key: cord-0957938-sq5z686p
authors: Emanuel, Wyler; Kirstin, Mösbauer; Vedran, Franke; Asija, Diag; Theresa, Gottula Lina; Roberto, Arsie; Filippos, Klironomos; David, Koppstein; Salah, Ayoub; Christopher, Buccitelli; Anja, Richter; Ivano, Legnini; Andranik, Ivanov; Tommaso, Mari; Simone, Del Giudice; Patrick, Papies Jan; Alexander, Müller Marcel; Daniela, Niemeyer; Matthias, Selbach; Altuna, Akalin; Nikolaus, Rajewsky; Christian, Drosten; Markus, Landthaler
title: Bulk and single-cell gene expression profiling of SARS-CoV-2 infected human cell lines identifies molecular targets for therapeutic intervention
date: 2020-05-05
journal: bioRxiv
DOI: 10.1101/2020.05.05.079194
sha: 0830b3286c606899cb594bd9617a5d086f6be59d
doc_id: 957938
cord_uid: sq5z686p

The coronavirus disease 2019 (COVID-19) pandemic, caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is an ongoing global health threat with more than two million infected people since its emergence in late 2019. Detailed knowledge of the molecular biology of the infection is indispensable for understanding of the viral replication, host responses, and disease progression. We provide gene expression profiles of SARS-CoV and SARS-CoV-2 infections in three human cell lines (H1299, Caco-2 and Calu-3 cells), using bulk and single-cell transcriptomics. Small RNA profiling showed strong expression of the immunity and inflammation-associated microRNA miRNA-155 upon infection with both viruses. SARS-CoV-2 elicited approximately two-fold higher stimulation of the interferon response compared to SARS-CoV in the permissive human epithelial cell line Calu-3, and induction of cytokines such as CXCL10 or IL6. Single cell RNA sequencing data showed that canonical interferon stimulated genes such as IFIT2 or OAS2 were broadly induced, whereas interferon beta (IFNB1) and lambda (IFNL1-4) were expressed only in a subset of infected cells. In addition, temporal resolution of transcriptional responses suggested interferon regulatory factors (IRFs) activities precede that of nuclear factor-κB (NF-κB). Lastly, we identified heat shock protein 90 (HSP90) as a protein relevant for the infection. Inhibition of the HSP90 charperone activity by Tanespimycin/17-N-allylamino-17-demethoxygeldanamycin (17-AAG) resulted in a reduction of viral replication, and of TNF and IL1B mRNA levels. In summary, our study established in vitro cell culture models to study SARS-CoV-2 infection and identified HSP90 protein as potential drug target for therapeutic intervention of SARS-CoV-2 infection.

Diseases caused by coronaviruses (CoVs) range from asymptomatic and mild infections of the upper respiratory tract to severe acute respiratory distress, when the lower respiratory tract is infected. In addition to the six previously-known CoVs affecting humans, a novel CoV termed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has recently emerged. The novel SARS-CoV-2, which causes coronavirus disease 2019 , is still an ongoing global health threat since the beginning of the outbreak in late 2019 and has, at the time of writing this text, infected more than three million people worldwide [1] . The SARS-CoV-2 life cycle initiates with the attachment of the virion to the cell surface and subsequent binding to the angiotensin converting enzyme 2 (ACE2), followed by proteolytic cleavage and internalization [2] [3] [4] . Non-structural proteins are then translated to form a replicasetranscriptase complex (RTC), in which the full genomic RNA, as well as subgenomic RNAs are generated within double membrane vesicles (DMV) [3] [4] [5] . Incoming viral RNA is detected by sensors such as IFIH1 (interferon induced with helicase C domain 1; also known as MDA5) and DDX58 (DExD/H-Box helicase 58; also known as RIG-I), which trigger the antiviral response. This sensing and signaling is impaired by a range of viral factors, e.g. replication within DMVs, RNA capping and methylation, or shortening of the poly-U tail on the minus strand RNA [6, 7] . Furthermore, inhibition of IRF activity [8] and a delayed induction of interferon-stimulated genes (ISGs) compared to Influenza virus infection or type I interferon treatment itself [9] was observed in SARS-CoV infection. Importantly, accessory genes in the SARS-CoV genome, like ORF6, may code for antagonists of interferon signaling [10] .

