key: cord-353612-9ux181xg
authors: Josset, Laurence; Menachery, Vineet D.; Gralinski, Lisa E.; Agnihothram, Sudhakar; Sova, Pavel; Carter, Victoria S.; Yount, Boyd L.; Graham, Rachel L.; Baric, Ralph S.; Katze, Michael G.
title: Cell Host Response to Infection with Novel Human Coronavirus EMC Predicts Potential Antivirals and Important Differences with SARS Coronavirus
date: 2013-04-30
journal: mBio
DOI: 10.1128/mbio.00165-13
sha: 
doc_id: 353612
cord_uid: 9ux181xg

A novel human coronavirus (HCoV-EMC) was recently identified in the Middle East as the causative agent of a severe acute respiratory syndrome (SARS) resembling the illness caused by SARS coronavirus (SARS-CoV). Although derived from the CoV family, the two viruses are genetically distinct and do not use the same receptor. Here, we investigated whether HCoV-EMC and SARS-CoV induce similar or distinct host responses after infection of a human lung epithelial cell line. HCoV-EMC was able to replicate as efficiently as SARS-CoV in Calu-3 cells and similarly induced minimal transcriptomic changes before 12 h postinfection. Later in infection, HCoV-EMC induced a massive dysregulation of the host transcriptome, to a much greater extent than SARS-CoV. Both viruses induced a similar activation of pattern recognition receptors and the interleukin 17 (IL-17) pathway, but HCoV-EMC specifically down-regulated the expression of several genes within the antigen presentation pathway, including both type I and II major histocompatibility complex (MHC) genes. This could have an important impact on the ability of the host to mount an adaptive host response. A unique set of 207 genes was dysregulated early and permanently throughout infection with HCoV-EMC, and was used in a computational screen to predict potential antiviral compounds, including kinase inhibitors and glucocorticoids. Overall, HCoV-EMC and SARS-CoV elicit distinct host gene expression responses, which might impact in vivo pathogenesis and could orient therapeutic strategies against that emergent virus.

partially explaining the large host range of HCoV-EMC. This was somewhat surprising, as coronaviruses generally show strict host specificity.

While recent identification of the crystal structure of HCoV-EMC protease suggests that a wide-spectrum CoV protease inhibitor could block the catalytic site (12) , there is currently no proven antiviral treatment for HCoV-EMC. Viruses rely on host factors to replicate and often hijack cellular processes initiated in response to infection to ensure efficient replication (13) . Targeting cellular responses has been shown to inhibit viral replication (13, 14) . Furthermore, immunomodulatory drugs that reduce the excessive host inflammatory response to respiratory viruses, as seen with influenza virus infections, have therapeutic benefit (reviewed in reference 15). Several genome-based drug repurposing strategies successfully identified known drugs that could be reused to treat lung cancer, inflammatory bowel disease (16) , and influenza virus infection (14) . Such an approach has the advantage of accelerating treatment availability, which could be crucial in case of an outbreak of an emerging pathogen.

Overall, differences in viral sequences, host cell receptor, and host range indicate that HCoV-EMC and SARS-CoV may have distinct strategies for interacting with their hosts. This fact could impact treatment strategies. To begin to assess this question, we compared the host response of human cells to HCoV-EMC and SARS-CoV infection using global transcriptomic profiling. Our goal was to gain a rapid and comprehensive assessment of the host response to HCoV-EMC infection that could guide research on this emerging virus. Importantly, we used this information to computationally predict antiviral treatment and identified a broad down-regulation of the antigen presentation pathway that may be important in vivo for the development of an adaptive immune response.

To characterize HCoV-EMC, we sought a cell line that mimics the human airway. Calu-3 cells, derived from epithelial cells lining the human conducting airway, can be differentiated into polarized ciliated cells and permit robust replication of several respiratory viruses, including SARS-CoV, influenza A virus, and respiratory syncytial virus (RSV) (17) (18) (19) (20) . Therefore, Calu-3 cells were infected with HCoV-EMC using a multiplicity of infection (MOI) of 5, and results were compared with those for Calu-3 cells infected with SARS-CoV Urbani at the same MOI. The viruses replicated to similar levels, with both peaking at Ͼ10 7 PFU at 24 h postinfection (hpi) (Fig. 1) . While SARS-CoV has been previously shown to maintain steady replication and cell viability to 72 hpi (19) , HCoV-EMC induced substantial cytopathic effect at 18 to 24 hpi, with significant cell rounding and detaching. Together, these data suggest that although HCoV-EMC and SARS-CoV exhibit similar replication kinetics, they elicit different host responses in lung epithelial cells.

