Thiopurines activate an antiviral unfolded protein response that blocks viral glycoprotein accumulation in cell culture infection model

Enveloped viruses, including influenza A viruses (IAVs) and coronaviruses (CoVs), utilize the host cell secretory pathway to synthesize viral glycoproteins and direct them to sites of assembly. Using an image-based high-content screen, we identified two thiopurines, 6-thioguanine (6-TG) and 6-thioguanosine (6-TGo), that selectively disrupted the processing and accumulation of IAV glycoproteins hemagglutinin (HA) and neuraminidase (NA). Selective disruption of IAV glycoprotein processing and accumulation by 6-TG and 6-TGo correlated with unfolded protein response (UPR) activation and HA accumulation could be partially restored by the chemical chaperone 4-phenylbutyrate (4PBA). Chemical inhibition of the integrated stress response (ISR) restored accumulation of NA monomers in the presence of 6-TG or 6-TGo, but did not restore NA glycosylation or oligomerization. Thiopurines inhibited replication of the human coronavirus OC43 (HCoV-OC43), which also correlated with UPR/ISR activation and diminished accumulation of ORF1ab and nucleocapsid (N) mRNAs and N protein, which suggests broader disruption of coronavirus gene expression in ER-derived cytoplasmic compartments. The chemically similar thiopurine 6-mercaptopurine (6-MP) had little effect on the UPR and did not affect IAV or HCoV-OC43 replication. Consistent with reports on other CoV Spike (S) proteins, ectopic expression of SARS-CoV-2 S protein caused UPR activation. 6-TG treatment inhibited accumulation of full length S0 or furin-cleaved S2 fusion proteins, but spared the S1 ectodomain. DBeQ, which inhibits the p97 AAA-ATPase required for retrotranslocation of ubiquitinated misfolded proteins during ER-associated degradation (ERAD) restored accumulation of S0 and S2 proteins in the presence of 6-TG, suggesting that 6-TG induced UPR accelerates ERAD-mediated turnover of membrane-anchored S0 and S2 glycoproteins. Taken together, these data indicate that 6-TG and 6-TGo are effective host-targeted antivirals that trigger the UPR and disrupt accumulation of viral glycoproteins. Importantly, our data demonstrate for the first time the efficacy of these thiopurines in limiting IAV and HCoV-OC43 replication in cell culture models. IMPORTANCE Secreted and transmembrane proteins are synthesized in the endoplasmic reticulum (ER), where they are folded and modified prior to transport. During infection, many viruses burden the ER with the task of creating and processing viral glycoproteins that will ultimately be incorporated into viral envelopes. Some viruses refashion the ER into replication compartments where viral gene expression and genome replication take place. This viral burden on the ER can trigger the cellular unfolded protein response (UPR), which attempts to increase the protein folding and processing capacity of the ER to match the protein load. Much remains to be learned about how viruses co-opt the UPR to ensure efficient synthesis of viral glycoproteins. Here, we show that two FDA-approved thiopurine drugs, 6-TG and 6-TGo, induce the UPR in a manner that impedes viral glycoprotein accumulation for enveloped influenza viruses and coronaviruses. These drugs may impede the replication of viruses that require precise tuning of the UPR to support viral glycoprotein synthesis for the successful completion of a replication cycle.

