Influenza A Virus Utilizes Low-Affinity, High-Avidity Interactions with the Nuclear Import Machinery To Ensure Infection and Immune Evasion

We used intracellular nanobodies to block influenza virus infection at the step prior to nuclear import of its ribonucleoproteins. By doing so, we were able to answer an important but outstanding question that could not be addressed with conventional tools: how many of the ∼500 available NLS motifs are needed to establish infection? Furthermore, by controlling the subcellular localization of the incoming viral ribonucleoproteins and measuring the cell’s antiviral response, we were able to provide direct evidence for the long-standing hypothesis that influenza virus exploits nuclear localization to delay activation of the innate immune response.

ABSTRACT

The incoming influenza A virus (IAV) genome must pass through two distinct barriers in order to establish infection in the cell: the plasma membrane and the nuclear membrane. A precise understanding of the challenges imposed by the nuclear barrier remains outstanding. Passage across is mediated by host karyopherins (KPNAs), which bind to the viral nucleoprotein (NP) via its N-terminal nuclear localization sequence (NLS). The binding affinity between the two molecules is low, but NP is present in a high copy number, which suggests that binding avidity plays a compensatory role during import. Using nanobody-based technology, we demonstrate that a high binding avidity is required for infection, though the absolute value differs between cell types and correlates with their relative susceptibility to infection. In addition, we demonstrate that increasing the affinity level caused a decrease in avidity requirements for some cell types but blocked infection in others. Finally, we show that genomes that become frustrated by low avidity and remain cytoplasmic trigger the type I interferon response. Based on these results, we conclude that IAV balances affinity and avidity considerations in order to overcome the nuclear barrier across a broad range of cell types. Furthermore, these results provide evidence to support the long-standing hypothesis that IAV’s strategy of import and replication in the nucleus facilitates immune evasion.

IMPORTANCE We used intracellular nanobodies to block influenza virus infection at the step prior to nuclear import of its ribonucleoproteins. By doing so, we were able to answer an important but outstanding question that could not be addressed with conventional tools: how many of the ∼500 available NLS motifs are needed to establish infection? Furthermore, by controlling the subcellular localization of the incoming viral ribonucleoproteins and measuring the cell’s antiviral response, we were able to provide direct evidence for the long-standing hypothesis that influenza virus exploits nuclear localization to delay activation of the innate immune response.

INTRODUCTION

An archetypal feature of influenza A virus (IAV) is its requirement for replication within the nuclear compartment (1, 2). To ensure productive infection, IAV must execute a complicated series of nuclear import and export events throughout its life cycle (3–6). Indeed, the first task for viral ribonucleoproteins (vRNPs) upon their entry into the cytoplasm is to mediate their own rapid translocation into the nucleus, whereupon primary transcription of the viral RNA (vRNA) genome is initiated (7–9). The recruitment of import machinery to the vRNPs is accomplished via a nuclear localization signal (NLS)-containing motif located on the N terminus of the viral nucleoprotein (NP), one copy of which binds the ∼13.5-kb genome every 24 to 30 bp, for a total of ∼566 to 452 NP molecules per virus particle (10–13). Other vRNP components contain NLS motifs, but their role, if any, during vRNP import remains unknown. NP is an essential protein and interacts with a multitude of host factors, including members of the importin alpha family here referred to as karyopherin alpha 1 [KPNA1] to KPNA6 (14–22). Because NP provides essential functions throughout the virus life cycle, it remains relatively intractable to analysis by conventional methods (e.g., mutagenesis, small interfering RNA [siRNA] knockdown) (23–25).

Single-domain antibodies, or nanobodies, provide an alternative method to study the function of essential viral proteins during infection (26–28). Nanobodies are the smallest antigen binding domain known (∼15 kDa) and are derived from the variable domain of the heavy chain from heavy-chain-only antibodies (VHH) expressed by members of the Camelidae family (e.g., camels, alpacas, llamas) (27, 29). Nanobodies exhibit two properties that are distinct from those of their conventional cousins and make them ideal tools for studying host-virus interactions during viral infection: (i) they can be expressed inside cells while retaining their full ligand binding capacity (these are also known as intrabodies), and (ii) binding often alters the ligand function due to the nanobodies’ proclivity for recognizing clefts/pockets/active sites. These properties make it possible to target and disrupt specific functions of essential proteins (26, 30–32).

Here we exploit anti-NP nanobodies to dissect out the function and importance of NP during the brief but critical window of time between virus-host membrane fusion and vRNP nuclear import. Specifically, we quantify, during live infection, how many of the ∼500 total NP-derived NLS motifs are required for successful vRNP import and infection in different cell types. Based on these results, we conclude that the NP panoply provided by incoming virus is needed to compensate for the low affinity for KPNAs during entry. Absent the necessary avidity, incoming genomes remain frustrated in the cytoplasm and trigger activation of the type I interferon (IFN) response, thus providing direct support for the hypothesis that IAV localizes to the nucleus for evasion of cytoplasmic vRNA sensors (33–35).

