key: cord-334624-chnibsa1
authors: Hayn, Manuel; Hirschenberger, Maximilian; Koepke, Lennart; Straub, Jan H; Nchioua, Rayhane; Christensen, Maria H; Klute, Susanne; Bozzo, Caterina Prelli; Aftab, Wasim; Zech, Fabian; Conzelmann, Carina; Müller, Janis A; Badarinarayan, Smitha Srinivasachar; Stürzel, Christina M; Forne, Ignasi; Stenger, Steffen; Conzelmann, Karl-Klaus; Münch, Jan; Sauter, Daniel; Schmidt, Florian I; Imhof, Axel; Kirchhoff, Frank; Sparrer, Konstantin MJ
title: Imperfect innate immune antagonism renders SARS-CoV-2 vulnerable towards IFN-γ and -λ
date: 2020-10-30
journal: bioRxiv
DOI: 10.1101/2020.10.15.340612
sha: 
doc_id: 334624
cord_uid: chnibsa1

The innate immune system constitutes a powerful barrier against viral infections. However, it may fail because successful emerging pathogens, like SARS-CoV-2, evolved strategies to counteract it. Here, we systematically assessed the impact of 29 SARS-CoV-2 proteins on viral sensing, type I, II and III interferon (IFN) signaling, autophagy and inflammasome formation. Mechanistic analyses show that autophagy and type I IFN responses are effectively counteracted at different levels. For example, Nsp14 induces loss of the IFN receptor, whereas ORF3a disturbs autophagy at the Golgi/endosome interface. Comparative analyses revealed that antagonism of type I IFN and autophagy is largely conserved, except that SARS-CoV-1 Nsp15 is more potent in counteracting type I IFN than its SARS-CoV-2 ortholog. Altogether, however, SARS-CoV-2 counteracts type I IFN responses and autophagy much more efficiently than type II and III IFN signaling. Consequently, the virus is relatively resistant against exogenous IFN-α/β and autophagy modulation but remains highly vulnerable towards IFN-γ and -λ treatment. In combination, IFN-γ and -λ act synergistically, and drastically reduce SARS-CoV-2 replication at exceedingly low doses. Our results identify ineffective type I and II antagonism as weakness of SARS-CoV-2 that may allow to devise safe and effective anti-viral therapies based on targeted innate immune activation.

1α) induce genes containing NF-κB sites in the promotor. Signaling of type I IFNs (IFN-α and IFN-β), 132

Type II IFN (IFN-γ), type III IFN (IFN-λ1) and pro-inflammatory cytokine signaling (TNFα and IL-133 1α) was quantified using quantitative firefly luciferase reporters controlled by the respective promotors 134 (Fig. 1c) . Stimulation with IFN-α2 and IFN-β (Fig. 1c) revealed that activation of the ISRE promotor 135 is strongly repressed by Nsp1, Nsp5, Nsp13, Nsp14, ORF6 and ORF7b. A similar set of viral proteins 136 interfered with type II IFN-γ and type III IFN-λ1 signaling, albeit much weaker (mean inhibition 18% 137 and 35%, respectively) compared to type I IFN signaling (mean inhibition 78% for IFN-α2 and 53% 138 for IFN-β). Activation of NF-κB signaling by TNFα or IL-1α was potently inhibited by the SARS-CoV-139 2 Nsp1, Nsp5, Nsp15, ORF3a, E, M, ORF6 and ORF7b proteins. These analyses revealed that a similar 140 set of proteins (Nsp1, Nsp5, Nsp15, ORF3a, E, M, ORF6 and ORF7b) antagonizes pro-inflammatory 141 cytokine signaling. 142

Since induction of autophagy does not depend on de novo gene expression 41 , we monitored autophagy 143 levels in SARS-CoV-2 protein expressing HEK293T cells by membrane-association of stably expressed 144 GFP-LC3B, a hallmark of autophagy induction (Fig. 1d, Supplementary Fig. 1e ) 42 . Autophagosome 145 numbers under basal conditions were strongly increased in the presence of ORF3a, E, M and ORF7a 146 suggesting either de novo induction of autophagy or blockage of turnover (Fig. 1d ). Upon induction of 147 autophagy using Rapamycin, a similar pattern was observed. To clarify whether these viral proteins 148 induce autophagy or block turnover, leading to accumulation of GFP-LC3B positive vesicles, we treated 149 cells with saturating amounts of Bafilomycin A1, which inhibits autophagic turnover. The increase of 150 autophagosome numbers by ORF3a, E, M and ORF7a was drastically reduced compared to non-151 blocking conditions (Fig. 1d) , indicating that these proteins block turnover, rather than induce it. 152