Secreted and transmembrane proteins are synthesized in the endoplasmic reticulum (ER), where 51 they are folded and modified prior to transport. During infection, many viruses burden the ER with 52 the task of creating and processing viral glycoproteins that will ultimately be incorporated into 53 viral envelopes. Some viruses refashion the ER into replication compartments where viral gene 54 expression and genome replication take place. This viral burden on the ER can trigger the cellular 55 unfolded protein response (UPR), which attempts to increase the protein folding and processing 56 capacity of the ER to match the protein load. Much remains to be learned about how viruses co-57 opt the UPR to ensure efficient synthesis of viral glycoproteins. Here, we show that two FDA-58 approved thiopurine drugs, 6-TG and 6-TGo, induce the UPR in a manner that impedes viral 59 glycoprotein accumulation for enveloped influenza viruses and coronaviruses. These drugs may 60 impede the replication of viruses that require precise tuning of the UPR to support viral 61 glycoprotein synthesis for the successful completion of a replication cycle. 62

INTRODUCTION 64
Enveloped viruses encode integral membrane proteins that are synthesized and post-65 translationally modified in the endoplasmic reticulum (ER) prior to transport to sites of virion 66 assembly. When ER protein folding capacity is exceeded, the accumulation of unfolded proteins 67 in the ER causes activation of the unfolded protein response (UPR) whereby activating 68 transcription factor-6 (ATF6), inositol requiring enzyme-1 (IRE1) and PKR-like endoplasmic 69 reticulum kinase (PERK) sense ER stress and trigger the synthesis of basic leucine zipper (bZIP) 70 transcription factors that initiate a transcriptional response (1). UPR gene expression causes the 71 accumulation of proteins that attempt to restore ER proteostasis by expanding ER folding capacity 72 and stimulating catabolic activities like ER-associated degradation (ERAD) (2). ERAD ensures 73 that integral membrane proteins that fail to be properly folded are ubiquitinated and 74 retrotranslocated out of the ER for degradation in the 26S proteasome. There is accumulating 75 evidence that bursts of viral glycoprotein synthesis can burden ER protein folding machinery, and 76 that enveloped viruses subvert the UPR to promote efficient viral replication (3,4). 77 Influenza A viruses (IAVs) encode three integral membrane proteins: hemagglutinin (HA) 78 neuraminidase (NA) and matrix protein 2 (M2). HA adopts a type I transmembrane topology in 79 the ER, followed by addition of N-linked glycans, disulfide bond formation, and trimerization prior 80 to transport to the Golgi and further processing by proteases and glycosyltransferases (5-11); NA 81 adopts a type II transmembrane topology in the ER, is similarly processed by glycosyltransferases 82 and protein disulfide isomerases, and assembles into tetramers prior to traversing the secretory 83 pathway to the cell surface (12, 13). The small M2 protein also forms disulfide-linked tetramers in 84 the ER, which is a prerequisite for viroporin activity (14-16). IAV replication causes selective 85 activation of the UPR; IRE1 is activated, but PERK and ATF6 are not (17), although the precise 86 mechanisms of regulation remain unknown. Furthermore, chemical chaperones and selective 87 chemical inhibition of IRE1 activity inhibit IAV replication, suggesting that IRE1 has pro-viral 88 effects. HA is sufficient to activate the UPR (18) and is subject to ERAD-mediated degradation 89 (19). By contrast, little is known about how NA and M2 proteins affect the UPR. However, these inhibitors triggered SG formation and cytotoxic effects in uninfected cells as well, 137 limiting their potential utility as antivirals. Because SG formation correlates with antiviral activity, 138 we conducted an image-based high-content screen to identify molecules that selectively induce 139 SG formation in IAV infected cells. We identified two FDA-approved thiopurine analogs, 6-140 thioguanine (6-TG) and 6-thioguanosine (6-TGo), that blocked IAV and HCoV-OC43 replication 141 in a dose-dependent manner. Unlike Pateamine A and Silvestrol, these thiopurines selectively 142 disrupted the processing and accumulation of viral glycoproteins, which correlated with UPR 143 activation. Synthesis of viral glycoproteins could be partially restored in 6-TG treated cells by the 144 chemical inhibition of the UPR or ISR. Our data suggest that UPR-inducing molecules could be 145 effective host-targeted antivirals against viruses that depend on ER processes to support efficient 146 replication. Induction of UPR by 6-TG and 6-TGo represents a novel host-directed antiviral 147 mechanism triggered by these drugs and reveals a previously unrecognized unique mechanism of 148 action that distinguishes them from other closely related thiopurines and nucleoside analogues.  