Global Interactomics Connect Nuclear Mitotic Apparatus Protein NUMA1 to Influenza Virus Maturation

Influenza A virus (IAV) infections remain a major human health threat. IAV has enormous genetic plasticity and can rapidly escape virus-targeted anti-viral strategies. Thus, there is increasing interest to identify host proteins and processes the virus requires for replication and maturation. The IAV non-structural protein 1 (NS1) is a critical multifunctional protein that is expressed to high levels in infected cells. Host proteins that interact with NS1 may serve as ideal targets for attenuating IAV replication. We previously developed and characterized broadly cross-reactive anti-NS1 monoclonal antibodies. For the current study, we used these mAbs to co-immunoprecipitate native IAV NS1 and interacting host proteins; 183 proteins were consistently identified in this NS1 interactome study, 124 of which have not been previously reported. RNAi screens identified 11 NS1-interacting host factors as vital for IAV replication. Knocking down one of these, nuclear mitotic apparatus protein 1 (NUMA1), dramatically reduced IAV replication. IAV genomic transcription and translation were not inhibited but transport of viral structural proteins to the cell membrane was hindered during maturation steps in NUMA1 knockdown (KD) cells.


Introduction
Influenza A virus (IAV) remains a significant pathogen that causes substantial amounts of contagious respiratory disease in humans. Recent human infections with avian influenza viruses such as H5N1 and H7N9 subtypes emphasize the ongoing threat of this virus to cause future epidemics and pandemics. Two classes of anti-influenza drugs, viral M2 and NA inhibitors, are available; however, emergence of drug-resistant IAVs is becoming a serious concern [1][2][3][4]. In the future, influenza virus may lose sensitivity to all available drugs due to its genetic plasticity caused by the segmented nature

Co-Immunoprecipitation (Co-IP)
Cytoplasmic and nuclear lysates were initially pre-cleared with non-coupled protein G Dynabeads (Invitrogen, Waltham, MA, USA) for 90 min at 4 • C. The pre-cleared lysates were clarified at 10,000× g for 7 min. Dynabeads were washed 3× with TBST (Tris-buffered saline supplemented with 0.05% Tween 20) and a mixture of anti-NS1 mAbs 3F5, 5F4 and 4E10, which recognize different NS1 epitopes [42], was added to the beads. The mAbs and beads were incubated at room temperature for 90 min in a rotator to allow Ab-bead binding. Monoclonal α-Emprin (IgG2a), monoclonal α-SYN (IgG2b) and monoclonal α-HSA (IgG1) antibodies (gift from Dr. Wilkins, Manitoba Centre for Proteomics and Systems Biology) also were bound to Dynabeads to serve as isotype controls. Ab-coupled beads were washed 4× with TBST to remove unbound mAbs and mixed with the pre-cleared cellular fractions in a rotator overnight at 4 • C. The unbound fractions were discarded and beads were washed 4× and resuspended in TBST. The washed and resuspended bead-Ab-antigen complex represented immunoprecipitated (IP) products. Co-IPs were also performed after coupling anti-NUMA1 (Bethyl Laboratory, A301-510A), anti-PRPF19 (Bethyl Laboratory, A300-101A) and anti-UTP6 (Thermo Fisher, PA5-21716) antibodies to Dynabeads.

Processing of IP Product for Western Blot Analysis and Mass Spectrometry
The IP products and beads were washed 2× with RIPA buffer, 1× with ammonium bicarbonate buffer supplemented with 0.1% NP40 and resuspended in ammonium bicarbonate buffer. 10% of the resuspended bead mixtures were dissolved in sodium dodecyl sulfate (SDS) running buffer and resolved in 4-12% gradient Novex NuPAGE Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE) Gels (Invitrogen, Waltham, MA, USA) for Western blot analysis and 90% of the resuspended beads were saved at −80 • C for subsequent mass spectrometry (MS) analysis. For MS analysis, the immunoprecipitated beads were digested overnight with 1 µg of trypsin in 100 mM ammonium bicarbonate solution at 37 • C. After tryptic digestion, equal volumes of trifluoroacetic acid (TFA)/Acetonitrile (ACN) (100% ACN & 1% TFA) were added to the digested IP products and vortexed 10-15 min. Digested peptides were separated from beads by centrifugation at 17,000× g for 5 min and were dried in a Savant SpeedVac vacuum dryer. Dried peptides were resuspended in 50 µL of 0.5% TFA and desalted with C18 ziptips. Eluted peptides were analyzed in an AB SCIEX (Concord, ON, Canada) Triple TOF 5600 mass spectrometer. Raw MS data were analyzed with Protein Pilot TM 3.0 (ABSciex, Concord, ON, Canada). The proteins were identified based on cumulative peptide numbers and scores (cut-offs of a minimum of 2 peptides with unused score ≥2.0).

