Immortalized cell lines

Human A549 (86012804; ECACC) and HEK293T (CRL-3216; ATCC) cells were cultured in DMEM (11960044; Gibco) supplemented with 10% FBS (0270106; Gibco), GlutaMAX (35050038; Gibco), sodium pyruvate (11360039; Gibco) and penicillin–streptomycin (15070063; Gibco) at 37 °C and 5% CO2. Calu-3 cells (no catalogue number available—provided by Petr Chlanda, Heidelberg University; originally from Prof. Ralf Bartenschlager, Heidelberg University) were maintained in the same medium with 20% FBS. MDCK (84121903; ECACC), MDCK.II (00062107; ECACC), MDCK-SIAT1 (05071502-1VL; Sigma-Aldrich) and MDBK (no catalogue number available—provided by Ervin Fodor, University of Oxford; Sir William Dunn School of Pathology Cell Bank stock) cells were grown in MEM supplemented with 10% FBS, GlutaMAX, sodium pyruvate and penicillin–streptomycin. MEFs (no catalogue number available—provided by Prof. Dr. Michael Bader, Max Delbrück Center) were cultured in DMEM with 10% FBS, GlutaMAX, sodium pyruvate, penicillin–streptomycin and non-essential amino acids (11140050; Gibco). HeLa (Kyoto; no catalogue number is available for this line, provided by the Mahamid laboratory, which had received them from the Sara Cuylen-Haering laboratory, originally from Daniel Gerlich, Institute of Molecular Biotechnology of the Austrian Academy of Sciences; the Gerlich laboratory reported this HeLa Kyoto stock as originally obtained from S. Narumiya, Kyoto University, RRID:CVCL_1922, authenticated by multiplex human cell line authentication testing) were cultured in DMEM (11960044; Gibco) supplemented with 10% FBS (0270106; Gibco), GlutaMAX (35050038; Gibco), sodium pyruvate (11360039; Gibco) and penicillin–streptomycin (15070063; Gibco) at 37 °C and 5% CO2. A549 and HEK293T cells were authenticated by short tandem repeat profiling. All cell lines were routinely tested for mycoplasma contamination and confirmed negative.

Primary cells

Primary HBEpCs (C-12640, donor lot number 499Z012.1; Promocell) were maintained in Airway Epithelial Cell Growth Medium (C-21060; Promocell) and cultured according to the manufacturer’s instructions in Airway Epithelial Cell Growth Medium at 37 °C with 5% CO2. HBEpCs were maintained as undifferentiated submerged cultures and were not differentiated at an air–liquid interface. Primary cells were used at passages 2–6 for all experiments. All cell lines were routinely tested for mycoplasma contamination (MycoplasmaCheck; Eurofins).

Virus production

WSN was produced in confluent MDBK cells infected at a multiplicity of infection (MOI) of 0.001 in DMEM containing 0.5% FBS. After 36 hours incubation at 37 °C and 5% CO2, when full cytopathic effect was observed, supernatants were collected, clarified by centrifugation (1,500g, 10 minutes, 4 °C) and aliquoted for storage at −80 °C. pdm09 and H3N2 were produced in MDCK cells infected at an MOI of 0.001 in serum-free DMEM supplemented with 0.2% BSA and 1 µg ml−1 TPCK-treated trypsin. After 36–48 hours incubation at 37 °C and 5% CO2 and confirmation of full cytopathic effect and successful HA assay, viral supernatants were clarified and stored as mentioned earlier.

Recombinant SC35M (H7N7) was generated by transfecting HEK293T cells with 8 plasmids (pHW2000) encoding all 8 influenza gene segments (1 μg each) and Lipofectamine 2000 (Invitrogen) according to the manufacturers’ protocol. After an incubation period of 6 hours, cell culture medium was removed and replaced with fresh DMEM. At 48 hours post-transfection, supernatants were collected to inoculate fresh MDCK. For a secondary infection, supernatants from previously infected MDCK cells were collected at 48 hours post-infection, stored at −70 °C and used for inoculation of fresh MDCK cells. After clear cytopathic effect development, supernatants were collected and processed as described earlier.