Blockage of autophagy and co-expression of Nsp1 and Nsp14 induced cell death, which may be 153 responsible for the low number of autophagosomes. Unexpectedly, in the presence of Nsp15 154 autophagosome numbers were consistently reduced, suggesting that it inhibits autophagy (Fig. 1d) . 155

Inflammasome responses were analyzed in stable THP-1 cell lines expressing SARS-CoV-2 proteins 156 upon doxycycline induction. To avoid any effects of transcription, assembly of ASC specks was 157 needle protein MxiH using the anthrax toxin delivery system ( Fig. 1 e) 43 . ASC speck assembly is 159 typically followed by caspase-1 activation and release of pro-inflammatory 44 . 160

Expression of the SARS-CoV-2 Nsp8, Nsp11 and ORF9c very weakly induced inflammasome activity 161 in the absence of inflammasome activators, although counterselection against cells prone to aberrant 162 inflammasome activation during selection cannot be ruled out. Activation of NLRC4 inflammasomes 163 was not significantly antagonized by any viral protein. 164

Taken together, our analysis reveals that SARS-CoV-2 encodes multiple proteins that strongly 165 antagonize innate immunity. Notably, there are differences in overall inhibition of the pathways with 166 IFN-γ, IFN-λ1 as well as inflammasome activity signaling being only weakly antagonized. However, 167 type-I IFN induction and signaling and autophagy are strongly repressed. 168

To analyses mechanistically why type-I IFN and autophagy are potently counteracted by SARS-CoV-170 2, we aimed at identifying the steps that are targeted in these pathways. We focused on the top 5 171 inhibitors as identified in Fig. 1b -d. Nsp1 was removed from the analysis as it prevents translation in 172 general 30 . To analyses IFN-β signaling, we monitored the levels of the type I IFN receptor, IFNAR 173 using western blotting in HEK293T cells overexpressing Nsp5, Nsp13, Nsp14, ORF6 or ORF7b. 174 Activation of the two major transcription factors of type I IFN signaling, STAT1 and STAT2 (Fig. 2a ) 175 was examined by phosphorylation status. Basal STAT1 and STAT2 levels were not significantly 176 affected by all proteins tested (Fig. 2b , quantification in Supplementary Fig. 2a-c) . (Fig. 2b ). In the 177 presence of Nsp5, activated STAT1 and to a lesser extend STAT2 accumulate ( Fig. 2b and 2d , 178 Supplementary Fig. 2a ). ORF6 and ORF7b did not affect IFNAR levels or STAT1 expression or 179 activation ( Fig. 2b-d) . This agrees with recent reports 26, 45, 46 suggesting that ORF6 instead prevents 180 trafficking of transcription factors. In the presence of Nsp14 and to a lesser extend for Nsp13 181 endogenous levels of IFNAR is prominently reduced (Fig. 2b, c) . Consequently, phosphorylation of 182 STAT1 was decreased upon Nsp14 co-expression (Fig. 2b, d) .

lipidated (LC3B-II) to decorate autophagosomal membranes 41, 42 . Upon fusion of autophagosomes with 185 lysosomes, the autophagic receptor p62 is degraded (autophagy turnover, Fig. 2e ). We analyzed the 186 effect of the top 5 autophagy modulating SARS-CoV-2 proteins: Nsp15, ORF3a, E, M and ORF7a (Fig.  187 1d) on autophagy markers. Levels of Beclin-1 and ULK1, which parts of the core machinery of 188 autophagy initiation remained constant (Fig. 2f, Supplementary Fig. 2d and 2e) . Overexpression of 189

Nsp15 leads to a very slight decrease of LCB3-II but accumulation of p62, suggesting that Nsp15 blocks 190 induction of autophagy . In line with this, the number of GFP-LC3B-puncta 191

(=autophagosomes) per cell in HeLa-GFP-LC3B cells is reduced upon Nsp15 expression to almost 0 192 ( Fig. 2i, j) . In the presence of ORF3a, E and ORF7a, the levels of processed LC3B (LC3B-II) were 4-193 to 7-fold increased (Fig. 2g) , and p62 levels are approximately 1.5-fold increased (Fig. 2h ). This 194

indicates that these three viral proteins block autophagic turnover. Consequently, the number of 195 autophagosomes is 10-fold increased upon ORF3a, E, M or ORF7a expression (Fig. 2i, j) . Curiously, 196 while accumulation of LC3B-II indicates that M blocks autophagic turnover or induces autophagy, the 197 levels of p62 are not significantly altered in the presence of M (Fig. 2f, h) . Notably, overexpression of 198 M resulted in an accumulation of LC3B in the perinuclear space, whereas for all other viral proteins 199 autophagosomes are normally distributed (Fig. 2i, j) . 200

