A strategy to suppress STAT1 signalling conserved in pathogenic poxviruses and paramyxoviruses

The induction of interferon-stimulated genes by signal transducer and activator of transcription (STAT) proteins, is a critical host defence to fight virus infections. Here, a highly expressed poxvirus protein 018 is shown to inhibit IFN-induced signalling by binding the SH2 domain of STAT1 to prevent STAT1 association with an activated IFN receptor. Despite the presence of additional inhibitors of IFN-induced signalling, a poxvirus lacking 018 was attenuated in mice. The 2.0 Å crystal structure of the 018:STAT1 complex reveals a mechanism for a high-affinity, pTyr-independent mode of binding to an SH2 domain. Furthermore, the STAT1 binding motif of 018 shows sequence similarity to the STAT1-binding proteins from Nipah virus, which like 018, block the association of STAT1 with an IFN receptor. Taken together, these results provide detailed mechanistic insight into a potent mode of STAT1 antagonism, found to exist in genetically diverse virus families.

(pSTAT) and undergo dimer rearrangement from an anti-parallel to an activated parallel 35 conformation, mediated by a reciprocal pTyr:SH2 interaction between two pSTATs (Wenta et 36 al., 2008). 37 38 Type I IFNs (IFN-I) signal through the IFNa/b receptor (IFNAR) to activate kinases that 39 phosphorylate STAT1 and 2. The pSTAT1:STAT2 heterodimer associates with IRF9 to form 40 a trimeric complex known as IFN-stimulated gene factor 3 (ISGF3) (Rengachari et al., 2018). 41 Type II IFN (IFN-II), of which IFNg is the sole member, signals through the IFNg receptor 42 (IFNGR) and activates kinases that phosphorylate STAT1 only. The pSTAT1 homodimer is 43 known as the gamma-activated factor (GAF). Nuclear ISGF3 and GAF drive the transcription 44 of ISGs with IFN-stimulated responsive element (ISRE) or gamma-activated sequence (GAS) 45 binding region of NiV-V protein can compete with a phosphorylated IFN receptor to bind 76 STAT1. This study reveals a conserved mechanism for targeting STAT1 utilised by members 77 of the poxvirus and paramyxovirus families, to subvert cellular anti-viral responses. 78

79
The 018 open reading frame (ORF) from VACV WR (gene VACWR018) is transcribed early 80 during infection and is one of the most abundant viral transcripts (Assarsson et al., 2008; 81 Wennier et al., 2013;Yang et al., 2015). Protein 018 is highly conserved within the 82 orthopoxvirus genus including human pathogens cowpox virus, monkeypox virus and variola 83 virus, the causative agent of smallpox ( Figure S1). The 018 ORF is also highly conserved in 84 ancient variola viruses dating from the Viking age (Mühlemann et al., 2020). To test if 018 modulates anti-viral immunity, reporter plasmids were used that express a 93 luciferase (Luc) gene upon activation of specific anti-viral signalling pathways. Activation of 94 IRF3, NF-kB and AP-1 pathways that induce IFNs (specifically IFNb) were measured using 95 an IFNb-Luc reporter after stimulation with Sendai virus (SeV), the prototypic paramyxovirus. 96 Downstream, IFN-induced pathways were measured using ISRE or GAS-Luc reporters after 97 stimulation with IFN-I (IFNa) or IFN-II (IFNg), respectively. 98 99 TAP-tagged (2 X Strep, 1 X FLAG epitope) 018 inhibited pathway activation induced by IFN-100 I and II (Figure 1A, B), whereas it had little effect on the activation of IFNb-Luc ( Figure 1C). 101 demonstrating a marginal loss in inhibitory activity, however, expression was undetectable 174 ( Figure S3D,E). All further C-terminal truncations showed <25% inhibitory activity, but 175 again, expression was undetectable (Figure S3D,E). The same pattern of inhibitory activity by 176 018 truncations mutants was observed for both IFN-I and -II signalling (Figure S3F), 177 indicating the same region of 018 is required to inhibit both pathways. 