HCoV-EMC induces earlier and different transcriptional changes than SARS-CoV. To assess the global host transcriptional response following infection with HCoV-EMC, samples of infected Calu-3 cells were collected throughout a 24-h time course postinfection. Genes that were differentially expressed (DE) compared to time-matched mock-infected controls were determined using the statistical cutoff of a q value of Ͻ0.01 and an absolute log 2 (FC) of Ͼ1. These genes were compared to DE genes from Calu-3 cells infected with SARS-CoV across a 72-h time course using the same statistical cutoff (Fig. 2) .

As previously observed (19) , SARS-CoV is able to replicate actively with a surprisingly small number of significant transcriptional changes before 24 hpi (from 5 DE genes at 0 hpi to 47 genes at 12 hpi) ( Fig. 2A) . Similarly, HCoV-EMC induced fewer transcriptional changes before 12 hpi than after. However, HCoV-EMC replication at early times postinfection induced more changes than SARS-CoV, with the number of DE genes ranging from 28 at 0 hpi to 206 genes at 12 hpi ( Fig. 2A) . At later times postinfection, a massive host response was observed during HCoV-EMC infection, with 6,532 DE genes at 18 hpi and 11,664 genes at 24 hpi, while SARS-CoV induced changes of only 792 genes at 24 hpi with maximum changes at 48 and 54 hpi of 6,496 and 6,498 genes, respectively.

To evaluate the similarity of transcriptional dysregulation between HCoV-EMC and SARS-CoV, we determined the percentage of overlap among DE genes changing in the same direction at each time point (Fig. 2B) . The intersection between up-or downregulated DE genes for each condition was calculated separately and then averaged to determine the percentage of intersecting genes (Fig. 2B) . The overlap between signatures at late times postinfection for a single virus was very high; for example, 89% of the genes DE by HCoV-EMC at 18 hpi were also DE at 24 hpi, and on average 77% of the DE genes at 24 hpi with SARS-CoV were also DE at later times postinfection. However, the overlap between SARS-CoV and HCoV-EMC at the same time point was low; for example, only 3% of the DE genes in response to HCoV-EMC at 24 hpi were also DE by SARS-CoV at this time point. Of note, the intersection of DE genes between the two infections was higher when later times postinfection for SARS-CoV were compared with earlier time points for HCoV-EMC. On average 22% of the FIG 1 HCoV-EMC replicates at a level similar to that of SARS-CoV in human epithelial cells. Triplicate wells of Calu-3 2B4 cells were infected with HCoV-EMC (MOI, 5). Medium from each well was collected and analyzed by plaque assay for viral growth kinetics in VeroE6 cells. The corresponding cells were harvested for transcriptomic analysis. SARS-CoV titers after infection of Calu-3 2B4 cells at an MOI of 5 were determined using the same method (19) . The error bars represent the standard deviations among triplicate cell samples. Josset et al. HCoV-EMC signatures from 12 to 24 hpi overlapped the signatures of SARS-CoV from 36 to 72 hpi. Together, these results indicate that HCoV-EMC induced both a more robust and a largely different host response compared to that induced by SARS-CoV at similar times postinfection. However, changes induced from 12 to 24 hpi after HCoV-EMC were more similar to changes induced late after SARS-CoV infection (24 to 72 hpi).

HCoV-EMC massively dysregulates the host transcriptome at late times postinfection. At late times postinfection, HCoV-EMC induced drastic changes in the host transcriptome with 12,392 DE genes at 18 hpi and/or 24 hpi. To characterize this late signature, we compared the log 2 fold change (log 2 FC) expression values of these 12,392 genes after infection with SARS-CoV or HCoV-EMC ( Fig. 3 and www.systemsvirology.org). There were 3,474 genes (28%) expressed similarly to those in SARS-CoVinfected samples at the same or later times postinfection and 8,918 genes (72%) expressed differently than during SARS-CoV infection. These genes were further clustered according to their expression pattern in HCoV-EMC infected cells (clusters I and III contain genes up-regulated after HCoV-EMC infection at 18 to 24 hpi; clusters II and IV contain down-regulated genes). Enrichment in canonical pathways for each of this cluster was performed and is shown Fig. 3B .

Genes that were up-regulated by both HCoV-EMC and SARS-CoV (cluster I), though later by SARS-CoV than by HCoV-EMC, were primarily related to viral recognition and the activation of innate immune pathways: the IL-17-related pathway and activation of interferon regulatory factor (IRF) by cytosolic pattern recognition receptors. Genes down-regulated by both viruses (cluster II) were related to metabolism.

