SARS-CoV-2 Exploits Sexually Dimorphic and Adaptive IFN and TNFa Signaling to Gain Entry into Alveolar Epithelium

Infection of the alveolar epithelium constitutes a bottleneck in the progression of COVID-19 to SARS presumably due to the paucity of viral entry receptors in alveolar epithelial type 1 and 2 cells. We have found that the male alveolar epithelial cells express twice as many ACE2 and TMPRSS2 entry receptors as the female ones. Intriguingly, IFN and TNF-α signaling are preferentially active in male alveolar cells and induce binding of the cognate transcription factors to the promoters and lung-active enhancers of ACE2 and TMPRSS2. Cotreatment with IFN-I and III dramatically increases expression of the receptors and viral entry in alveolar epithelial cells. TNFα and IFN-II, typically overproduced during the cytokine storm, similarly collaborate to induce these events. Whereas JAK inhibitors suppress viral entry induced by IFN-I/III, simultaneous inhibition of IKK/NF-κB is necessary to block viral entry induced by TNFα and IFN-II. In addition to explaining the increased incidence of SARS in males, these findings indicate that SARS-Cov-2 hijacks epithelial immune signaling to promote infection of the alveolar epithelium and suggest that JAK inhibitors, singly and in combination with NF-KB inhibitors, may exhibit efficacy in preventing or treating COVID-19 SARS.

males as compared to females ( Fig. 2A and 2B). In contrast, male ciliated bronchial cells 167 expressed lower levels of both ACE2 and TMPRSS2 as compared to their female counterpart, 168 suggesting that the enrichment of viral entry receptors in the lung epithelium of males is specific 169 to the alveolar epithelial cells ( Fig. 2A and 2B). In consonance with this observation, further 170 analysis indicated that the average level of expression of ACE2 in individual AT1 and AT2 cells 171 and TMPRSS2 in AT1 cells is significantly higher in males as compared to females ( Fig. 2C and 172 2D). Although we observed a trend towards elevated expression of TMPRSS2 also in male AT2 173 cells, the difference was not statistically significant (Fig. 2D). Finally, drawing from these 174 differences in the percentage and level of expression of ACE2 and TMPRSS2, we found that the 175 alveolar epithelium of males contains approximately twice as many ACE2+ TMPRSS2+ double-176 positive AT1 and AT2 cells as compared to its female counterpart (Fig. 2E). In agreement with 177 prior observations, smokers possessed a larger number of ACE2+ TMPRSS2+ double-positive 178 AT2 cells as compared to non-smokers ( fig. S2B) (37). In addition, individuals aged 65 or more 179 also exhibited more double-positive AT1 and AT2 cells, but the results did not reach statistical 180 significance. These observations suggest that the alveolar epithelium of normal adult males 181 contains a significantly larger number of cells potentially sensitive to SARS-Cov-2 infection as 182 compared to its female counterpart. 183 To examine the cell-type specificity of the sexually dimorphic expression of viral entry 184 receptors, we analyzed available single-cell datasets from organs that can be infected by [38][39][40][41]. The results indicated that the expression of ACE2 and TMPRSS2 in epithelial 186 reflective of IFN signaling, including the HALLMARK_INTERFERON_ALPHA_RESPONSE 249 signature, indicated that they are enriched in male AT1 and AT2 cells as compared to their 250 female counterparts (Fig. 4A-D). Examination of the expression of several IFN signaling target 251 genes confirmed their differential expression in the distal lung epithelium in the two genders (Fig. 252 4E,4F,S4G,and S4H). Taken together, these findings suggest that sexually dimorphic IFN 253 signaling controls the expression of SARS-CoV-2 entry receptors in the alveolar epithelium. 254 255 Type I Interferon Signaling Upregulates the Expression of ACE2 and TMPRSS2 in 256

Pulmonary Alveolar Epithelial Cells 257
To directly test the hypothesis that ACE2 and TMPRSS2 are IFN target genes in lung epithelium, 258 we initially surveyed the expression of IFN receptors and JAK family kinases in primary human 259 lung alveolar epithelial cells (AEpiC). qPCR indicated that these cells express IFNAR1, IFNAR2, 260 IL-10Rb, IFNGR1, IFNGR2, JAK1, TYK2, and lower levels of IFNLR1 and JAK2, suggesting that 261 they can be stimulated by IFN-I (a, b), II (g) and III (l) (fig. S5A, S5B and S5C). Immunoblotting 262 indicated that 20 nM IFNa promotes phosphorylation and accumulation of STAT1 and 2 in AEpiC 263 cells, consistent with the finding that STAT1 and 2 are IFN target genes (47,48). Tofacitinib, 264 which preferentially inactivates JAK1 as compared to JAK2 and 3 and inhibits TYK2 much less 265 efficiently (49), reversed both the phosphorylation and the accumulation of STAT1 and 2 ( Fig.  266 5A). These results are consistent with the conclusion that IFNa-stimulated JAK-STAT signaling 267 proceeds via the IFNAR1/2 heterodimer and associated JAK1/TYK2 kinases in AEpiC cells. 