slight decrease and a delay of viral production in the context of silencing of nucleolin (Fig. 4F).
Altogether, these results suggest a nuclear retention of vRNP complexes upon silencing of endogenous nucleolin and a consecutive delay in the dynamic trafficking of vRNPs towards the cytoplasm and consecutive viral production. They therefore provide evidence in support of our hypothesis that nucleolin is required for an efficient nucleocytoplasmic export of vRNPs during IAV infection.

Association of vRNPs with dense chromatin domain requires host nucleolin expression.
Several reports have demonstrated that vRNPs are associated with chromatin during the nuclear steps of the viral cycle 16,[61][62][63] . Recently, Chase and collaborators demonstrated that IAV gain privileged access to the host nuclear Crm1 export machinery by targeting vRNP complexes to the sites of Ran regeneration within dense transfected with si-Nucleolin or si-Control. Two days after transfection, cells were infected with H3N2 at MOI 0.2 and 2 for 24 to 30 hpi. Viral RNAs were extracted from culture supernatants and quantified by RTqPCR, using specific primers for M segment. * * * P-value < 0.001. (F) A549 cells treated by si-Control or si-Nucleolin (as previously described) were infected with viruses at a MOI of 2. Samples of supernatants were harvested at several time points post-infection and stored at − 80 °C until end point viral titration assays in MDCK cells using the Reed and Muench statistical method (TCID50/mL).
Scientific RepoRts | 6:29006 | DOI: 10.1038/srep29006 chromatin 13 . Taking into account our above results together with the functional role of histone chaperone played by nucleolin via its association with chromatin 64 , we examined the contribution of nucleolin towards the observed close association between vRNPs and the dense chromatin compartment.
To further support these results, we assessed the association of vRNP components with chromatin under conditions of endogenous nucleolin depletion during infection (Fig. 5B). Again, we infected A549 cells, previously transfected for 48 hours with si-Control or si-Nucleolin, with H3N2 virus. At 24 hpi we performed subnuclear fractionations to detect NP, nucleolin and subcellular markers by western blot, as described above. In the si-Control condition, NP localized in the cytoplasmic, nucleoplasmic and chromatin fractions (Fig. 5B, lanes 1,  3, 5, 7 and 9) and control subcellular markers were detected as expected, whereas nucleolin mainly accumulated in ch500 fraction (Fig. 5B, lane 9). In contrast, effective silencing of nucleolin led to a decrease in the level of NP found in Ch150 fraction a more pronounced decreased in Ch500 fraction at 24 hpi (Fig. 5B, compare lane 8 versus 7 and lane 10 versus 9), in agreement with the hypothesis of an alteration in the vRNP targeting to chromatin.
Altogether, these results indicate that nucleolin strongly accumulates in regions of high chromatin density during IAV infection and is required for an efficient association of vRNPs with this chromatin compartment during their nuclear trafficking. Discussion Influenza A are nuclear replicating viruses which have direct access to and thus hijack host nuclear machineries in order to achieve optimal infection 3 . In this context, several studies including ours have described the extensive modifications of nuclear ultrastructure and composition 7,5,30 with marked nucleolar and chromatin alterations in IAV infected cells 17,18,20 . In accordance with these observations, we report in this study the differential delocalization through the nucleus of the three major nucleolar components, nucleolin, B23 and fibrillarin during the time-course of IAV infection. Interestingly, the earlier and specific redistribution of nucleolin was of interest and in particular its dynamic colocalization with NP at the vicinity of the inner nuclear membrane, where vRNPs accumulate before their nuclear export 13,16,58 . Consequently, we then investigated the putative connection between nucleolin and vRNP traffic, notably in regard to a previous study that identified nucleolin as a possible interaction partner of reconstituted IAV vRNPs, without further exploring the functional significance of this interaction 7 . Interestingly, some functional interactions have been described in other viral models, such as feline calicivirus (FCV) and Norwalk virus (NV), in which infections also result in a relocalization of nucleolin from nucleoli to nucleoplasm and the perinuclear area. In this context, nucleolin was shown to act as a positive regulator of viral replication by forming RNP complexes with viral RNA and interacting with the viral NS6/7 proteins 54 .