Genes specifically up-regulated by HCoV-EMC were statistically enriched in only 2 pathways: cAMP-mediated signaling and protein ubiquitination. Because ubiquitination is involved in the innate immune antiviral response (21) , this could represent an interesting mechanism to explore further. Finally, genes specifically down-regulated after HCoV-EMC infection were largely related to the antigen presentation pathway. In addition, the other identified canonical pathways are strongly related to lymphocyte signaling processes. Together, these data suggest that HCoV-EMC may quickly move to interfere with elements of the adaptive immune response, in contrast to SARS-CoV.

The antigen presentation pathway is broadly downregulated after HCoV-EMC infection. As down-regulation of the antigen presentation pathway may have important implications for the development of the adaptive response, we more closely examined changes in this pathway after infection with HCoV-EMC (Fig. 4) . Twenty-two genes related to major histocompatibility complex (MHC) class I, MHC class II, or antigen presentation were found to be down-regulated after HCoV-EMC infection (Fig. 4A) , while in contrast, the vast majority of these genes were up-regulated by SARS-CoV infection after 36 hpi (Fig. 4B ). Interestingly, transcriptional regulators of MHC class I (NLRC5) and MHC class II (CIITA) were transiently up-regulated early after HCoV-EMC infection (at 12 and 0 hpi, respectively). Other genes were down-regulated starting at 12 hpi. Importantly, both MHC class I genes (HLA-A, -B, -C, -E, and -G) and class II genes (HLA-DMB, -DPA1, -DPB1, -DQA1, -DRA, -DRB1, -DRB3, -DRB4, and -DRB5) had decreased expression values after infection. Several genes from the MHC I pathway, including proteasome genes (PSMB8 and PSMB9), genes from the peptide-loading complex (PDIA3 and TAPBP), and B2M, were down-regulated, as well as the gene for the invariant chain (CD74), which belongs to the MHC II pathway. Down-regulation of genes within MHC I and II pathways was confirmed by quantitative reverse transcription-PCR (RT-PCR), with all 9 genes tested having significantly decreased expression in HCoV-EMC samples (Fig. 4C) . Together, the data show that HCoV-EMC and SARS-CoV induce opposite regulation of the entire antigen presentation pathway.

Early and sustained transcriptional changes may be reverted by kinase inhibitors and glucocorticoids. With the goal of identifying possible drugs that will modulate the host response throughout infection and from early times postinfection, we focused on characterizing the early transcriptional changes induced by HCoV-EMC that remained stable throughout the infection. Among the 348 genes DE early after infection (at 0, 3, 7, and/or 12 hpi), 207 (59%) remained dysregulated, with the same pattern at later times postinfection. We chose to focus on this 207-gene signature to avoid transient events that were specific to one time postinfection and to exclude genes with inconsistent expression patterns across the course of infection. Expression values for these genes are displayed in Fig. 5A and are available at www.systems virology.org. This early signature was enriched in genes related to the inflammatory response with IL-17, tumor necrosis factor 2 (TNFR2)-related pathways, and predicted activation of chemotaxis of leukocytes (see Table S1 in the supplemental material). None of the 207 genes were DE early after SARS-CoV infection; however, expression of 51 genes (24.5%) was changed after 24 hpi. This subset of 51 genes was enriched in cell viability molecules, glucocorticoid receptor signaling, and IL-17-related pathways.

In an effort to identify potential drugs that could block the host response to HCoV-EMC, we used the early 207-gene signature and IPA (Ingenuity pathway analysis) upstream regulator analysis (see Materials and Methods) to identify the upstream regulators that may be responsible for gene expression changes observed early after infection. Upstream regulators are defined as any molecule that can affect the expression of another molecule, including cules that are known to induce the activation of inflammatory genes that were also found up-regulated in the early signature. The top predicted inhibited regulators included four kinase inhibitors (SB203580, LY294002, U0126, and SP600125) and one corticosteroid (triamcinolone acetonide). These molecules are known to down-regulate the same genes that were up-regulated early and across HCoV-EMC infection. In addition, several kinase inhibitors (including SB203580, LY294002, and U0126) and a glucocorticoid (dexamethasone) were also predicted to be negative regulators of genes changed similarly after SARS-CoV and HCoV-EMC infection at late times postinfection (see Table S2 in the supplemental material). Therefore, we hypothesize that treating cells with these drugs might partially revert CoV signatures and inhibit deleterious host responses and/or viral replication in the case of both SARS-CoV and HCoV-EMC. In silico and in vitro evidence for kinase inhibitors as potential anti-CoV. To test these hypotheses, we first analyzed whether cells treated with identified kinase inhibitors have gene expression profiles opposite to the one induced after viral infection. To this end, we used Connectivity Map (cmap), which is a database of more than 1,309 drug transcriptional signatures in several cell lines (22) . This data-driven tool allows the identification of molecules that induce similar or opposite transcriptional changes relative to the query signature, based on their connectivity scores. The connectivity score is a value between ϩ1 and Ϫ1, where a high positive score indicates that the drug induces changes similar to those induced by viral infection, while a high negative score indicates that the drug reverses the expression of the HCoV-EMC signature. Cmap includes transcriptional profiles of several kinase inhibitors, including two compounds that were predicted to be potential negative regulators of viral response in IPA upstream regulator analysis (SB203580 and LY294002). SB203580 had a connectivity score of Ϫ0.733 in MCF7 cells (Data set S1 in the supplemental material), and LY294002 had negative scores in 4 of the 5 cell lines tested in cmap (Ϫ0.281 in HL60, Ϫ0.146 in MCF7, Ϫ0.252 in PC3, and Ϫ0.811 in SKMEL5), indicating that these two drugs reversed the expression profile of HCoV-EMC signature. Other related kinase inhibitors present in cmap are U0125 (a derivative of U0126) and SB202190, which both had negative scores in PC3 cells (Ϫ0.649 and Ϫ0.406, respectively).