Following production of subgenomic RNAs, during which a constant 5´ leader is prepended by a process called discontinuous transcription [11] , the viral genes are translated either in the cytoplasm (nucleocapsid protein, N), or at the endoplasmic reticulum (ER; envelope (E), membrane (M), spike (S) and open reading frame 3 (ORF3) proteins) [12, 13] . The substantial increase in ER translation causes ER stress, which triggers the unfolded protein response. This is then in turn integrated with double stranded RNA sensing at the level of eukaryotic initiation factor 2 alpha (eIF2alpha) phosphorylation [14] . The ER stress response is likely attenuated by the viral E protein [15, 16] . Accordingly, heat shock proteins (HSPs), which ameliorate ER stress, have been described to be generally relevant in virus infections [17] . Furthermore, ER stress induces autophagy, a cell recycling pathway which can be used by some viruses for productive replication [18] . Finally, dysregulation of microRNA (miRNA) expression and subsequent alterations in gene expression patterns have also been reported to play a role in infected cells [19, 20] .

Comprehensive profilings of SARS-CoV-2-mediated gene expression perturbations are just beginning. A recent in-depth analysis of the transcriptional response to SARS-CoV-2 in comparison to other respiratory viruses in cells and animal models revealed a virus-specific inflammatory response [21] . Of particular interest are methods such as single-cell RNAsequencing (scRNA-seq), which allow the characterization of heterogeneity over the course of infection, which may be masked at the population level [22] [23] [24] [25] [26] [27] , but also small RNA sequencing, which reveals miRNAs and other small RNAs [28, 29] .

Here, we performed a comprehensive analysis of three human cell lines infected with SARS-CoV or SARS-CoV-2, namely the gut cell line Caco-2, the lung cell lines Calu-3, and H1299.

We generated scRNA-seq, poly(A) + and total RNA transcriptomic data as well as small RNA profiling in infection time courses for both viruses.

Efficiency and productivity of infection as well as the interferon response was remarkably different between the cell lines. Interestingly, SARS-CoV-2 induced a two-fold higher expression of interferon stimulated genes (ISGs) than SARS-CoV. In addition, we found strong induction of miR-155 with both viruses, suggesting a role for this miRNA in the progression of infection. The scRNA-seq data showed that, while canonical ISGs such as interferon induced protein with tetratricopeptide repeats 1 and 2 (IFIT1/IFIT2) were broadly induced, interferon beta (IFNB1) was expressed only in a subset of infected cells. Furthermore, the transcriptional induction of nuclear factor-κB (NF-κB) targets could be temporally separated from the interferon gene induction. Detailed investigations of cellular gene expression programs suggest an involvement of the protein folding chaperone and autophagy regulator HSP90 in the viral infection cycle. Inhibition of HSP90 by 17-N-allylamino-17-demethoxygeldanamycin (17-AAG) resulted in reduced viral replication and TNF and IL1B mRNA levels. Overall, our study provides a detailed picture of the gene expression changes in cell line models for CoVs and particularly SARS-CoV-2, highlights the cell-type specificity of the transcriptional response to infection, and identifies potential targets for therapeutic interventions.

To establish cell culture systems for studying SARS-CoV-2 replication and host cell responses, we examined the epithelial lung cancer cell lines, H1299 and Calu-3, as well as the epithelial colorectal adenocarcinoma cell line, Caco-2, which is frequently used as a coronavirus cell culture model [30] [31] [32] . Transfection of poly-I:C RNA, resulted in induction of IFIT1, IFIT2 and OAS2 (2'-5'-Oligoadenylate Synthetase 2) genes in Calu-3 and H1299 cells, indicating sensing of cytoplasmic foreign RNA is active in these cell types. This response was not observed in By counting poly(A) + or total RNA-seq reads spanning the junction of the viral leader and its downstream gene (82.6% of virus-mapping split reads), we accurately quantified the relative amounts of subgenomic viral mRNAs [33, 34] . We observed a consistent hierarchy of gene expression across time, mostly dominated by viral mRNAs encoding the M gene (Fig. 1D, Fig.   S1D ), similar to a recent report for the alpha Human CoV-229E (HCoV-229E) [35] . At later time points post infection, the relative amount of ORF7a generally increased ( Figure S1E ).

Notably, this approach failed to detect expression of leaders immediately adjoining ORF7b or ORF10 (Supplementary Table 3) . 