In this way, our results obtained in GST-pull down and immunoprecipitation assays indicate that endogenous nucleolin interacts with NP and suggest a possible association with vRNP complexes in the context of IAV infection. This hypothesis is supported by several results obtained in conditions of infection such as (i) the relocalization of endogenous Nucleolin in subnuclear compartments in which vRNPs components NP (Fig. 2) and PB1 (Supplementary Figure 2C) are also accumulated, (ii) the detection of vRNA within immunoprecipitated Nucleolin-NP complex (Fig. 3C), and (iii) the silencing of endogenous nucleolin expression that led to an altered nuclear trafficking of vRNPs and a consecutive significant delay in viral production (Fig. 4).
An interesting way to explore should be the role of RNA in the interaction between two RNA binding proteins such as NP and nucleolin. Possibly, the flexibility of NP structure in promoting RNA binding which likely involves conformational change(s) of NP, as previously suggested 66 , would consecutively impact on its interactions with cellular factors.
Altogether, our results play in favor of a possible model in which nucleolin is involved in the nucleocytoplasmic export of vRNPs during IAV infection, even if we cannot exclude concomitant impact on other steps of viral replication (e.g. viral transcription, viral mRNA translation) due to the multifunctional properties of the host nucleolin.
Data from several other studies converge on a model for vRNP traffic illustrated by a vRNP nuclear export complex consisting of as a daisy chain of proteins and involving dynamic and intricate interactions between vRNP, viral and host factors 62,67,68,69 . Previous in vitro experiments demonstrated that both vRNPs and M1 interact with the chromatin components at two different sites, RNP to the histone tails and M1 to the globular domain of a histone octamer 62,70 , whereas NEP acts as an adapter and interacts with the host Crm1 export machinery 68 .
However, several questions remain unanswered, notably what really happens in the nucleus during IAV infection, what are the exact modifications of nuclear architecture and notably how does chromatin remodeling contribute towards vRNP trafficking? Chase and collaborators proposed that association between vRNPs and export machinery occurs in dense chromatin domains, where RCC1 exchanges GDP for GTP to produce export energy. They suggested that IAV gain preferential access to host nuclear export machinery by targeting vRNP export complexes to chromatin-dense territories which confer a physical proximity to the sites of Ran regeneration 13 . Our results support the model proposed by Chase and collaborators and also suggest the requirement of nucleolin in the close association between vRNPs and the host chromatin (Fig. 6A). Indeed, by subnuclear fractionation, we observed a strong accumulation of nucleolin within the condensed chromatin domain (ch500 fraction), which is targeted by vRNPs at 24 hpi. Moreover, depletion of endogenous nucleolin induced a decrease in accumulation of NP, a component of vRNP trafficking through the condensed chromatin fractions.
Our hypothesis relies on the chromatin remodeling and histone chaperone activities of the nucleolin 34,64,71 . Indeed, nucleolin is known to induce chromatin de-condensation by binding to histone H1 72 and thus has a function analogous to the family of high-mobility-group (HMG) proteins. In this context, the nucleolin-H1 interaction could cooperate to reorganize cellular chromatin in order to facilitate its efficient targeting by vRNPs. By interacting with NP, the main component of vRNPs, nucleolin would thus constitute a key mediator in the vRNP nuclear export (Fig. 6A). Alternatively and in regards to RNase sensitivity assay, nucleolin most probably also interacts with non-vRNP form of NP during infection, which could incidentally have an impact on several other steps of viral cycle, as well, including vRNP assembly and/or trafficking.
Analysis of enriched gene ontology terms corresponding to the interactants of NP and nucleolin provided molecular support to the pivotal role of NP-nucleolin association in linking chromatin organization and nucleo-cytoplasmic transport biological processes (Fig. 6B). Indeed, a significant proportion of nucleolin interactants are involved in chromatin organization, such as components of nucleosome (e.g.: H2A2C, H3F3A; Blue Squares, Benjamini P-value = 4.3 × 10 −2 ) or chromatin associated factors such as HMGB1 or HMGB2. Alternatively, NP interactants are strongly enriched in nucleo-cytoplasmic transport proteins, such as RAN or KPNA1, and protein complexes associated with the nuclear envelope such as Nup62 (Red squares, Benjamini P-value = 1.7 × 10 −2 ) (Fig. 6B). Interestingly, at the interface of chromatin organization and nucleo-cytoplasmic transport processes, and in addition to p53 (TP53) which we have previously shown to be involved in multiple levels of IAV-host interactions 73,74 , B23 (NPM1) represents another common interactant of NP and nucleolin as well as interacting with vRNP 7,28 .