To further validate the potential effectiveness of identified kinase inhibitors, we evaluated the ability of the top predicted negative upstream regulator, SB203580, to interfere with viral replication (Fig. 5C) . Importantly, this kinase inhibitor was predicted to regulate genes that were DE similarly by SARS-CoV and HCoV-EMC at late times postinfection (see Table S2 in the supplemental material) and could therefore inhibit both viruses' replication. Treating cells with 5 M SB203580 following SARS-CoV infection resulted in a significant decrease of 20% in the log 10 viral titer at 24 hpi (P Ͻ 0.01); these replication levels are similar to those in interferon (IFN)-treated cells. SB203580 was less effective against EMC-CoV, showing no efficacy with posttreatment; however, pretreatment of infection resulted in significant decreases of 15% and 7% of the log 10 viral titer, at 24 and 48 hpi, respectively (P Ͻ 0.01). Efficacy against SARS-CoV can be explained by the fact that all genes that were DE after SARS-CoV infection were predicted to be regulated by kinase inhibitors (cluster I and II genes [ Fig. 3 ] and additional genes specific to SARS-CoV [data not shown]), while the 8,918 genes specific to HCoV-EMC (clusters III and IV [ Fig. 3 ]) did not include any kinase inhibitor in their upstream negative regulators (see Table S2 ). In addition, host response dynamics to SARS-CoV and HCoV-EMC were different, with a delayed and more limited response to SARS-CoV allowing SB203580 posttreatment to be effective. In contrast, EMC-CoV rapidly induced host expression changes and thus required SB203580 pretreatment for measurable effect. As genes specific to HCoV-EMC were predicted to be negatively regulated by other classes of molecules (like gemfibrozil, which targets PPAR␣; z ϭ 2.96), it will be important to determine whether using such molecules in combination with kinase inhibitors like SB203580 could provide a better inhibition of viral replication. Together, these data demonstrate the efficacy of SB203580 against two unique CoVs and validate in silico approaches to predict effective drug treatment for emergent viruses.

HCoV-EMC was isolated from a patient who died from an acute respiratory disease similar to that caused by SARS-CoV. However, there are several indicators that the host responses to these two viruses may be significantly different. Several cases of HCoV-EMC infection have resulted in renal failure, which has rarely been observed in SARS-CoV infection. In addition, SARS-CoV and HCoV-EMC do not use the same cell receptor, and there are important differences in their genomic sequences. This study adds strength to the assertion that "HCoV-EMC is not the same as SARS-CoV" (23) . Indeed, even though we identified specific characteristics of the SARS-CoV response in the HCoV-EMC signatures, HCoV-EMC induced robust and specific transcriptional responses that were distinct from those induced by SARS-CoV, including the broad down-regulation of MHC molecules.