Through this screen, we identified two thiopurines, 6-thioguanine (6-TG) and 6-166 thioguanosine (6-TGo) (Fig. 1A), that triggered dose-dependent SG formation in IAV-infected 167 cells (Fig. 1B). Specifically, SGs formed in approximately 10 % of 6-TG-treated or 6-TGo-treated 168 infected cells; no SGs were detected in mock infected cells treated with either drug at the highest 169 concentration (Fig. 1B). These findings were confirmed in parental A549 cells infected with IAV 170 strain A/California/07/2009 (H1N1; IAV-CA/07); 6-TG treated cells displayed the formation of 171 foci that contained SG constituent proteins G3BP1 and poly A binding protein (PABP) (Fig. 1C). 172 These foci also contained canonical SG proteins TIAR and eIF3A (Fig. 1D), supporting their 173 identity as bona fide SGs. 174

Thiopurine analogs inhibit IAV replication 175
Next, we wanted to determine whether thiopurine-mediated SG formation indicated a 176 disruption of viral replication. A549 cells were infected with IAV strain A/PuertoRico/8/1934 177 (H1N1; IAV-PR8) and treated with 6-TG, 6-TGo or controls at 1 hpi. Cell supernatants were 178 harvested at 24 hpi and infectious virions enumerated by plaque assay. Despite SG induction in 179 only a fraction of virus-infected cells, we observed a sharp dose-dependent decrease in virion 180 production following treatment with either thiopurine analog. Treatment with 2 M 6-TG reduced 181 virion production by ~10-fold, whereas 2 M 6-TGo reduced virion production by ~100-fold (Fig.  182  2A). Furthermore, treatment of IAV infected cells with 10 µM concentrations of either 6-TG or 6-183 TGo led to even greater inhibition of IAV production ( Fig. 2A). This suggests that SG formation 184 correlates with the disruption of the viral replication cycle. However, the sharp decrease in 185 infectious virion production in 6-TG/6-TGo-treated cells suggests that SG formation is not 186 required for their antiviral effect. The nucleoside analog 5-fluorouracil (5-FU) had no effect on 187 IAV replication at 2 M and 10 M doses ( Fig. 2A). Using an alamarBlue assay, we observed a 188 ~30% reduction in A549 cell viability in the presence of 10 M doses of 6-TG/6-TGo (Fig. 2B). 189 Compared to SG-inducing translation inhibitor Silvestrol, which causes apoptosis in A549 cells 190 upon prolonged exposure, we did not observe significant disruption of cell monolayer by 6-TG 191 treatment ( Fig. 2C) or induction of apoptosis as measured by PARP cleavage (Fig. 2D). This is 192 consistent with a recent report of 6-TG-mediated cytostatic rather than cytotoxic effects on A549 193 cells (46). In Vero cells, 6-TG treatment partially protected cellular monolayers from IAV-induced 194 cell death over 72-h incubation (Fig. 2E). Taken together, our data suggest that 6-TG and 6-TGo 195 elicit a broad dose-dependent antiviral effect against IAV that was not shared by the nucleoside 196 analog 5-FU. faster-migrating, presumably un-glycosylated species, but these were difficult to visualize as they 211 migrated to the same position as NP on immunoblots probed with polyclonal anti-IAV antibodies 212 that concurrently detect NP, M1 and HA. 5-FU, which had no effect on viral replication over the 213 24 h time course in these cells ( Fig. 2A), likewise had no effect on the accumulation of these IAV 214 proteins (Fig. 3A). Consistent with the notion of selective inhibition of IAV glycoprotein synthesis 215 and maturation, we observed that 6-TG had no effect on the accumulation of IAV-PR8 HA or NA 216 transcripts or function of the RdRp in genome replication, as 6-TG had little effect on the 217 accumulation of HA and NA genome segments (Fig. 3B). Taken together, these data support a 218 novel mechanism of action for thiopurine analogs in selectively inhibiting processing and 219 accumulation of IAV glycoproteins and significantly impairing IAV replication. 220 6-TG and 6-TGo activate the UPR and chemical mitigation of ER stress restores synthesis of 221

HA glycoproteins 222