178 179 These observations map a putative minimal inhibitory region of 018 to aa 11-31. The C-180 terminal boundary was defined assuming the slight reduction in inhibitory activity after 181 deletion of residues 35-31 was due to lower protein expression levels, whereas further 182 truncation removed functional residues. Ser31 is included within the minimal inhibitory region 183 as it is highly conserved in orthopoxvirus orthologues of 018 ( Figure S1). 184 185 ITC measurements of the putative minimal fragment (018 T2 ) with STAT1 gave a KD of 235 186 nM, a value comparable to that of full length 018 (291 nM) ( Figure 3H). Removal of the C-187 terminal 28-TYTS-31 (018 T3 ) from the putative minimal fragment led to a large reduction in 188 affinity (>10 µM), thereby demonstrating the importance of these residues ( Figure 3I). 189 Collectively, these data show that a 21-residue segment of 018, aa 11-31, is sufficient for 190 maximal STAT1 binding and inhibitory activity. To study the role of 018 during infection, a VACV (strain WR) 018 deletion mutant (termed 194 vD018) was constructed. The wild-type sibling virus (termed v018) and vD018 were analysed 195 by PCR ( Figure S4A) and genomic sequencing, which showed no differences other than the 196 intended deletion of the 018 ORF. Comparison of v018 and vD018 in cell culture displayed no 197 was constructed by reintroduction of the 018 ORF fused to an N-terminal TAP-tag into vD018 199 at its natural locus. Pulldown of TAP-tagged 018 expressed from vTAP-018 confirmed the 200 018:STAT1 interaction during infection (Figure S4E, F). 201 202 Next, vD018's ability to inhibit IFN signalling was assessed. A549 cells were infected with 203 v018 or vD018 and, at the indicated times p.i., were stimulated with either IFN-I or -II, after 204 which the pSTAT1 level was determined by immunoblotting. Cells were washed once prior to 205 stimulation to remove the majority of soluble VACV IFN decoy receptors B8 and B18. This, 206 however, will not fully remove B18 (IFN-I decoy receptor) due to its ability to bind to the cell 207 surface (Alcamı́ et al., 2000). Although by 2 h p.i., both v018 and vD018 inhibited pSTAT1 208 induction after IFN-I stimulation, v018 inhibited earlier and to a greater extent. (Figure 4A). 209 In contrast, pSTAT1 induction was inhibited by v018 but almost fully rescued to mock levels 210 in vD018-infected cells after IFN-II stimulation ( Figure 4B). Consistent with this finding, 211 STAT1 translocation was blocked by v018 after IFN-II stimulation, whereas in vD018-infected 212 cells, STAT1 was predominantly nuclear ( Figure 4C). The impaired ability of vD018 to inhibit 213 IFN-II signalling was illustrated further by increased IRF1 levels (a canonical IFNg ISG) in 214 cells infected with v018 compared to vD018 after IFN-II stimulation at both the mRNA 215 ( Figure S4G) and protein level ( Figure 4D). 216 217 To evaluate if 018 contributes to virulence, BALB/c mice were infected via the intranasal route 218 with either v018 or vD018 and their weight was measured daily (Figure 4E). Mice infected 219 with vD018 lost significantly less weight than those infected with v018 ( Figure 4E) and 220 showed reduced virus titres at 7 and 9 days p.i. (Figure 4F). Furthermore, consistent with 018 221 functioning as an immunomodulator, mRNAs for several ISGs, chemokines and IFNs were 222 upregulated in the lungs of mice infected with vD018 compared to v018 ( Figure 4G). 223 Collectively, these data show VACV lacking 018 is defective in inhibition of IFN-induced 224 signalling and is attenuated in mice. been studied with NiV-V, which also only binds 31F (Rodriguez et al., 2004). 018 bound 234 STAT1 C-terminal truncations that lack the final 38 residues (STAT1b, a naturally occurring 235 isoform of STAT1) or the entire TAD ( Figure 5C). Lastly, 018 bound a chimera that contained 236 only the SH2 and TAD of STAT1 (Fus 1) but not a chimera that contained the LD of STAT1 237 (Fus 2) nor with STAT3 alone ( Figure 5D) To test this, a fluorescence polarisation (FP) assay was established using a fluorescent 12-mer 252 peptide corresponding to the pIFNGR1 sequence harbouring the STAT1 docking site 253 (pYDKPH) as a probe. Addition of 018 to a preformed STAT1-pIFNGR1 probe led to a dose-254 dependent displacement of probe and an IC50 value of 1.26 µM ( Figure 5E). IC50 values of 0.93 255 µM and 17.82 µM for 018 T2 and 018 T3 , respectively, were obtained, demonstrating that 018 T2 , 256 but not 018 T3 , has comparable inhibitory activity to full length 018, consistent with ITC data 257 ( Figure 5E). 258

