The ubiquitination landscape of the influenza A virus polymerase

Abstract

During influenza A virus (IAV) infections, viral proteins are targeted by cellular E3 ligases for modification with ubiquitin. Here, we decipher and functionally explore the ubiquitination landscape of the IAV polymerase proteins during infection of human alveolar epithelial cells by applying mass spectrometry analysis of immuno-purified K-ε-GG (di-glycyl)-remnant-bearing peptides. We have identified 59 modified lysines across the three subunits, PB2, PB1 and PA of the viral polymerase of which 17 distinctively affect mRNA transcription, vRNA replication and the generation of recombinant viruses via non-proteolytic mechanisms. Moreover, further functional and in silico analysis indicate that ubiquitination at K578 in the PB1 thumb domain is mechanistically linked to dynamic structural transitions of the viral polymerase that are required for vRNA replication. Mutations K578A and K578R differentially affect the generation of recombinant viruses by impeding cRNA and vRNA synthesis, NP binding as well as polymerase dimerization. Collectively, our results demonstrate that the ubiquitin-mediated charge neutralization at PB1-K578 disrupts the interaction to an unstructured loop in the PB2 N-terminus that is required to coordinate polymerase dimerization and facilitate vRNA replication. This provides evidence that IAV exploits the cellular ubiquitin system to modulate the activity of the viral polymerase for viral replication.

Fig. 1. Identification of site-specific UB modifications in the IAV polymerase.

a–c Western blot analysis of ubiquitin (UB), NEDD8 and ISG15 modification of the IAV polymerase subunits. A549 (UB) or HEK293-T (NEDD8, ISG15) were transfected with expression plasmids for strep-tagged polymerase subunits (PB2, PB1, or PA), respectively, together with plasmids for the expression of ubiquitin-like modifiers (UBLs) bearing an HA- or His-tag. UBL-modified polymerase subunits were precipitated under denaturing conditions using Strep-Tactin® bound sepharose beads. UBL modification was detected by western blot using the indicated antibodies. Representative blots of three independent experiments are shown. MW, molecular weight marker. d Experimental outline for the identification of di-glycylated lysines using a di-glycyl-specific immunoselection coupled to mass spectrometry analysis 5 h post infection (p.i). MOI, multiplicity of infection. This illustration was created with biorender. e Sequence alignment of UB/UBLs. The di-glycyl motif is highlighted in yellow, the N-terminal amino acid, which determines trypsin cleavage, is highlighted in gray. f–h Conservation analysis of the identified UBL acceptor lysines in PB2, PB1, and PA. PB2-K113, -K627; PB1-K278, -K586 are not depicted. PB2, PB1, and PA sequences from human, swine, and avian IAV isolates were downloaded from NCBI (PB1 and PB2: 09/15/2017; PA: 10/01/2017) and analyzed for sequence identity. Relative frequency of lysines at the respective position in PB2 (f), PB1 (g), and PA (h) is depicted. In case of a lysine frequency below 75%, the most abundant AA is shown (R = Arginine). Source data are provided as a Source data file.

Fig. 2. UB-modified lysines reside in functional domains of the IAV polymerase.

a, b 3D structural models of a WSN-adapted heterotrimeric IAV polymerase bound to vRNA (adapted from PDB: 4WSB; a) or cRNA (adapted from PDB: 5EPI; b) created by comparative homology modeling. Positions of lysines with diglycyl remnants are depicted in violet (PB2, yellow, upper panel), green (PA, violet, middle panel), and magenta (PB1, turquoise, lower panel). c Linear models of PB2 (upper panel), PA (middle panel), and PB1 (lower panel) depicting the location of the identified di-glycylated lysines (K-ε-GG position) in previously described functional domains. Boundaries of functional domains (gray boxes) are depicted in the upper lane (#AA). Interaction sites to viral proteins PA (violet), PB1 (light green), PB2 (orange), NP (dark orange), and NEP (brown) and the cellular proteins pol-II (dark green), importin (bright orange), and ANP32A (pink) are included. RNA interaction sites are depicted as follows: mRNA (gray) vRNA (white boxes) or cRNA (black boxes) and the 5′Cap-structures of cellular mRNAs (bright green boxes). Other functional motives include the catalytic domains of the polymerase in PB1 (light pink), NLS (violet), and the priming loop (dark green).