Transfection of Cells by siRNA
For initial screening, a reverse transfection format (RTF) SMART pool siRNA library was designed targeting 107 genes and purchased in 96-well format from Dharmacon (Lafayette, CO, USA). Reverse transfection of this siRNA array was carried out according to manufacturer's protocol. In brief, sets of siRNA plates were rehydrated with transfection reagent/DharmaFECT-1 cell culture media and incubated for 60 min at room temperature. 4 × 10 3 A549 cells were added to each well and incubated at 37 • C for 48 h to allow knockdown. Cell viability was determined in one set of plates with the Roche cell proliferation reagent WST1 and another set was infected with IAV PR8 at MOI 0.05 for 43 h. Cell viabilities of PR8-infected knockdown cells were also measured at the end of infection.
For specific gene knockdown, larger numbers of A549 cells were transfected with individual ON-Target plus siRNAs (Dharmacon) targeting NUMA1 gene or with scrambled non-targeting control siRNA (N-Si) according to the manufacturer's protocol. Stock siRNAs and DharmaFECT ® -1 transfection reagent (Dharmacon) were diluted separately with Opti-MEM ® I reduced serum medium (Life Technology, Waltham, MA, USA). A549 cells were transfected with siRNA every 24 h for 48 h. At 48 h post-transfection, cells were infected with IAV PR8 at MOI 0.05 or 3. Knockdown efficiencies of individual genes were checked by Western blot and IP using specific antibodies to relevant proteins.

Structured Illumination Microscopy (SIM)
SIM was conducted according to the procedure described by Righolt et al. [45]. A549 cells were grown overnight on 18 × 18 mm high performance cover glasses with restricted thickness-related tolerance, D = 0.17 mm ± 0.005 mm (Zeiss, cat # 474030-9000-000). The next day, cells were treated with individual siRNA. After 48 h knockdown, cells were infected at MOI 3, incubated 20 h, and fixed with 3.7% formaldehyde for 20 min. Fixed cells were washed 4× with PBS, permeabilized with 0.2% Triton-X 100 for 10 mins, washed again 4× with PBS, blocked with 3% BSA/PBS for 1 hour and probed with rabbit polyclonal α-M2 or mouse monoclonal α-M1 antibodies diluted in 3% BSA/PBS for 6 h at 4 • C. Cover glasses were then washed 4× with PBT (PBS, 3% BSA and 0.05% Tween20), treated with Alexa Fluor 488-conjugated goat α-rabbit or Alexa Fluor 546-conjugated goat α-mouse secondary antibodies diluted in 1% BSA/PBS (1:250) for 1 h, washed 4× with PBT, treated with DAPI for 5 min and mounted with Vectashield or Prolong gold (Life Technology) mounting medium. The cover glasses were sealed with nail polish. Images were taken with a Zeiss ELYRA PS1 in SIM mode equipped with a Plan-Apochromat 63 ×/1.40 oil immersion objective and 2D and 3D images were processed as described [45].

Electron Microscopy (EM)
A549 cells were grown in P100 cell culture plates overnight and treated with individual siRNAs. At 48 h post-transfection, A549 cells were infected with PR8 at MOI 3. At 20 hpi, cells were harvested, and washed 3× with ice-cold PBS. Cell pellets were resuspended in EM grade Karnovsky fixative (2.5% glutaraldehyde and 4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.2) and sent to the Histology Lab, Department of Human Anatomy and Cell Science, University of Manitoba for further processing. Images were obtained on a Philips CM10 Electron Microscope.

Real-Time Polymerase Chain Reaction (PCR)
PR8-infected A549 cells were harvested, washed 3× with cold PBS and total cellular mRNA was extracted with RNeasy Mini Kit (Qiagen, Venlo, Netherlands) according to the manufacturer's protocol; 250 ng of purified mRNA was used to synthesize cDNA with the Go Script TM Reverse Transcription System kit (Promega, Madison, WI, USA). Real time polymerase chain reaction (PCR) was performed using Luminaries Color HiGreen High ROX qPCR kit (Thermo Fisher). According to the manufacture's protocol, PCR master mix (10 µL) consisted of: 5 µL 2× Luminaries Color HiGreen High ROX qPCR master mix, 4.4 µL (100 ng) template cDNA and 0.3 µL each of 10 µM forward and reverse primers (Supplementary Table S2). PCR reactions were run in triplicate on an Applied Biosystems 7300 Real-Time PCR System. The program of cycle condition was 50 • C for 2 min, 95 • C for 2 min, and 50 cycles of (95 • C for 15 s and 60 • C for 30 s). The Ct values were normalized to 18S rRNA control and compared to non-targeting siRNA controls.

Bioinformatics and Statistical Analyses
Lists of IAV-NS1 interacting host proteins were generated from Protein Pilot analysis. The gene symbol and Uniprot IDs of all proteins were uploaded into the Database for Annotation, Visualization and Integrated Discovery (DAVID) [46] for functional tool analysis. NS1-interacting proteins were also uploaded into Consensus Path Database (CPDB) [47] for pathway and enrichment analysis. Groups of proteins identified in individual pathways in DAVID were uploaded into the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) [48] to visualize the protein-protein interactions. Interaction among viral proteins and host factors were also analyzed by using VirHostNet 2.0 [31] database. All statistical analyses were calculated in Microsoft-Excel and SigmaPlot ® software. p-values were determined using Student's t-test.

Co-Immunoprecipitation of Influenza A Virus (IAV) Non-Structural Protein 1 (NS1) from Infected Cells
We previously generated and characterized a panel of nine different broadly cross-reactive anti-NS1 monoclonal antibodies (mAbs) that detect five different epitopes on native and denatured forms of IAV NS1 [42]. For this NS1 interactome study, a mixture of anti-NS1 mAbs 3F5, 4E10 and 7D11, which recognize three of these different epitopes [42], were used, as described in Materials and Methods, to increase potential molecular interaction coverage. The anti-NS1 mAb mixture successfully pulled down NS1 protein (~26 kDa) from both PR8-infected cytosol and nuclei ( Figure 1A). The 6 hpi nuclear input NS1 band intensity was higher than the cytosol input ( Figure 1A), which correlated with previous observations of NS1 subcellular distribution [42]. No non-specific NS1 binding was seen in PR8-infected cytosolic or nuclear IPs with the isotype controls, nor in mock-infected control samples ( Figure 1A).