268 To examine the capacity of IFN and androgen to regulate ACE2 and TMPRSS2 269 expression, we performed qPCR assays with LNCaP prostate adenocarcinoma cells, Calu-3 270 lung adenocarcinoma cells, and AEpiC cells, which were treated with IFNa (20 nM), androgen 271 (10 nM DHT), or a combination of the two. Of note, the AEpiC cells consist of AT2 and, to a 272 smaller extent, AT1 cells (50). The results indicated that DHT induces expression of TMPRSS2 273 within 16 hours in LNCaP cells, as anticipated (Fig. 5B and 5C,left panel). However, DHT did 274 not promote rapid expression of either ACE2 or TMPRSS2 in Calu-3 or AEpiC cells and 275 treatment with the potent AR inhibitor enzalutamide (5 µM) did not reduce their level of 276 expression in unstimulated cells ( Fig. 5B and 5C, middle and right panels, and S5D). Notably, 277 treatment with DHT did not enhance, but instead suppressed IFNa's induction of TMPRSS2 in 278 Calu-3 and AEpiC. In contrast, IFNa induced expression of both ACE2 and TMPRSS2 in these 279 two cell lines ( Fig. 5B and 5C, middle and right panels). These results provide direct evidence 280 that IFN signaling elevates the expression of SARS-CoV-2 entry receptors in lung alveolar 281 epithelium. In addition, the lack of response to enzalutamide in AEpiC cells indicates that 282 TMPRSS2 is not an AR target gene in alveolar epithelial cells. 283 To examine the mechanism through which IFN signaling induces expression of 284 TMPRSS2, we performed chromatin immunoprecipitation (ChIP)-qPCR assays using AEpiC 285 cells treated with either IFN-a, DHT, or the combination. The results revealed that activated 286 STAT1 (P-STAT1) binds to the promoter and enhancer 2 of TMPRSS2 in IFNa-but not DHT-287 stimulated cells (Fig. 5D). This event was accompanied by a decrease of the suppressive mark 288 H3K27me3, an increase of the activation mark H3K4me3, and a selective enrichment of 289 activated Pol II (S2P-Pol II) at the promoter of TMPRSS2 (Fig. 5E). Notably, treatment with DHT 290 decreased the binding of P-STAT1 to the promoter and enhancer 2 of TMPRSS2 and 291 transcriptional activation of the gene in response to IFN (Fig. 5D,5E and S5E) and ChIP-qPCR 292 assays indicated that this suppression did not involve direct binding of the AR to the lung-active 293 enhancers or the promoter of TMPRSS2 ( fig. S5F). Therefore, IFN stimulation induces binding 294 of activated STAT1 to the enhancer 2 and promoter of TMPRSS2 and transcriptional activation 295 of the gene. In contrast, DHT stimulation interferes with the IFN-stimulated expression of 296

TMPRSS2. 297
By performing similar ChIP-qPCR assays, we found that P-STAT1 binds also to the 298 promoter and intronic enhancer ACE2 in IFNa-stimulated AEpiC cells and induces de-repression 299 and transcriptional activation of the promoter ( Fig. 5F and 5G). Interestingly, treatment with DHT 300 decreased the binding of P-STAT1 to the promoter and enhancer of ACE2 and transcriptional 301 activation of the gene in response to IFN ( Fig. 5F and G) as we had observed for TMPRSS2. 302 However, ChIP-qPCR assays indicated that this suppression involved direct binding of the AR 303 to the lung-active intronic enhancer and the promoter of ACE2 (fig. S5G). These findings indicate 304 that IFN signaling promotes the binding of P-STAT1 to the promoter and lung-specific enhancers 305 of both TMPRSS2 and ACE2, corroborating the hypothesis that sexually dimorphic IFN signaling 306 leads to elevated levels of SARS-Cov-2 entry receptors in the male alveolar epithelium. In 307 contrast, androgen does not exert these effects in lung epithelium. 308 309 IFN-I and III Conspire to Upregulate the Expression of ACE2 and TMPRSS2 and Induce 310

Robust Viral Entry 311
To identify the types of interferon that upregulates ACE2 and TMPRSS2 expression, we 312 examined the effects of IFN-I (IFN⍺ and IFNβ), IFN-II (IFNɣ) and IFN-III (IFNλ) on the expression 313 of ACE2 and TMPRSS2 in AEpiC cells. qPCR indicated that IFNb and g induce higher expression 314 of ACE2 as compared to IFNa and l (Fig. 6A). In contrast, IFNl induced higher expression of 315 TMPRSS2 as compared to IFNb, and IFNa and g proved ineffective (Fig. 6B). 316 Immunofluorescent staining of non-permeabilized Calu-3 cells and qPCR analysis corroborated 317 the preferential upregulation of ACE2 by IFNa and TMPRSS2 by IFNl (Fig. 6C, S6A, and S6B). 318 Immunoblotting analysis revealed that IFNa and IFNl induce activation of JAK1 and TYK2 and, 319 downstream, phosphorylation of STAT1 and 2 and formation of the ISGF3 complex ( fig. S6C). 