RESULTS

What is NP2’s mechanism of action?

We previously described the generation of the anti-NP nanobody NP2 and showed that its expression in cells enables restriction to IAV infection (36) (Fig. 1A). This restriction was specific to IAV and was not observed after challenge with vesicular stomatitis virus (36) or Sindbis virus (Fig. 1B). Our previous study indicated that restriction occurs at a step prior to nuclear import. Since KPNAs are known to be required during this time, we tested whether NP2 expression interferes or competes with their binding to the incoming vRNPs. To do this, we first exposed vRNPs to saturating levels of NP2 or HA68 by forcing IAV fusion at the plasma membrane of NP2- or HA68-expressing cells, followed immediately by lysis in NP-40-containing buffer (HA68 is a nanobody that recognizes the IAV hemagglutinin and that served as a negative control in our experiments). Next, the lysate was incubated with a second lysate obtained from 293T cells overexpressing FLAG-tagged KPNA6. The IAV vRNPs were then precipitated from the mixture, and coprecipitating proteins were analyzed by Western blotting. Figure 1C demonstrates that NP2-exposed vRNPs coprecipitated significantly less KPNA6 than the negative control (HA68-exposed vRNPs). This indicates that NP2 binding to the vRNPs interferes with KPNA recruitment, which is consistent with blocked nuclear import of vRNPs in NP2-expressing cells.

FIG 1.

NP2 restricts IAV infection by interfering with KPNA recruitment. (A) NP2-expressing cells restrict IAV infection. MDCK cells stably expressing NP2-HA or HA68-HA were challenged with IAV (MOI, 1.0). At 8 hpi, cells were collected, lysed, and analyzed by SDS-PAGE and immunoblotting using antibodies against the HA epitope (to monitor VHH expression) and NP (as a readout for infection levels). (B) NP2-expressing cells do not restrict replication of an irrelevant virus. 293T cells stably expressing NP2-HA or HA68-HA were challenged with Sindbis virus (MOI, 0.001). Supernatant was collected at 0, 12, and 24 hpi and subsequently titrated over BHK cells. (C) NP2 interferes with KPNA recruitment to incoming vRNPs. HA68-HA- and NP2-HA-expressing MDCK cells were infected with PR8 at an MOI of 1,000 using a low pH to force fusion of the virus at the plasma membrane, and samples were immediately lysed. Lysates were then mixed with lysate from 293T cells transfected with FLAG-tagged KPNA6. After incubation for 1 h at 4°C, NP was immune precipitated, and bound fractions were analyzed by SDS-PAGE and immunoblotting using antibodies against the FLAG epitope, NP, and the HA epitope (left). Expression levels of all proteins in the total cell extract. (D) NP2 binds an internal region of NP (middle and right). Expression plasmids for NP2-Ch or HA68-Ch and NP-GFP (WT or truncations) were transfected into MDCK cells. At 24 hpt, cells were imaged using live fluorescence microscopy. All results are representative of those from experiments performed three or more times. In panels A and C, the numbers to the left of the gels are molecular weights (in kilodaltons). tub and TUB, tubulin; Ch, mCherry.

Next, we mapped the region on NP that is necessary for NP2 binding using an intracellular binding assay, which takes advantage of two points: (i) NP2 binds to NP alone but does not block its import (36), and (ii) NP localizes to the nucleus via an N-terminal NLS (NLS-I). Thus, we coexpressed NP2-mCherry alongside either full-length NP (residues 1 to 498) [NP(1–498)] or C-terminal truncations fused to green fluorescent protein (GFP) and then imaged their localization via live microscopy. Figure 1D demonstrates that coexpression of NP-GFP(1–498) with NP2-mCherry, but not HA68-mCherry, results in binding and nuclear colocalization. Nuclear colocalization was maintained in cells coexpressing NP2-mCherry with NP-GFP(1–478) or NP-GFP(1–418) but lost with the further truncations NP-GFP(1–376) and NP-GFP(1–340). This indicates that NP residues 377 to 418 are required for binding NP2.

Can KPNA recruitment to incoming vRNPs be “outsourced” to NP2 containing an NLS?

The results thus far suggest that NP2 restriction may be due solely to interference with KPNA recruitment during viral entry. If true, then bypassing this restriction would rescue virus infection and replication. We wanted to test this by fusing an NLS to NP2, thereby “outsourcing” KPNA recruitment to the nanobody. We chose to use the NLS-I sequence (derived from NP), since it is well-defined and its fusion to a foreign protein can mediate nuclear localization (15).

Before testing whether NP2 fused to NLS-I (NP2-NLS) can be used for outsourcing, we first measured if the fusion affected its binding to NP or its localization in cells (in the absence of NP). Figure 2A demonstrates that NP2 and NP2-NLS exhibit a similar binding affinity for NP. Furthermore, Fig. 2B demonstrates that, despite containing an NLS motif, NP2-NLS localization remains diffuse throughout the cell, similar to what is observed with NP2 or HA68. This overlapping localization pattern is most likely due to their small size (∼15 kDa), which enables easy diffusion through the nuclear pore.

