key: cord-0978782-j34xqrs4
authors: Puray-Chavez, Maritza; LaPak, Kyle M.; Schrank, Travis P.; Elliott, Jennifer L.; Bhatt, Dhaval P.; Agajanian, Megan J.; Jasuja, Ria; Lawson, Dana Q.; Davis, Keanu; Rothlauf, Paul W.; Liu, Zhuoming; Jo, Heejoon; Lee, Nakyung; Tenneti, Kasyap; Eschbach, Jenna E.; Mugisha, Christian Shema; Cousins, Emily M.; Cloer, Erica W.; Vuong, Hung R.; VanBlargan, Laura A.; Bailey, Adam L.; Gilchuk, Pavlo; Crowe, James E.; Diamond, Michael S.; Hayes, D. Neil; Whelan, Sean P.J.; Horani, Amjad; Brody, Steven L.; Goldfarb, Dennis; Major, M. Ben; Kutluay, Sebla B.
title: Systematic analysis of SARS-CoV-2 infection of an ACE2-negative human airway cell
date: 2021-06-23
journal: Cell Rep
DOI: 10.1016/j.celrep.2021.109364
sha: b21c9e564a8e4c15fb73b4dbbef559961a22bdf1
doc_id: 978782
cord_uid: j34xqrs4

SARS-CoV-2 spike (S) variants govern transmissibility, responsiveness to vaccination and disease severity. In a screen for new models of SARS-CoV-2 infection, we identified human H522 lung adenocarcinoma cells as naturally permissive to SARS-CoV-2 infection despite complete absence of ACE2 expression. Remarkably, H522 infection requires the E484D S variant; viruses expressing wild-type S are not infectious. Anti-S monoclonal antibodies differentially neutralize SARS-CoV-2 E484D S in H522 cells as compared to ACE2-expressing cells. Sera from vaccinated individuals block this alternative entry mechanism, whereas convalescent sera are less effective. Though the H522 receptor remains unknown, depletion of surface heparan sulfates block H522 infection. Temporally resolved transcriptomic and proteomic profiling reveal alterations in cell cycle and the antiviral host cell response, including MDA5-dependent activation of type-I interferon signaling. These findings establish an alternative SARS-CoV-2 host cell receptor for the E484D SARS-CoV-2 variant, which may impact tropism of SARS-CoV-2 and consequently human disease pathogenesis.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the causative agent (Fig. 1A, Fig. S1 and Table S1 ).

Normalized RNA-seq read counts for established SARS-CoV-2 cell models Caco-2,

Calu-3 and Vero E6 enabled comparative analysis (Fig. 1A, Fig. S1 and Table S1 ).

Validation by qRT-PCR and protein quantification by immunoblotting showed ACE2 154 protein levels ranging from undetectable to 2-3-fold lower than Vero E6 cells, currently 155 one of the most permissive cell models for SARS-CoV-2 infection (Fig. 1B, C) . The 156 observed cell line-dependent differences in ACE2 protein migration are possibly due to expressing ACE2 and TMPRSS2 at relatively high levels (i.e. Detroit562 and H596),

were not permissive to SARS-CoV-2 replication (Fig. 1D) . H522, and to a lesser degree 163 HCC827 cells, supported virus replication (Fig. 1D) . Unexpectedly, neither ACE2 nor 164 TMPRSS2 was detected in H522 cells (Fig. 1A-D) .

Given the possibility of ACE2-independent infection, we focused our efforts on H522 

RNA levels in the cell culture supernatants (Fig. 1F) . We confirmed replication 171 competency of virus released from H522 cells through plaque assays on Vero cells (up 172 to 2.2x10 5 PFU/mL, data not shown). Although permissive, infection progressed slower 173 in H522 cells than in Vero E6 cells, and higher doses of the virus were required to 174 achieve similar percentages of infected cells (Fig. 1G) . Viruses formed plaques on H522 175 cells, and plaque sizes were comparable to those obtained on Vero E6 cells, albeit the 176 effective MOI was ~20-fold lower (Fig. 1H) .