320 As anticipated by JAK2's exclusion from type I and type III IFN-Rs (51), IFNa and IFNl did not including IFNl or g did not stimulate expression of ACE2 more than either IFNa or IFNb alone 332 (about 25-fold over control for both IFNa or IFNb). Conversely, combinations of IFNl with IFNa 333 or IFNb and triple combinations including IFNg did not induce higher expression of TMPRSS2 334 as compared to IFNl alone (about 4-fold over control for IFNl) ( fig. S6E). In addition, treatment 335 with JAK inhibitors such as fedratinib, ruxolitinib and tofacitinib, but not prednisone 336 (corticosteroid), camostat or nafamostat (protease inhibitors) inhibited the IFNa dependent 337 induction of ACE2 and TMPRSS2 (Fig. 6D). Together with the results of ChIP studies, these 338 results provide strong evidence that ACE2 and TMPRSS2 are IFN target genes and that ACE2 339 is predominantly controlled by IFN-I and TMPRSS2 by IFN-III. 340 13 IFN-I and III cooperate to restrict viral infection of epithelial cells, including those lining 341 the airways of the lung. Newly infected cells produce IFNs in response to activation of pattern 342 recognition receptors, including cGAS-STING (51, 52). To examine if activation of such 343 pathways and, hence endogenous production of IFNs, can upregulate expression of ACE2 and 344 TMPRSS2 in alveolar epithelial cells, we initially transduced dsDNA or LPS into AEpiC cells. 345 qPCR revealed that dsDNA induces expression of both entry receptors, but LPS only 346 upregulates TMPRSS2 (Fig. 6E). Since LPS is recognized by the Toll-like Receptor 4 (TLR4), 347 which predominantly impinges on NF-kB signaling, the latter observation implies that TMPRSS2 348 may be induced also by NF-kB (53). To better model the effect of the viral RNA of SARS-CoV-349 2, we used Poly IC. As shown in Figure 6F, poly IC rapidly induced expression of ACE2 and 350 TMPRSS2, suggesting that initial viral entry stimulates expression of viral entry receptors, 351 potentially facilitating the entry of additional viruses. The expression of ACE2 and TMPRSS2 352 induced by Poly IC was suppressed by treatment of the cells with an inhibitor of TBK kinase 353 (GSK8612), which controls expression of IFNs via NF-kB (54), the JAK1/2 and TYK2 inhibitor 354 ruxolitinib (49) and the NF-kB inhibitor BAY-11-70-82 (55) (Fig. 6G). This pattern of inhibition is 355 consistent with the signaling mechanisms that enable cGAS-STING to induce the expression of 356 IFNs. Together, these observations suggest that SARS-CoV2 exploits the ability of its RNA to 357 induce the production of IFN to radically increase the expression of its entry receptors on alveolar 358

epithelial cells. 359
To determine if IFN can enhance the entry of SARS-CoV2 into alveolar epithelium, we 360 used replication-defective lentiviral particles bearing coronavirus S proteins. Pseudotyped viral 361 particles have been shown to enter into cells by binding to ACE2 and TMPRSS2 (9). Firstly, we 362 tested if knocking down ACE2 and TMPRSS2 could hinder the entry of SARS-2S pseudotyped 363 lentivirus in CALU3 cells. As show in Figure S6F and S6G, reduced expression of either one of 364 the receptors significantly reduced the viral entry. We then stimulated AEpiC cells with IFN⍺ for 365 16 hours and infected them with LVM-SARS-CoV-2_S, a SARS-2S pseudotyped lentivirus 366 encoding luciferase. Stimulation with IFNa increased viral entry by 3-fold, and pre-treatment with 367 the TMPRSS2 protease inhibitors camostat and nafamostat inhibited it, confirming the 368 dependency of this process on S protein binding and activation by TMPRSS2 (Fig. 6H). Pre-369 treatment with the JAK inhibitors tofacitinib and ruxolitinib also suppressed the entry of the virus 370 induced by IFN (Fig. 6H). To model the microenvironment of the early infection of alveolar 371 epithelium, we repeated the experiment by exposing the AEpiC cells to a mixture of IFNa and 372 IFNl. In agreement with its capacity to induce maximal expression of both ACE2 and TMPRSS2, 373 the combination promoted a 5-fold increase in viral entry, which was largely reversed by 374 ruxolitinib (Fig. 6I). These findings suggest that IFN-I and IFN-III can substantially upregulate the 375 expression of viral entry receptors during the early phase of infection of the alveolar epithelium 376 and that JAK inhibitors can interfere with this process. 377 378 TNF⍺ Models Severe SARS-COV2 Disease and Cytokine Storm in Pulmonary Alveolar 379

Epithelial Cells 380