Further studies are required to elucidate the mechanism through which nucleolin contributes to IAV vRNP trafficking and to identify other host factors involved in chromatin remodeling and post-translational modifications of histones in the context of IAV infection. Another important consideration is the constant shuttling of nucleolin between subcellular compartments which may reflect a dependence of nucleolin on post-translational modifications that determine its localization 36 . Therefore, the possibility that the phosphorylation and methylation of nucleolin, which influence its interaction with nucleic acids or ADP-ribosylation 40 , also contribute to its nuclear redistribution during IAV infection, remains to be elucidated/investigated.
Beyond the role of nucleolin in IAV vRNP trafficking, its binding to NP could also contribute to viral transcription/replication in accordance with previous observations of nucleolin's impact on IAV polymerase activity [31][32][33] . This further hypothesis is consistent with the facts that (i) IAV transcription is dependent on the cellular transcription machinery 61 and (ii) vRNP binds nucleosomes 62 and chromatin-associated factors 7 . Similarly to HMGB1 75 , putative nucleolin binding to NP/vRNA complexes could facilitate the recruitment of vRNPs at transcriptionally active sites of chromatin and also contribute in this way to efficient IAV replication.
Additionally, the NP-nucleolin association could also restrict nucleolar functions of nucleolin (ribosome biogenesis, ribosome transport) and thereby contribute to the strong remodeling of nucleolar compartment observed during IAV infection 17,18,20 . Concomitantly, by directly binding to nucleolin in the nucleolus 28,29 , NS1 may contribute towards nucleolar disruption resulting in the subsequent redistribution and availability of nucleolin for vRNP trafficking and viral replication. However, further studies are needed to investigate the specific sequence of events occurring in such a viral hijacking of nucleolin. An interesting approach could be to further investigate the histone mimic of NS1 which was recently reported to be able to recruit several host chromatin-associated proteins 76 . Figure 6. Model for the nucleolin contribution in the vRNPs export at the crossroads between chromatin organization and nucleocytoplasmic transport. (A) Influenza viruses gain preferential access to host cell export machinery by targeting chromatin at the sites of Ran regeneration, as proposed by Chase et al. 13 . As a chromatin remodeler and an interaction partner of histone H1, nucleolin should contribute to the chromatin targeting of vRNPs. (B) The interaction between nucleolin (NCL, yellow diamond) and NP (orange diamond) is placed on the center of the graph (thick edge). Proteins interacting with nucleolin or NP and involved in chromatin organization are in blue squares. Proteins interacting with nucleolin or NP and involved in nucleo-cytoplasmic transport are in red squares. NPM1 (B23) and TP53 involved in both processes are in green squares.
In conclusion, our results highlight nucleolin as an essential host actor for the chromatin targeting of the nucleocytoplasmic export complex of vRNPs during infection. An interesting avenue to explore would now be the potential of targeting host nucleolin, as already proposed in cancer therapy 77 , and more broadly chromatin and host-associated factors, in identifying novel anti-influenza treatments.

Methods
Viruses and cells. IAV strain A/Moscow/10/99 (H3N2), was obtained from the French national influenza monitoring network GROG (Groupes Régionaux d'Observation de la Grippe, Lyon, France). Virus was cultivated and titered in MDCK cells and stored at − 80 °C, as previously described 60 . MDCK cells were purchased from Lonza (ATCC, CCL34) and were passaged twice weekly in serum-free Ultra-MDCK medium (Lonza) supplemented with 2 mM L-glutamine (Sigma Aldrich), penicillin (225 units/ml) and streptomycin (225 μ g/ml) (Lonza). Human lung epithelial cells (A549) were cultivated in Dulbecco's Modified Eagle's Medium (DMEM, Lonza, Biowhittaker) supplemented with 200 units/ml penicillin, 200 μ g/ml streptomycin, 2 mM L-glutamine and 10% fetal calf serum (Dutscher). All cells were maintained at 37 °C with 5% CO2. For infection, viruses were inoculated at a multiplicity of infection (MOI) of 1 for 1 hpi to 24 hpi, in DMEM supplemented with 200 units/ml penicillin, 200 μ g/ml streptomycin, 2 mM L-glutamine and 0.5 μ g/ml TPCK-trypsin (Roche diagnostics), as previously described 17 . For viral kinetics on A549 cells, confluent cells were infected with viruses at a MOI of 2 for one hour under a minimal volume of medium at 37 °C. Cells were then overlaid with fresh medium and incubated at 37 °C. Samples of supernatants were harvested at several time points post-infection and stored at − 80 °C until end point viral titration assays in MDCK cells using the Reed and Muench statistical method (TCID50/mL), as previously described 78 . Light microscopy observations. The Confocal Microscope Laser Scanning allows an optical section image acquisition in two dimensions X and Y. For 3D reconstruction of a stack of images, with a motorized stage, a