Although the infection is comparable in Caco-2 and Calu-3 cells, judging based on the amount of intracellular viral RNA and virion yield (see above), the host transcriptome responses were markedly different ( Fig. 2A, 2B ). In case of the SARS-CoV-2-infected Caco-2 cells, an increase in expression of a number of genes (activating transcription factor 3, ATF3; early growth response 1, EGR1; immediate early response 3, IER3;) was detected that are typically activated in response to ER stress. In contrast, Calu-3 cells showed a strong increase in expression of ISGs such as the IFIT1, IFIT2, and interferon genes. This response was absent in Caco-2 cells likely due to the reduced expression of pattern-recognition receptors, IFIH1 and DDX58 that activate transcription of target genes via interferon regulatory factors (IRF) and NF-κB signaling [36, 37] , which supports similar results for SARS-CoV infection [38] (Supplementary data 1).

When comparing the expression of ISGs in virus-infected Calu-3 cells, our data revealed that the expression levels of ISGs were on average about twice as high for SARS-CoV-2 compared to SARS-CoVinfected cells (Fig. 2C ) at similar amounts of viral RNAs present in the cells (Fig.   1C , right panel). This difference in the extent of the ISG response may be of clinical relevance, since cytokines [39] are among the induced ISGs (Fig. S2B) , and their expression might be connected with pathologies such as the acute respiratory distress syndrome (ARDS) in CoV infections [40, 41] .

Next, we compared the induction of ISG in SARS-CoV-2 between experiments, and with a recently published gene expression study of normal human bronchial epithelial (NHBE) cells, A549 cells with and without ACE2 expression, and Calu-3 cells upon infection with SARS-CoV-2 [21] (Supplementary Table 2 

To identify genes that might be altered independently of RNA sensor-triggered signal cascades, we compared gene expression changes between the Caco-2 and Calu-3 cell lines at 12 hpi with SARS-CoV-2 ( Fig. 2D, Fig S2A) . Two genes, arrestin-related domain-containing protein-3 (ARRDC3) and thioredoxin-interacting protein (TXNIP), stood out among the few that were significantly dysregulated upon infection with either viruses. Both genes encode proteins that are involved in regulation of signaling pathways [42] . ARRDC3 mediates G protein-coupled receptor lysosomal sorting and apoptosis-linked gene 2-interacting protein X (ALIX) ubiquitination [43] . ALIX is a Lys63-specific polyubiquitin binding protein that functions in retrovirus budding and Dengue virus propagation [44, 45] . TXNIP is involved in the regulation of glucose and lipid metabolism [46] , and has been shown to be involved in initiation and perpetuation of NLRP3 (nucleotide-binding domain and leucine-rich repeat and pyrin domain containing 3) inflammasome activation [47, 48] .

To conclude, most gene expression changes in response to SARS-CoV-2 infection are likely triggered by RNA sensors. However, there are a few exceptions, and the mechanism of induction, and the function of these proteins during virus replication remains to be elucidated.

In addition to assessing mRNA changes, we have also profiled small RNAs in the context of Calu-3 infections. Both viruses trigger a close to 16-fold upregulation of miR-155-3p, the "star" form, and an almost 3-fold upregulation of miR-155-5p ( Fig. 3A , 3B, S3A). Importantly, the primary miRNA precursor gene, miR-155 host gene (MIR155HG), was also upregulated in polyA-seq and total RNA-seq data by about 10-fold, suggesting that the increase of two miRNAs was primarily driven by transcription (Fig. S3B) .

Interestingly, the miRNA profiling identified small RNAs mapping to vault RNA (VTRNA) genes ( Fig. 3C, S3C ). The function of vtRNAs has not been fully elucidated, but the mature VTRNA1-1 was recently discovered as a negative regulator of autophagy [49] . VTRNAderived sRNAs can be processed by DICER and bound by Argonaute proteins [50, 51] . The analysis of scRNA-seq data showed that cellular transcriptomes grouped by infection and type of virus (Fig. 4A, S4C ). Two small groups made up of cells (cluster 13), derived from all time points and primarily, but not only SARS-CoV-2 infected cells showed high expression of the IFNB1 (Fig. 4B, 4D , S4D, S4E, S4F). As described above, we observed RNA sensingindependent expression of ARRDC3 in cells infected with either virus. ARRDC3, as well as the pro-apoptotic gene protein phosphatase 1 regulatory subunit 15A (PPP1R15A, also known as growth arrest and DNA-damage-inducible 34; GADD34), which had previously been related to ER stress [52] , were highly expressed in the interferon gene expression cluster, but also in cells with very high levels of SARS-CoV-2 RNA (cluster 11; Fig. 4B , S4G). The mechanism leading to interferon expression in cluster 13 remains to be investigated.