To model the effect of cytokines present in the lung of COVID-19 patients affected by advanced 381 lung disease, we tested IFNg, TNFa, and additional cytokines, which have been found to 382 correlate with or participate in disease progression (56-59). Notably, TNFa induced expression 383 of TMPRSS2 but not ACE2, whereas IFNg induced expression of ACE2 but not TMPRSS2. The 384 two cytokines in combination did not exert a higher effect as compared to either one singly (Fig. 385 7A and S7A). None of the 14 additional cytokines tested induced a significant increase in the 386 expression of ACE2 or TMPRSS2 ( fig. S7B). Whereas the IFNg receptor signals via JAK1/2 and 387 STAT3, the TNFa receptor signals predominantly through activation of 60). 388 Accordingly, ChIP experiments revealed that IFNɣ promotes the binding of p-STAT3 to the 389 enhancer and promoter of ACE2 but not of TMPRSS2 ( Fig. 7B and S7C). In contrast, TNF⍺-390 promotes binding of the NF-kB-p65 complex to the enhancer and promoter of TMPRSS2 but not 391 of ACE2 ( Fig. 7C and S7C). These findings indicate that at concentrations inferior to those 392 inducing inflammatory cell death (56), TNF⍺ and IFNg induce a robust upregulation of SARS-393 CoV-2 entry receptors in the alveolar epithelium. 394 We next asked if inhibition of JAK-STAT and NF-kB signaling would suppress the 395 expression of ACE2 and TMPRSS2 induced by a combination of IFNg and TNFa. Notably, BAY-396 11-7082 suppressed the expression of ACE2 induced by IFNg or IFNg and TNFa, but the JAK1 397 inhibitor tofacitinib did not (Fig. 7D). While the inhibition of the effect of IFNg by BAY-11-7082 398 may arise from the crosstalk between IRF, NF-kB, and JAK-STAT pathways (61), the inability of 399 tofacitinib to block the effect of IFNg was unexpected. We therefore tested optimal concentrations 400 of tofacitinib (JAK1i), fedratinib (JAK2i), and ruxolitinib (JAK1/2i) and found that only the latter 401 two compounds suppress the upregulation of ACE2 induced by IFNg (Fig. 7E). These results 402 suggest that the upregulation of ACE2 induced by IFNg depends more on JAK2 than JAK1 under 403 our experimental conditions. As anticipated, BAY-11-7082 suppressed the expression of 404 TMPRSS2 induced by TNFa and tofacitinib inhibited that induced by IFNg (Fig. 7D). These 405 results suggest that combined inhibition of JAK-STAT and NF-kB signaling may inhibit the 406 expression of SARS-CoV-2 entry receptors induced by TNFa and IFNg during the cytokine storm 407 that typifies advanced disease. 408 We finally reasoned that IFN⍺ and λ would continue to be produced during advanced 409 disease as long as new alveolar epithelial cells are infected. We therefore asked if a combination 410 of IFN⍺ and λ and TNFa could promote a simultaneous and substantial upregulation of both 411 ACE2 and TMPRSS2. Notably, we found that the triple combination induces higher expression 412 of both TMPRSS2 and ACE2 as compared to all double combinations or each cytokine alone 413 In this study, we provide evidence that sexually dimorphic IFN and, possibly, NF-kB signaling 424 upregulates expression of the viral entry receptors ACE2 and TMPRSS2 in male alveolar type I 425 and II cells, potentially explaining why progression to SARS occurs much more frequently in this 426 gender. Examination of primary alveolar epithelial cells indicates that IFN-I and III, which are 427 found that ACE2 is expressed in only a very small proportion of AT1 and, as previously reported 442 (11, 15), AT2 cells (<1%). In contrast, TMPRSS2 is expressed in a large fraction of both cell 443 types (ca. 36%). Since approximately 50% of ACE2+ cells also express TMPRSS2, double-444 positive AT1 and AT2 cells are extremely rare in the normal lung (0.22% and 0.44%, 445 respectively). We thus posited that additional signaling mechanisms upregulate the expression 446 of ACE2 and TMPRSS2 in these cells and thereby facilitate initial viral entry as well as the 447 progression of the infection, especially in the face of protective innate immunity mechanisms. 448 The mechanisms underlying the male prevalence of SARS are poorly understood. 449 Intriguingly, we found that the expression of ACE2 and TMPRSS2 is significantly higher in 450 normal alveolar epithelial cells from males as compared to females. Previous studies have 451 shown that women mount more robust immune responses against viruses and vaccines and 452 exhibit superior immune-mediated tissue repair as compared to males (33). In addition, clinical 453 studies have shown that male patients with moderate COVID-19 have defective T cell responses 454 that correlate with disease severity (30). While it is likely that differences in immune responses 455 between the genders contribute to the higher disease severity in males, our finding that males 456 possess about twice as many ACE2+ and TMPRSS2+ AT1 and AT2 cells as compared to 457 females suggests that the male alveolar epithelium is more prone to SARS-CoV-2 infection 458 because it contains a larger number of cells co-expressing the viral entry receptor and co-459

receptor. 460