Virus infection

Cells were infected with IAV strains A/WSN/33 (H1N1), A/California/07/2009 (pdm09, H1N1), A/Aichi/2/1968 (H3N2) or SC35M (mouse-adapted variant of the highly pathogenic avian influenza virus strain SC35 derived from A/Seal/Massachusetts/1/80 (H7N7)) at the indicated MOI. Before infection, virus inocula were diluted in infection medium consisting of serum-free DMEM supplemented with 0.2% BSA and antibiotics. Cells were washed once with PBS and incubated with the virus inoculum for 30 minutes at 4 °C, followed by an additional 30 minutes at 37 °C with gentle rocking to allow virus attachment and entry. For SC35M, cells were incubated with virus inoculum for 30 minutes at 37 °C with gentle shaking to allow even distribution of virus particles on the cell layer. After infection, the inoculum was removed and replaced with fresh infection medium.

For infections with A/WSN/33 and SC35M, infection medium did not contain trypsin. For infections with A/California/07/2009 and A/Aichi/2/1968, infection medium was supplemented with TPCK-treated trypsin (1 µg ml−1) as required for these strains to support multicycle replication. Infected cells were incubated at 37 °C and 5% CO2 and collected at the indicated time points for downstream analyses.

Reconstitution of recombinant IAV overexpressing NanoLuc on PB2

A recombinant IAV expressing NanoLuc fused to PB2 was generated using a 12-plasmid reverse genetics system as described previously34. The system included eight plasmids encoding the vRNA segments and four expression plasmids for PB2, PB1, PA and NP under control of a human Pol I promoter. The NanoLuc reporter gene was inserted in-frame at the amino terminus of PB2, separated by a self-cleaving T2A peptide to allow independent translation.

HA assay

The HA assay was performed in 96-well round-bottom plates using 2-fold serial dilutions of virus in PBS (50 µl per well). An equal volume (50 µl) of 0.5% chicken erythrocyte suspension was added to each well. Plates were incubated for 30 minutes at 4 °C, followed by 30 minutes at room temperature. HA titres were determined visually as the highest virus dilution producing complete haemagglutination, defined by a uniform lattice across the well. Results are expressed in haemagglutination units.

Plaque assay

For WSN, confluent MDCK cells in 6-well plates were infected with 330 µl of serially diluted virus in infection media (DMEM with 0.2% BSA). Plates were incubated for 30 minutes at 4 °C, followed by 30 minutes at 37 °C with gentle rocking. Cells were overlaid with Avicel-containing media (DMEM, 0.2% BSA) and incubated for 72 hours. After removing the overlay, cells were washed with PBS and stained with 1% crystal violet for 5 minutes. Plaques were counted to calculate viral titre in plaque-forming units per millilitre (PFU ml−1). For pdm09 and H3N2, MDCK-II cells were infected as above and overlaid with DMEM containing Avicel, 0.2% BSA and 1 µg ml−1 TPCK-treated trypsin. After 72 hours, cells were fixed with 4% paraformaldehyde for 30 minutes at 4 °C and blocked with 0.3% hydrogen peroxide. Viral plaques were detected by immunostaining with anti-NP antibodies (Aichi, MA5-42364, Thermo Fisher; pdm09, ab43821, Abcam; 1:1,000) for 1 hour at room temperature, followed by HRP-conjugated secondary antibody and KPL TrueBlue substrate (15 minutes). Viral titres were calculated as PFU ml−1:

$$\mathrm{Titre}\,\left(\frac{\mathrm{PFU}}{\mathrm{ml}}\right)=\frac{\left(\mathrm{number}\,\mathrm{of}\,\mathrm{plaques}\,\times \,\mathrm{dilution}\,\mathrm{factor}\right)}{\mathrm{volume}\,\mathrm{of}\,\mathrm{virus}\,\mathrm{innoculum}\,(\mathrm{ml})}$$

Luciferase assay

A549 cells were seeded at 1 × 105 cells per well in 24-well plates and transfected 12 hours later in duplicate with 10 µM siRNAs using RNAiMAX (Thermo Fisher Scientific), according to the manufacturer’s instructions. After 72 hours incubation at 37 °C and 5% CO2, cells were washed with infection media and infected with PB2-T2A-NanoLuc recombinant WSN at an MOI of 0.01. At 48 hours post-infection, cells were lysed in 200 µl Passive Lysis Buffer (Promega) per well and incubated at room temperature for 20–30 minutes on a rocking platform. For luciferase activity, 25 µl of lysate was mixed with 25 µl NanoGlo substrate (Promega) in a white 96-well plate. Luminescence was measured after 5–10 minutes equilibration using a luminometer (1,000 ms integration time).