In agreement with the bulk RNA-seq data, we observed a strong increase of expression of ISGs, IFIT1 and IFIT2, in infected cells, and particularly those exposed to SARS-CoV-2 (Fig. 4D,   S4H ). Likewise, MIR155HG, though poorly detected, resembled the expression pattern of IFIT1 and IFIT2 genes (compare Fig. S4I with 4B, and S4J with 4C/S4H). In order to identify genes co-regulated with IFNB1, we performed a correlation analysis of cells binned by increasing IFNB1 and ARRDC3 expression level (Fig. S4L ). Using this approach we found a putative co-regulation of the four interferon lambda genes (IFNL1-4), the chemokine genes, C-X-C Motif Chemokine Ligand 9 (CXCL9) and C-C Motif Chemokine Ligand 5 (CCL5), and the cholesterol-25-hydroxylase gene, CH25H. This enzyme, as well as its product 25-hydroxycholesterol, have been shown to act against a range of viruses [53] . In addition, two other genes were found in this group, the sodium voltage-gated channel alpha subunit 3 gene (SCN3A) and the dual oxidase 1 gene (DUOX1) (Fig. S4M) , which are poorly expressed in cells outside cluster 13. Whereas SCN3A has previously not been described in the context of virus infections, DUOX1 promotes the innate immune defense to pathogens via the production of reactive oxygen species in mucosae [54] .

For a better visualization of the effect of infection on cell clustering, we also analyzed the cells without taking viral transcripts into account, confirming stronger alterations of the cellular transcriptome in SARS-CoV-2 compared to SARS-CoV infections (Fig. S4N ). Of note, the cells bearing interferon mRNAs clustered together independent of the virus.

To better understand the nature of interferon gene induction in the context of CoV infection, we performed RNA velocity, which uses sequencing reads originating from introns to measure the amount of nascent mRNA [55] . This method applies additional filtering and embeddings, 

The involvement of HSP90AA1, a highly-conserved molecular chaperone, in viral infections has since a long time been discussed to be involved in the infection of a range of viruses [59] . S7A ) and a cell viability assay (Fig. S7B ).

We performed gene expression profiling of three different human cell lines infected with SARS-CoV and SARS-CoV-2 at bulk and single-cell level. We show a particularly strong induction of ISGs in Calu-3 cells, including cytokines, by SARS-CoV-2 in both bulk and scRNA-seq experiments. For various CoVs, a range of mechanisms that interfere with interferon signaling has been reported [60] . For SARS-CoV, it was shown that ORF6 inhibits signal transducer and activator of transcription (STAT) signaling [61] and that IRF3 activity is impaired [8] . Since RNA levels per cell (Fig. S4) and on the population level were comparable ( Fig. 1) , it is tempting to speculate that such mechanisms could be less efficient in SARS-CoV-2 compared to SARS-CoV. Indeed, cytokine production was described to be connected to pathogenesis [21, 40, 41] . However, the extent to which observations in cell lines and other models are reproducible in animal models and humans will require further investigations [62] [63] [64] .

By comparing cell lines, we distinguish genes induced independently of the RNA sensing system, such as TXNIP and ARRDC3. Both genes are involved in signaling processes, and further investigations into their role in SARS-CoV-2 infection is warranted.

Small RNA profiling showed a strong induction of miR-155 in the infected cells. This miRNA has been associated with various virus infections [65] [66] [67] [68] . miR-155-3p is also a well-known regulator of immune cells, in particular T-cell differentiation [69, 70] . Involvement of this miRNA in the regulation of innate immunity has also been reported [71] . Recently, miR-155-5p expression was shown to be induced in mice infected with influenza A virus [72] .

Importantly, in this study, lung injury by ARDS was attenuated by deletion of miR-155, making this miRNA a potential therapeutic target in the context of COVID-19. Its role in SARS-CoV-2 infection and pathogenesis, whether it has a biological role in epithelial cells, and the potential for therapeutic interventions, e.g. through antisense oligonucleotide approaches, remains to be explored. The same holds true for the sRNAs generated from vtRNAs, and for other non-coding RNAs, which remain to be investigated in SARS-CoV-2 infection.

The scRNA-seq experiments provided a rich dataset to analyze host cell expression changes in response to infection. Surprisingly, the percentage of cells containing viral RNA was much higher than expected based on the MOI used for the infection experiments. This finding could also be explained by spreading of the cell-to-cell fusions, facilitated by the S protein on the cell surface [73] . Furthermore, the analysis of the scRNA-seq data of infected Calu-3 cells indicated a sequential activation of IRF and NF-κB target genes, and in particular, a putatively strong but transient induction of interferon genes. This could be due to a relatively short time window during the progression of infection, in which both IRF and NF-kB activity is sufficiently high to trigger interferon gene transcription [74, 75] .