Analysis of the ENCODE database identified distal enhancers of ACE2 and TMPRSS2 461 active in lung tissue. In contrast, the canonical AR-regulated enhancer of TMPRSS2, which is 462 active in prostate epithelial cells, is located proximally and was not active in the lung. 463 Transcription factor binding motif and GSEA suggested that several JAK-activated transcription 464 factors, including STAT1 and 2, and IFN-induced transcription factors, including IRF1, are 465 induced and bind to these novel enhancers and the promoter of TMPRSS2 and ACE2 to a larger 466 extent in male AT1 and AT2 cells as compared to their female counterparts. In contrast, we did 467 not detect a statistically significant difference in the level of expression or activity of AR between 468 male and female AT1 and AT2 cells. Provocatively, GSEA revealed that several IFN-regulated 469 signatures are amongst the top upregulated in male AT1 and AT2 cells as compared to their 470 female counterparts. Consistently, several canonical IFN target genes were expressed at higher 471 levels in male AT1 and AT2 cells as compared to their female counterpart. These results suggest 472 that both ACE2 and TMPRSS2 are IFN target genes in alveolar epithelial cells and that sexually 473 dimorphic IFN signaling leads to higher levels of expression of both viral entry receptors in males, 474 potentially explaining the increased susceptibility of males to SARS. 475 To directly confirm these findings, we examined the effect of IFN signaling on the 476 expression of ACE2 and TMPRSS2 in primary human lung alveolar epithelial cells. We found 477 that these cells respond to stimulation with IFNa by activating STAT1 and STAT2 in a JAK-478 dependent manner. Consistent with the notion that STAT1 and 2 are not only transcriptional 479 activators but also target genes in the IFN signaling pathway (47, 48), IFNa robustly activated 480 their expression, providing a feed-forward mechanism for amplification of signaling in the 481 alveolar epithelium. Notably, treatment of human lung alveolar epithelial cells with IFN⍺ 482 promoted binding of p-STAT1 to the promoter and the enhancer of ACE2, transcriptional 483 activation of the gene, and expression of ACE2. Similarly, IFNa induced binding of p-STAT1 to 484 the promoter and the lung-active distal enhancer of TMPRSS2, transcriptional activation of the 485 gene, and expression of TMPRSS2. In contrast, treatment with androgen did not stimulate 486 expression of the two entry receptors; in fact, it interfered with IFN-induced p-STAT1 binding to 487 the regulatory elements of ACE2 and TMPRSS2 and therefore reduced their expression. 488 Moreover, we found that the second-generation AR inhibitor enzalutamide does not reduce the 489 expression of ACE2 and TMPRSS2 in primary alveolar epithelial cells. In fact, enzalutamide 490 increased the expression of TMPRSS2 in these cells, presumably by interfering with the ability 491 of autocrine IFN signaling to induce STAT1/2 binding to the promoter and distal enhancer of Although alveolar epithelial cells express type I, II, and III IFN-Rs, IFNa and b induce 498 expression of ACE2 to a much larger extent (>20 fold over control) as compared to other IFNs. 499 In contrast, IFNl is the most potent inducer of TMPRSS2 (>7 fold over control). Since the basal 500 expression of ACE2 is much lower than that of TMPRSS2, co-stimulation with type I and II IFN 501 induces a relative overexpression of TMPRSS2 as compared to ACE2, suggesting that multiple 502 TMPRSS2 co-receptors may assist a single ACE2 receptor in mediating viral entry in alveolar 503 epithelial cells. It is hypothesized that initial viral entry into naïve epithelial cells induces 504 expression of both type I and III IFNs through pattern recognition receptors (62, 63). Intriguingly, 505 we found that transduction of Poly IC, mimicking viral RNA, robustly upregulates the expression 506 of ACE2 and TMPRSS2 in alveolar epithelial cells, suggesting that viral entry and the ensuing 507 liberation of viral RNA into the cytoplasm can induce production of IFNs and upregulation of viral 508 entry receptors in a feed-forward mechanism. Importantly, our results also suggest that secreted 509 type I and III IFN will upregulate the expression of viral entry receptors in adjacent epithelial 510 cells, facilitating their infection. Consistent with the role of JAK-STAT signaling in this process, 511 type I and III IFN stimulated entry of an S protein-pseudotyped virus in alveolar epithelial cells 512 and JAK inhibitors reversed this process. These results suggest that JAK inhibitors may be 513 effective in preventing the transition of COVID-19 disease to SARS in addition to ameliorating 514 disease progression in SARS patients, as suggested by recent clinical trials (68, 69). 515 The cytokine shock syndrome often underlies the progression of SARS-CoV-2 to the 516 lethal stage (64). We found that TNF⍺ upregulates the expression of TMPRSS2, whereas INFg 517 controls the expression of ACE2. In contrast, 14 other cytokines potentially involved in the 518 cytokine storm do not affect the expression of either one of the two viral entry receptors.  