Identifying NS1 Interacting Host Proteins by Mass Spectrometry (MS)
Once specific NS1 IP was confirmed, 90% of the IP products were processed and analyzed by MS for protein identifications. Three different biological replicates were performed and the last replicate was technically examined twice for a total of four replicates. Positive identification of NS1-interacting proteins was only considered when each protein was identified by at least 2 non-redundant peptides at unused scores ≥2 (p < 0.01). All protein IDs detected in the PR8-infected IP probed with isotype controls, and in mock-infected IP probed with anti-NS1 mAb mixture, were considered non-specific background binding and were subtracted from proteins in the PR8-infected IP probed with anti-NS1. 233, 138, and 324 NS1 interacting host factors were identified in the cytosolic and nuclear fractions of biological replicates 1, 2, and 3, respectively ( Figure 1B). Overall, 183 unique NS1 interacting host factors were identified in at least two different biological replicates (Table 1 and Figure 1B). According to a recent analysis of the VirHostNet 2.0 [31] database, 59 of these 183 host factors (32%) had been previously reported to interact with IAV NS1 protein (indicated by # in Table 1), whereas 124 host proteins represent potentially novel NS1 interacting partners. PR8 viral matrix protein 1 (M1) and nucleoprotein (NP) were also detected in the NS1 IP with high peptide and high unused scores, confirming numerous other reports and their interaction with NS1 in the VirHostNet 2.0 database.

Bioinformatic Analysis of NS1 Interacting Proteins
The multifunctional IAV-NS1 protein is expected to interact with a wide variety of host factors to serve its multiple functions. Therefore, categorizing the additional novel NS1-interacting host factors identified in our study may identify additional potential pathways and functions where NS1 plays vital roles during viral replication. DAVID and Panther analyses of the 183 genes showed various biological and molecular classes (Figure 2A,B). The top enriched biological processes were nucleoside, nucleotide and nucleic acid metabolism, mRNA splicing, pre-mRNA processing and protein biosynthesis. Similar molecular functions such as nucleic acid binding, mRNA splicing factor, mRNA processing factor and ribosomal protein were also enriched. The NS1-interacting host factors were involved in eight different pathways; gene expressions, 3 -UTR-mediated translational regulation, processing of capped intron-containing pre-mRNA, spliceosome, Influenza infection and metabolism of proteins ( Figure 2C). Gene Ontology analyses of the 183 NS1 interacting host factors also showed diverse biological processes

Identifying NS1 Interacting Host Proteins by Mass Spectrometry (MS)
Once specific NS1 IP was confirmed, 90% of the IP products were processed and analyzed by MS for protein identifications. Three different biological replicates were performed and the last replicate was technically examined twice for a total of four replicates. Positive identification of NS1interacting proteins was only considered when each protein was identified by at least 2 nonredundant peptides at unused scores ≥2 (p < 0.01). All protein IDs detected in the PR8-infected IP probed with isotype controls, and in mock-infected IP probed with anti-NS1 mAb mixture, were considered non-specific background binding and were subtracted from proteins in the PR8-infected IP probed with anti-NS1. 233, 138, and 324 NS1 interacting host factors were identified in the cytosolic  Figure S1E). Tight interactions among the proteins of different pathways, such as 3 -UTR-mediated translational regulation, ribosome and metabolism indicate the strong associations of these pathways with NS1 during Influenza virus replication. * Represents the -log 10 probability of a false positive (for example, Unused score of 2.0 corresponds to p = 0.01; Unused score of 3 corresponds to p = 0.001); † Genes depicted in bold and slightly larger font were knocked down using gene-specific siRNA and the knockdown effects on cell viability and virus production determined ( Figure 3)

Assessing the Necessity of NS1-Interacting Proteins for Viral Replication
In order to test the roles of these potentially novel NS1-interacting host factors, we designed a high-throughput, 96-well-based custom siRNA array; 107 proteins were targeted based on their novelty, functions and high MS scores. These 107 proteins were identified as NS1 interacting partners in 2 or more different biological replicates with at least 2 peptides and p < 0.05 (unused score ≥2.0). Cell viabilities were determined after 48 h of knockdown (KD) and most remained >80% viable compared to non-targeting control (N-Si) (Figure 3, left). The lowest viability (62.8%) was seen in SNRNP200 siRNA treatment. The KD cells were infected with IAV PR8 at MOI 0.05 and the supernatants were harvested at 43 hpi to detect infectious viral yields to determine how efficiently the virus replicated. Cell viabilities after 48 h KD and 43 h infection were also measured (Figure 3, middle). Infected cell cultures remained >70% viable compared to N-Si. In 34 KD cases, PR8 infection reduced the cell viability to 69.6-26.9%. The lowest cell viability was seen in PR8-infected NOL6 KD cells. In some KD cases, cell viabilities were highly increased after viral infection indicating that the KD may protect the cell from cytopathic effect (Figure 3, middle). Knocking down 11 genes significantly reduced the infectious virus production to between 2.6-30% of the N-Si controls (Figure 3, right). The lowest IAV yields were generated in NUMA1 KD cells. Among these 11 KD gene candidates, ILF2, HNRNPUL2 and PRPF19 were previously detected as NS1-interacting factors in the VirHostNet 2.0 database. NUMA1, RBM28, RBM34, RPF1, SYNCRIP, SF3B3, UTP6 and SNRNP200 were novel discoveries, and the role of PRPF19 was recently reported [40] while our study was underway. These 11 gene candidates play roles in various biological processes, molecular functions and pathways (Supplementary Table S1). Some important biological processes and molecular functions include cellular macromolecular complex assembly and subunit organization; chromatin assembly, packaging and remodeling; nucleoside, nucleotide and nucleic acid metabolism; mRNA splicing; mRNA processing; nucleic acid binding; structural molecule activity; and ribonucleoprotein complex biogenesis. Some important pathways include spliceosome, gene expression and processing of capped intron-containing pre-mRNA.