Taken together, our data demonstrates that SARS-CoV-2 synergistically targets type-IFN signaling and 201 autophagy. The major type I IFN antagonists Nsp5, Nsp13, Nsp14, ORF6 or ORF7b block the signaling 202 cascade at different levels. E, ORF3a and ORF7a use similar mechanism to block autophagic turnover, 203 while M may have evolved a different mechanism and Nsp15 inhibits de novo autophagy induction. 204

Our data showed that ORF3a and ORF7a are the most potent autophagy antagonists of SARS-CoV-2 206 ( Fig. 1d, Fig. 2f-j) . To determine their molecular mechanism(s), we performed proteome analysis of 207

HEK293T cells overexpressing SARS-CoV-2 ORF3a and ORF7a ( Supplementary Fig. 3a) . As a 208 control, we used S, Nsp1 and Nsp16 overexpressing cells which show little to no effect on autophagy 209 (Fig. 1d ). In addition, we analyzed the proteome of Caco-2 cells infected with SARS-CoV-2 for 24 or 210 Table 1) . Analysis of the data revealed that in the presence of 212 Nsp1, cellular proteins with a short half-life are markedly reduced (Supplementary Fig. 3f) 47 . This 213 supports our previous finding that Nsp1 globally blocks translation 30 and confirming the validity of the 214 proteome analysis. PANTHER-assisted Gene Ontology Analysis of the proteins regulated more than 4-215 fold by the overexpression of individual SARS-CoV-2 proteins revealed that ORF3a and ORF7a target 216 the late endosome pathway (GO:0005770) (Fig. 3c, Supplementary Table 2) . A similar analysis for the 217 SARS-CoV-2 samples showed that the late endosome pathway is also affected during the genuine 218 infection. Thus, we had a closer look at the subcellular localization of ORF3a and ORF7a and their 219 effect on intracellular vesicles. In line with the proteome analysis, ORF7a and ORF3a both localized to 220 the late endosomal compartment, co-localizing with the marker Rab9 (Fig. 3d ,e). In contrast, 221 localization to Rab5a-positive early endosomes was not apparent ( Supplementary Fig. 3g ). Disturbance 222 of the integrity of the trans-Golgi network (TGN) at the interface with the late endosomes 48,49 by viral 223 proteins is a well-known strategy to block autophagy 50 . Immunofluorescence analysis revealed that the 224 localization of ORF3a or ORF7a partially overlap with a TGN marker (R = 0.5, Fig. 3g) indicating 225 close proximity. ORF6, which is known to localize to the Golgi apparatus 45 was used a positive control 226 (R=0.7). Nsp8, which displayed a cytoplasmic localization was used as a negative control (R=0.3). 227

Importantly, analysis of free TGN-marker positive vesicles in SARS-CoV-2 ORF3a or ORF7a 228 expressing cells revealed that both viral proteins cause significant fragmentation of the TGN (Fig. 3f,  229 h). 230

These data indicate, that both ORF3a and ORF7a disturb the proteome at the late endosomes eventually 231 causing the TGN to fragment, which leads to a block of autophagic turnover [49] [50] [51] [52] . 232

To examine the conservation of innate immune antagonism, we functionally compared Nsp1, Nsp3, 234

Nsp7, Nsp15, M, N, ORF3a, ORF6 and ORF7a of SARS-CoV-2, the closest related CoV, CoV and the previous highly pathogenic SARS-CoV-1. RaTG13-CoV was isolated from the 236 Measles virus V protein and TRIM32 expression served as positive controls. Overall, proteins of SARS-240

CoV-1 and RaTG13 behave similar to their SARS-CoV-2 counterparts, suggesting that many functions 241 are conserved. Importantly, however, this is not the case for Nsp15, Nsp3 and to a lesser extend ORF6 242 ( Fig. 4a-c) . SARS-CoV-1 ORF6 is about 4-fold less potent in antagonizing type I IFN signaling (Fig.  243 4b) but induces higher levels of autophagy (Fig. 4c) . However, expression levels of SARS-CoV-1 ORF6 244

were also higher than that of its SARS-VoV-2 and RaTG13 counterparts ( Supplementary Fig. 4g) , 245 which may explain the differences in activity. Differences between SARS-CoV, RaTG13 and SARS-246