259
The mechanism was further validated by competition ITC. A 5-mer peptide corresponding to 260 the pIFNGR1 docking region (pYDKPH) was titrated into U-STAT1, giving a KD value of 7.6 261 µM ( Figure 5F). In contrast, inclusion of excess 018 in the calorimeter cell resulted in complete 262 loss of detectable binding ( Figure 5G) Figure S5). Furthermore, ITC titration of purified GB1-fused 018 AGA into STAT1 resulted in 282 no detectable binding ( Figure 6B). Loss in STAT1 binding ability correlated with a loss of 283 inhibitory activity because 018 AGA was unable to inhibit IFN-I and -II signalling by reporter 284 gene assay (Figure 6C,D). Consistent with this, 018 AGA did not interfere with 285 STAT1:pIFNGR1 12-mer interaction by FP ( Figure 6G). In addition, 018 AGA showed no 286 inhibition of STAT1-pIFNGR1 binding via ITC ( Figure 6H). 287 288 Consistent with the idea that 018 and NiV-V harbour analogous motifs, recently the site for 289 NiV-V binding to STAT1 was mapped to the SH2 domain of STAT1 (Keiffer et al., 2020). To 290 assess if these viral proteins target the same SH2 interface, the ability of 018 to outcompete the 291 NiV-V:STAT1 interaction was tested. In cells transfected with TAP-tagged NiV-V, NiV-V co-292 precipitated with endogenous STAT1, however, this was decreased in a dose-dependent 293 manner by expression of HA-tagged 018 ( Figure 6E). In contrast, HA-tagged 018 AGA did not 294 affect the NiV-V:STAT1 interaction ( Figure 6F). These data show that 018 and NiV-V utilise 295 a shared motif to bind a common interface on the SH2 domain of STAT1. 296 297 Previous reports show NiV-V sequesters STAT1 and 2 within the cytoplasm and prevents 298 STAT1 phosphorylation (Rodriguez et al., 2002). The finding that 018 and NiV-V bind STAT1 299 via the same interface prompted us to assess if, like 018, NiV-V competes with pIFNGR1 to 300 bind STAT1. To test this, NiV-V STAT1-binding fragment residues 110-140 (NiV-V 110-140 ) 301 fused to a GB1 tag was purified together with a mutant in which His117 and His119 of the 302 HxH motif were mutated to Ala (NiV-V ADA ). By FP assay, addition of NiV-V to the preformed 303 STAT1-pIFNGR1 12-mer complex led to a modest reduction in polarisation, whereas addition 304 of NiV-V ADA was non-competitive ( Figure 6I). Consistent with these data, preincubation of 305 STAT1 with NiV-V abolished any detectable binding between STAT1 and the pIFNGR 5-mer 306 by ITC ( Figure 6J). In contrast, preincubation with NiV-V ADA did not prevent 307 STAT1:pIFNGR binding ( Figure 6K) Figure S6A). 318

319
The 018 peptide forms a β-hairpin fold with a β-turn midway through the sequence ( Figure  320 7A,B) and the two strands of the peptide augment the central β-sheet of the SH2 domain, with 321 Val14-His17 backbone hydrogen-bonding to the βD strand of the SH2 domain ( Figure 7C). 322 There is spatial overlap with published binding modes of pTyr peptides from pIFNGR1 and 323 pSTAT1 homodimer ( Figure 7B). The 680 Å 2 interface is formed by a large number of shallow 324 contacts exclusively within the SH2 domain. Residues Trp12, Val14, Ile16 comprise a 325 continuous hydrophobic interface with STAT1 helix αA and strand βD ( Figure 7D). This is 326 followed by a HxH motif, in which His17 forms an imidazole-to-imidazole hydrogen bond 327 with His629 of STAT1 (Figure 7D, E). The His17 rotamer is stabilised intramolecularly by a 328 second hydrogen bond with the backbone carbonyl of 018 Gly21. Gly18 carbonyl forms a 329 hydrogen bond with Tyr651 hydroxyl of STAT1, similar to pIFNGR1 Pro443 (PDB: 1YVL). 