This study is the first global transcriptomic analysis of the cellular response to HCoV-EMC infection. Kindler et al. performed RNA-Seq on human airway epithelium (HAE) cells infected with HCoV-EMC (24). However, their analysis was focused on viral sequences and did not include a genome-wide analysis of the host response. They did, however, use RT-qPCR (quantitative PCR) to compare expression levels of a set of 15 genes, including IFN, RNA sensor molecules, and IFN-stimulated genes (ISGs), following infection with HCoV-EMC, SARS-CoV, or HCoV-229E (MOI 0.1). In our study, we confirm that SARS-CoV and HCoV-EMC induce a similar up-regulation of RNA sensor molecules, such as RIGI, MDA5, and two of three genes of ISGF3 (IRF9 and STAT1) (genes in cluster I [ Fig. 3] ). Of note, HCoV-EMC titers were up to 10 2fold higher than those of SARS-CoV in HAE cells (24) Activation of similar innate viral-sensing pathways by HCoV-EMC and SARS-CoV is not surprising given the conservation of this mechanism to detect foreign RNA and familial relationships of the viruses. We also found that both viruses induced proinflammatory cytokines related to IL-17 pathways. It has previously been shown that IL-17A-related gene expression exacerbates severe respiratory syncytial virus (RSV) or influenza virus infection (25, 26) . IL-17A was predicted to be activated throughout infection with HCoV-EMC and may induce immune-mediated pathology that possibly contributes to a high mortality rate. IL-17A is known to be produced by T-helper cells, but its expression in Calu3 cells was increased up to 2-fold at 24 hpi after HCoV-EMC infection. Interestingly, IL-17C and IL-17F, which can be produced by epithelial cells under certain inflammatory conditions and which activate pathways similar to IL-17A-mediated responses (27) , were increased earlier and to a greater extent following HCoV-EMC infection (up to 3-fold at 18 hpi for IL-17C and 4-fold at 7 hpi for IL-17F). Therefore, further study of the IL-17 response may provide interesting targets to limit lung injury (26) .

A main difference between responses to HCoV-EMC and SARS-CoV was the specific down-regulation of the antigen presentation pathway after HCoV-EMC infection. In contrast, these genes were found to be up-regulated after SARS-CoV infection. Several viruses have evolved mechanisms to inhibit both the MHC class I (reviewed in references 28 and 29) and class II (reviewed in reference 30) pathways. While expression of MHC class II is usually limited to professional antigen-presenting cells, human lung epithelial cells constitutively express this complex (31) . Our data demonstrated down-regulation of the MHC class II transactivator (CIITA) after HCoV-EMC infection, a finding that possibly explains decreases in MHC class II molecule expression; this is a common viral strategy used to block that pathway (30) . MHC class II inhibition can prevent class II-mediated presentation of endogenous viral antigens produced within infected cells and impair the adaptive immune response. Similarly, MHC class I genes were also down-regulated after HCoV-EMC infection; decreasing expression of MHC class I can attenuate CD8 T-cell-mediated recognition of infected cells and could allow immune evasion by HCoV-EMC. Finally, PSMB8 and PSMB9, parts of the immunoproteasome, were also down-regulated by HCoV-EMC; these components replace portions of the standard proteasome and enhance production of MHC class I binding peptides (32) . In their absence, proteins targeted for degradation may not generate peptides that robustly bind MHC class I, thus limiting their presentation. Down-regulation of PSMB8 and PSMB9 could counteract the host response to viral infection, including up-regulation of ubiquitins and ubiquitin ligases observed during HCoV-EMC infection (Fig. 3B ) that may ineffectively target viral protein for degradation. Together, the inhibition of MHC class I and II as well as immunoproteasome construction may have an important impact on the in vivo adaptive immune response against HCoV-EMC.

While there is no proven effective antiviral therapy against SARS-CoV (33), several molecules have in vitro antiviral activity, including ribavirin, lopinavir, and type I IFN, but their benefits for patients are unclear (33) . IFN-␣ pretreatment of cells has been shown to inhibit HCoV-EMC replication (24) , but no direct antiviral therapies have been reported. Targeting host factors impor-tant for the virus, instead of the virus itself, has been investigated for HIV (34) and influenza virus (13) . For example, inhibiting upstream regulators (such as NF-B) that control the host response to influenza virus infection has been shown to reduce virus replication in vitro and in mice (35) . Inhibition of immunophilins that interact with the viral nonstructural protein 1 (Nsp1) resulted in potent inhibition of SARS-CoV replication (36, 37) . In this study, we characterized upstream regulators predicted to be activated (e.g., NF-B and IL-17, which could be targeted with specific inhibitors) and upstream regulators predicted to be inhibited.