By inhibiting N-linked glycosylation, TM impedes proper processing of secreted and 223 transmembrane proteins in the lumen of the ER, which elicits ER stress and activates the UPR 224 (47). Indeed, we observed that TM treatment of A549 cells caused accumulation of XBP1s and 225 the ER chaperone binding immunoglobulin protein (BiP) (Fig. 4A). BiP upregulation is an 226 excellent measure for UPR activation because it requires both ATF6(N)-dependent transcription 227 of BiP and PERK-mediated activation of the ISR and uORF-skipping-dependent translation (1). 228 We observed that both 6-TG and 6-TGo caused BiP and XBP1s accumulation in A549 cells, 229 whereas the chemically similar thiopurine 6-mercaptopurine (6-MP) did not. Nucleoside analogs 230 5-FU and ribavirin also did not affect BiP or XBP1s levels (Fig. 4A). These data demonstrate that 231 6-TG and 6-TGo, but not all thiopurines, activate the UPR in A549 cells. 232 To determine whether thiopurines could activate the UPR during IAV infection, A549 cells 233 infected with IAV-PR8 were treated with 6-TG. We observed that 6-TG caused strong 234 accumulation of BiP which coincided with diminished accumulation of HA (Fig. 4B). By contrast, 235 co-administration of 6-TG and the chemical chaperone 4-phenylbutyrate (4-PBA) (48) diminished 236 accumulation of BiP and partially restored HA levels in infected cells, without affecting levels of 237 NP and M1 proteins. This suggests that thiopurine-mediated activation of the UPR/ISR is at least 238 partially responsible for the diminished accumulation of HA glycoproteins in infected cells. 239 To corroborate our observation of 6-TG-mediated UPR activation, A549 cells were mock 240 infected or IAV-PR8 infected, and treated with 6-TG, 6-MP, or TM for 24 h before harvesting 241 RNA for RT-qPCR analysis of UPR gene expression. We analyzed transcripts produced from 242 target genes linked to each arm of the UPR; ATF4 target gene CHOP, XBP1s target genes EDEM1 243 and ERdj4, and ATF6(N) target genes BiP and HERPUD1. As expected, TM treatment caused 244 strong induction of all 3 arms of the UPR and increased transcription of all five target genes in 245 mock-infected cells and infected cells alike (Fig. 4C). This strong and consistent transcriptional 246 output between mock-infected and infected cells suggests that the UPR remains largely intact 247 during IAV-PR8 infection. Treatment with 6-MP had little effect on UPR gene expression (Fig.  248   4C). By contrast, 6-TG treatment caused statistically significant increases in transcription from all 249 five UPR target genes (Fig. 4C). These observations confirm that 6-TG activates all three arms of 250 the UPR, whereas the chemically similar thiopurine 6-MP does not. 251 Inhibition of the integrated stress response does not restore NA processing and 252 oligomerization in 6-TG treated cells 253 IAV glycoproteins are translocated into the ER, where they are modified with N-linked 254 glycans and organize into oligomeric complexes. Upon synthesis in the ER, the type II 255 transmembrane protein NA is glycosylated and forms dimers linked by intermolecular disulfide 256 bonds in the stalk region (13) that then assemble into tetramers (49). We investigated the effect of 257 6-TG on NA processing and oligomerization using SDS-PAGE/immunoblotting procedures in the 258 presence or absence of the disulfide bond reducing agent dithiothreitol (DTT). Since NA tetramers 259 are known to dissociate into dimers during electrophoresis (12, 50) we annotated the ~120 kDa 260 band as dimers/tetramers (Fig. 5A). We observed intact glycosylated NA dimers/tetramers and 261 monomers in mock-treated IAV-PR8-infected cells, which were resolved into ~60 kDa 262 glycosylated NA monomers in the presence of DTT (Fig. 5A). Unglycosylated NA monomers 263 were undetectable in mock-treated cells at steady state, confirming that N-glycosylation is a rapid 264 initial step in NA processing in the ER. TM treatment eliminated NA dimers/tetramers, leaving a 265 minor fraction of unglycosylated NA monomers. The 6-TG treatment diminished accumulation of 266 all forms of NA, yielding a distinct residual band that migrated closer to the size of the 267 unglycosylated NA monomers from TM-treated cells; this suggests that 6-TG treatment interferes 268 with proper N-glycosylation of nascent NA. Treatment with Integrated Stress Response Inhibitor 269 (ISRIB), which prevents ISR-mediated translation arrest by maintaining eIF2B activity (51, 52), 270 rescued accumulation of NA monomers in both TM-and 6-TG-treated cells. However, ISRIB was 271 not able to restore NA glycosylation and oligomerization. These data provide further evidence that 272 6-TG inhibits IAV glycoprotein accumulation via UPR/ISR activation and extends our 273 understanding by demonstrating that ISR suppression does not fully reverse these effects. This is 274 further supported by our observations that administration of ISRIB alone had no impact on IAV 275 replication while co-administration of ISRIB with TM-or 6-TG failed to restore virion production 276 in single-cycle infection assays (Fig. 5B). 277