Viral growth kinetics

For viral growth kinetics, cells were infected at low multiplicity to allow multicycle replication. A549 cells and HBEpCs were seeded at 2 × 105 cells per well, transfected with the indicated siRNAs for 72 hours, and subsequently infected with WSN at an MOI of 0.001 or pdm09 at an MOI of 0.01. For experiments performed in KO or overexpressed cell lines, cells were seeded at 7.5 × 105 cells per well before infection. Following infection, cells were washed and maintained in the appropriate infection medium.

At the indicated time points post-infection, 50 µl of supernatant was collected and stored at −70 °C until analysis. Viral titres were determined by plaque assay on MDCK cells (WSN) or MDCK-SIAT1 cells (pdm09) as described earlier. Growth curves are presented as PFU ml−1 at 24 hpi, 48 hpi and 72 hpi.

FISH

Cells (1 × 105 per well) were seeded onto sterilized glass coverslips in 24-well plates and grown to ~70–80% confluence. Cells were washed with DPBS, fixed in 4% paraformaldehyde for 10 minutes at room temperature and permeabilized in 70% ethanol overnight at −20 °C. Hybridization and amplification were performed using the hybridization chain reaction (HCR v.3.0) protocol adapted from ref. 79, with optimizations for detecting IAV vRNA, mRNA and antisense RNA (see Supplementary Table 5 for probe sequences and design). Briefly, fixed cells were prehybridized with Probe Hybridization Buffer (Molecular Instruments) and incubated overnight with 2 pmol of each probe at 37 °C. Following washes in Probe Wash Buffer and SSCT, amplification was carried out overnight at room temperature using snap-cooled HCR hairpins (h1 and h2). Cells were washed in 5× SSCT, stained with DAPI and mounted using Dako antifade mountant (Agilent). Samples were imaged using a confocal microscope with appropriate filter settings for DAPI and HCR fluorophores. Acquisition settings were optimized to reduce background and enhance signal detection.

siRNA transfection

All siRNAs were purchased from Qiagen (Supplementary Table 6). A549 cells at ~60% confluence were transfected with 5 pmol siRNA using Lipofectamine RNAiMAX (13778075; Thermo Fisher Scientific) according to the manufacturer’s protocol. Knockdown efficiency was assessed 72 hours post-transfection by qPCR or western blot.

PLA

PLA was performed using the Duolink In Situ PLA kit (Sigma-Aldrich) following the Sigma protocol. A549 cells were seeded on glass coverslips and infected as described earlier. At the indicated time point post-infection, cells were washed with PBS, fixed with 4% paraformaldehyde for 10 minutes at room temperature and permeabilized with PBS containing 0.2% Triton X-100 for 10 minutes at room temperature, according to the manufacturer’s instructions. Next, cells were blocked in the supplied blocking solution and incubated with the primary antibodies. After washing, PLA probes were applied, followed by ligation and rolling-circle amplification steps. PLA signals were detected using Duolink detection reagents coupled to Alexa Fluor 488. Nuclei were counterstained with DAPI, and coverslips were mounted using Dako mounting medium (Agilent). Samples were imaged by fluorescence microscopy using identical acquisition settings across conditions, and PLA puncta and signal intensity were quantified from maximum z-projection images.