The top five inhibited regulators included one glucocorticoid and four kinase inhibitors; these drugs may be able to directly block part of the host response and impact viral replication/pathogenesis. Among them, LY294002, a potent inhibitor of phosphatidylinositol 3 kinase (PI3K), has known antiviral activity, inhibiting the replication of influenza virus (38) , vaccinia virus (39) , and HCMV (40) . SB203580, an inhibitor of p38 MAPK, is also an effective antiviral against the encephalomyocarditis virus (41), RSV (42) , and HIV (43) . LY294002 and SB203580 were also identified in Connectivity Map, a database of drug-associated gene expression profiles (22) , as molecules reversing components of the HCoV-EMC gene expression signature. Finally, SB203580 showed promising antiviral results against both HCoV-EMC and SARS-CoV in our in vitro assay (Fig. 4C) . Further extensive studies, including dose-response tests and tests of other kinases inhibitors, are ongoing. Nonetheless, these results validate our genomebased drug prediction, which allows rapid identification of effective antivirals. Despite central roles of PI3K and MAPK pathways in regulating multiple cellular processes, many kinase inhibitors targeting these pathways have been shown to be safe and well tolerated in vivo (reviewed in references 44 and 45). It has been hypothesized that mitogenic MAPK and survival PI3K/Akt pathways may be of major importance only during early development of an organism and may be dispensable in adult tissues (13) . Several drugs targeting JNK, PI3K, and MEK have shown promising therapeutic potential in humans against a variety of diseases, including cancer and inflammatory disorder (44, 45) . p38 MAPK inhibitors have also been evaluated in humans, but the first generation of molecules, including SB203580, has a high in vivo toxicity (liver and/or central nervous system). However, development of novel nontoxic inhibitors (e.g., ML3403) (46), more selective molecules (e.g., AS1940477) (47) , and administration via inhalation (48) are promising strategies for use of this class of inhibitor for treatment of pulmonary disease. Overall, these results indicate that kinase inhibitors could be used as broad anti-CoV agents which might be combined with other host-targeting molecules, like peroxisome proliferator-activated receptor ␣ (PPAR␣) agonists, to better inhibit HCoV-EMC replication.

In conclusion, using global gene expression profiling, we have shown that HCoV-EMC induces a dramatic host transcriptional response, most of which does not overlap the response induced by SARS-CoV. This study highlights the advantages of highthroughput "-omics" to globally and efficiently characterize emerging pathogens. The robust host gene expression analysis of HCoV-EMC infection provides a plethora of data to mine for further hypotheses and understanding. Host response profiles can also be used to quickly identify possible treatment strategies, and we anticipate that host transcriptional profiling will become a gen-eral strategy for the rapid characterization of future emerging viruses.

Calu-3 2B4 cells, a clonal population of Calu-3 cells sorted for ACE2 expression, were grown in minimum essential media (MEM; Gibco) supplemented with 20% fetal bovine serum (HyClone) and 1% antibiotic antimycotic (Gibco). Viral titrations and propagation HCoV-EMC were performed in VeroE6 cells using standard methods. Human coronavirus EMC 2012 (HCoV-EMC) was received from Bart L. Haagmans (Erasmus Medical Center, Rotterdam, Netherlands) via MTA. Experiments with SARS-CoV were previously performed under the same conditions using the same cells (19) .

Calu-3 cell infections. All work was performed in a biosafety level 3 (BSL3) facility supported by redundant exhaust fans, and personnel protective equipment was worn, including Tyvek suits, hoods, and HEPAfiltered powered air-purifying respirators as previously described (49) . Cells were washed with phosphate-buffered saline (PBS) and inoculated with virus at an MOI of 5 or mock diluted in PBS for 40 min at 37°C. Following inoculation, cells were washed 3 times, and fresh medium was added to signify time zero. Triplicate samples of mock-infected and virusinfected Calu-3 2B4 cells were harvested between 0 and 72 hpi.

RNA isolation and microarray processing. RNA isolation from Calu-3 2B4 cells, subsequent hybridization to Agilent 4ϫ44K human HG arrays, and processing of raw data were performed as previously described (18) . We analyzed triplicates for each time postinfection except for 12 hpi, for which one replicate did not pass RNA quality control and was therefore excluded. For direct comparison with SARS-CoV-infected cells, raw data from HCoV-EMC experiments were quantile normalized together with the SARS-CoV data set (GEO series accession number GSE33267). All probes were required to meet Agilent Feature Extraction QC criteria for all replicates at at least one infection time point (33,773 probes passed QC filtering). For each sample, a log 2 fold change (log 2 FC) value was calculated as the difference between log 2 normalized data for this sample and the average of log 2 normalized data for time-and data set-matched mock-infected samples.