Concomitantly with the interferon induction, we observed a mild activation of ACE2 transcription in cells infected with SARS-CoV-2. Changes in ACE2 mRNA levels in the context of interferon treatment and coronavirus infection have been described before [58, 76] .

Whereas the transcriptional induction (RNA velocity) was detectable, changes in mature mRNA levels were moderate in comparison to ISGs like IFIT1. Interestingly, SARS-CoV was shown to downregulate ACE2 protein levels in Vero cells [77] .

Coronaviruses induce ER stress and activate the unfolded protein response (UPR) in infected cells [78, 79] . We observed transient induction of the stress-responsive heat shock protein gene HSP90AA1 [80] , in the "slow-motion" infection model, H1299 cells, and 4hpi in Calu-3 cells.

HSP90 modulates UPR by stabilizing the ER stress sensor transmembrane kinases IRE1α [81] .

Inhibition of the HSP90 has previously been shown to slow down the replication of several viruses [59, 82, 83] . The reduction of SARS-CoV-2 growth by HSP90 inhibition was recently proposed based on a computational analysis of patient RNA sequencing data [84] . Here, we

show that inhibition of HSP90 by 17-AAG at high nanomolar concentrations can reduce virus replication in an in vitro infection model. Interestingly, IFIT2 mRNA levels seemed unaffected by HSP90 inhibition, supporting the "on-off-switch" independent of the amount of viral RNA observed in the scRNA-seq data. In addition, mRNA expression of the pro-inflammatory cytokines TNF and IL1B, which are implied in the progression of COVID-19 [85] , were strongly reduced. Since 17-AAG can induce apoptosis in cancer cells, which was not observed here for Calu-3 cells, the inhibitory effect of 17-AAG on viral replication needs to be expanded to studies in primary tissue infection models. Since several inhibitors of HSP90 with higher affinities have been in clinical development as anticancer agents [86] and advanced to phase 2 and 3 clinical trials, some of these compounds could be readily available to become part of a therapeutic strategy for COVID-19. 

Vero E6 (ATCC CRL-1586), Calu-3 (ATCC HTB-55), Calu-2 (ATCC HTB-37) and H1299 (ATCC CRL-5803) were cultivated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal calf serum, 1% non-essential amino acids, 1% L-glutamine and 1% sodium pyruvate (all Thermo Fisher Scientific) in a 5% CO2 atmosphere at 37 °C.

Transient transfection of eukaryotic cells was performed using X-tremeGENE™ siRNA transfection reagent (Roche) according to the manufacturer´s instructions. Briefly, 2x10^5 cells/ml were grown in 6-well plates for 24 h and fresh DMEM without antibiotics was added.

OptiPRO SFM™ (Gibco) was supplemented with 0.25 µg poly(I:C) (Invivogen) and 0.75 µl X-tremeGENE™ siRNA reagent, incubated for 15 min, and 100 µl transfection mix was added to the cells.

RNA was isolated from Trizol using the RNA clean and concentrator kit (Zymo). The RNA was reverse transcribed using maxima RT and subjected to qPCR as described [26] . Primers 

24 h prior to infection, the different cell lines were seeded to 70% confluence. The cells were washed once with PBS before virus (diluted in OptiPro serum-free medium) adsorption. After incubation for 1 h at 37 °C, 5% CO2 the virus-containing supernatant was discarded and cells were washed twice with PBS and supplied with DMEM as described above.

To determine the amount of infectious virus particles in the supernatant a plaque titration assay was performed. For the assay Vero E6 cells were seeded to confluence and infected with serial dilution of virus-containing cell culture supernatant diluted in OptiPro serum-free medium. One hour after adsorption, supernatants were removed and cells overlaid with 2.4% Avicel (FMC BioPolymers) mixed 1:1 in 2xDMEM. Three days post-infection the overlay was removed, and cells were fixed in 6% formaldehyde and stained with a 0.2% crystal violet, 2% ethanol and 10% formaldehyde.