6-TG and 6-TGo inhibit HCoV-OC43 replication 278
In vitro studies have shown that thiopurines 6-TG and 6-MP can reversibly inhibit SARS-279 CoV-1 and MERS-CoV papain-like cysteine proteases PL(pro) (53-55); however, whether these 280 thiopurines could inhibit viral replication was not assessed. Our observations of UPR activation 281 and selective inhibition of IAV HA and NA processing and accumulation by low micromolar doses 282 of 6-TG and 6TGo, but not 6-MP, suggest a distinct antiviral mechanism of action for these 283 thiopurines. If true, the antiviral activity of 6-TG and 6-TGo may be broadly applicable to other 284 viruses with envelope glycoproteins like coronaviruses. To test this directly, we performed HCoV-285 with the strong effect on infectious virion production, we also observed significant, dose-294 dependent reductions in viral protein accumulation due to 6-TG and 6-TGo treatment in HCoV-295 OC43-infected HCT-8 cells (Fig. 6C, the main band recognized by the anti-OC43 antibody is 296 consistent with the size of the nucleocapsid N protein). Treatment with higher dose of 6-MP caused 297 detectable decline in N protein levels in HCoV-OC43-infected cells, but not to the levels observed 298 with 6-TG or 6-TGo treatment (Fig. 6C). In HCoV-OC43-infected HCT-8 cells, BiP and CHOP 299 expression was upregulated following thiopurine treatment (Fig. 6C). This is consistent with the 300 significant induction of BiP proten and CHOP mRNA levels that was observed in thiopurine-301 treated A549 cells (Fig. 4C). To test the effects of 6-TG on CoV replication, we analysed HCoV-302 OC43 mRNA synthesis by harvesting RNA from infected cells treated with 6-TG or vehicle 303 control. We observed that 6-TG treatment caused significant decreases in steady-state levels of (+) 304 genomic RNA (ORF1ab) as well as (+) subgenomic RNA (sgRNA) that encodes N (Fig. 6D). 305 Thus, despite the previously reported effects of 6-TG and 6-MP on HCoV cysteine protease 306 activity in vitro, we observed that 6-MP had only modest effects on HCoV-OC43 replication 307 whereas 6-TG and 6-TGo had clear antiviral effects similar to our previous observations of 308 inhibition of IAV replication. Thiopurine antiviral activity in our HCoV-OC43 infection assays 309 correlated with UPR activation and hampered viral genome synthesis and viral protein production. 310

6-TG inhibits SARS-CoV-2 Spike protein accumulation via UPR activation 311
Because 6-TG activates the UPR/ISR and inhibits the processing and accumulation of IAV 312 glycoproteins, we reasoned that coronavirus glycoproteins would be similarly affected by 6-TG 313 treatment. Due to the ongoing SARS-CoV-2 pandemic, numerous reagents and constructs have 314 been rapidly developed to study this virus, including expression plasmids. We therefore sought to 315 determine if SARS-CoV-2 S glycoprotein is sensitive to 6-TG in ectopic expression experiments. 316 S is first translated as full-length S0 proprotein, before cleavage to S1 and S2 domains by cellular 317 proprotein convertases like furin (56). We observed that the S protein co-localised with the ER 318 marker calnexin when expressed alone or co-expressed with M protein (Fig. 7A). M also caused 319 some of the S protein to accumulate in distinct regions of the cytoplasm proximal to, but not 320 overlapping with, calnexin-stained ER, which likely represents the ERGIC (Fig. 7A). Ectopic 321 expression of S led to accumulation of ~230 kDa full-length N-glycosylated S0 monomers; 322 detection of ~110 kDa S1 ectodomains demonstrated efficient S N-glycosylation and trimerization 323 in the ER and transport to the Golgi for furin cleavage (Fig. 7B). We observed that ectopic S 324 expression was sufficient to activate the UPR/ISR, as indicated by accumulation of BiP (Fig. 7B), 325 consistent with previous reports of SARS-CoV-1 S (26, 31). 