FIG 2.

KPNA recruitment to incoming vRNPs can be mediated in trans by NP2-NLS. (A) The NP2 affinity for NP is unchanged upon fusion with an NLS. (Left) Lysates from cells transfected with an empty vector or vectors expressing the NP2-HA and NP2-NLS-HA nanobodies were used as probes for a dot blot assay on a membrane that contained immobilized NP. (Right) The signal intensity was quantified using ImageJ software. Error bars represent the standard deviation from three independent experiments. a.u., absorbance units. (B) NP2 localization in cells is unchanged upon fusion with an NLS. WT and VHH-expressing MDCK cells were fixed and permeabilized. DNA was stained with DAPI (4′,6-diamidino-2-phenylindole), and VHHs were labeled using anti-HA antibody. Images were obtained using a 40× oil immersion lens and are representative of those from experiments performed three or more times. (C) NP2-NLS-expressing cells support IAV infection. MDCK cells stably expressing NP2-HA, NP2-NLS-HA, and HA68-HA were challenged with IAV (MOI, 1.0). At 8 hpi, cells were collected, lysed, and analyzed by SDS-PAGE and immunoblotting. The image is representative of the images from experiments performed more than three times. (D) NP2-NLS-expressing cells support IAV infection. WT and VHH-expressing MDCK cells were infected with PR8 (MOI, 1.0), and at 16 hpi, cells were fixed and permeabilized. Cells were stained with DAPI (DNA) and antibodies against NP and the HA epitope. Images were obtained using a 40× oil immersion lens and are representative of those from experiments performed at least three times. (E) NP2-NLS-expressing cells support IAV production in a single-cycle assay. Supernatants removed at 24 hpi from VHH-expressing cells infected at an MOI of 1.0 were analyzed for the presence of infectious virus using a plaque assay. Error bars represent the standard deviation from three independent experiments. The figure is representative of the figures from experiments performed at least three times. (F) NP2-NLS supports IAV replication in a multicycle assay. Supernatants removed over time from VHH-expressing cells infected at an MOI of 0.01 were analyzed for the presence of infectious virus using a plaque assay. Error bars represent the standard deviation from three independent experiments. (G) NP2 restriction is specific for influenza virus. MDCK cells were infected with NDV-GFP (MOI, 1.0). At 24 hpi, cells were lysed and samples were analyzed using SDS-PAGE and immunoblotting using antibodies against GFP, the HA epitope, and tubulin. The image is representative of the images from experiments performed at least three times. In panels A, C, and G, the numbers to the left of the gels are molecular weights (in kilodaltons).

Since the NP2 function is not altered by its fusion to NLS-I, we next tested whether it could mediate the nuclear import of incoming vRNPs. For this, we compared IAV infection in cells stably expressing NP2, NP2-NLS, and HA68. Figures 2C to F demonstrate that cells stably expressing NP2-NLS support IAV replication at levels similar to those for the HA68-expressing control cells. NP2-NLS expression has no effect on the replication of a distinct virus, Newcastle disease virus (NDV)-GFP, indicating that its effects are specific to IAV (Fig. 2G).

Since NP2-NLS-expressing cells support the multicycle growth of IAV, we tested whether the viral progeny, obtained from the supernatant during infection, contain NP2-NLS protein. Indeed, Fig. 3A demonstrates that significant amounts of NP2-NLS become incorporated into the virus prior to virion biogenesis. We quantified the amount of NP2-NLS incorporated on a per particle basis and concluded that each virion contained ∼438 NP2-NLS molecules. Similar analysis of NP levels yielded an estimate of ∼501 NP molecules per virion (Fig. 3B to D). Based on this result, we conclude that a majority of the NP molecules inside the virus particle (∼88%) remain bound to NP2. We note that, due to our virus purification strategy, it is possible that not all of the NP2-NLS/NP complexes observed in Fig. 3A are derived from bona fide virus particles (e.g., they may be inside copurified exosomes). Nevertheless, we believe that our approach provides a relatively accurate estimation of the quantity of NP2-NLS/NP2 complexes inside both the particle and the cell.

FIG 3.