The expression of human ACE-2 (hACE-2) was confirmed by Western blot analysis. For the preparation of total cell lysate cells were washed with PBS and lysed in RIPA Lysis Buffer 

Total and poly(A) + RNA-seq reads were mapped with STAR 2.7.3a to a combined genome comprised of GRCh38 and GenBank MN908947 (SARS-CoV-2) or AY310120 (SARS-CoV) using permissive parameters for noncanonical splicing [33, 91] . Viral genes were quantified by taking the top eight noncanonical splice events called by STAR across all total RNA-seq datasets according to the numbers of uniquely-mapping reads spanning the junction (Supplemental Table 3 ). To estimate levels of ORF1ab, insertions, soft-clipping events and split reads were filtered from virus-mapping reads, followed by intersection with positions 53-83 of the virus using bedtools, requiring a minimum of 24 nucleotides overlap to reflect the parameters STAR requires to call a noncanonical splice junction [92] . These counts were either combined with a count matrix of the human genes quantified by STAR and TMM/CPM normalized with edgeR ( Figure S1D ) or normalized by the total number of viral junctionspanning reads per time point ( Figure S1E ) [93] . Coverage plots were made from merged STAR-mapped BAM files, or from Bowtie-mapped small RNA-seq BAM files using ggsashimi [94] . This workflow was implemented with custom Python scripts in a Snakemake pipeline [95] .

Raw reads were preprocessed by trimming with cutadapt (version 2.9) in two passes, first trimming i) the Illumina TruSeq adaptor at the 3' end and allowing for one mismatch, ii) all 3'end bases with mean Phred score below 30 and iii) the three 5'end overhang nucleotides associated with the template-switching Clontech library preparation protocol.

In the second pass, poly(A)-tails were trimmed. Trimmed reads were mapped using bowtie (version 1.2.2) to a SARS genome consisting of the combined SARS-CoV and SARS-CoV-2 genomes using the non-standard parameters (-q -n 1 -e 80 -l 18 -a -m 5 -beststrata). Reads that did not align to the SARS-CoV genome were aligned to the GRCh38 genome. The expression of known miRNAs (miRBase 22 annotation) was estimated using mirdeep2 (version 2.0.0.7) and standard parameters.

The differential expression analysis used the limma [96] and edgeR [93] packages after applying the voom transformation to the TMM-normalized count data produced by mirdeep2.

For the different viral infections we contrasted SARS-CoV2-24h -SARS-CoV-2-4h with mock-24h -mock-4h in order to test for those miRNAs differentially expressed long after the infection having removed any effects seen in mock as well.

Starting from count tables, RNA sequencing results were analysed on a per run basis comparing infected samples to time matched mock experiments unless otherwise specified using DESeq2 [97] version 1.22.2. Genes with a maximum read count across samples of less than two were filtered out. Differentially expressed genes were defined as genes with an absolute fold change in mRNA abundance greater than 1.5 (log2 fold change of 0.58 -using DESeq2 shrunken log2 fold changes) and an adjusted p-value of less than 0.05 (Benjamini-Hochberg corrected).

Genes whose mRNAs were found to be differentially expressed were subjected to gene set overrepresentation analysis using the clusterProfiler package in R [98] . Specifically, gene sets from Gene Ontology (Molecular Function, Biological Process, Cellular Component) and

Methanol-fixed cells were centrifuged at 2,000 x g for 5 min, rehydrated in 1 mL rehydration buffer containing 0.01% PBS/BSA and 1:100 Superasein (Thermo Fisher), and resuspended in 400 µL rehydration buffer followed by passing through a 40 µm cell strainer. Encapsulation was done on the Nadia system (Dolomite biosystems) using the built-in standard procedure.

For library preparation, we followed the version 1.8 of the manufacturer's protocol, with adding a second-strand synthesis step [99] . 

After trimming one nucleotide from the 3' end of read one, count tables were generated using the pigx scRNA-seq pipeline [100] version 1.1.4.

All analysis shown in figure 4-6 was done similar as described previously [26] using Seurat and ggplot2 packages [101, 102] .

The HSP90 inhibitor 17-AAG was purchased from Sigma () and dissolved in DMSO. Cells 

The viral RNA from supernatant of infected cells was isolated using the NucleoSpin RNA virus isolation kit (Macherey-Nagel) according to the manufacturer's instructions. To determine the amount of viral genome equivalents the previously published assay specific for both SARS-CoV and SARS-CoV-2 Envelope gene [103] was used. Data analysis was done using LightCycler Software 4.1 (Roche).

Raw sequencing is available at the Gene Expression Omnibus database (GEO), identifier GSE148729

(https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE148729).