minimum of twenty optical sections were acquired with the confocal microscope described above with a 0.1 μ m z-step. JAcoP plugin for imageJ 80 was used to highlight the colocalization of both stainings by thresholding the signal above the background. The confocal image stacks were treated by three-dimensional visualization software: Amira version 3.1 (Mercury Inc). To create three-dimensional animations and pictures, two methods were applied. Isosurface generated a geometric surface representing points of constant value as a limit between shadow and brightness of a staining. This limit was defined through an adjusted threshold. This representation was used to define the limit of the nucleus staining (DAPI). Volume rendering visualization technique was used to simulate the casting of light rays through the volume and give a 3D impression of the stained structure. Whole cell lysates were cleared by centrifugation for 20 min at 13,000 rpm at 4 °C and soluble protein complexes were purified using glutathione sepharose 4B beads (Amersham), according the manufacturer's instructions. Beads were then washed extensively four times with lysis buffer, and pull-down proteins displayed on NuPAGE acrylamide gels before being transferred to nitrocellulose membrane. GST-tagged viral protein and endogenous nucleolin were detected using standard immunoblotting techniques with mouse anti-GST (Covance, 1:1000) and rabbit anti-nucleolin (1:1000) antibodies. All results were reproduced in at least three independent experiments.

Immunoprecipitation. 1 × 10 6 cells were harvested 8 or 24 h after infection with the human virus
A/Moscow/10/99 (H3N2). Then, cell pellets were lysed at 4 °C for 30 min in 500 μ L of RIPA buffer (50 mM Tris-HCl pH 7.2, 0.1% Triton X-100, 0.1% CHAPS) that contained a cocktail of protease inhibitors (cOmplete ® , ROCHE) and was supplemented with 150 mM NaCl. Cell extracts were centrifuged at 13000 rpm for 30 min at 4 °C. Inputs correspond to 10% of supernatants (50 μ L/500 μ L). Supernatants were incubated overnight at 4 °C in the presence of anti-nucleolin antibody. Immuno-complexes were immobilized for 30 min on protein A/G agarose beads (Santa-Cruz, sc-2003). Bound complexes were then washed 3 times in RIPA buffer supplemented with 150 mM NaCl 43 . Then, for protein analysis, bound complexes were eluted with NuPAGE LDS sample buffer (Invitrogen, NP0008) and boiled for 5 min. Proteins were then separated on a 10% polyacrylamide/SDS gel. For viral RNA analysis, RNAs were extracted and then quantified by RT-qPCR 60 .
Scientific RepoRts | 6:29006 | DOI: 10.1038/srep29006 Western blot analysis. The level of expression of viral or cellular proteins was assessed by western blotting.
Cell fractionation. For nucleo-cytoplasmic fractionation, 6 × 10 5 cells were scraped into 1 mL of ice-cold PBS. After one wash in PBS, cells were fractionated with PBS, 0.1% NP-40 and protease inhibitor. One-tenth of the suspension was kept aside as whole cell lysate. After centrifugation, the supernatant was kept for the cytoplasmic fraction. The nuclei pellets were washed with PBS, and then nuclei lysed in LDS 1X (Method adapted from 81 ). The protocol used for cytoplasm/nucleoplasm/chromatin150/chromatin500 fractionation, has already been described 13,65 . Briefly, 1.5× 10 7 A549 cells were washed and pelleted in PBS. The cell pellet was resuspended in 5 ml sucrose buffer (10 mM HEPES pH 7.9, 10 mM KCl, 2 mM Mg acetate, 3 mM CaCl2, 340 mM sucrose, 1 mM DTT and protease inhibitor), and cells were incubated 10 min on ice. NP-40 was then added to a final concentration of 0.5%, followed by vortexing for 20 sec. Lysate was triturated 6 times and vortexed for 20 sec. One-tenth of the suspension was removed for the whole extract. After centrifugation for 10 min at 4 °C at 4,500 g, supernatant was saved as the cytoplasmic fraction, and the nuclear pellet was washed once with 2.5 ml sucrose buffer. The nuclear pellet was then resuspended in 750 μ L of nucleoplasm extraction buffer (50 mM HEPES pH 7.9, 150 mM potassium acetate, 1.5 mM MgCl2, 0.1% NP-40, 1 mM DTT and protease inhibitor), transferred to an all-glass 1 ml Dounce homogenizer, and homogenized with 60 strokes. Homogenate was then incubated 20 min at 4 °C on a rotating wheel, followed by a second homogenization with 70 strokes of Dounce, followed by a centrifugation for 10 min at 4 °C at 14,000 rpm in a microcentrifuge. The supernatant was saved as the nucleoplasmic fraction, and the pellet was resuspended in 750 μ l nuclease incubation buffer (50 mM HEPES pH 7.9, 10 mM NaCl, 1.5 mM MgCl 2 , 1 mM DTT and protease inhibitor) containing 100 U/ml Benzonase nuclease (Novagen). Chromatin was digested for 10 min at 37 °C, followed by addition of NaCl to 150 mM and further incubation for 20 min on ice. Digested lysate was then centrifuged 10 min at 4 °C at 14,000 rpm, and supernatant was saved as the low-salt chromatin fraction. The pellet was resuspended in 750 μ l chromatin extraction buffer (50 mM HEPES pH 7.9, 500 mM NaCl, 1.5 mM MgCl 2 , 0.1% Triton X-100, 1 mM DTT and protease inhibitor) and incubated 20 min on ice, followed by centrifugation as above. The supernatant was saved as the high-salt chromatin fraction 13 .