Quantification of NP2-NLS incorporated into the virus particle. (A) NP2-NLS incorporates into viral particles. MDCK cells were infected with PR8 (MOI, 0.01). Supernatant containing virus was recovered at 72 hpi, concentrated by ultracentrifugation over a 20% sucrose cushion, and subjected to SDS-PAGE and immunoblot analysis using antibodies against NP and the HA epitope. The image is representative of the images from experiments performed at least three times. (B) Development of standards to measure NP and NP2-NLS in the virus. Purified NP (top) and NP2 (middle) were run on an SDS-PAGE gel alongside known quantities of BSA or RNase A and quantified using densitometry analysis. (Bottom) The resulting concentrations for each standard are shown in the table. Average values and errors were obtained from analysis of two independent SDS-PAGE gels. IP, immune precipitation. (C) Quantification of NP2-NLS incorporated into the virus particle. (Top) Defined amounts of each standard described in the legend to panel B were run alongside defined volumes of WT PR8 and PR8 (NP2-NLS) on an SDS-PAGE gel and then analyzed by immunoblot analysis using antibodies against NP and the HA epitope. (Bottom) The amounts of NP and NP2 were then calculated using densitometry analysis. The amount of NP per microliter was then used to back calculate the virus particle concentration (assuming ∼550 NP molecules per particle). The resulting calculations for NP and NP2 levels incorporated into both WT and NP2-NLS virus particles are shown in the table at the bottom of panel B. Histogram data are from two independent SDS-PAGE gels. In panels A to C, the numbers to the left of the gels are molecular weights (in kilodaltons).

How many NLS motifs are needed for IAV infection?

The NP NLS-I exhibits a low affinity for KPNAs, especially compared to that of the classical simian virus 40 (SV40) NLS (37). We hypothesized that this low affinity may be compensated for by a high avidity (provided by the ∼500 NP-encoded NLS motifs) and set out to measure the number of NLS motifs required for infection.

To do this, we transfected 293T cells with various ratios of NP2 and NP2-NLS and then challenged them with IAV. Figure 4A demonstrates that NP2 restriction cannot be overcome until NP2-NLS comprises ∼20% of the total intracellular nanobody pool. In other words, nearly one-fifth of the nanobody-occupied sites must contain an accessible NLS motif (provided in this case by NP2-NLS) to ensure infection levels similar to those observed in control cells transfected with only HA68 or only NP2-NLS.

FIG 4.

Quantification of NLS requirements for vRNP import and infection. (A) 293T cells require 10 to 20% of total NLS motifs for infection. 293T cells were cotransfected with NP2 and NP2-NLS in various ratios. At 24 hpt, cells were challenged with PR8 using an MOI of 1.0. Samples were lysed at 8 hpi and analyzed by SDS-PAGE and immunoblotting using antibodies against NP, the HA epitope, and tubulin. The immunoblot is representative of immunoblots from three independent experiments. The top row of numbers above the gel in panels A and C corresponds to the percentage transfected of either HA68 or NP2, and the bottom row of numbers corresponds to the percentage transfected of NP2NLS. (B) Susceptibility to IAV infection varies across cell types. The relative susceptibility of 293T, MDCK, BHK, and HeLa cells to infection by GFP-expressing forms of IAV (PR8) and NDV was tested by challenging each with a defined volume of inoculum and then counting the number of GFP-positive cells at 12 hpi by live microscopy. The percentage of cells infected was set relative to what was observed in 293T cells. The results are representative of those from two independent experiments. (C) (Top) MDCK cells require 30 to 40% of total NLS motifs for infection. MDCK cells were transfected with various ratios of NP2 and NP2-NLS and then challenged with IAV (MOI, 1.0). Samples were lysed at 8 hpi and analyzed by SDS-PAGE and immunoblotting using antibodies against NP, the HA epitope, and tubulin. The immunoblot is representative of immunoblots from three independent experiments. (Middle) BHK cells require 25 to 50% of total NLS motifs for infection. The samples were analyzed as described for the top panel. (Bottom) HeLa cells require 50 to 75% of total NLS motifs for infection. The samples were analyzed as described for the top panel. In panels A and C, the numbers to the left of the gels are molecular weights (in kilodaltons).

Is the number of NLS motifs required an intrinsic property of the incoming virus or dependent on external (host) factors? To address this, we repeated the transfected ratio experiment in three additional cell types (MDCK, HeLa, and BHK cells) that were chosen based on the fact that they could be transfected with a high efficiency and also exhibited a range of susceptibility to IAV infection (Fig. 4B), which may be due, in part, to the import requirements of the incoming vRNPs. As Fig. 4C demonstrates, NP thresholds varied across all three of the cell types (∼40% to 50% for MDCK and BHK cells and >50% for HeLa cells), indicating that import requirements are cell type dependent and correlate with their relative susceptibility to infection. Overexpression of KPNAs or other host factors known to interact with incoming vRNPs (RIG-I, MxA) did not alter NLS requirements (data not shown). From this we conclude that import requirements are not static and depend on as yet undetermined cell type-specific influences.

Does increasing the KPNA affinity alter the number of NLS motifs needed for infection?

We next tested the relationship between KPNA affinity and avidity in KPNA recruitment. For this we outsourced KPNA binding to an NP2 fused to the high-affinity SV40 NLS (NP2-NLS^hi). Figure 5A demonstrates that NP2-NLS^hi expression could support infection in 293T and HeLa cells but not in MDCK or BHK cells. We confirmed that this effect was due to differences in affinity by repeating the assay in BHK cells using a variant SV40 NLS with an affinity level closer to that of NLS-I (NP2-NLS^lo), which supported IAV infection in both 293T and BHK cells (Fig. 5B).