Supplementary data and supporting files such as scRNA-seq Seurat objects are available at www.mdc-berlin.de/singlecell-SARSCoV2. IFNB1   CXCL10  OAS2  TNFAIP3  IFNL2  RSAD2  MX1  USP18   IFNL3   NFKBIA  IFIT2  CXCL11  ATF3  IFIT1  IFI44L   0   2 

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Coronaviruses: an overview of their replication and pathogenesis

Fields Virology

SARS-coronavirus replication/transcription complexes are membrane-protected and need a host factor for activity in vitro

Coronavirus non-structural protein 16: evasion, attenuation, and possible treatments

Coronavirus endoribonuclease targets viral polyuridine sequences to evade activating host sensors

Inhibition of Beta interferon induction by severe acute respiratory syndrome coronavirus suggests a two-step model for activation of interferon regulatory factor 3

Pathogenic influenza viruses and coronaviruses utilize similar and contrasting approaches to control interferon-stimulated gene responses

SARS coronavirus accessory proteins

Continuous and Discontinuous RNA Synthesis in Coronaviruses

Membrane binding proteins of coronaviruses

Human coronavirus NL63 open reading frame 3 encodes a virion-incorporated N-glycosylated membrane protein

Coronavirus infection, ER stress, apoptosis and innate immunity

Severe acute respiratory syndrome coronavirus envelope protein regulates cell stress response and apoptosis

Regulation of the ER Stress Response by the Ion Channel Activity of the Infectious Bronchitis Coronavirus Envelope Protein Modulates Virion Release, Apoptosis, Viral Fitness, and Pathogenesis

Role of Heat Shock Proteins in Viral Infection

Porcine Epidemic Diarrhea Virus Induces Autophagy to Benefit Its Replication. Viruses

Verheije MH: miRNA repertoire and host immune factor regulation upon avian coronavirus infection in eggs

microRNAs in viral acute respiratory infections: immune regulation, biomarkers, therapy, and vaccines

Imbalanced host response to SARS-CoV-2 drives development of COVID-19

The use of single-cell RNA-Seq to understand virus-host interactions

HSV-1 single-cell analysis reveals the activation of anti-viral and developmental programs in distinct sub-populations

Extreme heterogeneity of influenza virus infection in single cells

Defining the Transcriptional Landscape during Cytomegalovirus Latency with Single

Single-cell RNA-sequencing of herpes simplex virus 1-infected cells connects NRF2 activation to an antiviral program

Single-cell transcriptional dynamics of flavivirus infection

MicroRNAs: genomics, biogenesis, mechanism, and function

Rajewsky N: miRDeep2 accurately identifies known and hundreds of novel microRNA genes in seven animal clades

Isolation of human pathogenic viruses from clinical material on CaCo2 cells

SARS-CoV-2 and SARS-CoV differ in their cell tropism and drug sensitivity profiles

Comparative tropism, replication kinetics, and cell damage profiling of SARS-CoV-2 and SARS-CoV with implications for clinical manifestations, transmissibility, and laboratory studies of COVID-19: an observational study

The architecture of SARS-CoV-2 transcriptome

High-Resolution Analysis of Coronavirus Gene Expression by RNA Sequencing and Ribosome Profiling

Direct RNA nanopore sequencing of full-length coronavirus genomes provides novel insights into structural variants and enables modification analysis

Interferon-stimulated genes: a complex web of host defenses

Interferon-Stimulated Genes: What Do They All Do?

Dynamic innate immune responses of human bronchial epithelial cells to severe acute respiratory syndrome-associated coronavirus infection

Global landscape of mouse and human cytokine transcriptional regulation

Pathogenic human coronavirus infections: causes and consequences of cytokine storm and immunopathology

Cytokine release syndrome in severe COVID-19

The arrestin fold: variations on a theme

The alpha-arrestin ARRDC3 mediates ALIX ubiquitination and G protein-coupled receptor lysosomal sorting

ALIX is a Lys63-specific polyubiquitin binding protein that functions in retrovirus budding

Dengue virus requires apoptosis linked gene-2-interacting protein X (ALIX) for viral propagation

TXNIP in Metabolic Regulation: Physiological Role and Therapeutic Outlook

Initiation and perpetuation of NLRP3 inflammasome activation and assembly

Thioredoxin-interacting protein mediates ER stress-induced beta cell death through initiation of the inflammasome

The Small Non-coding Vault RNA1-1 Acts as a Riboregulator of Autophagy

The noncoding RNA of the multidrug resistance-linked vault particle encodes multiple regulatory small RNAs

Assessing the gene regulatory properties of Argonaute-bound small RNAs of diverse genomic origin

Hairy and enhancer of split 1 (HES1) protects cells from endoplasmic reticulum stress-induced apoptosis through repression of GADD34

Interferon-inducible cholesterol-25-hydroxylase broadly inhibits viral entry by production of 25-hydroxycholesterol