Quantified RNA-in situ hybridization (ISH) revealed the kinetics of SARS-CoV-2 viral 178 replication and spread in H522 cells (Fig. 1I, S2A, B) . Incoming virions were readily 179 detected at 4 hours post-infection (hpi) in cells infected with an MOI of 1 (white arrows to 180 green puncta; Fig. 1I ). As expected, both the number of cells positive for viral RNA and 181 the number of viral RNA puncta/cell were MOI-dependent (Fig. S2A, B) . We tracked 182 viral spread over time and observed increased staining for both viral RNA and N, and 183 increased number of infected cells per field (Fig. S2A) . Furthermore, similar to Vero 

The inability of VSV-GFP-SARS-CoV-2-S ∆21 to infect H522 cells prompted us to analyze 211 sequence variation within S in our virus stocks and in infected H522 cells. This revealed 212 the presence of the E484D substitution within the receptor-binding domain (RBD) and 213 the less prevalent R682W substitution within the furin cleavage site. No other spike 214 mutations were observed. Unexpectedly, we found that virus stocks containing 215 sequence-validated WT S were unable to infect H522 cells (Fig. 2E) . We next tested the 216 contribution of S mutations to H522 infection using chimeric VSV-GFP-SARS-CoV-2-217 S ∆21 and lentiviral pseudoparticles. Chimeric VSV-GFP-SARS-CoV-2-S ∆21 bearing the 218 E484D substitution specifically enhanced viral entry in H522 cells but did not affect the 219 infection of 293T-ACE2 cells (Fig. 2F-H) . In contrast, E484K/R685S substitutions did not 220 enable VSV-GFP entry into H522 cells (Fig. 2F, H) . Despite the enhanced viral entry, we 221 observed no viral spread with VSV-GFP-SARS-CoV-2-S ∆21 E484D in H522s (Fig. 2H) 

suggesting that additional viral factors may be needed to facilitate viral egress and 223 spread. Consistent with these findings, we found that lentivirus pseudoparticles bearing 224 the E484D substitution allowed low but reproducibly detectable levels of infection, 225 whereas pseudoparticles bearing WT or R682W S were unable to infect H522s (Fig. 2I ).

Combination of the E484D and R682W substitutions did not further promote viral entry 227 (Fig. 2I) . In contrast, WT and E484D lentivirus pseudoparticles efficiently infected 293T-228 ACE2 cells and the R682W substitution further enhanced entry (Fig. 2J) .

Given the requirement for S E484D in H522 but not in ACE2-expressing cells, we (Table S2 ) did not affect infection (Fig. 2K) . Remarkably, all eight mAbs blocked 234 infection of H522 cells, demonstrating differential usage of S for the alternative entry 235 pathway in H522 cells (Fig. 2L) . Taken together, these data show that S is necessary for 236 infection of H522 cells but not sufficient. Moreover, the E484D mutation within the RBD 237 of S is required in H522 cells, but not ACE2-expressing cells. 

In a second approach to test ACE2 involvement, we inactivated the ACE2 genetic locus 249 by CRISPR gene editing in H522 and Calu-3 cells. Polyclonal cell populations containing 250 CRISPR-edited loci were inoculated with SARS-CoV-2 and viral replication was 251 monitored at 4 and 72 hpi (Fig. 3B, S3A) . While viral RNA levels increased at similar 252 levels in H522 and H522 ACE2 -/cells, the lack of ACE2 expression significantly reduced 253 SARS-CoV-2 replication in Calu-3 cells (Fig. 3B) . Moreover, the addition of an ACE2 254 blocking antibody did not impair virus replication in H522 or H522 ACE2 -/cells, but 255 completely abolished replication in Calu-3 cells (Fig. 3C) . We next isolated monoclonal 256 cell populations of H522 ACE2 WT (n=6), H522 ACE2 -/-(n=2) and H522 ACE2 +/-(n=1) 257 J o u r n a l P r e -p r o o f cells to corroborate these findings (Fig. S3B) . Sanger sequencing of the edited loci 258 revealed unique 5 bp deletions in Exon 3 of ACE2, resulting in a truncated ACE2 protein 259 lacking the C-terminal 672 amino acids (Fig. S3B) . SARS-CoV-2 infection of monoclonal 260 cell lines from H522 control and H522 ACE2 -/resulted in similar levels of infection ( Fig.   261   3D ).