Fig. 3. Mutational screen of modified lysines reveals distinct effects on mRNA transcription and vRNA replication.

a Schematic of the polymerase reconstitution assay. Cells are co-transfected with plasmids for PB2, PB1, PA, and NP together with a firefly luciferase-encoding vRNA minigenome under control of the pol-I promoter and a Renilla luciferase under control of the pol-II promoter. FF activity correlates with the activity of the viral polymerase. Renilla activity serves as an internal transfection control. b Generation of recombinant influenza viruses. Cells are co-transfected with 8 bi-directional pHW2000 plasmids encoding the viral genome segments. Viral mRNA and vRNA is generated from pol-II and pol-I promoters, respectively. Illustrations created with biorender. c–e Polymerase reconstitution assay with A/R-substitutions of the modified lysines in PB2 (c), PA (d), and PB1 (e). Relative FF activities are presented as the mean percentage activity using the wild type (WT) polymerase (±SEM), n = 3 independent biological replicates. P values < 0.05 compared to WT from Dunnett’s multiple comparison one-way ANOVA test are indicated. Success (+) or failure (−) to generate recombinant viruses is depicted below. Virus rescue was considered negative if three independent rescue attempts failed. Sites located in functional domains (white), interaction regions with RNA templates (light red), or viral/host proteins (gray) are highlighted. f–h Quantification of FF mRNA and cRNA levels from the polymerase reconstitution assay for mutations in PB2 (f), PA (g), or PB1 (h) using qRT-PCR using segment and RNA species-specific primers. Values are depicted as mean relative to WT (±SEM), n = 3 independent biological replicates. GAPDH served as housekeeping control. P values < 0.05 compared to WT from Dunnett’s multiple comparison two-way ANOVA are indicated. i–k Viral titers after rescue (i) or passaging (j, k), n = 3 independent biological replicates (±SEM). Viral titers were determined 48 h p.i. as plaque forming units (PFU) per ml. l–n Immunofluorescence of WT or mutated versions of PB2 (l), PA individually (m), or in combination with HA-tagged PB1 (n). 24 h post transfection A549 cells were fixed and analyzed using the indicated antibodies. Cell nuclei were visualized using DAPI. Representative cells from one experiment are shown. Scale bar = 10 µm.

Fig. 4. PB1-K578 is ubiquitinated and interacts with a loop in the PB2 N-terminus.

a A549 cells were co-transfected with strep-tagged PB1 or PB1-K578A and HA-tagged UB. UB-modified PB1 subunits were strep-purified using denaturing conditions and analyzed by western blot. For de-ubiquitination, bound PB1 constructs were treated with USP2. Co-precipitated UB-HA levels were quantified and presented as the mean fold change of WT (±SEM), n = 4 independent biological replicates. P values compared to WT from Welch’s corrected two-tailed t-test are indicated. b, c Illustration of PB1-K578 in the 3D structures of the WSN polymerase (vRNA-bound conformation; b) and the 3D structure of the symmetric dimer of the H3N2 polymerase (PDB: 6QNW; c), showing the K578 containing PB1-helix (cyan), the flexible loop in the PB2 N-terminus (yellow) harboring PB2-E72 and the helix in the PA C-terminus that participates in dimer formation. Distances between PB1-K578 and PB2-E72 are indicated in Angstrom (Å). Amino acids involved in the dimer interface are indicated. Created with ChimeraX. d Mean lifetime of ionic interactions with PB2-E72 for 100 ns simulations, n = 4 independent biological replicates for each condition (±SEM). P values < 0.05 compared to WT from Welch’s corrected two-tailed t-test are indicated. e, f RMSD distribution of PB2-E72 (e) and PB2-Q73 (f) for WT, K578R and K578A simulations. Each 100 ns simulation generated 1000 RMSD values leading to a total number of 4000 values for each condition. Equal bin sizes with 0.5 Å steps were used for of each histogram. g RMSF values of PB2 residues (±SEM). Significance of mean differences was analyzed by Welch corrected two-sided t-test, n = 4 independent experiments. P values are summarized in Supplementary Table 2. Source data are provided as a Source data file.