The following primary antibodies were used for PLA: anti-IAV nucleoprotein (NP) (clone C43; Abcam), anti-NS1 (EPR28247-51; Abcam), anti-HA (clone 1.B.408; Abcam), anti-PB2 (PA5-32220; Thermo Fisher Scientific), anti-M1 (clone GA2B, MA1-80736; Thermo Fisher Scientific), anti-hnRNPK (F45P9C7; Thermo Fisher Scientific), anti-RAB11A (67902-1-Ig; Proteintech), anti-hnRNPA1 (67844-1-Ig; Proteintech), anti-NF45 (67685-1-Ig; Proteintech), anti-hnRNPA2B1 (67445-1-Ig; Proteintech), anti-CCAR2 (66497-1-Ig; Proteintech), anti-SFPQ (67129-1-Ig; Proteintech), anti-nucleolin (D4C7O, number 14574; Cell Signaling Technology), anti-B4GALT1 (PA5-52744; Thermo Fisher Scientific), anti-VIP36 (PA5-90437; Thermo Fisher Scientific), anti-hnRNPM (26897-1-Ig; Proteintech), anti-ERP29 (24344-1-AP; Proteintech), anti-GLUT1 (21829-1-AP; Proteintech), anti-RPL35 (14826-1-AP; Proteintech), anti-ERGIC-53 (13364-1-AP: Proteintech), anti-hnRNPC (11760-1-AP; Proteintech) and anti-ANP32B (10843-1-AP; Proteintech).

Cell viability

Cell viability was measured using the CellTiter-Glo 2.0 Assay Kit (G9241; Promega). Treated cells in 24-well plates were incubated with reagent directly added to the culture medium, followed by 10 minutes of shaking at room temperature. Luminescence, proportional to intracellular ATP levels, was measured using a plate reader and normalized to untreated controls.

CRISPR–Cas9-mediated KO cell lines

Single guide RNAs were designed using Benchling or purchased as validated guides (IDT) and resuspended at 1 µM. Recombinant Cas9 protein (EMBL Protein Production Facility) was diluted to 1 µM in Opti-MEM. A549 cells were seeded at 1.5 × 105 cells per well in 6-well plates at <50% confluence to ensure high KO efficiency. Cas9–single guide RNA ribonucleoprotein (RNP) complexes were formed in Opti-MEM and transfected using Lipofectamine RNAiMAX. After 96 hours, cells were either collected for genomic DNA (NEB kit) or seeded at 0.5 cells per well in 96-well plates for clonal selection. Target loci were PCR-amplified, gel-purified and sequenced (Sanger). Indel frequency was analysed using TIDE, and KO was confirmed by western blotting.

Plasmid transfection

Plasmid transfections were performed in HEK293T cells at 70–80% confluence using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Plasmid DNA and Lipofectamine 3000 reagent were each diluted in Opti-MEM (Thermo Fisher Scientific), combined and incubated for 15 minutes at room temperature to allow DNA–lipid complex formation. The complexes were added dropwise to the cells and incubated for 4–6 hours at 37 °C in 5% CO2, after which the medium was replaced with fresh complete DMEM. Cells were incubated for an additional 16–48 hours before analysis. Transfection efficiency was assessed by cotransfection with a GFP-expressing plasmid or by fluorescence microscopy following immunofluorescence or FISH staining.

Lentiviral cell line generation

Lentiviral vectors encoding mEGFP-tagged NONO, PSPC1 or SFPQ were cotransfected into HEK293T cells with packaging plasmids using a calcium phosphate transfection protocol. Supernatants containing lentiviral particles were collected after 48 hours, filtered and stored at −70 °C. A549 cells were transduced with viral supernatants in the presence of polybrene and incubated for 48–72 hours to allow stable genomic integration. Transduced cells were selected using 10 µg ml−1 Blasticidin and maintained in complete DMEM. Expression of tagged proteins was confirmed by fluorescence microscopy and western blotting. Protein functionality was verified by their colocalization with NEAT1 in paraspeckles.

RNA extraction, reverse transcription and qPCR

Total RNA was extracted using the RNeasy Kit (Qiagen) according to the manufacturer’s instructions. Cells were lysed in buffer RLT supplemented with 5 mM TCEP, and RNA was purified on RNeasy columns with RW1 and RPE wash buffers. RNA was eluted in RNase-free water, quantified using a NanoDrop spectrophotometer and stored at −70 °C.