Recent studies have shown that even single virions can productively infect AT2 cells in 531 alveolar organoids. Notably, whereas high levels of IFN limit further infection, resulting in modest 532 viral burden, low levels of IFN exerts the opposite effect, suggesting that IFN signaling can exert 533 a bimodal effect depending on its dose (16, 66). We propose that viral mechanisms enable newly 534 infected cells to titrate the production of type I and III IFNs to a level that is insufficient to mediate 535 viral restriction but is sufficient to upregulate viral entry receptors. Consistently, it has been 536 observed that SARS-CoV-2 infection drives lower antiviral transcription marked by low IFN-I and 537 IFN-III levels as compared to common respiratory viruses (73). By necessity, most of the studies 538 on viral proteins suppressing IFN signaling have been conducted by using overexpression of 539 individual viral genes and, thus, cannot provide definitive information on the level of IFN signaling 540 induced in naturally infected cells (74-78). Future studies using a recently developed trans-541 complementation system will contribute to a better understanding of how SARS-CoV-2 fine-542 tunes IFN signaling to facilitate its spreading in the host (79). We further propose that type I IFN 543 immunity plays a similar bimodal role also in late-stage disease. In fact, it has been reported that 544 inborn errors of IFN-I immunity or autoantibodies to IFN-I are present, at a low rate, in patients 545 with a life-threatening disease, but they are completely absent in those with mild disease (80, 546 81). In contrast, immunophenotyping of patients suggest that that the IFN-I response exacerbate 547 inflammation in severe COVID-19 (28) and IFN-I and III disrupt lung epithelial repair during 548 recovery from viral infection (82). 549 In conclusion, our study indicates that sexually dimorphic IFN-I and III signaling 550 upregulates the expression of the SARS-CoV-2 entry receptors ACE2 and TMPRSS2 in lung 551 alveolar epithelium, providing a potential mechanism for the male-biased incidence of SARS in 552 COVID-19 patients. We also demonstrate that SARS-CoV-2 hijacks IFN-I and III signaling to 553 upregulate the viral entry receptors and hence facilitate viral spreading during initial infection of 554 the alveolar epithelium. Furthermore, induced as components of the cytokine storm, TNF-a and 555 IFN-II produce a similar upregulation of viral entry receptors and a direct cytopathic effect during 556 the late phase of the disease. Based on these results, we suggest that JAK1/TYK2 inhibitors, 557 such as ruxolitinib, may be used in conjunction with antivirals to halt viral spread from the upper 558 respiratory tract to the distal lung, reducing the incidence of SARS. Furthermore, combinations 559 of JAK1/2, such as baricitinib, and NF-kB inhibitors, such as salicylate, may be more efficacious 560 as compared to current treatments during advanced SARS in COVID-19. In addition to informing 561 our ongoing understanding of COVID-19 pathophysiology, these findings suggest novel 562 therapeutic strategies and regimens for the prevention and treatment of COVID-19 SARS. 563 We were unable to include several scRNAseq datasets in this study due to unavailability 564 of sex information. Our results on sexually dimorphic IFN signaling and expression of ACE2 and 565 TMPRSS2 in alveolar epithelial cells are therefore based on a comparison of a relatively small 566 number of individual samples. In addition, we did not validate these results by using an 567 independent approach. Finally, although our results suggest that SARS-Cov-2 does not 568 suppress IFN signaling in infected cells as profoundly as other viruses, this model will need to 569 be validated by using the wild-type virus or a trans-complementation system.

Study Design 578
We have used public scRNAseq datasets to compare the level of expression of SARS-CoV-2 579 entry receptors in male and female lung alveolar epithelial cells. By using transcription factor 580 binding motif analysis and gene set enrichment analysis, we have then identified the signaling 581 pathways potentially able to regulate the expression of TMPRSS2 and ACE2 in alveolar type I 582 and II cells. To confirm these observations, we have examined the ability of IFN and NFkB 583 signaling to elevate the expression of both entry receptors in primary alveolar epithelial cells by 584 using ChIP-Q-PCR, Q-PCR, and immunoblotting with antibodies to activated JAK and STAT 585 isoforms. Finally, we have used a pseudo-typed lentiviral reporter vector to study the effect of 586 IFN and NFkB signaling and specific inhibitors on viral entry in primary alveolar epithelial cells. 587 588 Reagents 589 A list of reagents, including antibodies, probes, cell lines, chemical reagents and software, can 590 be found in the Resources table (Supplementary Table 1).