CoV-2 Nsp3 were reanalyzed in a dose-dependent manner, and only in the range of 2-3-fold which may 247 also explained by differential expression (Supplementary Fig. 4j) . 248

The most striking, statistically significant difference was observed for Nsp15. SARS-CoV-1 Nsp15 is 249 over 10-fold more potent in suppression of type I IFN induction and signaling than RaTG13 and SARS-250

CoV-1 Nsp15 (Fig. 4a, b) . Notably, expression levels of SARS-CoV-2, RaTG13 and SARS-CoV-1 251

Nsp15 are similar, with SARS-CoV-1 Nsp15 even being slightly less expressed ( Supplementary Fig.  252 4c). Notably, all Nsp15 variants still inhibit autophagy (Fig. 4c) . Dose-dependent effect of SARS-CoV-253 2 Nsp15, RaTG13-CoV Nsp15 and SARS-CoV-1 Nsp15 on type I IFN induction (Fig. 4d ) and signaling 254 ( Fig. 4e) showed that on average SARS-CoV2 Nsp15 performed 32-fold worse than SARS-CoV-1 255 Nsp15, and RaTG13 Nsp15 inhibited type I IFN induction 7.8-fold less (Fig. 4d) . Similarly, SARS-256

CoV-1 Nsp15 outperformed RaTG13 and SARS-CoV-2 Nsp15 by 15-and 5.7-fold, respectively, in 257 inhibition of type I IFN signaling (Fig. 4e) . 258

Taken together, this data indicates, that while most IFN antagonist activities are conserved between 259 SARS-CoV, RaTG13 and SARS-CoV-2, there is an exception: Nsp15 of SARS-CoV-1 is considerably 260 more potent than SARS-CoV-2 Nsp15 in counteracting both IFN-β induction and signaling. 261

Inefficient antagonism by SARS-CoV-2 proteins is predictive for efficient immune control immune activation , albeit with different efficiency. The mean inhibition of IFN-γ and IFN-264 λ1 signaling was 18% and 35%, respectively, compared to type I IFN signaling with a mean inhibition 265 of only 78% for IFN-α2 and 53% for IFN-β. Consequently, we assessed whether IFN-α2, IFN-β, IFN-266 γ and IFN-λ1 have a different impact on SARS-CoV-2 (Fig. 5a, Supplementary Fig. 5a , b). Treatment 267 with the type I IFN-α2 was the least efficient. In contrast, at the same concentration IFN-γ (500 U/mL) 268 reduced viral RNA in the supernatant almost 300-fold more efficiently. All agents caused little if any 269 cytotoxic effects ( Supplementary Fig. 5c ). Altogether, we observed a good correlation (r= 0.89) 270 between average inhibition of the respective signaling pathway (Fig. 1c) antagonized by the 29 SARS-271

CoV-2 proteins and IFN susceptibility at 5 U/mL (Fig. 5b) . 272

In contrast to type II and II IFN signaling, autophagic turnover was strongly repressed by at least four 273 SARS-CoV-2 proteins ( Fig. 1c and Fig. 2) . Thus, based on our inhibition data ( Fig. 1c) we would expect 274 that modulation of autophagy only weakly affects SARS-CoV-2 replication. Indeed, treatment with 275

Rapamycin, which induces autophagy, reduced viral replication to a maximum of 4-6-fold 276 ( Supplementary Fig. 5e ). Bafilomycin A1, which blocks autophagy, had little to no effects 277 ( Supplementary Fig. 5e ). Both drugs were used at concentrations that only marginally affected cell 278 survival ( Supplementary Fig. 5f ). 279

Thus, our results indicate that the overall efficiency of SARS-CoV-2 proteins in counteracting specific 280 signaling pathway is predictive for the overall antiviral potency of the pathway, as illustrated by 281 different types of IFNs. 282

IFN therapy is commonly associated with significant adverse effects, due to induction of inflammation. 284

To minimize detrimental pro-inflammatory effects of IFNs, doses required for efficient viral restriction 285 should be reduced. Thus, we analyzed the impact of the most potent IFNs, IFN-γ and IFN-λ1 and their 286 combination of SARS-CoV-2. To mimic prophylactic and therapeutic treatment we examined pre-287 treatment for 24 h before infection with SARS-CoV-2 and treatment 6 h post-infection. Overall, the treatment but consistent (Fig. 5c, d) . Expression analysis of SARS-CoV-2 S and N confirmed the qPCR 290 results, and equal GAPDH levels exclude effects on viral replication by cytotoxicity (Fig. 5d ). While 291 treatment with a single dose of IFN-γ and IFN-λ1 alone reduced viral RNA production 50-100-fold, the 292 combinatorial treatment at the same concentration potentiated the effect to about 1000-fold reduction 293 in SARS-CoV-2 RNA (Fig. 5c) . 294