Roles of DUOX-mediated hydrogen peroxide in metabolism, host defense, and signaling

RNA velocity of single cells

Transcriptional profiling of interferon regulatory factor 3 target genes: direct involvement in the regulation of interferon-stimulated genes

SARS-CoV-2 receptor ACE2 is an interferon-stimulated gene in human airway epithelial cells and is detected in specific cell subsets across tissues

Broad action of Hsp90 as a host chaperone required for viral replication

Interaction of SARS and MERS Coronaviruses with the Antiviral Interferon Response

Severe acute respiratory syndrome coronavirus ORF6 antagonizes STAT1 function by sequestering nuclear import factors on the rough endoplasmic reticulum/Golgi membrane

Shotgun Transcriptome and Isothermal Profiling of SARS-CoV-2 Infection Reveals Unique Host Responses, Viral Diversification, and Drug Interactions

Identification of Drugs Blocking SARS-CoV-2 Infection using Human Pluripotent Stem Cell-derived Colonic Organoids

SARS-CoV-2 productively infects human gut enterocytes

Environmental pollutants modulate RNA and DNA virus-activated miRNA-155 expression and innate immune system responses: Insights into new immunomodulative mechanisms

MicroRNA 155 and viral-induced neuroinflammation

Roles of microRNAs in the life cycles of mammalian viruses

An update on the role of miRNA-155 in pathogenic microbial infections

MicroRNAs as regulatory elements in immune system logic

Regulation of the germinal center response by microRNA-155

and its star-form partner miR-155* cooperatively regulate type I interferon production by human plasmacytoid dendritic cells

Increased expression of microRNA-155-5p by alveolar type II cells contributes to development of lethal ARDS in H1N1 influenza A virus-infected mice

The molecular biology of coronaviruses

Cell fate in antiviral response arises in the crosstalk of IRF, NF-kappaB and JAK/STAT pathways

NF-kappaB and IRF pathways: cross-regulation on target genes promoter level

Molecular pathology of emerging coronavirus infections

Differential downregulation of ACE2 by the spike proteins of severe acute respiratory syndrome coronavirus and human coronavirus NL63

Regulation of Stress Responses and Translational Control by Coronavirus

The coronavirus spike protein induces endoplasmic reticulum stress and upregulation of intracellular chemokine mRNA concentrations

Regulation and function of the human HSP90AA1 gene

Heat shock protein 90 modulates the unfolded protein response by stabilizing IRE1alpha

Inhibition of HSP90 attenuates porcine reproductive and respiratory syndrome virus production in vitro

Heat Shock Protein 90 Ensures Efficient Mumps Virus Replication by Assisting with Viral Polymerase Complex Formation

Drug Repositioning Suggests a Role for the Heat Shock Protein 90 Inhibitor Geldanamycin in Treating COVID-19 Infection

COVID-19: A New Virus, but a Familiar Receptor and Cytokine Release Syndrome

Inhibition of Heat Shock Protein 90 as a Novel Platform for the Treatment of Cancer

Autophagy is involved in regulating influenza A virus RNA and protein synthesis associated with both modulation of Hsp90 induction and mTOR/p70S6K signaling pathway

Hsp90 inhibitor induces autophagy and apoptosis in osteosarcoma cells

Binding of the pathogen receptor HSP90AA1 to avibirnavirus VP2 induces autophagy by inactivating the AKT-MTOR pathway

Analysis of SARS-CoV-2-controlled autophagy reveals spermidine, MK-2206, and niclosamide as putative antiviral therapeutics

STAR: ultrafast universal RNA-seq aligner

BEDTools: a flexible suite of utilities for comparing genomic features

Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation

ggsashimi: Sashimi plot revised for browser-and annotation-independent splicing visualization

Snakemake--a scalable bioinformatics workflow engine

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

Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

clusterProfiler: an R package for comparing biological themes among gene clusters

Highly Efficient, Massively-Parallel Single-Cell RNA-Seq Reveals Cellular States and Molecular Features of Human Skin Pathology

PiGx: reproducible genomics analysis pipelines with GNU Guix

Wickham H: ggplot2: Elegant Graphics for Data Analysis

Integrating single-cell transcriptomic data across different conditions, technologies, and species

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

The authors wish to thank Melanie Brinkmann, Leif Sander, Marco Hein, Joseph Luna, Friedemann Weber and Robert Zinzen for comments and discussion; Jeannine Wilde, Tatiana Borodina, Daniele Franze and Nouhad Benlasfer for sequencing and technical assistance; and