Reverse transcription was performed with up to 2 µg RNA per 10 µl reaction using the High-Capacity cDNA Reverse Transcription Kit (4368814; Thermo Fisher Scientific). qPCR was carried out using PowerUp SYBR Green Master Mix (A25742; Thermo Fisher Scientific) with gene-specific primers (Supplementary Table 7). Thermal cycling and melt curve analysis were performed according to the manufacturer’s protocol. Relative gene expression was quantified using the ΔΔCt method and normalized to GAPDH.

Immunoblotting

Whole-cell lysates were resolved by SDS–PAGE using 4–12% precast TG PRiME gels (SERVA) and transferred to nitrocellulose membranes (Bio-Rad). Membranes were blocked in 2.5% BSA (in TBS with 0.01% Tween-20) for 1 hour at room temperature, followed by overnight incubation at 4 °C with primary antibodies diluted in blocking buffer. After 3 washes in TBS-Tween, membranes were incubated for 1 hour at room temperature with HRP-conjugated anti-mouse or anti-rabbit secondary antibodies (1:5,000), washed again and developed using enhanced chemiluminescence (A38555; Thermo Fisher Scientific). Antibodies used for immunoblotting are listed in Supplementary Table 8. Uncropped blots are available in the Source data.

Affinity pulldown of Twin-Strep-tagged proteins

HEK293T cells were transfected with Twin-Strep-tagged constructs (NONO, M2 or empty vector control) using PEI at a 1:3 DNA:PEI ratio. At 24 hours post-transfection, cells were collected by scraping in PBS, washed and lysed in IP lysis buffer (Thermo Fisher) supplemented with HALT protease inhibitor cocktail. Lysates were clarified (17,000g, 20 minutes, 4 °C) and incubated with Strep-Tactin magnetic beads (IBA Lifesciences) pre-equilibrated in Buffer W. After 1 hour incubation at room temperature with rotation, beads were washed and bound proteins were eluted in Laemmli buffer at 95 °C for 2 minutes. Eluates were analysed by SDS–PAGE and western blot.

IP

A549 cells were seeded in 10 cm dishes (4 × 106 cells per dish) and infected with WSN (MOI 3) for 14 hours. Cells were collected and lysed in IP buffer (Thermo Fisher) supplemented with HALT protease inhibitor (Thermo Fisher). For certain IPs (for example, NP, NONO, SFPQ and PSPC1), 0.1% SDS was added to reduce non-specific binding. Lysates were cleared by centrifugation (max speed, 20 minutes, 4 °C). IP was performed with 10 µg of antibody and 500 µl of cleared lysate for 4 hours at 4 °C, followed by incubation with Protein A/G or G Dynabeads (Thermo Fisher) for 1 hour at room temperature. Isotype-matched control antibodies were used in all experiments. Beads were washed with TBS + 0.05% Tween-20 and lysis buffer and finally with TBS alone. For AP–MS, bound proteins were eluted in 8 M guanidine HCl at 95 °C and precipitated in cold ethanol overnight. Inputs were processed by methanol–chloroform precipitation. For western blotting, elution was performed with 0.2 M glycine (pH 2.0), followed by Tris neutralization and/or SDS loading buffer. Eluates were analysed by SDS–PAGE, western blot or mass spectrometry as described later. All AP–MS experiments were performed in triplicate.