We next tested the involvement of published alternative SARS-CoV-2 receptors NRP1,

AXL and heparan sulfate. Depletion of NRP1 and AXL by siRNAs and CRISPR did not 264 diminish SARS-CoV-2 replication in H522 cells (Fig. 3E, Fig. S3C-E) . In contrast, 265 removal of surface heparan sulfates by propagation of H522 cells in sodium chlorate or 266 heparinase treatment substantially reduced virus replication (Fig. 3F, G) . Taken 267 together, these data establish that H522 cells are permissive to SARS-CoV-2 infection in 268 a manner that is independent of ACE2, NRP1, and AXL, but dependent on surface 269 heparan sulfates. 

These findings were corroborated in comparative analysis of H522, Vero E6 and H522-284 ACE2 cells. Bafilomycin A significantly decreased cell-associated viral RNA levels in 285 H522 and H522-ACE2 cells but did not affect viral entry in Vero E6 cells (Fig. S4) .

Inhibition of both AAK1 and endosomal cathepsins B/L significantly decreased viral RNA 287 levels in H522 cells but did not impact ACE2-dependent replication at appreciable levels 288 in Vero E6 and H522-ACE2 cells at this concentration (Fig. S4) . While apilimod 289 decreased viral entry in Vero E6 cells, the effect was modest in H522 cells and trended 290 towards significance (p=0.07; Fig. S4 ). Finally, camostat mesylate did not decrease, and 291 on the contrary, increased the amount of cell-associated viral RNA in H522s, highlighting 292 the TMPRSS2 independence of viral entry (Fig. S4) .

Western blot analysis of H522 cells infected with SARS-CoV-2 revealed transient 294 induction of AP2M1 phosphorylation 12-24 hpi, further supporting the involvement of 295 CME in H522 viral infections (Fig. 4B) . AAK1 inhibitors are highly specific and have 296 been considered as therapeutic options for treatment of SARS-CoV-2 (Richardson et al., 297 2020). Consistent with our observations in H522 cells, inhibition of AAK1 kinase activity 298 in differentiated primary HBECs grown at air-liquid interface resulted in a 10-to 20-fold 299 decrease in cell-associated SARS-CoV-2 RNA in a dose responsive manner (Fig. 4C) .

Together, these data support a role for CME and endosomal cathepsins in SARS-CoV-2 301 infection of H522 cells.

The E484D S variant has been found to circulate within the human population and given 303 the alternative mechanism of entry and the requirement for the E484D substitution, we well as numerous ISGs including ISG15, MX1, IFI35 and OAS3 (Fig. 5D, Table S3 ).

Hierarchical consensus clustering of the 2,631 DEGs (|logFC|>2 and q<0.005) 319 generated 7 temporally resolved clusters (Fig. 5E, F) . Over-representation analysis of 320 each cluster revealed an initial sharp increase of cell cycle regulatory and inflammatory 321 genes followed by decreasing levels as the infection proceeded (cluster 1) (Fig. 5F , G, 322 To define the impact of SARS-CoV-2 infection on the H522 proteome, we conducted 332 whole cell quantitative proteomics experiments over the course of 4 days (Fig. 6A) .