6-TG causes a loss of membrane-326 bound S0 and S2, but spared the cleaved S1 subunit (Fig 7B). Co-expression of S with M altered 327 S processing leading to different accumulation of S1 protein species, possibly due to altered S 328 trafficking by M and retention at the ERGIC compartment (57). PNGase F treatment of the lysates 329 to remove N-glycosylations confirmed that M and 6-TG altered glycosylation of S, but did not 330 affect cleavage (Fig. 7C). Treatment with either the chemical chaperone 4PBA or DBeQ, a 331 selective chemical inhibitor of the p97 AAA-ATPase, led to partial restoration of S0 and S2 (Fig  332   7D). Together, these observations suggest that like IAV glycoproteins, SARS-CoV-2 S 333 glycoprotein is vulnerable to 6-TG mediated activation of the UPR/ISR and suggest a mechanism 334 involving accelerated turnover of membrane-anchored S0 and S2 proteins by ERAD. 335 336 DISCUSSION 337 Compared to current direct-acting antiviral drugs, effective host-targeting antivirals may 338 provide a higher barrier to the emergence of antiviral drug-resistant viruses. However, it remains 339 challenging to identify cellular pathways that can be targeted to disrupt viral replication without 340 causing adverse effects on bystander uninfected cells. Here, we report that two chemically similar 341 FDA-approved thiopurine analogues, 6-TG and 6-TGo, have broad antiviral effects that result 342 from activation of UPR and disruption of viral glycoprotein synthesis and maturation. Importantly, 343 our data demonstrate for the first time that 6-TG and 6-TGo are effective antivirals against 344 influenza virus and coronavirus and may be effective against other glycoprotein-containing 345 viruses. 6-TG is currently used in clinical settings to treat acute lymphoblastic leukemia and other 346 hematologic malignancies, with the main mechanism of action involving conversion into 347 thioguanine nucleotides and subsequent incorporation into cellular DNA, which preferentially kills 348 cycling cancer cells (58, 59). Furthermore, the active 6-TG metabolite, 6-thioguanosine 5′-349 triphosphate, was shown to inhibit small GTPase Rac1 (60), which is believed to be largely with minimal effects on other viral proteins suggests that UPR induction by 6-TG and 6-TGo is 357 the main antiviral mechanism. Indeed, chemical chaperones and the ISR inhibitor ISRIB partially 358 restored IAV HA, NA, and SARS-CoV-2 S protein accumulation in cells treated with 6-TG. 359 Inhibition of the ERAD pathway with DBeQ also restored accumulation of ER membrane-360 anchored subunits of SARS-CoV-2 S protein (uncleaved precursor S0 and cleaved S2) in 6-TG-361 treated cells. This indicates that the 6-TG-induced UPR causes both the phospho-eIF2a dependent 362 decrease in viral glycoprotein mRNA translation and the ERAD-mediated degradation of newly 363 synthesized ER-anchored proteins. In the case of IAV which replicates in the nucleus of infected 364 cells, depletion of viral envelope glycoproteins blocks infectious virion production but minimally 365 affects replication of viral nucleic acids. By contrast, synthesis of coronavirus genomic RNA is 366 inhibited by thiopurine-induced UPR. This reduction could be due to inhibition of PL(pro) activity 367 as previously suggested (53-55); however, we suggest that thiopurines likely inhibit viral 368 replication due to its occurrence on the multivesicular network generated from virus-rearranged 369 ER membranes which may be highly sensitive to UPR-induced alterations. Despite multiple mechanisms deployed by IAV to block SG formation, 6-TG and 6-TGo 382 treatment induced SGs in infected cells, which allowed us to identify these molecules in our image-383 based screen. We previously reported that in A549 cells, IAV inhibited SG formation triggered by 384 treatment with thapsigargin, a potent inducer of ER stress, with only 9% of infected cells forming 385 SGs compared to 35% of mock-infected cells (44). Thus, induction of SGs in approximately 10% 386 of infected cells by 6-TG and 6-TGo treatment appears consistent with our previous observations. 387 However, unlike thapsigargin, 6-TG and 6-TGo did not trigger SG formation in uninfected cells. 388 The levels of UPR induction by these drugs were similar between infected and uninfected cells, 389 highlighting that in IAV-infected cells SG formation may not be triggered exclusively by ER stress 390 and PERK activation and may only partially contribute to antiviral effects of thiopurines. Indeed, 391 SGs formed in a fraction of infected cells while accumulation of viral glycoproteins HA and NA 392 was nearly completely blocked by 6-TG and 6-TGo. 393 Consistent with previous reports, ectopic expression of SARS-CoV-2 S protein was 394 enhanced during M co-expression and S alone was sufficient to trigger ER stress. In this system, 395 6-TG treatment further potentiated UPR responses, as measured by increased BiP accumulation. 