Real-time quantitative RT-PCR. Viral RNA was extracted from the supernatant of infected MDCK cell cultures using the QIAmp viral RNA minikit (Qiagen), according manufacturer's instructions. On Cellular viral RNAs were extracted using Trizol reagent method (Josset et al., 2008). The real-time PCR was developed to detect and quantify the number of genome copies from all subtypes of influenza type A viruses, as previously described 60 . Primers and TaqMan probe (5′ FAM/3′ BHQ) were designed in a highly conserved sequence of viral M gene 82 . The quantification standard was pHW2000 plasmid containing influenza A/Moscow/10/99 H3N2 M1 gene. Amplification was performed in an ABI Prism 7500 system (Applied Biosystems) using the Superscript III/ Platinum Taq One-step qRT-PCR kit (Invitrogen). The 50 μL reaction mixture contained 5 μL extracted RNA or standard plasmid, 25 μL Superscript III/Platinum Taq One-step qRT-PCR reaction mix, 1 μL Superscript III/ Platinum Taq One-step qRT-PCR enzyme mix, 0.1 μL ROX reference dye as a passive reference, 200 nmol Taqman probe, and 800 nmol of each primer. The following thermal profile was used: an initial reverse transcription at 50 °C for 15 min, followed by reverse transcriptase inactivation and DNA polymerase activation at 95 °C for 2 min and 40 cycles of amplification (15 s at 95 °C and 30 s at 60 °C). The reporter dye (FAM) signal was measured against the internal reference dye (ROX) signal to normalize the signals for non-PCR-related fluorescence fluctuations that occur from well to well. Data were collected at the annealing step of each cycle and the threshold cycle (CT) for each sample was calculated by determining the point at which the fluorescence exceeded the threshold limit. siRNA transfection. A pool of ON-TARGETplus SMARTpool siRNA duplexes specific for human nucleolin (si C23), siRNA J-003854-05: 5′ -GCAAAGAAGGUGGUCGUU-3′ , siRNA J-003854-06: 5′ -GAUAGUUACUGACCGGGAA-3′ , siRNA J-003854-07: 5′ -CAAAUCUGCUCCUGAAUUA-3′ , siRNA J-003854-08: 5′ -GAAAGAAGAUGAAGUUUGA-3′ (from Dharmacon, ThermoFisher Scientific) were used to silence endogenous nucleolin expression, as previously shown 83 . Transfection was performed using Oligofectamine (Invitrogen), as previously described 60 . Two rounds of siRNA transfection were performed -with a 24 h interval -to achieve a sufficient level of nucleolin silencing. Twelve hours after the second transfection, cells were subjected to H3N2 infection at the indicated MOIs and incubated for different time periods.
Functional analysis, network reconstruction and representation. DAVID database was used for functional annotation 84 . DAVID functional annotation chart tool was used to perform Gene Ontology biological process analysis. Gene Ontology terms with a Benjamini-Hochberg corrected p-value smaller than 5.10-2 were considered as significantly overrepresented. NP-human protein-protein interactions were retrieved from the VirHostNet database 85 . Human protein interactions were retrieved from iREF 86 . Network is visualized with Cytoscape 87 .