To further decrease inflammatory side-effects by IFN treatment, anti-inflammatory pathways like 295 autophagy could be induced 53-55 . Treatment with Rapamycin, which induces autophagy, already reduces 296 viral replication ~ 4-6-fold at 125nM (Fig. 5c ). Treatment of Rapamycin (125nM) in combination with 297 either IFN-γ or IFN-λ1 was found to be additive (Fig. 5c, d) . Triple treatment with IFN-γ, IFN-λ1 and 298

Rapamycin showed the most potent anti-viral effect of all combinations for pre-treatment and post-299 treatment, reducing viral RNA in the supernatant by 2100-fold and 85-fold, respectively (Fig. 5c) . 300

In summary, our data shows that the anti-SARS-CoV-2 effect of combinatorial treatments of IFN-γ, 301 IFN-λ1 are synergistic. Additional activation of anti-inflammatory autophagy by Rapamycin further 302 decreased SARS-CoV-2 replication. This suggests that concerted activation of innate immunity may be 303 an effective anti-viral approach, exploiting vulnerabilities of SARS-CoV-2 revealed by analysis of its 304 innate immune antagonism. 305

Viruses drastically alter our innate immune defenses to establish an infection and propagate to the next 307 host 13, 14, 21, 27, 45, 56 . Our data reveal the extend of immune manipulation SARS-CoV-2 employs. We 308 determined the major antagonists of type I IFN induction and signaling as well as pro-inflammatory 309 NF-κB activity encoded by SARS-CoV-2 (Nsp1, Nsp5, Nsp13, Nsp14, ORF6 and ORF7b). Type II and 310 III IFN signaling is targeted by a similar set of proteins, although much less efficient. Autophagy is 311 majorly targeted by Nsp15, ORF3a, E, M and ORF7a. Inflammasome activity is very weakly induced 312 by Nsp8, Nsp11 and ORF9c, but none of the SARS-CoV-2 proteins block formation of the NLR4C 313 inflammasome. Subsequent mechanistic studies revealed that SARS-CoV-2 proteins synergistically Nsp14 lowers the cellular levels of the IFN receptor, IFNAR, thus blocking activation of the crucial 316 transcription factors STAT1 and STAT2. Both ORF3a and ORF7a cause fragmentation of the TGN via 317 disturbing the late endosomal pathway. This is a common strategy of viruses to block autophagic 318 turnover. Examination of the functional conservation showed that SARS-CoV-2 Nsp15 was less 319 efficient in blocking innate immune activation, both type I IFN induction and signaling, than SARS-320

CoV-1 Nsp15. This may ultimately cause SARS-CoV-2 to be better controlled by the innate immune 321 system than SARS-CoV-1, explaining higher numbers of subclinical infections and thus overall lower 322 mortality rates of the current pandemic CoV. Overall, the combined analysis of IFN antagonism allowed 323 us to deduce that treatment with type-I IFNs and regulation of autophagy is only weakly anti-viral. In 324 contrast, treatment with IFN-γ and IFN-λ1 drastically reduced SARS-CoV-2 replication. Finally, 325 combinatorial treatment of SARS-CoV-2 with these two IFNs potentiated the effects of the individual 326 treatments. This may pave the way for future anti-viral therapies against SARS-CoV-2 based on rational 327 innate immune activation. 328

Why would multiple effective proteins target the same pathway? For example, type I IFN signaling 329 could have been shut down by Nsp1, Nsp5, Nsp13, Nsp14, ORF6 and ORF7b alone, each reducing the 330 activation of the innate immune pathways to below 10%. However, our assays revealed (Figs. 1-3) that 331 the targeting mechanisms are often not redundant and may act synergistically. This could allow the 332 virus to better control the targeted pathway, thus minimizing the effect of the signaling on its replication. 333

In addition, a viral protein majorly targeting one pathway may affect other connected immune pathways 334 at once. For example, disturbance of the kinase TBK1 activation may affect primarily IFN induction 335 and to a lesser extend also impact autophagy 57 . Proteome analyses revealed the late endosome/Golgi 336 network as a target of ORF3a and ORF7a. Our data suggests, that both ORF3a and ORF7a of SARS-337