Biological triplicates for each time point were processed, and the abundance of 7,469 334 proteins was analyzed across samples. PCA highlights the high level of reproducibility 335 and clustering of samples by infection and time post-infection (Fig. 6B) . Similar to viral 336 RNAs, abundance of viral proteins increased substantially within the first 24 hours of 337 infection and plateaued thereafter (Fig. 6C) . At 96hpi versus 96h mock, 492 338 differentially-expressed proteins were identified (Fig. 6D) . Unsupervised clustering 339 defined seven unique clusters that characterize the temporal regulation of the H522 340 proteome (Fig. 6E, Table S5 ). Overall, the majority of the differentially expressed 341 proteins increased following SARS-CoV-2 infection, with proteins in cluster 4 displaying 342 the greatest fold changes (Fig. 6F, Table S5 ). Over-representation analysis revealed 343 that cluster 4 proteins include those involved in the IFN-α and IFN-γ responses, which 344 were the most significantly altered pathways (Fig. 6G, Table S6 ). Of note, all viral 345 proteins were present in Cluster 2, and their accumulation preceded the induction of type 346 I/III IFNs (Fig. 6E) . Cell cycle regulators were increased at early time points but declined 347 thereafter, matching what was seen at the RNA level with a 12-24 hour delay (Cluster 1; replication/repair and microtubule organization but tended to remain upregulated during 351 SARS-CoV-2 infection (Fig. 6E-G) . Finally, clusters 6 and 7 included proteins that were 352 downregulated and included plasma membrane proteins such as Semaphorins

(SEMA3A, C, D), APOE, ERBB4, LRP1 and SLIT2 with potential roles in viral entry 354 J o u r n a l P r e -p r o o f pathways ( Fig. 6E-G) . We next looked for genes that correlated between our 355 transcriptomic and proteomic datasets. Correlations were evenly distributed for the entire 356 gene set; the subset of differentially expressed proteins shows increased correlation 357 (Fig. 6H) . Among these genes, only the IFN-α and IFN-γ signaling pathways were 358 identified by GSEA for enrichment in correlation, further supporting an IFN response in 359 H522 cells to SARS-CoV-2 infection (Fig. 6I) .

To further illuminate pathways altered by SARS-CoV-2 infection, we mined the CORUM 361 database for protein complexes consisting mostly of differentially-expressed proteins 362 ( Fig. 6J, S5 ). In total, 27 complexes were found and involved IFN signaling, cell To validate the IFN response, we measured the levels and activation of STAT1 and 379 downstream ISGs in infected H522 cells. We found that SARS-CoV-2 replication 380 induced STAT1 expression and its phosphorylation, as well as downstream ISGs, MX1, 381 and IFIT1 by 48 hpi (Fig. 7A) . MX1 and IFIT1 were upregulated further as the infection 382 progressed at 72 and 96 hpi (Fig. 7A) . Upregulation of the type I IFN response was 383 delayed relative to the accumulation of viral N protein expression which peaked by 24 384 hpi, possibly due to antagonism of host responses at early times in infection.

We next sought to define the mechanism by which H522 cells sense and respond to 386 SARS-CoV-2 replication. Components of TLR and RLR-dependent sensing pathways 387 were depleted by siRNA transfection, and upregulation of ISGs were assessed following 388 SARS-CoV-2 infection. Most targets remained efficiently silenced up to 120 hours post 389 siRNA transfection (Fig. S6) . The low silencing efficiency of TLR3, TLR7, TLR8 and 390 TLR9 is most likely due to their low to undetectable basal expression levels (Fig S6) . 

Our data implicates CME in SARS-CoV-2 infection of H522 cells. Specific inhibition of a 482 kinase directly involved in CME, AAK1, reduced SARS-CoV-2 infection in H522 cells and 483 patient-derived HBECs (Fig. 4, S4) . Though inhibition of AAK1 and CME have been 

See also Figure S4 Table S7 . Table S7 . Table S7 .

Samples were prepared for RNA-seq using the Truseq stranded mRNA kit (Illumina) and 892 subjected to sequencing on a Next-seq platform (1x75bp) at the Center for Genome 

The R statistical programming language was used for data processing and figure   1141 generation.

In order to standardize RNA-seq data from 3 different protocols in Figure 1A 

The lower fold-change threshold was employed in order to capture proteins whose 1200 differential expression peaked at other time points. 

Gene set enrichment analysis was performed on RNA-seq logCPM values ( Figure 5G ).