Statistical analysis. Differential expression was determined by comparing SARS-CoV-or HCoV-EMC-infected replicates to time-and data set-matched mock-infected controls, based on a linear model fit for each probe using the R package Limma (50). Criteria for differential expression were an absolute log 2 FC of 1 and a q value of Ͻ0.01 calculated using a moderated t test with subsequent Benjamini-Hochberg correction. Differentially expressed (DE) genes after HCoV-EMC infection at early times postinfection were defined as all genes DE at 0, 3, 7, or 12 hpi, and genes DE at late times postinfection were defined as all genes DE at 18 or 24 hpi. To identify genes with similar patterns of variation at early and late times postinfection, early and late signatures were intersected considering upregulated and down-regulated genes separately and then combined to define the signature of 207 stable genes. To cluster the genes DE at late times postinfection following HCoV-EMC infection, differential expression was also calculated directly between SARS-CoV-and HCoV-EMCinfected samples using the same criteria. A comparison of similar times postinfection was performed, except for HCoV-EMC-infected samples at 18 hpi, which were compared to SARS-CoV samples at 24 hpi. HCoV-EMC samples at 24 hpi were compared with SARS samples at 24 hpi and later times postinfection (30, 36, 48, 54, 60 , and 72 hpi). Genes similarly changed between HCoV-EMC and SARS-CoV (cluster I and II) were defined as genes DE by both viruses compared to their time-matched mocks and that were also not differentially changed between HCoV-EMC and SARS-CoV in at least one of the previously listed comparisons. Genes of cluster III and IV were genes DE at late times postinfection following HCoV-EMC infection that were not similarly changed between HCoV-EMC and SARS-CoV. To further cluster the genes based on their expression value after HCoV-EMC infection, log 2 FC values at 18 and 24 hpi were averaged, and clusters I and III were genes with average log 2 FC values of Ͼ0, while cluster II and IV genes had average log 2 FC values of Ͻ0.

Functional enrichment and upstream regulator analysis. Functional analysis of statistically significant gene expression changes was performed using Ingenuity Pathways Knowledge Base (IPA; Ingenuity Systems). For all gene set enrichment analyses, a right-tailed Fisher's exact test was used to calculate a P value determining the probability that each biological function assigned to that data set was due to chance alone. All enrichment scores were calculated in IPA using the probes that passed our QC filter as the background data set.

Upstream regulator analysis, which was used to predict regulators and infer their activation state, is based on prior knowledge of expected effects between regulators and their known target genes according to the IPA database. A z score is calculated and determines whether gene expression changes for known targets of each regulator are consistent with what is expected from the literature (z Ͼ 2, regulator predicted to be activated) or are anti-correlated with the literature (z Ͻ 2, regulator predicted to inhibited).

Quantitative reverse transcription PCR (RT-PCR). RNA was reverse transcribed using the QuantiTect reverse transcription kit (Qiagen). The resulting cDNA samples were diluted 50ϫ. Primer sets for SYBR green quantitative RT-PCR were designed using Primer3 (51) . Primer sequences are available in Table S3 in the supplemental material. Relative gene expression in infected samples compared to that in mock-infected cells was calculated using the 2 Ϫ⌬⌬CT method (52), using the MYL6 gene as a calibrator, as the expression of MYL6 did not significantly change in HCoV-EMC-infected cells in the microarray data.

Connectivity map. To determine whether some drugs can reverse the HCoV-EMC infection signature, we used the publicly available Connectivity Map (cmap) database (build 02) (22) . Cmap is a collection of genome-wide transcriptional data from cultured human cells treated with 1,309 different compounds. We used the list of 207 DE genes that are stable after infection with HCoV-EMC. Agilent probes were mapped to Affymetrix U133A probe sets using the BioMart ID converter tool (53) in order to query the cmap database. Results were filtered based on their negative connectivity enrichment score.

Antiviral in vitro assay. For in vitro evaluation of the p38 MAPK inhibitor SB203580 (Sigma), VeroE6 cells were incubated for 4 h with the drug at 5 M prior infection with HCoV-EMC at an MOI of 0.01 (pretreatment condition), medium was replaced during the 40-min infection, and medium containing SB203580 was added back following inoculation (54) . For the posttreatment condition, cells were treated with SB203580 at 5 M or type I IFN (100 U/ml; PBL Interferon Source) at 0 hpi following the infection inoculation.

Publically available data. Normalized matrixes and expression values can be found at https://www.systemsvirology.org/project/home/ Data%20%26%20Resources/Experimental%20Metadata/HCoV-EMC%20 infection%20in%20Calu-3%20cells/begin.view?. The other data set utilized was Calu-3 cells infected with SARS-CoV at MOI 5 (GSE33267).

Microarray data accession numbers. Raw microarray data have been deposited in NCBI's Gene Expression Omnibus under accession number GSE45042.

Supplemental material for this article may be found at http://mbio.asm.org /lookup/suppl/doi:10.1128/mBio.00165-13/-/DCSupplemental. 