Fig. 5. Mutation of PB1-K578 affects cRNP stabilization and vRNA synthesis.

a, b Detection of packaged vRNA segments isolated from released virus particles in the supernatants of the rescue experiments at 48 h p.t. vRNA segments were amplified with segment-specific primers, separated on 2% agarose gel (b). Gels from all three replicates of K578R and representative gels of three replicates (R) with consistent results are shown for WT and PB1-K578A. vRNA was analyzed using deep sequencing. Percentage of nucleotides encoded at position 1732–34 encoding for K578 (a). c–e Analysis of cRNA stabilization, cRNA and vRNA synthesis. HEK293-T cells were transfected with plasmids expressing NP, PB2, PA, inactive PB1 [PB1a] (D445A/D446A^,; c) or active PB1 (d, e). At 24 h p.t., cells were infected with WT WSN virus (MOI: 5) and cultured in infection medium with cycloheximide (CHX). At 6 h p.i. cRNA (c, d) or vRNA levels (e) of the NA segment were assessed using strand-specific RT-PCR and are shown as mean n-fold of WT (±SEM), n = 6 independent biological replicates. P values < 0.05 compared to WT from Dunnett’s multiple comparison one-way ANOVA-test are indicated. f, g Accumulation of vRNA segments in HEK293-T cells co-transfected with pHW2000 plasmids at the indicated time points. vRNA levels were quantified using segment-specific primers for RT-PCR and presented as the mean n-fold of WT (±SEM), n = 3 independent biological replicates. Cellular GAPDH mRNA levels were used as housekeeping control. P values < 0.05 compared to WT from Šidák’s corrected multiple comparison two-way ANOVA test are indicated. Source data are provided as a Source data file.

Fig. 6. Mutation of PB1-K578 affects the binding of free polymerase to NP and polymerase dimerization.

a, b Detection of NP binding to trimeric polymerase by co-affinity precipitation. Cells were transfected with PB1, PB2, HA-tagged PA, strep-tagged NP (a) and a firefly-encoding vRNA reporter (b) followed by strep-purification 24 h p.t. Co-precipitated proteins were detected by western blot. c Polymerase dimerization assessed by co-affinity precipitation. Cells were transfected with PB1, PB2 and both HA- and strep-tagged PA. 24 h p.t. the polymerase components bound to strep-tagged PA were purified. Co-precipitated proteins were detected by western blot. (Input Strep-PA signals derive from a blot analyzed in parallel). Quantification of the interaction is shown below the panels. Levels of HA-tagged PA were normalized to strep-tagged NP (a, b) or strep-tagged PA (c) and depicted below as the mean n-fold of WT (±SEM). The number of independent biological replicates (n) is provided in the figure. P values < 0.05 compared to WT from Dunnett’s multiple comparison one-way ANOVA are indicated. Source data are provided as a Source data file. d Model depicting the biological function of PB1-K578 ubiquitination during vRNA replication. Incoming vRNPs perform viral mRNA transcription for viral protein expression including the viral polymerase proteins and NP. A subpopulation of newly synthesized trimeric polymerase complexes acquires ubiquitination at PB1-K578 (cyan polymerase complexes with UB in yellow) which disrupts the interaction between PB1-K578 and the loop in the PB2-N1 domain, prevents early formation of the symmetric dimer and reduces the affinity to NP. The PB1-K578 ubiquitinated polymerase interacts with the vRNP-bound polymerase to form the ANP32A-stabilized asymmetric dimer for cRNA synthesis and encapsidation. For vRNA synthesis from the cRNA template, a non-ubiquitinated polymerase (violet polymerase complexes) interacts with the cRNP-associated polymerase under formation of the symmetric dimer. Ultimately, progeny vRNPs may perform secondary mRNA transcription, further rounds of vRNA replication or get exported out of the nucleus for progeny virion assembly. Substitution of the UB-acceptor site PB1-K578 with both alanine and arginine affects vRNA replication (right panel). While PB1-K578A retains the pool of free polymerases, PB1-K578 aborts progression of vRNA replication due to a depletion of free polymerases.