Validation of NS1-Host Factor Interaction by Reciprocal Immunoprecipitation (IP)
Host factor NUMA1 was selected for reciprocal IP, based on its ability to significantly reduce viral replication (Figure 3, right), its functional properties such as biological process, molecular functions and pathways and the availability of antibodies. A549 cells were infected with IAV PR8 at MOI 5 and harvested at 24 hpi. Cellular proteins were extracted from harvested cells and analyzed by co-IP using Protein G Dynabeads coupled to anti-NUMA1 antibody. IP products were analyzed by Western blot analysis using both anti-NUMA1 and anti-NS1 Abs. No NUMA1 signal was detected in non-concentrated cell lysates but a strong NUMA1 signal was seen near the 225 kDa region in immunoprecipitated cell lysates ( Figure 4A, upper panel). NS1 was also detected in the input cell lysate and in the NUMA1 IP but not in isotype control, which confirmed the NS1-NUMA1 interaction ( Figure 4A, lower panel).

Role of Nuclear Mitotic Apparatus Protein 1 (NUMA1) in Influenza Virus Replication
Nuclear mitotic apparatus protein 1 (NUMA1) was detected in all four biological and technical NS1 interactome replicates (Table 1). In the initial siRNA screens, IAV production was significantly reduced compared to the N-Si in all independent studies (Figure 3, right), which indicated an important role for NUMA1 in IAV replication. To validate the siRNA array results, KD was performed in larger dishes of A549 cells with NUMA1 Smart Pool siRNA (Dharmacon, Lafayette, CO, USA). Optimization experiments indicated sufficient KD (NUMA protein expression reduced to~16%) when cells were treated twice with 25 nM siRNA 24 h apart and infected 48 h post-initial KD. Transfected A549 cells were~100% viable at 48 h post-transfection compared to untreated A549. Comparative IP of NUMA1and N-Si-KD cells, using identical concentrations of cell lysates, Protein G beads and anti-NUMA Ab, confirmed NUMA1 protein expression was reduced to~15% of the N-Si treated cells ( Figure 4B,C). A 1/40th dilution of input cell lysates confirmed cell lysate loading control ( Figure 4B) and equivalent band intensities of anti-NUMA Ab heavy chains confirmed equal Ab treatment ( Figure 4B). After optimization and KD validation, both NUMA1 KD and N-Si cells were infected with IAV PR8 at MOI 0.05 and incubated 43 h. Knocking down NUMA1 significantly reduced infectious PR8 production to~21% compared to the N-Si in four independent experiments ( Figure 4D), which validated the NUMA1 KD results from the siRNA array. Host factor NUMA1 was selected for reciprocal IP, based on its ability to significantly reduce viral replication (Figure 3, right), its functional properties such as biological process, molecular functions and pathways and the availability of antibodies. A549 cells were infected with IAV PR8 at MOI 5 and harvested at 24 hpi. Cellular proteins were extracted from harvested cells and analyzed by co-IP using Protein G Dynabeads coupled to anti-NUMA1 antibody. IP products were analyzed by Western blot analysis using both anti-NUMA1 and anti-NS1 Abs. No NUMA1 signal was detected in nonconcentrated cell lysates but a strong NUMA1 signal was seen near the 225 kDa region in immunoprecipitated cell lysates ( Figure 4A, upper panel). NS1 was also detected in the input cell lysate and in the NUMA1 IP but not in isotype control, which confirmed the NS1-NUMA1 interaction ( Figure 4A, lower panel).   43 hpi were determined by plaque assay on canine kidney (MDCK) cells. All values were normalized to the corresponding non-silencing (N-Si) controls, which were set as 100%. Error bars represent standard error of the mean (SEM) from three independent replicates. *: p < 0.05; **: p < 0.005. The 11 genes, knockdown (KD) of which significantly reduced the infectious virus production to between 2.6-30% of the N-Si controls, are depicted as black bars, and cell viabilities after 91 h of knockdown, and the ratios of virus production to cell viability for these 11 genes are shown in Supplementary Figure S2. to the corresponding non-silencing (N-Si) controls, which were set as 100%. Error bars represent standard error of the mean (SEM) from three independent replicates. *: p < 0.05; **: p < 0.005. The 11 genes, knockdown (KD) of which significantly reduced the infectious virus production to between 2.6-30% of the N-Si controls, are depicted as black bars, and cell viabilities after 91 h of knockdown, and the ratios of virus production to cell viability for these 11 genes are shown in Supplementary Figure S2. antibody in Western blot. NUMA1 could not be detected in non-concentrated cell lysates. Cell lysates were prepared from P100 dishes of A549 cells, reacted with Dynabeads to which α-NUMA1 Abs had been coupled, and ½ of the reaction dissolved in SDS-PAGE sample buffer, resolved in 10% SDS-PAGE, proteins transferred to Immobilon-P PVDF membranes, and probed with α-NUMA1 antibody. Lower panel: Cell extracts prepared from P100 dishes of A549 cells infected with PR8 at MOI = 5 PFU/cell were probed for NS1 prior to immunoprecipitation (Input; 30 µg), or were immunoprecipitated with beads to which NUMA1 or an irrelevant isotype control antibody had been bound. Co-precipitated products were resolved by SDS-PAGE and blots were immunoprobed with α-NS1 antibody. (B) Confirmation of NUMA1 KD efficiency in A549 cells. Sets of P100 dishes were treated with 25 nM of non-silencing (N-Si) control or with NUMA1-specific siRNA twice, 24 h apart for a total of 48 h treatment. Cell extracts were prepared and a 1/40th dilution probed for β-actin to confirm equivalent starting amounts (middle panel). Extracts were then immunoprecipitated with α-NUMA1-Dynabeads. After washing, beads were dissolved in SDS-PAGE sample buffer, proteins antibody in Western blot. NUMA1 could not be detected in non-concentrated cell lysates. Cell lysates were prepared from P100 dishes of A549 cells, reacted with Dynabeads to which α-NUMA1 Abs had been coupled, and 1 2 of the reaction dissolved in SDS-PAGE sample buffer, resolved in 10% SDS-PAGE, proteins transferred to Immobilon-P PVDF membranes, and probed with α-NUMA1 antibody. Lower panel: Cell extracts prepared from P100 dishes of A549 cells infected with PR8 at MOI = 5 PFU/cell were probed for NS1 prior to immunoprecipitation (Input; 30 µg), or were immunoprecipitated with beads to which NUMA1 or an irrelevant isotype control antibody had been bound. Co-precipitated products were resolved by SDS-PAGE and blots were immunoprobed with α-NS1 antibody. (B) Confirmation of NUMA1 KD efficiency in A549 cells. Sets of P100 dishes were treated with 25 nM of non-silencing (N-Si) control or with NUMA1-specific siRNA twice, 24 h apart for a total of 48 h treatment. Cell extracts were prepared and a 1/40th dilution probed for β-actin to confirm equivalent starting amounts (middle panel). Extracts were then immunoprecipitated with α-NUMA1-Dynabeads. After washing, beads were dissolved in SDS-PAGE sample buffer, proteins resolved by SDS-PAGE, and immunoprobed for NUMA1 (upper panel)