SHVIP sample preparation

Cells were infected at an MOI of 5 and, at 5 hpi, incubated in labelling medium containing 500 µM L-homopropargylglycine (L-HPG) in methionine-free DMEM with 0.2% BSA. At 14 hpi, cells were washed with PBS and collected by scraping. For cross-linking, cell pellets were resuspended in PBS and cross-linked with 5 mM DSSO for 1 hour at room temperature with constant agitation. The reaction was quenched with 30 mM Tris (pH 8.0) for 20 minutes. Cells were lysed in buffer containing 200 mM Tris (pH 8.0), 4% CHAPS, 1 M NaCl, 8 M urea and protease inhibitors. Lysates were frozen at −80 °C until enrichment. After thawing, lysates were treated with Benzonase and sonicated (10 × 30 second on/off cycles; Bioruptor Pico) and then clarified by centrifugation (10,000g, 10 minutes). A 50 µl aliquot was saved as input. Enrichment of HPG-labelled proteins was performed as described previously80: 800 µl of lysate was incubated with 200 µl picolyl-azide agarose beads (Click Chemistry Tools), and copper(I)-catalysed click chemistry was performed overnight at room temperature with rotation. Proteins were reduced with 10 mM DTT (70 °C, 15 minutes) and alkylated with 40 mM chloroacetamide (room temperature in the dark). Beads were washed sequentially in gravity columns with: 1% SDS, 250 mM NaCl and 5 mM EDTA in 100 mM Tris (pH 8.0); 8 M urea in 100 mM Tris (pH 8.0); 80% acetonitrile (ACN) in water; 5% ACN in 50 mM triethylammonium bicarbonate (TEAB); and 5% ACN with 2 M urea in 50 mM TEAB. Bound proteins were digested on-bead overnight at 37 °C with trypsin and LysC in the final wash buffer. Peptides were collected, acidified with 1% formic acid, desalted using C18 stage tip (or Sep-Pak C8 columns for cross-linked peptides), dried by vacuum centrifugation and stored at −20 °C until LC–MS or further processing. Input samples were processed by methanol–chloroform precipitation and digested in buffer containing 1% sodium deoxycholate, 5 mM TCEP and 40 mM 2-chloroacetamide in 50 mM TEAB. The resulting peptides were desalted as described earlier.

Off-line fractionation of cross-linked peptide samples

To enrich cross-linked peptides, HPG-enriched samples were first subjected to strong cation exchange (SCX) chromatography using a polysulfethyl A column. Peptides were separated with a 95-minute linear gradient and collected in 45-second intervals. Fractions were desalted, dried and stored at −20 °C. After initial measurement, selected SCX fractions were pooled and further separated by size exclusion chromatography (SEC) using a Superdex 30 column. Fractions were collected every 2 minutes, dried and stored at −20 °C before LC–MS/MS.

LC–MS/MS of bottom-up proteomics and AP–MS samples

Bottom-up samples and AP–MS eluates were analysed using Orbitrap Fusion Tribrid or Exploris 480 instruments coupled to nano-LC systems. Peptides were loaded onto a 50 cm in-house packed C18 column and separated by 120–180 minute gradients.

For bottom-up and AP–MS samples, MS1 scans were acquired at 120,000 resolution, dynamic exclusion was set to 40 seconds, precursors (charge +2 to +4) were isolated (1.6 m/z window) and fragmented by higher-energy collisional dissociation (HCD) at 30% normalized collision energy (NCE), with MS2 detection performed in the ion trap (Fusion) or Orbitrap (Exploris).

LC–MS/MS of cross-linked samples

Cross-linked peptides were analysed on an Orbitrap Fusion Lumos equipped with a FAIMS Pro Duo interface. MS1 scans were performed at 120,000 resolution using FAIMS voltages (−50 V, −60 V and −75 V). Precursors (charge +4 to +8) were fragmented by stepped HCD (21%, 27% and 33% NCE) and MS2 scans acquired in the Orbitrap at 60,000 resolution.

Bottom-up proteomics data analysis

Raw files were searched using MaxQuant v.1.6.2.6a81 against the human SwissProt database (release 2020) and the IAV/WSN/1933 (H1N1) proteome. Trypsin was used for in silico digestion (max 2 missed cleavages). Carbamidomethylation (C) was set as fixed modification and variable modifications, including oxidation (M), acetylation (N-term) and methionine replacement with HPG, where appropriate. iBAQ and label-free quantification (LFQ) quantification were enabled. Contaminants, reverse hits and proteins identified only by site were removed. For LFQ, missing values were imputed as described previously16. Fold changes were log2-transformed and P values were computed using 2-sided t tests. All AP–MS experiments were performed in triplicates.

Raw data were searched using Scout (v.1.5.1)82 with DSSO cross-linker settings and residue specificity for K, S, T and Y. Search parameters included: minimum peptide length, 6; max missed cleavages, 3; precursor mass tolerance, 10 ppm; fragment mass tolerance, 20 ppm; static mod, carbamidomethylation (C); and variable mod, oxidation (M).