Genes were ranked according to signal to noise ratio as defined by the Broad Institute 

WNT Activates the AAK1 Kinase to Promote Clathrin-Mediated Endocytosis of LRP6 1264 and Establish a Negative Feedback Loop

Gene 1267 expression and in situ protein profiling of candidate SARS-CoV-2 receptors in human 1268 airway epithelial cells and lung tissue

A Potently Neutralizing Antibody Protects 1271 Mice against SARS-CoV-2 Infection

CD209/DC-SIGN act as receptors for SARS-CoV-2 and are differentially expressed in 1275 lung and kidney epithelial and endothelial cells

Imbalanced Host Response to 1278 SARS-CoV-2 Drives Development of COVID-19

Proteomics of SARS-CoV-2-infected host cells reveals therapy 1281 targets

The Global 1284

Phosphorylation Landscape of SARS-CoV-2 Infection

SARS-CoV-2 cellular tropism

Neuropilin-1 1288 facilitates SARS-CoV-2 cell entry and infectivity

Neutralizing Antibody and Soluble ACE2 Inhibition 1291 of a Replication-Competent VSV-SARS-CoV-2 and a Clinical Isolate of SARS-CoV-2

Clinical and immunological features of severe and moderate 1295 coronavirus disease 2019

Resistance of SARS-CoV-2 variants 1298 to neutralization by monoclonal and serum-derived polyclonal antibodies

Comparative replication and immune activation profiles of 1302

SARS-CoV-2 and SARS-CoV in human lungs: an ex vivo study with implications for the 1303 pathogenesis of COVID-19

Genetics Reveals a Variable Infection Gradient in the Respiratory Tract

SARS coronavirus nsp1 protein induces template-dependent 1365 endonucleolytic cleavage of mRNAs: viral mRNAs are resistant to nsp1-induced RNA 1366 cleavage

Clinical features of patients infected with 2019 novel coronavirus in Wuhan

Human Coronavirus HKU1 Spike Protein Uses 1372 O-Acetylated Sialic Acid as an Attachment Receptor Determinant and Employs 1373

Protein as a Receptor-Destroying Enzyme

Human coronaviruses OC43 1377 and HKU1 bind to 9-O-acetylated sialic acids via a conserved receptor-binding site in 1378 spike protein domain A

The reactome pathway knowledgebase

A 1383 two-pronged strategy to suppress host protein synthesis by SARS coronavirus Nsp1 1384 protein

Inhibition of PIKfyve kinase prevents 1387 infection by Zaire ebolavirus and SARS-CoV-2

Identification of Coronavirus

Korea with COVID-19

7a protein of 1393 severe acute respiratory syndrome coronavirus inhibits cellular protein synthesis and 1394 activates p38 mitogen-activated protein kinase

Shared and Distinct 1396 Functions of Type I and Type III Interferons

Functional assessment of cell entry and 1398 receptor usage for SARS-CoV-2 and other lineage B betacoronaviruses

Identification of sialic acid-binding 1402 function for the Middle East respiratory syndrome coronavirus spike glycoprotein

The Molecular Signatures Database (MSigDB) hallmark gene set collection

Molecular signatures database (MSigDB) 3.0

Receptor usage and cell entry of porcine epidemic diarrhea coronavirus

Systematic 1414 identification of type I and type II interferon-induced antiviral factors

Identification of SARS-CoV-2 1418 spike mutations that attenuate monoclonal and serum antibody neutralization

Type I Interferon 1422 Susceptibility Distinguishes SARS-CoV-2 from SARS-CoV

Severe acute 1424 respiratory syndrome coronavirus protein nsp1 is a novel eukaryotic translation inhibitor 1425 that represses multiple steps of translation initiation

Host and viral 1427 determinants of influenza A virus species specificity

HIV Restriction Factors and Mechanisms of 1429

Cold Spring Harb Perspect Med 2, a006940

The molecular biology of coronaviruses

Reproducible workflow for 1434 multiplexed deep-scale proteome and phosphoproteome analysis of tumor tissues by 1435 liquid chromatography-mass spectrometry