FIG 5.

Outsourcing to a high-affinity NLS alters avidity requirements. (A) NP2-NLS^hi cannot support replication in BHK and MDCK cells. Cells expressing HA68 (68), NP2, NP2-NLS^hi (hi), and NP2-NLS (nls) were challenged with IAV (MOI, 1.0). Samples were lysed at 8 hpi and analyzed by SDS-PAGE and immunoblotting using antibodies against NP, the HA epitope, and GAPDH. The immunoblots are representative of those from two independent experiments. (B) NP2-NLS^lo supports replication in BHK cells. BHK or 293T cells expressing HA68, NP2, NP2-NLS, NP2-NLS^hi, and NP2-NLS^lo were challenged with PR8-NS1-GFP (MOI, 1.0). At 8 hpi, samples were either live imaged (for 293T cells) or fixed and probed using antibody against NP (for BHK cells). Parallel samples for both cell types were lysed at the same time point and analyzed by SDS-PAGE and immunoblotting using antibodies against NP, the HA epitope, and GAPDH. The immunoblots and images are representative of those from three independent experiments. (C) Outsourcing to a high-affinity NLS lowers avidity requirements in 293T and HeLa cells. 293T cells (left) and HeLa cells (right) were transfected with various ratios of NP2/NP2-NLS^hi and then challenged with IAV (MOI, 1.0). Samples were lysed at 8 hpi and analyzed by SDS-PAGE and immunoblotting using antibodies against NP, the HA epitope, and tubulin or GAPDH. The immunoblot is representative of immunoblots from at least two independent experiments. In all panels, the numbers to the left of the gels are molecular weights (in kilodaltons).

Next, we repeated the transfected ratio experiment in 293T and HeLa cells, using NP2 and NP2-NLS^hi. Figure 5C demonstrates that expression of a high-affinity NLS in both cell types resulted in lower avidity requirements (∼10% and ∼25%, respectively) for infection. Based on these results, we conclude that while increasing affinity can decrease avidity requirements during vRNP import, there is a significant trade-off for IAV in terms of cell tropism.

What are the consequences for affinity/avidity imbalance?

We next wanted to determine what happens when avidity requirements are not met in cells during virus challenge. We hypothesized that incoming genomes that remain frustrated in the cytoplasm would become substrates for cytoplasmic RNA sensors and trigger IFN expression. To test this, we examined IFN expression in wild-type (WT) MDCK cells or cells expressing HA68, NP2, or NP2-NLS. Figure 6A demonstrates that NP2-expressing cells exhibit a robust burst of IFN mRNA relative to WT or NP2-NLS- or HA68-expressing cells by 4 h postinfection (hpi). Interestingly, the levels then decreased in NP2-expressing cells, while they began increasing in WT and NP2-NLS- and HA68-expressing cells (Fig. 6B).

FIG 6.

Efficient nuclear import is necessary for vRNP evasion of the innate immune response. (A) Cytoplasmic trapping of vRNPs triggers the IFN response. MDCK cells were challenged with PR8 using an MOI of 10. At 4 hpi, the total RNA was extracted and used to analyze the levels of NP-vRNA (left), NP-mRNA (middle), and IFN mRNA (right). The values are normalized (Norm) to those for WT-infected cells (or mock-treated cells for the right panel). Error bars represent the standard error from two independent experiments. (B) Importation of vRNPs delays the IFN response. NP2- and HA68-expressing cells were infected with PR8 using an MOI of 10. Samples were removed at 4, 8, and 24 hpi for isolation of total RNA and subsequent analysis of IFN mRNA levels. Values are normalized to those for the mock-infected cells at each time point. Error bars represent the standard error from three independent experiments. (C) The delayed IFN response is independent of NS1 or virus replication. NP2- and HA68-expressing cells were treated with UV-inactivated dNS1 (MOI, 10), and at 4 hpi, IFN mRNA levels were analyzed by qRT-PCR. Error bars represent the standard deviation from three independent experiments. (D) Trapped vRNPs trigger an effective antiviral response. NP2- and HA68-expressing cells were treated with UV-inactivated dNS1 (MOI, 10). A 24 hpt, cells were challenged with NDV-GFP, and the cells were imaged at 12 hpi by live microscopy. NDV-GFP is an IFN-sensitive virus and is unable to replicate efficiently in cells with an induced antiviral state. Images are representative of those from experiments performed at least three times. (E) NLS availability correlates with evasion of the IFN response. MDCK and 293T cells were cotransfected with various ratios of NP2/NP2-NLS and an IFN-RFP reporter. At 24 hpt, cells were treated with UV-dNS1 (MOI, 100) and imaged at 24 h posttreatment by live microscopy and immunoblotting. Images are representative of those from experiments performed at least three times. The numbers to the left of the gels are molecular weights (MW; in kilodaltons). (F) NP2 expression does not affect the IFN response after challenge with an irrelevant virus. 293T cells were cotransfected with IFN-RFP and either an empty plasmid or plasmids expressing HA68, NP2, or NP2-NLS. (Left column) At 24 hpt, cells were challenged with Sendai virus and the RFP signal was imaged at 8 hpi by live microscopy. (Right three columns) In addition, supernatant from each cell was collected at 24 hpi and treated with UV to inactivate infectious virus. The relative levels of endogenous secreted IFN in each sample were then measured by titrating over naive MDCK cells. At 24 h posttreatment, cells were challenged with NDV-GFP and imaged at 12 hpi. NDV-GFP is an IFN-sensitive virus and is unable to replicate efficiently in cells with an induced antiviral state. Images are representative of those from experiments performed at least three times. Spnt, supernatant; IFNb, beta interferon.