NUMA1 Knockdown (KD) Does not Affect IAV Transcription and Translation
Since infectious virus production was consistently reduced in NUMA1 KD cells, we next tested specific steps in virus replication. NUMA1 is mainly a nuclear protein, but it is also found in the cytoplasm [49]. After infection, viral ribonucleoproteins (vRNPs) need to enter the nucleus for transcription. We initially checked whether the import of incoming vRNPs or the transcription process was inhibited in NUMA 1 KD cells by analyzing viral RNA production. NUMA1 KD and N-Si KD cells were infected with IAV-PR8 at MOI 3, infected cells were harvested at 18 hpi and total RNA was extracted. Synthesis of viral NP and NS1 RNAs were not reduced in NUMA1 KD cells ( Figure 4E). Thus, NUMA1 KD did not affect early vRNP entry into the nucleus and transcription of NS1 and NP. After transcription, viral mRNAs are transported to the cytoplasm and translated by host cell machinery. Whole cell lysates were extracted from PR8-infected NUMA1 KD and N-Si cells at 18 hpi and analyzed by Western blot. Early (NS1 and NP) and late (M2) viral protein translation was not reduced in the infected NUMA1 KD A549 cells compared to the N-Si cells ( Figure 4F). Therefore, NUMA1 KD did not impede the transport of viral mRNAs to the cytoplasm and the translation process.

NUMA1 KD Affects IAV Maturation
Although viral proteins were synthesized efficiently in NUMA1 KD A549 cells, infectious virus production was significantly reduced ( Figure 4D). To determine the generality of this observation, we then knocked down NUMA1 in human bronchial epithelial cells (HBEC-3KT; ATCC cat # CRL-4051; "HBEC"), and in additional A549 cells, and determined the capacity of additional H1N1 and H3N2 IAV strains to replicate in these KD cells. All tested IAV clones were reduced 55-99% in the NUMA1 KD cells compared to non-silencing control cells ( Figure 4E). These data suggest that NUMA1 depletion might affect infectious virus production. To examine this possibility, NUMA1 KD and N-Si cells were infected with PR8 at MOI 0.05 and supernatants were harvested at 43 hpi. Viruses were pelleted from the N-Si and NUMA1 KD supernatants by ultracentrifugation and resuspended in small equal volumes of PBS. Titration of the pre-and post-concentrated viruses confirmed that viral titers from the NUMA1 KD cells were significantly reduced to~20% of titers from N-Si cells ( Figure 4H). To differentiate between the possibilities that equal numbers of particles were released from the NUMA1 KD cells but they were less infectious than particles released from N-Si-treated cells, or that fewer particles were matured from NUMA1 KD cells, equal volumes of purified and resuspended viruses generated from the N-Si and NUMA1 KD cells were analyzed by Western blot to detect viral structural proteins NP, M1 and M2. The amounts of all tested proteins from NUMA1 KD cells were reduced to 40%, 26% and 16%, respectively compared to viral proteins produced from the N-Si (Figure 4I,J). Therefore, the average structural protein reduction was similar to the infectivity reduction ( Figure 4D,E,I,J), suggesting no significant difference in particle-to-PFU ratios, but that NUMA1 is involved in viral maturation.
As an alternate method to assess differences in virus maturation, we infected non-silencing and NUMA1 KD cells with PR8 at MOI = 3 and examined infected cells at 20 hpi by electron microscopy. Numerous small spherical structures, consistent with the~100 nm IAV virion size, were clearly visible outside but near the cell membrane, presumably after budding ( Figure 5, middle, arrows) in N-Si infected cells. In contrast, production of these virus-sized particles was greatly reduced in infected NUMA1 KD cells ( Figure 5, right), in agreement with previous observations of reduced extracellular viral protein and infectious virus from these NUMA1 KD cells.