396 Our data also highlight the sensitivity of membrane anchored viral proteins to 6-TG treatment. It 397 is currently unknown if host glycoproteins will be similarly affected by 6-TG treatment, but this 398 is an important question to answer due to the prevalent use of thiopurines clinically. While we 399 suspect that 6-TG and 6-TGo will be effective against a wide-range of enveloped viruses, our 400 future studies will investigate if SARS-CoV-2 replication can be negatively impacted following What is the mechanism of UPR induction by 6-TG and 6-TGo? Our results suggest that 406 the effects are unlikely to be mediated through DNA or RNA incorporation of 6-TG because 1) 407 replicative stress does not specifically induce UPR; 2) among viral proteins, glycoprotein 408 accumulation and processing was preferentially disrupted; 3) messenger RNA levels of HA and 409 NA were not affected. Furthermore, the closely related thiopurine 6-MP that can be converted 410 into 6-thioguanosine triphosphate and incorporated into nucleic acids did not induce UPR and 411 had no effect on IAV glycoproteins or OC43 replication. Another nucleoside analogue, 5-FU, 412 that is also incorporated into nucleic acids and can even trigger SG formation upon prolonged 413 48-hour incubation (62), was similarly inactive in our assays. The second previously described 414 antiviral mechanism of action of 6-TG and 6-MP that involves direct inhibition of viral cysteine 415 proteases is similarly unlikely to have major contribution to the observed phenotypes because 416 UPR induction was triggered in both infected and uninfected cells and because, as mentioned 417 above, 6-MP was not active in our assays. Thus, by process of elimination, we speculate that the 418 mechanism of UPR induction by 6-TG and 6-TGo could involve GTPase inhibition. Numerous 419 GTPases regulate ER homeostasis, including Rab GTPases that govern vesicular trafficking 420 events and dynamin-like GTPases that regulate homotypic ER membrane fusion events required 421 for the maintenance of branched tubular networks (63). Future studies will focus on identifying 422 specific molecular targets of these UPR-inducing thiopurines using orthogonal biochemical and 423 genetic screens. 424

RT-qPCR target
Primer sequences (5'-3') 5s rRNA hpi, cells were treated with 0, 1, 10 and 30 uM doses of thiopurine analogs 6-thioguanine (6-TG) or 6-thioguanosine (6-TGo). At 8 hpi, cells were fixed and stained with Hoeschst 33342. Automated image capture was performed using a Cellomics Arrayscan VTI HCS reader. 15 images were captured for each well and average punctate EGFP-G3BP1 intensity was calculated. (C) A549 cells were infected with IAV-CA/07 at a MOI of 1. At 4 hpi, cells were treated with 6-TG or mock-treated. At 8 hpi, cells were fixed and immunostained with antibodies directed to stress granule marker proteins G3BP1 (red), PABP (green) and a polyclonal IAV antibody (blue) that detects antigens from NP, M1, and HA, followed by staining with Alexa-conjugated secondary antibodies. (D) A549 cells were infected with IAV-CA/07 at a MOI of 1. At 4 hpi, cells were treated with 6-TG (10µM). At 8 hpi, cells were fixed and immunostained with antibodies directed to stress granule marker proteins G3BP1 (red), TIAR (green) and eIF3A (green), followed by staining with Alexa-conjugated secondary antibodies. Images captured on a Zeiss Axioimager Z2 fluorescent microscope. Representative images shown. Scale bars represents 20 µm. A549 cells were treated with 6-thioguanine (6-TG), 6-thioguanosine (6-TGo), 6-mercaptopurine (6-MP), 5-fluorouracil (5-FU) or ribavirin at the indicated concentrations for 6 h (XBP1s) or 24 h (BiP) prior to harvesting lysates for immunoblotting. 5 µg/ml tunicamycin (TM) served as positive control for UPR activation, whereas DMSO was mock treatment. Membranes were probed with anti-BiP and anti-XBP1s antibodies to measure UPR activation. -actin served as a loading control. (B) A549 cells were mock-infected or infected with IAV-PR8 at MOI of 1. After 1 h, cells were washed and incubated with 20 µM 6-TG or vehicle control, with or without 10 mM 4PBA, a chemical chaperone. At 20 hpi, cell lysates were harvested and probed with antibodies for the indicated target proteins. Western blots are representative of 3 independent experiments. (C) A549 cells were infected with IAV-PR8 at MOI of 1, washed and overlaid with media containing 6-MP, 6-TG, or TM. Cell lysates were collected at 24 hpi and RNA was isolated and processed for RT-qPCR. Changes in CHOP, BiP, EDEM1, ERdj4, and HERPUD1 mRNA levels were calculated by the ΔΔCt method and normalized using 18S rRNA as a reference gene and standardized to mock. Error bars represent the standard deviation between biological replicates (N=3); Circles represent biological replicates; Lines represents the average value. Statistical significance was calculated via a two-way ANOVA followed by a Dunnett multiple comparisons test.