Upper airway gene expression 1438 reveals suppressed immune responses to SARS-CoV-2 compared with other respiratory 1439 viruses

Functional role of type I and type II interferons in antiviral defense

Viral and Cellular mRNA 1444 Translation in Coronavirus-Infected Cells

Severe acute respiratory syndrome coronavirus nsp1 suppresses 1447 host gene expression, including that of type I interferon, in infected cells

Consensus transcriptional 1450 regulatory networks of coronavirus-infected human cells

The genome landscape of the african green monkey kidney-1453 derived vero cell line

Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its 1456 immune cross-reactivity with SARS-CoV

The BioGRID database: A comprehensive 1459 biomedical resource of curated protein, genetic, and chemical interactions

Type I and Type III Interferons -Induction

Evasion, and Application to Combat COVID-19

Structures of MERS-CoV spike glycoprotein in 1465 complex with sialoside attachment receptors

Salmon 1467 provides fast and bias-aware quantification of transcript expression

Resistance to influenza 1470 virus and vesicular stomatitis virus conferred by expression of human MxA protein

SARS-CoV-2 1473 one year on: evidence for ongoing viral adaptation

Multiplex single-cell visualization of 1476 nucleic acids and protein during HIV infection

Single cell RNA sequencing of 13 1478 human tissues identify cell types and receptors of human coronaviruses

Baricitinib as potential treatment for 2019-nCoV 1482 acute respiratory disease

TRIM69 Inhibits Vesicular Stomatitis 1485 Indiana Virus

limma powers differential expression analyses for RNA-sequencing and microarray 1488 studies

edgeR: a Bioconductor 1490 package for differential expression analysis of digital gene expression data

Reconstruction of the cell 1494 entry pathway of an extinct virus

Species-Specific Host-Virus Interactions: 1496 Implications for Viral Host Range and Virulence

Convenient assay for interferons

In situ detection of SARS-CoV-2 in lungs and airways of patients 1501 with COVID-19

Measuring SARS-CoV-2 neutralizing antibody activity using pseudotyped and chimeric 1505 viruses

Coronavirus envelope protein: current 1507 knowledge

Recent advances in antiviral interferon-stimulated gene biology

SARS-CoV-2 Nsp1 binds 1512 the ribosomal mRNA channel to inhibit translation

Transmissible gastroenteritis coronavirus, but not the related porcine 1515 respiratory coronavirus, has a sialic acid (N-glycolylneuraminic acid) binding activity

Comprehensive characterization of N-and O-glycosylation of SARS-CoV-2 1519 human receptor angiotensin converting enzyme 2

Cell entry 1521 mechanisms of SARS-CoV-2

A 1524 Simplified Quantitative Real-Time PCR Assay for Monitoring SARS-CoV-2 Growth in 1525 Cell Culture

A facile 1527 Q-RT-PCR assay for monitoring SARS-CoV-2 growth in cell culture

Differential analyses for RNA-seq: 1529 transcript-level estimates improve gene-level inferences

Multi-level proteomics reveals host-1532 perturbation strategies of SARS-CoV-2 and SARS-CoV. bioRxiv

SARS-CoV-2 entry 1536 factors are highly expressed in nasal epithelial cells together with innate immune genes

Exploring the pathogenesis of severe acute respiratory 1539 syndrome (SARS): the tissue distribution of the coronavirus (SARS-CoV) and its putative 1540 receptor, angiotensin-converting enzyme 2 (ACE2)

Structural basis for human coronavirus 1543 attachment to sialic acid receptors

The MaxQuant computational platform for 1545 mass spectrometry-based shotgun proteomics

UniProt: a worldwide hub of protein knowledge

Endothelial cell 1550 infection and endotheliitis in COVID-19

Human and bovine 1552 coronaviruses recognize sialic acid-containing receptors similar to those of influenza C 1553 viruses

J o u r n a l P r e -p r o o f * **