Data were filtered at 2% or 5% FDR on cross-link spectrum match, residue-pair and PPI levels. For performance evaluation (Extended Data Fig. 1m–r), SCX fractions were searched per replicate. For the full dataset, SCX and SEC files were combined for global analysis.

Glycosylation analysis

For each condition, 5 million A549 cells were seeded in a 15 cm dish. After 12 hours, cells were transfected with siRNAs targeting B4GALT1 or LMAN2 (2 independent siRNAs per target) using Lipofectamine RNAiMAX (Thermo Fisher Scientific) according to the manufacturer’s instructions. At 72 hours post-transfection, cells were infected with WSN at an MOI of 3 as described earlier. At 14 hpi, cells were collected by scraping, washed with PBS and frozen.

Glycoproteomics experiments were performed as described previously48 with minor adjustments: after lysis, digestion and glyco-enrichment, the samples were labelled with tandem mass tag (TMT) reagents and pooled. The samples were prefractionated into 24 fractions. After analysis of the glycopeptide samples on a Exploris 480, generated raw files were searched using MSFragger v.4.0 in Fragpipe v.20.0 against the Swissprot Homo sapiens database (UP000005640, 20,443 entries) combined with the influenza A database (UP000009255, 13 entries).

For the quantitative analysis of proteomics and glycoproteomics data, we first normalized the TMT reporter intensities for glycopeptides and proteins with complete quantification across all conditions (that is, TMT reporter ion intensity different from zero for all channels) using the normalizeVSN function (variance-stabilizing normalization; VSN) of the limma R package. We corrected glycopeptide intensities per protein by removing abundance-derived intensity using a linear regression for all glycopeptides with a matched total protein intensity. The corrected and normalized glycopeptide intensities and the normalized protein intensities were used for the differential expression analysis using the limma R package. For the design of the differential expression analysis, we considered the knockdown condition and the sample replicate. Contrasts were set for the comparison between the non-targeting control and the respective knockdown conditions.

Glycosylation at the N27/N28 sites of HA could not be assessed, as the corresponding amino-terminal region (residues 3–43) lacks lysine and arginine residues and therefore does not generate detectable peptides under the trypsin-based proteomics and glycoproteomics workflows used in this study.

Confocal microscopy

All imaging was conducted using a Nikon Ti2 microscope equipped with a CSU-W1 spinning disk confocal unit and an Andor DU-888 X-11633 camera. The microscope was controlled by NIS-Elements software and used a 100×/1.49 SR Apo TIRF AC oil immersion objective (Nikon). For live-cell imaging, cells were seeded in Ibidi μ-Slide 4-well chambers (NC0685967; Ibidi) and maintained at 37 °C with 5% CO2. Imaging was performed using the 405 nm (DAPI channel), 488 nm (GFP channel) and 561 nm (Cy5 channel) lasers, with exposure times optimized for each experiment. The camera was maintained at −69.4 °C, with a binning of 1 × 1 and a readout speed of 30 MHz. Z-stacks were captured with an interval size of 130.6 nm and images were acquired at a resolution of 1,024 × 1,024 pixels.

Image analysis

All images were processed as maximum intensity projections. Nuclei (DAPI) and paraspeckles (NEAT1_2) were segmented using CellProfiler v.4.2.6 with the Otsu thresholding method. The CV of the NEAT1_2 signal within nuclei and the number of segmented paraspeckles were used as independent metrics. Both metrics have their respective advantages and disadvantages. Although CV relies on more reliable nuclei segmentation, it is an indirect measure of paraspeckle integrity and does not provide precise structural information. In contrast, paraspeckle quantification directly assesses paraspeckle formation but is more error prone, particularly at later time points when the NEAT1_2 signal diminishes.

Line profile analysis and preparation of representative images were performed using Fiji (ImageJ v.2.16.0/1.54p).