Two possibilities could explain the delayed IFN activation observed in WT or HA68- or NP2-NLS-expressing cells: (i) nuclear import facilitates evasion of the host sensors during entry, or (ii) imported vRNPs express the NS1 IFN antagonist, which blocks subsequent activation of the IFN pathway. To determine which possibility explains the results in Fig. 5B, we compared IFN levels in NP2-expressing versus HA68-expressing cells challenged with UV-irradiated IAV that lacks the NS1 gene (UV-dNS1). Figure 6C demonstrates that IFN activation was observed only in the NP2-expressing cells, indicating that nuclear import facilitates innate immune evasion. Furthermore, treatment with UV-dNS1 induced an effective antiviral state in NP2-expressing cells but not HA68-expressing cells (Fig. 6D).

Finally, we examined how the balance between affinity and avidity impacts innate sensing in 293T and MDCK cells. Both cell types were transfected with various ratios of NP2 and NP2-NLS and an IFN reporter that expresses red fluorescent protein (IFN-RFP). Transfected cells were then challenged with UV-dNS1 and assessed for RFP expression at 24 hpi by microscopy and immunoblotting. Figure 6E demonstrates that IFN expression inversely correlates with the ability of the incoming virus to bind KPNAs and import to the nucleus. This effect was specific to IAV and was not observed in 293T cells challenged with an irrelevant virus that was not recognized by NP2 (Fig. 6F). Based on these results, we conclude that the affinity/avidity balance is essential for infection and immune evasion.

DISCUSSION

Our understanding of the early stages in IAV infection prior to translation of primary transcripts has been hamstrung by the lack of specific tools available. The use of chemical inhibitors (cycloheximide, actinomycin D, leptomycin B) or siRNA/CRISPR-based methods (targeting KPNAs, for example) induces pleiotropic effects, while more targeted approaches (i.e., reverse genetic engineering of NP) frequently result in replication-incompetent virus (23, 24, 38–41). This difficulty is underscored by the brevity of the time period in question, with nuclear import occurring in minutes and primary transcription being initiated shortly thereafter (7).

We have exploited a recently developed tool that specifically targets the NP function during vRNP import to slow the process, which opens it to unprecedented analysis during the earliest stages of infection. By fusing the NP-derived NLS motif to NP2, we successfully outsourced KPNA binding to the nanobody, a trick that we applied to measure the importance of avidity and affinity during infection. Numerous studies preceding this have shown that a single NLS motif can mediate the import of individual proteins. However, the ∼500 identical NLS motifs present on the incoming vRNP complex provide a unique and biologically relevant system to measure NLS requirements for successful import of a large protein complex. Using infection as our readout, we demonstrated that incoming vRNPs required that a substantial fraction of their total NLS motifs be made available for import to occur. Moreover, the absolute number varied among the four cell types that we examined.

We calculated an NP2 occupancy rate of ∼88% inside the virus particle. This calculation can likely be extended to describe the occupancy rate inside NP2- and NP2-NLS-expressing cells during virus entry and replication. Indeed, this calculation may even be an underestimate of the occupancy rate for incoming vRNPs, since actively replicating IAV shuts down host gene expression, and also, particle-associated vRNPs bind M1, which may interfere with the NP2-NLS interaction (both events would reduce the NP2-NLS ability to bind vRNPs during virion biogenesis). Nevertheless, assuming that (i) incoming vRNPs import as a complex, as has been reported previously (42), (ii) ∼501 NP-derived NLS motifs are present on the complex, (iii) the vRNP occupancy rate is ∼88%, and (iv) the threshold NLS requirement is ∼20% (in 293T cells), we calculate that incoming IAV requires ∼88 NP molecules to make their NLS available for infection. We note that our calculations are not meant to be a precise accounting of NLS requirements at the single-cell level. However, they do indicate that (i) a substantial fraction of NPs is required for import and (ii) the number of NPs required is not static and differs between cell types. We also examined the role of binding affinity during infection using the NP2-NLS^hi construct and show that it directly affects avidity in 293T and HeLa cells. Notably, no infection was observed when we outsourced to NP2-NLS^hi in MDCK or BHK cells. We think that this result is due to excessive levels of affinity and avidity that combine to limit the postimport disassociation of the KPNAs, thus interfering with subsequent transcription and replication steps.