Localization of NUMA1 and Viral NS1, M1 and M2 Proteins in NUMA1-Deficient Cells
To further assess NUMA1 KD and to determine localization of various viral non-structural and structural proteins under our control and KD conditions, we infected A549 cells with PR8 at MOI = 3 and examined cells for various protein markers. Viral NS1 protein was found in both nuclei and dispersed throughout the cytoplasm in most infected cells, but was primarily localized to perinuclear regions in the NUMA1 KD cells ( Figure 6A). NUMA1 was found predominantly in nuclei, co-localizing with DAPI, except in the NUMA1 KD cells where the signal intensity was clearly reduced and the protein occupied only some of a few nuclei. As expected from the initial identification of NUMA1 (Table 1) and reciprocal immunoprecipitations (Figure 4) indicating NS1/NUMA1 interaction, NS1 and NUMA1 co-localized in the nuclei as indicated by the pale blue/white color in wild-type and N-Si KD cells. However, this interaction was greatly reduced in the NUMA1 KD cells as indicated by the few pale blue/white patches ( Figure 6A, bottom right). Although viral proteins were synthesized efficiently in NUMA1 KD A549 cells, infectious virus production was significantly reduced ( Figure 4D). To determine the generality of this observation, we then knocked down NUMA1 in human bronchial epithelial cells (HBEC-3KT; ATCC cat # CRL-4051; "HBEC"), and in additional A549 cells, and determined the capacity of additional H1N1 and H3N2 IAV strains to replicate in these KD cells. All tested IAV clones were reduced 55-99% in the NUMA1 KD cells compared to non-silencing control cells ( Figure 4E). These data suggest that NUMA1 depletion might affect infectious virus production. To examine this possibility, NUMA1 KD and N-Si cells were infected with PR8 at MOI 0.05 and supernatants were harvested at 43 hpi. Viruses were pelleted from the N-Si and NUMA1 KD supernatants by ultracentrifugation and resuspended in small equal volumes of PBS. Titration of the pre-and post-concentrated viruses confirmed that viral titers from the NUMA1 KD cells were significantly reduced to ~20% of titers from N-Si cells ( Figure 4H). To differentiate between the possibilities that equal numbers of particles were released from the NUMA1 KD cells but they were less infectious than particles released from N-Si-treated cells, or that fewer particles were matured from NUMA1 KD cells, equal volumes of purified and resuspended viruses generated from the N-Si and NUMA1 KD cells were analyzed by Western blot to detect viral structural proteins NP, M1 and M2. The amounts of all tested proteins from NUMA1 KD cells were reduced to 40%, 26% and 16%, respectively compared to viral proteins produced from the N-Si ( Figure 4I,J). Therefore, the average structural protein reduction was similar to the infectivity reduction ( Figure 4D,E,I and J), suggesting no significant difference in particle-to-PFU ratios, but that NUMA1 is involved in viral maturation.
As an alternate method to assess differences in virus maturation, we infected non-silencing and NUMA1 KD cells with PR8 at MOI = 3 and examined infected cells at 20 hpi by electron microscopy. Numerous small spherical structures, consistent with the ~100 nm IAV virion size, were clearly visible outside but near the cell membrane, presumably after budding ( Figure 5, middle, arrows) in N-Si infected cells. In contrast, production of these virus-sized particles was greatly reduced in infected NUMA1 KD cells ( Figure 5, right), in agreement with previous observations of reduced extracellular viral protein and infectious virus from these NUMA1 KD cells.   The IAV M1 protein plays important roles in assembly of progeny virions. It helps in transporting vRNPs from the nucleus to the budding sites [50,51]. Therefore, we determined whether virus production was reduced due to the relocation of viral M1 in infected NUMA1 KD cells. To investigate this, NUMA1 KD and N-Si KD A549 cells were infected with PR8 at MOI 3, fixed at 20 hpi, treated with anti-M1 Ab and analyzed by super resolution structured illumination microscopy (SIM). M1 proteins accumulated and formed clusters adjacent to the nucleus in infected-NUMA1 KD A549 cells ( Figure 6B (upper right), white arrow). However, similar M1 clustering was not observed in PR8-infected N-Si cells, where M1 was more evenly distributed in extra-nuclear regions ( Figure 6B (upper middle)). Thus, the SIM results suggested that M1 protein trafficking was interrupted in NUMA1 KD cells and NUMA1 may be involved in M1 trafficking. M2 is another IAV protein that plays important roles in viral assembly and budding steps near the cytoplasmic membrane [50,52]. SIM was also conducted with PR8-infected NUMA1 KD and N-Si cells. In NUMA1 depleted cells, most of the M2 proteins accumulated near the plasma membrane ( Figure 6B (lower right), white arrow). However, M2 proteins were more evenly distributed in extra-nuclear regions in PR8-infected N-Si cells ( Figure 6B (lower middle)).