Subcellular localization analysis of viral protein interactors

To compare subcellular localizations of host interactors across studies, we compiled viral–host protein interactions from SHVIP (this study), refs. 13,15 and the meta-analysis in ref. 26. For all datasets, only viral proteins that were identified in SHVIP (PB1, PB2, NP, NS1, M1, M2, HA and NA) were considered.

Host protein localizations were annotated based on data from the Human Protein Atlas. Only primary localization assignments were retained. A simplified set of localization categories was defined that grouped terms as follows: plasma membrane, cytosol, cytoskeleton, vesicular system, ER, mitochondrion, nucleoplasm, nucleoli, nucleus (other) and other, followed by manual curation. For ambiguous or multilocalized proteins, the most specific localization according to UniProt was assigned.

For SHVIP, we included only high-confidence host interactors filtered at 2% FDR at both the cross-linking PPI levels. For ref. 13, host interactors were taken directly from their published AP–MS dataset in HEK293 cells expressing individual FLAG-tagged viral proteins. For ref. 15, we used the high-confidence filtered dataset provided by the authors, based on AP–MS of 13 IAV proteins expressed in 3 cell types across 3 different IAV strains. For ref. 26, only host interactors identified in at least three independent studies or databases were considered, as described in their meta-analysis.

Viral protein localization was manually curated from published literature (Supplementary References) and database annotations (UniProt). Where conflicting evidence for localization existed, fractional assignments were split evenly across compartments (for example, a score of 1 distributed as 0.5 and 0.5 across 2 compartments).

Validation of cross-links by host structures and AlphaFold 2 models

For each cross-linked human protein, all available structures in the PDB were retrieved using the corresponding UniProt identifier via the UniProt83 REST API and RCSB PDB Search API84. All cross-linked pairs were mapped to the retrieved structures, and if the corresponding pair was found among structures, Cα–Cα was calculated. If one pair was mapped to multiple structures, the shortest Cα–Cα distance was taken. The same mapping procedure was used for the random cross-linked pairs generated by taking arbitrary lysine–lysine pairs for individual proteins and protein pairs detected in PPIs. AlphaFold 2 (ref. 44) models were built using AlphaPulldown85. Structural figures were rendered using UCSF ChimeraX86.

Virus–host interaction structural modelling

AlphaFold 2 models were built using AlphaPulldown85 v.2.0 using 24 recycles and 4 predictions per model. The sequence and template databases as of August 2024 were used. AlphaFold 3 was run using the original AlphaFold 3 code and using the default sequence database (downloaded in November 2024) and PDB database from November 2021 using the download script provided in AlphaFold 3 (ref. 31). AF3x32, a modelling protocol implemented within the AlphaFold 3 code that allows incorporation of cross-links as restraints in the form of explicit covalent ligands, was run with default settings and using version #dfb94a3 and the same databases as AlphaFold 3 (ref. 31). For AlphaFold 3 and AF3x, 20 random seeds were used per prediction (in some cases up to 1,000). The quality of models was assessed using the interface-predicted template modelling (ipTM) score, predicted template modelling (pTM), residue’s predicted local distance difference test (pLDDT) score and predicted alignment error, as returned by all the above tools. GRASP33 was run in normal mode using version 2a381bc, with four predictions per model, ranking by pTM. The multiple sequence alignment (MSA) and templates were the same as for AlphaFold2. Distance restraints in GRASP were applied using a 30 Å threshold and 0.05 FDR. Structural figures were rendered using UCSF ChimeraX86.

Statistical analysis

Statistical analyses, plotting and graph preparation were performed using GraphPad Prism 10 (GraphPad Software). Exact P values for statistical analyses depicted in the figures are provided in the Source data.

Other bioinformatics analyses

Functional enrichment analysis was performed using the Enrichr web platform87. The default Fisher’s exact test with the Benjamini–Hochberg method for correction for multiple hypotheses testing was used. Overlap with prior host–host interactions was calculated using BioGRID v.2.0.18 (ref. 88). For cross-links to viral proteins where alternative proteins with identical cross-linked peptide sequences were detected ambiguously (for example, histone proteins), only the first protein from the list of ambiguous candidates was selected for the analysis. Cross-link diagrams were drawn using xiNET89.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.