While the specific factors that influence NLS requirements remain unknown, we propose that it reflects the need for rapid import, particularly under unfavorable conditions, like those that would occur when crossing species boundaries or in the context of an antiviral response. Indeed, by delaying or preventing KPNA recruitment, incoming vRNPs become substrates for cytoplasmic host sensors, as discussed in more detail below.

Finally, members of the Orthomyxoviridae family, which includes IAV, are relatively unique in their requirement for replication in the nuclear compartment. Other negative-strand viruses have the same general needs for replication (e.g., cap snatching, splicing) but accomplish their tasks in the cytoplasm, thereby avoiding the complex nuclear import/export cycle described in the introduction. There is a long-standing (yet untested) hypothesis which states that the atypical strategy used by IAV to replicate in the nucleus contributes to its evasion of the innate immune response (33–35). Rigorous validation of this hypothesis requires an ability to dictate the localization of the vRNPs in the cell during infection without disruption of cellular processes like transcription, nuclear export, or translation (which would interfere with the host response). Recently developed reagents, like nucleozin, permit the more nuanced control of vRNP function (43). In our case, by using HA68- or NP2-expressing cells for IAV challenge, we were able to exclusively place a defined amount of vRNPs in the cytoplasm or nucleus, respectively. Furthermore, by using live and UV-inactivated virus, we were able to uncouple the influence of viral products like newly synthesized vRNPs (which would export to the cytoplasm for packaging) and the NS1 protein (which blocks IFN activation) from the influence of incoming vRNPs on IFN activation. Our results unambiguously demonstrate that the ability of incoming vRNPs to trigger the IFN pathway is dependent on its cytoplasmic localization. Incoming vRNPs trapped in the cytoplasm exhibited rapid activation of the IFN pathway. In contrast, incoming vRNPs that imported exhibited a delayed activation of the IFN pathway, with the levels still increasing by the 24-hpi time point. Indeed, these kinetics correlate with both the arrival of newly synthesized vRNPs and their increased concentration over time and may resolve the contrasting results observed by others (which relied on the use of chemical inhibitors) (44, 45). Finally, the absence of a response to incoming vRNPs in HA68-expressing cells is independent of NS1-mediated antagonism, as challenge with UV-inactivated NS1-deficient virus recapitulated this effect.

The properties that make NP difficult to study, its essential nature and multitasking functionality, also provide ample opportunity for drug development. Drugs targeting NP function have been reported and show promising results (46). Our results demonstrate that NP-mediated import is a critical function which is highly sensitive to disruption and should therefore continue to be a target for antiviral drug development.

MATERIALS AND METHODS

Antibodies.

Anti-hemagglutinin (HA) tag antibody was purchased from Sigma (catalog number H3663), anti-NP was purchased from Thermo Scientific (catalog number PA5-32242), anti-NP-fluorescein isothiocyanate was purchased from Abcam (catalog number ab20921), anti-NS1 was purchased from Thermo Scientific (catalog number PA5-32243), antitubulin was purchased from Abcam (catalog number ab28439), anti-M1 was purchased from Abcam (catalog number ab20910), anti-influenza virus HA was purchased from Pierce (catalog number MA1-22684), anti-GFP was purchased from Sigma (catalog number G1544), anti-neuraminidase (NA; antibody 5H3) was obtained and purified at the Icahn School of Medicine at Mount Sinai, anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was purchased from Thermo Scientific (catalog number MA5-12738), and anti-FLAG was purchased from Sigma (catalog number F3165). Protein G-agarose was purchased from Roche Diagnostics. Anti-Sindbis virus ascitic fluid was purchased from ATCC (catalog number VR-1284AF).

Virus.

Influenza strain A/Puerto Rico/8/1934 (PR8), NS1-deficient A/Puerto Rico/1934 (dNS1) (47), and PR8-NS1-GFP were all kind gifts from the Adolfo Garcia-Sastre lab, and the viruses were propagated and titers were determined using MDCK cells. PR8 (NP2-NLS) was generated by passaging PR8 through NP2-NLS-expressing cells. NDV-GFP was grown in eggs, and titers were determined on MDCK cells. Sindbis virus was grown in BHK cells. Infections with NDV-GFP and Sindbis virus were performed similarly to those with IAV, except that they were in the absence of trypsin. UV-inactivated virus was prepared by exposure to UV irradiation at a distance of 6 in. for 10 min.

Infections were performed as follows, unless otherwise noted: cells were washed twice in PBS(+) (phosphate-buffered saline [PBS] containing MgCl₂ and CaCl₂ at 100 mg/liter) and incubated with virus in PBS(+) containing 0.2% bovine serum albumin (BSA) and 1 μl/μg of tosylsulfonyl phenylalanyl chloromethyl ketone-treated trypsin. After 1 h at room temperature, the inoculum was removed and cells were washed twice with PBS(+) prior to addition of Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum albumin (FBS). Cells were then placed at 37°C with 5% CO₂. For infections performed in the presence of cycloheximide, cells were preincubated for 30 min in DMEM supplemented with 10% FBS with inhibitor and maintained for the remainder of the experiment.