CoV-2 cause fragmentation of Golgi apparatus and thus blockage of autophagy. SARS-CoV-1 ORF3a 338 was previously already implicated in Golgi fragmentation, thus our data suggests that SARS-CoV-2 339

ORF3a uses a similar strategy 51,58 . Notably, fragmentation of the Golgi is for example triggered by 340

Hepatitis C virus viruses to block anti-viral autophagic turnover 50 and thus may represent a common studies will see more mechanistic data to explain the molecular details of the impact of SARS-CoV-2 343 proteins on innate immune activation. Notably, several proteins including ORF6, ORF3a, ORF7a, M 344 and E accumulate at the Golgi network or in perinuclear spaces, alluding to the emerging role of the 345

Golgi as a hub for immune manipulation 52, 59 . 346

Our results demonstrate that ORF6, ORF3a, ORF7a and ORF7b are the strongest innate immune 347

antagonists among the accessory genes of SARS-CoV-2 (Fig. 1) . Besides the accessory genes, which 348 classically encode immune antagonists, a surprising number of non-structural proteins manipulate 349 innate immunity. Nsp1, which targets cellular translation and thus broadly inhibits any response enzymatic functions may impact their activity against innate immunity. Except for Nsp3, as its activity 357 as a de-ISGlase may inactivate the transcription factor IRF3 and thus reduce IFN induction 62 . According 358 to our analysis the structural proteins E and M strongly manipulated autophagy (Fig. 1d ). This suggests 359 that the incoming virion may already block autophagic turnover to prevent their own degradation by 360

However, while we may pick up most counteraction strategies, our screening approach may miss 362 immune evasion strategies employed by SARS-CoV-2. For example, many non-structural proteins form 363 complexes, that are not formed during single overexpression and may only be functional as a full 364 assembly. Evasion mechanisms based on RNA structures and sequences are lost due to usage of codon-365 optimized expression plasmids. Finally, the virus itself may employ strategies to hide itself from 366 recognition, thus not activating innate immune defenses in the first place. One example is the capping would be immediately recognized by the cytoplasmic sensor RIG-I. 369

Our analyses further revealed that the human innate immune antagonism is largely conserved in the 370 SARS-CoV-2 closest related bat isolate, RaTG13 (Fig. 4) . This indicates that the bat virus is capable of 371 counteracting the human immune defenses, which may have facilitated successful zoonotic 372 transmission from bat eventually to humans. Currently, the intermediate animal host of SARS-CoV-2 373 is under debate 3,68-70 , however it is likely, that the virus isolated from it is even closer related to SARS-374

CoV-2 than RATG13. Thus, any immune evasion mechanisms conserved between SARS-CoV-2 and 375 RATG13, is likely to be conserved in the direct progenitor virus of SARS-CoV-2. The previous 376 epidemic and related human SARS-CoV-1 and the current pandemic SARS-CoV-2 differ in 377 susceptibility towards IFN s with SARS-CoV-1 being more resistant 26 . Furthermore, infection with 378 SARS-CoV-2 is often asymptomatic and likely controlled by the host 26 as lower mortality rates and 379 higher subclinical infections suggest 4 . Paradoxically, this may support the fast spread of the virus. Thus, 380 SARS-CoV-2 may have found the 'perfect' balance. Intermediate immune evasion and thus 381 intermediate pathogenicity to support spread, but not kill the host. Our data shows that SARS-CoV-2 382

Nsp15 is strikingly less in efficient in IFN evasion than Nsp15 of SARS-CoV. These data are the first 383 mechanistic evidence why SARS-CoV-1 is less susceptible towards IFN treatment than SARS-CoV-2. 384

It may be tempting to speculate that common cold CoVs counteract the innate immune system less 385 efficiently than SARS-CoV-2. 386

Our analysis indicates that during a SARS-CoV-2 infection less cytokines than expected are released, 387 autophagic turnover is blocked and general immune activation is perturbed. This is supported by a large 388 amount of data from COVID19 patients [24] [25] [26] [27] [28] 45, 62, [71] [72] [73] . However, an important question remains: Why 389 are some innate immune pathways, such as IFN-γ signaling less antagonized (Fig. 1) ? Are the viral 390 immune manipulation strategies ineffective? Indeed, IFN-γ is most active against SARS-CoV-2 among 391 the IFNs 72 (Fig. 5) . One possible explanation would be that there was no need for the virus to antagonize 392 them. Indeed, in COVID19 patients and in vitro infections with SARS-CoV-2, IFN-γ levels are 393 surprisingly low 28,73 . Furthermore, despite high IFN-γ levels being a hallmark of cytokine storms IFN-γ expression in CD4+ T cells is associated with severe COVID19 4,74,75 . It is tempting to speculate 396

that T-cells which confer pre-existing immunity against SARS-CoV-2 76,77 could, upon activation, 397 release IFN-γ, whose innate immune signaling may also contribute to increased clearance of the 398 infection. Strikingly, our work thus shows that analysis of the innate antagonism may be predictive for 399 therapeutic opportunities. 400