330 His19 occupies the same cleft as His444 of pIFNGR1, forming an identical π-stacking 331 interaction with STAT1 Tyr634. The Asp20 sidechain stabilises the β-turn by hydrogen 332 bonding with the Ser21 backbone and forms an intramolecular salt bridge with Lys24 ( Figure  333   7D). An inter-strand hydrogen bond between the hydroxyl groups of Ser13 and Thr28 act as a 334 non-covalent bridge that may stabilise the β-hairpin fold ( Figure 7D). 335 336 Strikingly, 018 does not interact with the pTyr pocket. The only tyrosine in the peptide, Tyr29, 337 hydrogen bonds with the ζ-amine of STAT1 Lys584 through its hydroxyl and makes van der 338 Waals contacts with the alkyl chain of the same lysine ( Figure 7D) Pulldown of 018 demonstrated that 018 binds STAT1 and STAT4, but not other STATs 347 ( Figure 7F). STAT2 revealed that both NS5 proteins overlap the IRF9 binding site to prevent ISGF3 364 assembly (Wang et al., 2020). A similar mechanism was described for measles V protein 365 (Nagano et al., 2020). Here, the structure of 018, an uncharacterised poxvirus protein, with 366 STAT1, shows 018 occupies the STAT1 SH2 domain to block STAT1 association with the 367 active pIFNGR. subsequently is phosphorylated at Tyr690 (Yan et al., 1996). The pTyr690 of STAT2 serves as 388 a docking site for STAT1 to present STAT1 for proximal phosphorylation at Tyr701 by JAKs 389 on IFNAR2 are also important for ISGF3 formation and could serve as docking sites for STAT1 391 and 2 functioning in a cell type-or species-dependent manner (Zhao et al., 2008). Thus, we 392 rationalise that during IFN-I signalling, occupancy of the STAT1 SH2 domain by 018 would 393 diminish STAT1 engagement of either STAT2 pTyr690 or IFNAR to prevent STAT1 394 phosphorylation.
) and U6A (STAT2 -/-) cells (B) by transfection and were affinity purified by Strep-Tactin. 562 Whole cell lysate (Input) and affinity purified proteins (AP:Strep) were analysed by 563 immunoblotting. (C) ITC data for GB1-018 (100 µM) titrated into U-STAT1 (10 µM). 564 Fitting of the isotherm (bottom) to a one site model gave a KD of 290 nM. Initial low volume 565 injection is excluded from analysis. Complete fitted ITC parameters are provided in Table  566 S5.   Table S5. 619   The 018 open reading frame was codon-optimised for expression in human cells and was 741 synthesised by Gene Art (Thermo Fisher Scientific). All plasmids are described in recombinant 742 DNA key resource table and primers used for construction in Table S1. After 1 h 30 min, the inoculum was removed and replaced with a MEM, 1% (w/v) low gelling 787 temperature agarose (Sigma Alrich), supplemented with MPA, HX, and X. After three days, 788 EGFP-expressing plaques were picked, representing virus that had integrated the pUC13-789 Ecogpt-EGFP-D018, and then further plaque purified in the absence of MPA, HX and X. The 790 genotype of these plaques was then determined by PCR using primers that flank the 018 ORF 791 (Table S3) and VACVs containing the desired mutation (vD018) or wild type genotype were 792 isolated. vTAP-018 was produced using the same strategy as described above, except that 793 vD018 was used as the parental VACV into which the TAP-018 ORF was inserted at its natural 794 locus. Stocks of VACVs were grown in RK13 cells and titrated by plaque assay on BS-C-1 795

cells. 796 797
Purification of VACVs by sedimentation through sucrose 798 VACVs were purified by two rounds of ultracentrifugation through a sucrose cushion as 799 described (Joklik, 1962) and stocks were resuspended in 1 mM Tris-HCl pH 9.0 for cell culture 800 were determined by plaque assay. Oligonucleotide primers (Table S2) targeting HRPT and IRF1 were designed using 898 PrimerQuest Tool (IDT). RT-qPCRs were carried out using a real-time PCR system (Thermo 899 Fisher Scientific) and fold-inductions of ISG levels were calculated using 2 -∆∆Ct taking mock 900 non-stimulated readings as the basal level sample and HRPT as the control housekeeping gene.