Cells.

MDCK (canine), BHK (hamster), HeLa (human), and 293T (human) cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and incubated at 37°C. VHH-HA constructs were stably integrated into MDCK cells by cotransfection with a plasmid that confers Zeocin resistance. After cotransfection, the cells were maintained in the presence of Zeocin to select VHH-expressing cells. Clonal isolates for each VHH were obtained using cloning cylinders and subsequently expanded for use in the experiments. VHH sequences can be obtained from the original publication (36). The NLS sequences used for NP2 and their KPNA affinity fusions (based on those described in references 37 and 48) are as follows: for NLS-I, MASQGTKRSYEQM (1.7 μM); for NLS^hi, PKKKRKV (9 nM); and for NLS^lo, PAAKRVALD (650 nM).

Immune fluorescence microscopy.

Cells were fixed for 20 min with PBS containing 4% (vol/vol) p-formaldehyde, washed 3 times with PBS, and incubated for 5 min in PBS containing 0.5% (vol/vol) NP-40 to permeabilize the cells. Samples were blocked using PBS containing 1% (vol/vol) cold fish gelatin. Preparations were mounted using Prolong Gold antifade reagent (Life Technologies). Images were obtained using a Zeiss Axioplan 2 with Apotome microscope and treated using ImageJ software.

Karyopherin immune precipitation assay.

A 293T cell lysate containing FLAG-KPNA6 was generated by transfection and lysing the cells at 24 h posttransfection (hpt). NP2- or HA68-saturated vRNPs were generated by incubating each cell type with IAV at a multiplicity of infection (MOI) of 1,000 on ice for 1 h. Infection was then triggered using prewarmed fusion buffer (DMEM, 2% BSA, MES [morpholineethanesulfonic acid] buffer, pH ∼5.5). Samples were immediately lysed in NP-40-containing lysis buffer. The lysates from 293T cells and HA68- or NP2-expressing cells were then coincubated for 1 h at 4°C prior to immune precipitation using antibody against NP.

VHH and NP quantification in viral particles.

Purified NP2-HA or NP was run on an SDS-PAGE gel alongside a defined amount of RNase A or BSA and stained using Coomassie brilliant blue. Densitometry analysis of each band was performed using ImageJ software. These samples were then used as standards for determining the levels of VHH or NP in cells or virus using immunoblot analysis.

qRT-PCR.

MDCK cells expressing, or not, VHHs were infected with PR8 or UV-inactivated PR8 (MOI, 10), and samples were collected over time. Total RNA was purified following the TRIzol extraction protocol (Life Technologies). cDNA was obtain using a SuperScript IV first-strand synthesis system (Life Technologies), and quantitative reverse transcription-PCR (qRT-PCR) was carried out using iQ SYBR green supermix (Bio-Rad) following the manufacturer’s instructions. The primers used for detection of NP were 5′-CTCAATATGAGTGCAGACCGTGCT-3′ (forward) and 5′-TAAGCGGTGGTACTGCCGG-3′ (reverse) for vRNA and 5′-CGATCGTGCCTTCCTTTG-3′ (forward) and 5′-TGCTGAGCTTGCTAGACC-3′ (reverse) for mRNA. For beta interferon (IFN-β) quantification we used 5′-GTCAGAGTGGAAATCCTAAG-3′ (forward) and 5′-ACAGCATCTGCTGGTTGAAG-3′ (reverse). All reactions were performed in triplicate, and actin was used as a control for normalization. qRT-PCR profile reactions and analysis were performed as described previously (49, 50).

Transfected ratio experiment.

293T, MDCK, BHK, and HeLa cells were chosen based on their transfectability and their permissiveness to IAV infection. Plasmids were mixed in the stated ratios, and DNA was transfected using the Lipofectamine 2000 reagent (Thermo Fisher) following the manufacturer’s recommendation. At 24 hpt, cells were challenged with virus. At 8 hpi, samples were then imaged by live fluorescence microscopy and/or lysed for immunoblot analysis.

Antiviral state on infected cells.

MDCK cells expressing NP2 or HA68 VHH were infected with UV-inactivated IAV (MOI, 10). At 24 hpi, the medium was removed and the cells were infected with NDV-GFP (MOI, 2). NDV-GFP replication was analyzed 16 h later via microscopy using an Olympus IX70 microscope. Fluorescent images were treated using ImageJ software.

Affinity assay for NP2/NP2-NLS.

IAV from allantoic fluid was lysed in NP-40-containing buffer. Then, decreasing amounts of lysed virus were UV cross-linked onto nylon strips. After blocking, the strips were incubated at 4°C overnight with lysate from 293T cells transfected with NP2, NP2-NLS, or empty plasmid. Strips were probed with anti-HA antibody and visualized with horseradish peroxidase-linked secondary antibody. Densitometry analysis was performed using ImageJ software.