Discussion
The goal of this study was to detect novel host factors that interact with native IAV NS1 during natural infection and viral replication. To target a wide range of NS1-interacting host factors during viral replication, we selected early and late time points. We had previously detected NS1 as early as 5-6 hpi [42]. Therefore, we selected 6 and 24 hpi as early and late time points for our co-IP experiments. 183 NS1-interacting host proteins were detected in at least two different biological replicates and most of these are involved in different cellular biological processes, molecular functions and pathways.
IAV utilizes the host cell system for its gene expression. To initiate IAV transcription, viral RNA-dependent RNA polymerase (RdRp) executes cap-snatching from host pre-mRNA and NS1 may play roles during this step [20,53]. Influenza viruses use the host's splicing machinery to produce M1, M2, NS1 and NS2/NEP proteins [54,55]. NS1 interacts with spliceosomal subunits U2 and U6 during viral replication, potentially favouring their splicing [56,57]. The 5' UTR of viral mRNAs can mediate selective viral mRNA translation over cellular mRNA translation [20,58,59]. NS1 plays important roles in initiating viral mRNA translation without affecting host mRNAs [60]. NS1 interacts with the 5' UTR of viral mRNAs and with translation initiation factors PABP1 and eIF4GI, which enhance viral mRNA translation [20,58,61,62]. NS1 also interacts with factors involved in mRNA export, such as NXF1/Tap, and blocks host mRNA export [63]. In addition, NS1 interacts with viral RNP and regulates IAV replication [62,64]. Numerous NS1-interacting host factors we identified were enriched in gene expression, splicing, protein metabolism, nucleic acid binding, translational regulation and mRNA processing (Figure 2A-E). IAV may use these host factors to favour viral replication. We also identified large numbers of RNA-binding proteins. NS1 contains two functional domains: an N-terminal RNA-binding domain and C-terminal effector domain. The RNA binding and effector domains interact with different cellular RNAs and host factors, respectively [20,26,65], potentially accounting for the large numbers of RNA binding proteins we identified.
Many of our NS1-interacting host proteins interact with each other according to STRING analysis ( Figure 2F-H). Thus, some of the newly identified proteins may not directly interact with NS1 but may have been pulled down in the IPs by secondary interactions. However, we validated the interactions of NUMA1 with IAV-NS1 by reciprocal co-IP ( Figure 4A), identifying NUMA1 as one of several proteins not currently known to interact with other identified proteins (Supplementary Figure S1, red circle) according to STRING, suggesting that the NUMA1-NS1 interaction is a direct one.
We knocked down 107 of the most promising NS1-interacting host proteins to evaluate their impact on IAV replication. Several groups have used viral gene luciferase assays or viral protein immunostaining in 96 well formats to detect viral replication, but these are surrogates for RNA production or protein, respectively, and do not address progeny viral infectivity directly. We measured viral titers to assess virus replication and infectious virus production directly from infected-KD cells. Knocking down 11 NS1-interacting host proteins significantly reduced IAV replication at least 3-fold compared to the control cells in three different experimental replicates. Although some genes' KD, such as DHX30, DDX54, ELAVL1, FXR1, H2AFJ, NOP56 and RRS1 increased viral titers ( Figure 3C), we focused this study on the 11 gene candidates that significantly reduced virus titer. Most cells were ≥80% viable after 48 h KD; however, many infected-KD cells were less viable 43 h later ( Figure 3B). Influenza causes virus-induced cell death during viral replication [66][67][68]. To differentiate between virus-mediated versus prolonged KD-mediated cell death, we examined cell viabilities of non-infected cells at 91 h post KD (Supplementary Figure S2A). Viability of these cells was uniformly higher than for the KD-infected cells, indicating loss of cell viability at prolonged time was primarily caused by viral-induced cell death. Lower viral titer in many of the KD cells also corresponded with lower cell viability post-infection. For example, ILF2 KD resulted in~50% cell viability at the end of infection. Therefore, we compared viral yields to post-infection cell viabilities for the N-Si and all 11 KD A549 cells. Even when accounting for reduced cell viability, viral yields were substantially lower in many of the KD cells (Supplementary Figure S2B), including NUMA1 KD, suggesting the lower titers were caused by the KD, not by the lower cell viability. Among these 11 candidates, ILF2, HNRNPUL2, and PRPF19 were previously identified as NS1 interacting proteins in the VirHostNet 2.0 database. The PRPF19 deficient cells were demonstrated to reduce influenza virus production in a recent study [40] while our study was ongoing. NUMA1, RAI14, RBM28, RBM34, RPF1, SF3B3, SNRNP200, and UTP6 proteins represent newly discovered potential NS1 interacting proteins, knocking down of which significantly reduces infectious IAV production. These novel NS1-interacting host factors may have roles in controlling viral/host mRNA maturation, splicing and gene expression to favour viral replication. NUMA1, also known as NuMA, is an important structural component of both the nucleus and spindle poles, and it plays essential roles during the assembly and maintenance of the mitotic spindle in the cell cycle (reviewed in [69]). The NuMA protein is solubilized and modified extensively, including phosphorylation at unknown sites, during Herpes Simplex Virus (HSV) infection [70]. Knocking down NuMA expression in Hep-2 cells also decreased HSV production. NUMA1 has strong connections with microtubules. During mitosis, NUMA1 plays an important role in connecting the microtubules to the spindle poles [69,71] Tubulin (a monomer of microtubules) directly binds with the C-terminus of NUMA1 [72]. Other studies suggested that NUMA1 binds with microtubules in association with dynein, a microtubule motor protein [73,74]. Microtubules and dynein contribute in transporting newly synthesized proteins through the Golgi complex in the exocytic pathway [75,76] to the cell surface.