Severe side effects are prevalent for treatments with IFNs 35-37 . However, the side-effects are dose-401 dependent 78 . Thus, minimizing the dose required for treatment is paramount. Our data indicates that 402 effects of treatment with multiple IFNs is additive but synergistic and potentiates each other (Fig. 5) . 403

Thus, a promising anti-viral approach may be a combinatorial treatment of different cytokines, 404 effectively also reducing the burden of side-effects. The side effects of IFN therapy are mainly caused 405 by inflammation. Combined with anti-inflammatory approaches such as autophagy activation by 406

Rapamycin 54,55 , this approach may even be more successful, as our in vitro data suggests. Future studies 407 are highly warranted to study rational, concerted innate immune activation against SARS-CoV-2 in 408 vivo. These studies may eventually pave the way for novel therapies, which may not only work against 409 SARS-CoV-2, but also against other pathogenic viruses, including potentially future CoVs. 410

In summary, our results reveal the extend of innate immune manipulation of SARS-CoV-2. Comparison 411 to SARS-CoV-1 revealed that mutations in Nsp15 may be responsible for the higher susceptibility of 412 SARS-CoV-2 against IFNs. Finally, our data allowed us to deduce a potent immune activation strategy 413 against SARS-CoV-2: combinatorial application of IFN-γ and IFN-λ. RaTG13, and SARS-CoV-1 were synthesized by Twist Bioscience and subcloned into the pCG 559 vector using restriction cloning using the restriction enzymes XbaI and MluI (New England 560 Biolabs). Firefly luciferase reporter constructs, harboring binding sites for NF-κB or IRF3, 561 ISRE or GAS sites, or the genomic promoter of IFNA4 or IFNB1 in front of the reporter were 

A pneumonia outbreak associated with a new coronavirus of probable bat 758 origin

A new coronavirus associated with human respiratory disease in China

The proximal 762 origin of SARS-CoV-2

Quantifying SARS-CoV-2 transmission suggests epidemic control with 766 digital contact tracing

SARS : epidemiology CUMULATIVE NUMBER OF CASES AND 770 DEATHS IN VARIOUS COUNTRIES IN

Middle East respiratory syndrome

Comparing SARS-CoV-2 with SARS-CoV and influenza pandemics

Hosts and Sources of Endemic Human 776

Advances in Virus Research

RIG-I in RNA virus recognition

SARS coronavirus pathogenesis: Host innate immune responses 780 and viral antagonism of interferon. Current Opinion in Virology

SARS coronavirus and innate immunity

Innate immunity to virus infection

The antiviral activities of TRIM proteins. Current 787 Opinion in Microbiology

Intracellular detection of viral nucleic acids. Current Opinion 789 in Microbiology

TRIM proteins and their roles in antiviral host 791 defenses

CoV-2 infection: Apossible smart targeting of the autophagy pathway

Involvement of autophagy in coronavirus replication

TRIM proteins: New players in virus-induced autophagy

Autophagy during viral infection -A double-edged 800 sword

Regulation of adaptive immunity by the innate immune system

Control of adaptive immunity by the innate immune system

Inborn errors of type I IFN immunity in patients with life-threatening COVID-806 19

Auto-antibodies against type I IFNs in patients with life-threatening

Type I interferon susceptibility distinguishes SARS-CoV-2 from 810

Multi-level proteomics reveals host-perturbation strategies of SARS-CoV-2 28

SARS-CoV-2 ORF3b Is a Potent Interferon Antagonist Whose Activity Is 816 Increased by a Naturally Occurring Elongation Variant

Structural basis for translational shutdown and immune evasion by the Nsp1 819 protein of SARS-CoV-2. Science (80-. ). (2020)

Activation and evasion of type I interferon responses by SARS-CoV-2. Nat

Evasion of Type I Interferon by SARS-CoV-2

SARS-CoV-2 ORF6 disrupts nucleocytoplasmic transport through 825 interactions with Rae1 and Nup98. bioRxiv (2020)