592
Single cell RNA sequencing datasets 593 Three publicly available scRNA-seq datasets were obtained as follows: 1) processed data 594 including count and metadata tables of healthy lung tissue was downloaded from Figshare 595 (https://doi.org/10.6084/m9.figshare.11981034.v1); 2) h5 files of normal lungs were extracted 596 from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) under 597 accession number GSE122960; and 3) processed data including count and metadata tables of 598 human lung tissue was acquired from GSE130148. All three datasets were generated on 599 Illumina HiSeq 4000. Characteristics of all the samples containing sample name, sex, age and 600 smoking status are provided in Supplementary Table 2.

602
Single cell RNA sequencing data analysis 603 Count matrix was used to create a Seurat object for each dataset, and the three Seurat objects 604 were further merged into a new Seurat object with the resulting combined count matrix. The 605 merged matrix was first normalized using a global-scaling normalization method "LogNormalize" 606 in Seurat v.3.2.0 with default parameters. To detect the most variable genes used for principal 607 component analysis (PCA), variable gene selection was performed and the top 2,000 variable 608 genes were then selected using the 'vst' selection method in Seurat FindVariableFeatures 609 function. All genes were scaled in the scaling step and PCA was performed using the selected 610 top 2,000 informative genes. To do batch effect correction, Harmony algorithm was run on the 611 top 50 PCA embeddings using RunHarmony function. Then UMAP calculation was performed 612 on the top 30 Harmony embeddings for visualizing the cells. Meanwhile, graph-based clustering 613 was done on the harmony-reduced data. The final resolution was set to 0.2 to obtain a better 614 clustering result.

616
Generation of histone modification enrichment profiles and transcription factor binding 617 motif analysis 618 H3K27ac and H3K4me3 binding profiles were constructed using publicly available ENCODE 619 datasets. Accession numbers for all ENCODE datasets used can be found in Encode Data Sets 620 Table (Supplementary Table 3). The data were visualized with UCSC genome browser (83, 84). 621 Predicted enhancer regions of TMPRSS2 were identified using the GeneHancer tool within 622 Dynabeads. The tubes were gently mixed and placed on a rocker at 4°C. The tubes were then 670 placed in magnetic stand, inverted several times, and beads were allowed to clump. The 671 supernatant was then discarded. Beads were flicked to resuspend and were then washed with 672 1X RIPA-150, 1X RIPA-500, 1X RIPA-LiCl, and 2X TE buffer (pH 8.0), for 5 minutes each on a 673 rocker at 4°C. After each wash, the tubes were again placed on a magnetic stand and 674 supernatant was discarded after the beads clumped. Beads were then resuspended in 200µl of 675 Direct Elution Buffer (10mM Tris-HCl pH8.0, 0.3M NaCl, 5mM EDTA, 0.5% SDS). 1µl of RNaseA 676 was added and incubated at 65°C to reverse crosslink. The tubes were quickly centrifuged, place 677 on a magnetic stand, and supernatant was transferred to a new low-bind tube after beads 678 clumped. 3µl of Proteinase K was added and incubated for 2hrs at 55°C. The sample was 679 purified using phase lock tubes and ethanol precipitation. Samples were resuspended in 25µl of 680 Qiagen elution buffer. DNA was amplified by real-time PCR (ABI Power SYBR Green PCR mix). 681 682 Co-immunoprecipitation (co-IP) 683 Calu3 cells were incubated with either IFN-α and IFN-β, or IFN-λ for 3 hours. After IFN 684 stimulation, cells were lysed in 1 ml of ice-cold non-denaturing lysis buffer (20 mM, Tris-HCl pH8, 685 137 mM NaCl, 10% glycerol, 1% Nonidet P-40, and 2 mM EDTA) supplemented with a protease 686 inhibitor cocktail (Thermo Scientific, 78429 Statistical analysis used R and GraphPad Prism 8 software. At least three biologically 720 independent samples were used to determine significance. Results are reported as mean ± SD. 721 Non-parametric two-sided Wilcoxon rank sum tests were used to identify differentially expressed 722 genes in all the comparisons discussed in scRNA-seq analysis. Comparisons between two 723 groups were performed using an unpaired two-sided Student's t test (p < 0.05 was considered 724 significant). Comparison of multiple conditions was done with One-way or two-way ANOVA test. 725 The Fisher's exact test was used to compare the ratio of double positive cells between groups. 726 Only p values of 0.05 or lower were considered statistically significant (   After treatment, quantitative reverse transcription PCR (RT-qPCR) analysis of TMPRSS2 and 1105 ACE2 was performed to assess the expression of SARS-CoV-2 receptors, with 18S ribosomal 1106 RNA as an endogenous control. 1107 1108 (D and F) In order to directly examine whether p-STAT1 is capable of transactivates SARS-CoV-1109 2 receptors in response to interferon or AR-dependent stimulation, 1x10 6 AEpiC cells were either 1110 treated with DHT alone (10 nM, 24 hours), IFN⍺ alone (20 nM, 16 hours), or in combination.