NUMA1 KD significantly reduced replication of several IAV strains in both A549 and HBEC cells ( Figure 4E), which has not been previously reported. We found no significant inhibition of PR8 viral mRNA synthesis in NUMA1 deficient cells compared to the control cells ( Figure 4F,G). This suggests that viral entry into the cell, import of incoming vRNPs to the nucleus and viral transcription steps do not depend on NUMA1 proteins. NUMA1 deficient cells and N-Si control cells synthesized similar amounts of IAV structural and non-structural proteins, which indicates that viral mRNA transport to the cytoplasm and translation were not affected by NUMA1 KD (Figure 4H). The levels of IAV structural proteins released into the supernatant were significantly reduced in NUMA1 KD cells compared to the N-Si control cells ( Figure 4I,J). Thus, viral transcription and translation steps are not inhibited, but viral maturation steps are inhibited within NUMA1 deficient cells.
During viral maturation, newly produced vRNPs need to be exported to the cytoplasm, and then transported to the cell membrane for progeny virus assembly and budding. Influenza M1 directly interacts with vRNPs and plays vital role in exporting these vRNPs into the cytoplasm [51,77]. The viral envelope proteins, HA, NA and M2, are independently transported to the cell membrane, where eventual interactions among the viral proteins and vRNPs lead to progeny virion assembly and budding. M1 acts as a linker between vRNPs and envelope proteins (reviewed in [51,78]). In addition, our previous study showed NS1 proteins were detected in the nucleus at early times of infection and later spread into the cytoplasm [42]. HA, NA and M2 use exocytic pathways to reach the cell membrane for assembly and budding through the trans Golgi network [51,79]. After the export of vRNPs-M1 complexes from the nucleus to the cytoplasm, it is suggested that the vRNPs-M1 complexes can be transported to the cell membrane by piggy-backing on the HA and NA cytoplasmic domains or via cytoskeleton elements [51,52,[80][81][82]. Our SIM images of PR8-infected NUMA1 KD cells showed IAV M1 proteins formed clusters within the cytoplasmic regions, in close proximity to the nucleus, whereas no clustering was seen within the N-Si cells ( Figure 6B). In PR8-infected NUMA1 KD cells, NP levels were also higher than in the PR8-infected N-Si cells ( Figure 4H). Therefore, transport of M1-associated viral proteins to the cell membrane assembly and budding site was inhibited in NUMA1 deficient cells, which ultimately reduced the infectious virus production compared to the N-Si cells. However, an increase of IAV M2 proteins was also deposited near the NUMA1 KD cytoplasmic region ( Figure 6B), which suggested that M2 was produced but could not participate in viral maturation with other structural proteins. EM results also showed significant reduction of virus budding in NUMA1 KD cells compared to the N-Si cells ( Figure 5). We propose a model illustrating the role of NUMA1 in IAV maturation (Figure 7), in which the viral M1-RNP complexes transport from the nucleus into the cytoplasm and M1 interacts with NS1. We propose a model illustrating the role of NUMA1 in IAV maturation (Figure 7), in which the viral M1-RNP complexes transport from the nucleus into the cytoplasm and M1 interacts with NS1. We report in the current study that NS1 interacts with host NUMA1 protein during IAV replication. NS1 works as a bridge between NUMA1 and M1 proteins. After that, the strong link between NUMA1 and microtubules facilitates the transfer of viral M1-associated proteins (vRNPs) to the cytoplasm using the exocytic pathway via trans-Golgi network for assembly and budding. We also propose that M1-vRNPs are not able to interact with the microtubule network in NUMA1 deficient cells; thus, the transportation of these essential proteins to the assembly site is obstructed. IAV RNPs can be routed to the cell periphery through the Rab11-dependent vesicular transport system [83]. Moreover, M1 and NP vRNP also can reach the assembly site through cytoskeletal We report in the current study that NS1 interacts with host NUMA1 protein during IAV replication. NS1 works as a bridge between NUMA1 and M1 proteins. After that, the strong link between NUMA1 and microtubules facilitates the transfer of viral M1-associated proteins (vRNPs) to the cytoplasm using the exocytic pathway via trans-Golgi network for assembly and budding. We also propose that M1-vRNPs are not able to interact with the microtubule network in NUMA1 deficient cells; thus, the transportation of these essential proteins to the assembly site is obstructed. IAV RNPs can be routed to the cell periphery through the Rab11-dependent vesicular transport system [83]. Moreover, M1 and NP vRNP also can reach the assembly site through cytoskeletal microfilaments [82]. These Rab11 and microfilament-mediated transport systems might contribute to the generation of some viruses (2.6-20% of the N-Si) in our PR8-infected NUMA1 KD cells. Our proposed model elucidates the mechanism performed by NUMA1 protein to help influenza virus replication by the exocytic pathway.
In conclusion, novel and essential NS1-interacting host factors including NUMA1 identified in this study shed further light on the detailed mechanism of influenza virus replication and may identify alternative non-viral targets to develop new antiviral therapies.