Novel coronavirus-Saudi Arabia: human isolate

Isolation of a novel coronavirus from a man with pneumonia in Saudi Arabia

Novel coronavirus infection in the United Kingdom. Global alert and response

Severe respiratory illness caused by a novel coronavirus

Novel coronavirus infection---update. Global alert and response

Background and summary of novel coronavirus infection-as of

The severe acute respiratory syndrome

Identification of a novel coronavirus in patients with severe acute respiratory syndrome

Genomic characterization of a newly discovered coronavirus associated with acute respiratory distress syndrome in humans

Human coronavirus EMC does not require the SARS-coronavirus receptor and maintains broad replicative capability in mammalian cell lines

Dipeptidyl peptidase 4 is a functional receptor for the emerging human coronavirus-EMC

The newly emerged SARS-Like coronavirus HCoV-EMC also has an "Achilles' heel": current effective inhibitor targeting a 3C-like protease

Disruption of virus-host cell interactions and cell signaling pathways as an anti-viral approach against influenza virus infections

Gene expression signature-based screening identifies new broadly effective influenza A antivirals

Into the eye of the cytokine storm. Microbiol

Discovery and preclinical validation of drug indications using compendia of public gene expression data

Evaluation of the Calu-3 cell line as a model of in vitro respiratory syncytial virus infection

Host regulatory network response to infection with highly pathogenic H5N1 avian influenza virus

Release of SARS-CoV nuclear import block enhances host transcription in human lung cells

Dynamic innate immune responses of human bronchial epithelial cells to severe acute respiratory syndromeassociated coronavirus infection

Emerging role of ubiquitination in antiviral RIG-I signaling. Microbiol

The connectivity map: a new tool for biomedical research

Human coronavirus EMC is not the same as severe acute respiratory syndrome coronavirus

Efficient replication of the novel human betacoronavirus EMC on primary human epithelium highlights its zoonotic potential

IL-17 boosts proinflammatory outcome of antiviral response in human cells

Critical role of IL-17RA in immunopathology of influenza infection

Regulation of epithelial immunity by IL-17 family cytokines

The MHC class I antigen presentation pathway: strategies for viral immune evasion

MHC class I antigen presentation: learning from viral evasion strategies

Viral inhibition of MHC class II antigen presentation

A comparison of the antigen-presenting capabilities of class II MHC-expressing human lung epithelial and endothelial cells

Immunoproteasome assembly: cooperative incorporation of interferon gamma (IFN-gamma)-inducible subunits

SARS: systematic review of treatment effects

Novel HIV-1 therapeutics through targeting altered host cell pathways

The NF kappaB inhibitor SC75741 efficiently blocks influenza virus propagation and confers a high barrier for development of viral resistance

The SARScoronavirus-host interactome: identification of cyclophilins as target for Pan-coronavirus inhibitors

Replication of human coronaviruses SARS-CoV, HCoV-NL63 and HCoV-229E is inhibited by the drug FK506

Effect of the phosphatidylinositol 3-kinase/Akt pathway on influenza A virus propagation

Activation of the PI3K/Akt pathway early during vaccinia and cowpox virus infections is required for both host survival and viral replication

Human cytomegalovirus Up-Regulates the phosphatidylinositol 3-kinase (PI3-K) pathway: inhibition of PI3-K activity inhibits viral replication and virusinduced signaling

Effect of p38 mitogen-activated protein kinase on the replication of encephalomyocarditis virus

The respiratory syncytial virus small hydrophobic protein is phosphorylated via a mitogen-activated protein kinase p38-dependent tyrosine kinase activity during virus infection

Activation of the HIV-1 long terminal repeat by cytokines and environmental stress requires an active CSBP/p38 MAP kinase

Mitogen activated protein kinase inhibitors: where are we now and where are we going?

Development of PI3K inhibitors: lessons learned from early clinical trials

Novel p38 MAPK inhibitor ML3403 has potent antiinflammatory activity in airway smooth muscle

Anti-inflammatory effect and selectivity profile of AS1940477, a novel and potent p38 mitogen-activated protein kinase inhibitor

Design and synthesis of inhaled p38 inhibitors for the treatment of chronic obstructive pulmonary disease

Reverse genetics with a full-length infectious cDNA of severe acute respiratory syndrome coronavirus

Limma: linear models for microarray data

Primer3 on the WWW for general users and for biologist programmers

Analysis of relative gene expression data using real-time quantitative PCR and the 2(ϪDelta delta C(T)) method

Biomart-biological queries made easy

Roles of p38 MAPK and JNK in TGF-beta1-induced human alveolar epithelial to mesenchymal transition

We thank Lynn Law and Marcus Korth for valuable feedback on the manuscript.This project was funded in part by federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health,

Department of Health and Human Services, under contract HHSN272200800060C and grant U54AI081680. The findings and conclusions in this report are those of the authors and do not necessarily reflect the views of the funding agency.