Review of trials currently testing treatment and prevention of COVID-19

The safety of pegylated interferon alpha-2b in the treatment of 829 chronic hepatitis B: Predictive factors for dose reduction and treatment discontinuation

A Multidisciplinary Therapeutic Approach for Reducing the Risk of Psychiatric 832

Side Effects in Patients With Chronic Hepatitis C Treated With Pegylated Interferon α and 833

Side effects of therapy of hepatitis C and their management

Triple combination of interferon beta-1b, lopinavir-ritonavir, and ribavirin 837 in the treatment of patients admitted to hospital with COVID-19: an open-label, randomised, 838 phase 2 trial

A SARS-CoV-2 protein interaction map reveals targets for drug 840 repurposing

ORF8 and ORF3b antibodies are accurate serological markers of early and 842 late SARS-CoV-2 infection

Guidelines for the use and interpretation of assays for monitoring 844 autophagy

An improved method for high-throughput quantification of autophagy in 846 mammalian cells

Type III secretion needle proteins induce cell signaling and cytokine 848 secretion via toll-like receptors

A single domain antibody fragment that recognizes the adaptor ASC 850 defines the role of ASC domains in inflammasome assembly

Activation and evasion of type I interferon responses by SARS-CoV-2. Nat

SARS-CoV-2 nsp13, nsp14, nsp15 and orf6 function as potent interferon 855 antagonists

Systematic analysis of protein turnover in primary cells

Organization of the ER-Golgi interface for membrane traffic 859 control

Bidirectional traffic between the Golgi and the endosomes -861 machineries and regulation

Hepatitis C virus triggers Golgi fragmentation and autophagy through the 863 immunity-related GTPase M

The Open Reading Frame 3a Protein of Severe Acute Respiratory 866

Syndrome-Associated Coronavirus Promotes Membrane Rearrangement and Cell Death

Golgi Apparatus: An Emerging Platform for Innate 869

Autophagy in immunity and inflammation

Viral unmasking of cellular 5S rRNA pseudogene transcripts induces RIG-I-877 mediated immunity article

TRIM23 mediates virus-induced autophagy via activation of TBK1

Severe acute respiratory syndrome coronavirus Orf3a protein interacts with 881 caveolin

Signaling organelles of the innate immune system

Nsp3 of coronaviruses: Structures and functions of a large 888 multi-domain protein

Papain-like protease regulates SARS-CoV-2 viral spread and innate immunity

Chimeric Exchange of Coronavirus nsp5 Proteases (3CLpro) Identifies 892

Common and Divergent Regulatory Determinants of Protease Activity

SARS-CoV 3CL protease cleaves its C-terminal autoprocessing site by 895 novel subsite cooperativity

Human Coronavirus 229E Nonstructural Protein 13: 898 Characterization of Duplex-Unwinding, Nucleoside Triphosphatase, and RNA

SARS-Coronavirus-2 Nsp13 Possesses NTPase and RNA Helicase Activities 901

That Can Be Inhibited by Bismuth Salts

Old" protein with a new story: Coronavirus endoribonuclease is 903 important for evading host antiviral defenses

Probable Pangolin Origin of SARS-CoV-2 Associated with the 905 COVID-19 Outbreak

Identifying SARS-CoV-2-related coronaviruses in Malayan pangolins

Interplay between SARS-CoV-2 and the 911

type I interferon response

The Zinc Finger Antiviral Protein restricts SARS-CoV-2. bioRxiv (2020)

Impaired type I interferon activity and inflammatory responses in severe 915 COVID-19 patients

The trinity of COVID-19: 917 immunity, inflammation and intervention

Clinical and immunological features of severe and moderate coronavirus 920 disease 2019

Pre-existing immunity to SARS-CoV-2: the knowns and unknowns

SARS-CoV-2-specific T cell immunity in cases of COVID-19 and SARS, 924 and uninfected controls

Side effects of 926 interferon-α therapy

Beta Interferon Transcription in the Nucleus

Measles Virus V Protein Is a Decoy Substrate for IκB 930

Kinase α and Prevents Toll-Like Receptor 7/9-Mediated Interferon Induction

Hiv-1 vpu is a potent transcriptional suppressor of nf-kb-elicited antiviral 933 immune responses

TRIM32 Senses and Restricts Influenza A Virus by Ubiquitination of PB1

IFITM proteins promote SARS-CoV-2 infection of human lung cells

Assessment of inflammasome formation by flow cytometry