1111
Cells were lysed in a non-denaturing lysis buffer and subjected to chromatin immunoprecipitation 1112 (ChIP) of p-STAT1, so as to investigate the enrichment of p-STAT1 on the enhancer and the 1113 promoter region of both TMPRSS2(Top) and ACE2(Bottom). 1114 1115 (E and G) The occupancy of both active transcription markers (H3K4me3, PoI II, and PoI II S2p) 1116 and a suppressive transcription marker (H3K27me3) on the regulatory region of SARS-CoV-2 1117 receptors was also applied to investigate the transcriptional alteration of TMPRSS2(Top) and 1118 ACE2(Bottom) in copying with interferon response and AR-dependent signaling in AEpiC cells.

1119
Results are reported as mean ± SD. Comparisons between two groups were performed using 1120 an unpaired two-sided Student's t test (p < 0.05 was considered significant). Comparison of 1121 multiple conditions was done with One-way or two-way ANOVA test. All experiments were 1122 reproduced at least three times, unless otherwise indicated. or 1 µg/ml IFNλ (lower right) for the indicated times (6, 12, 24 ,48 hours  (I) AEpiC were cells incubated for 12 hours with different concentrations of ruxolitinib and were 1168 stimulated with either PBS or 10 nM IFN⍺ plus 1 ug/ml IFNλ for 16 hours before viral infection. 1169 The transduction efficiency of the virus was quantified 48 hours post-transduction by measuring 1170 the activity of firefly luciferase in cell lysates. Results are reported as mean ± SD. Comparisons 1171 between two groups were performed using an unpaired two-sided Student's t test (p < 0.05 was 1172 considered significant). Comparison of multiple conditions was done with One-way or two-way 1173 ANOVA test. All experiments were reproduced at least three times, unless otherwise indicated. cells were pre-treated with either 500 nM tofacitinib, 500 nM fedratinib, or 500 nM ruxolitinib for 1198 12 hours and then stimulated with 10 ng/ml IFNɣ for another 24 hours. RT-qPCR analysis was 1199 performed to assess the mRNA expression level of ACE2 (top) and TMPRSS2 (bottom), with 1200 18S as an endogenous control. 1201 1202 (F) AEpiC cells were treated with either PBS, 10 ng/ml TNF⍺, 10 nM IFN⍺, 10 ng/ml IFNλ, or in 1203 combinations for 24 hours. TMPRSS2 mRNA expression measured by RT-qPCR, with 18S as 1204 an endogenous control. 1205 1206 (G) The efficacy of NF-kB inhibitor and JAK1/2 inhibitor in the blockade of SARS-CoV-2 1207 pseudotype entry in the COVID-19 advanced lung disease model. AEpiC cells were incubated 1208 for 12 hours with different concentrations of BAY-11-7082, ruxolitinib, or in combination, and 1209 stimulated without or with 10 nM IFN⍺, 1 ug/ml IFNλ, and 10 ng/ml TNF⍺ for 16 hours before 1210 viral infection. The transduction efficiency of the virus was quantified 48 hours post-transduction 1211 by measuring the activity of firefly luciferase in cell lysates. Results are reported as mean ± SD. 1212 Comparisons between two groups were performed using an unpaired two-sided Student's t test 1213 (p < 0.05 was considered significant  Sample  A9LCTZng  ASK428  ASK440  ASK452  ASK454  GSM3489182  GSM3489185  GSM3489187  GSM3489189  GSM3489191  GSM3489193  GSM3489195 chr21:42,874,874,830 TGGTTCTCTCTCTTTCTGTTT IRF1: chr21:42,881,881,784 CCCTTCCTGGGAAGC STAT6: chr21:42,874,874,753 CAAGTTTCTTTTCCT chr21:42,893,893,995 CTTCCCAGAAA chr21:42,857,857,607 AAAAAAAAATAGAAAGAAAAG IRF1: chr21:42,857,858,002 GGTTTCTACAGAAAA STAT6: chr21:42,857,857,750 TTTCTGGGAAG chr21:42,852,852,580 TCTTTCTTTTTTTTTCTTTTT IRF1: chr21:42,850,850,762 ATTCCTAGAAC chr21:42,849,849,665 CATTTCCTAAGAATC STAT6: chr21:42,850,850,817 AGGAAAGAAAACTAA chr21:42,893,893,155