The inflammatory character of SARS-CoV-2 infection is something the virus requires rather than something it fails to suppress
lab-led
Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19
lab-ledDiscovery
Across cell lines, primary bronchial epithelium, ferrets and patient material, SARS-CoV-2 infection produces a transcriptional response distinguishable from that to other respiratory viruses, combining low type I and type III interferon induction with a moderate interferon-stimulated gene response and strong chemokine and IL-6 expression.
- Senior authors
- Benjamin R. tenOever
- Correspondence
- Robert E. Schwartz; Jean K. Lim; Randy A. Albrecht; Benjamin R. tenOever
Research areas & themes
Innate Immune Signaling and the Interferon ResponsePandemic Host Response and DiseaseCalibration of the Interferon Response In VivoThe Imbalanced Host Response to SARS-CoV-2
Citation
Blanco-Melo D, Nilsson-Payant BE, Liu W-C, Uhl S, Hoagland D, Møller R, Jordan TX, Oishi K, Panis M, Sachs D, Wang TT, Schwartz RE, Lim JK, Albrecht RA, tenOever BR. Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19. Cell 2020, volume 181, issue 5, pages 1036-1045.e9.
DOI 10.1016/j.cell.2020.04.026. PMID 32416070. PMCID PMC7227586.
One-sentence contribution
Across cell lines, primary bronchial epithelium, ferrets and patient material, SARS-CoV-2 infection produces a transcriptional response distinguishable from that to other respiratory viruses, combining low type I and type III interferon induction with a moderate interferon-stimulated gene response and strong chemokine and IL-6 expression.
Executive summary
Early in the COVID-19 pandemic there was little information on how host cells respond to SARS-CoV-2 relative to other respiratory viruses. This study profiles the transcriptional response across four levels of system. In respiratory cell lines, SARS-CoV-2, SARS-CoV-1, MERS-CoV, influenza A virus, human parainfluenza virus 3 and respiratory syncytial virus were compared by sequencing, with A549 cells rendered permissive by adenoviral delivery of ACE2. In primary human bronchial epithelial cells, SARS-CoV-2 was compared with wild-type influenza A virus, an influenza A virus lacking its NS1 antagonist and interferon beta treatment. Ferrets were infected and followed longitudinally by nasal wash sequencing. Post-mortem lung from two COVID-19 patients was compared with healthy lung, and serum from 24 patients who tested positive was compared with 24 controls. Across these systems the recurring pattern is low type I and type III interferon with only a subset of interferon-stimulated genes induced, alongside pronounced chemokine and interleukin expression. In primary bronchial cells SARS-CoV-2 and influenza A virus produced responses similar in magnitude but sharing only eight significantly induced genes, and interferon-stimulated gene induction by SARS-CoV-2 was not restored by blocking interferon signalling with ruxolitinib, indicating that the chemokine response is interferon-independent. High multiplicity infection in permissive cells did induce interferons, which the authors read as antagonism being overcome or as artificial pattern generation. Serum from patients showed no detectable interferon beta or lambda but significantly raised IL-6, IL1RA, CCL2, CCL8, CXCL2, CXCL8, CXCL9 and CXCL16.
Scientific context
Three highly pathogenic betacoronaviruses had emerged from zoonotic events in two decades, and the newest, SARS-CoV-2, was spreading with an undetermined mortality rate and an incompletely characterised host interaction. The general framework was established, in which intracellular pattern recognition receptors detect aberrant viral RNA, activate interferon regulatory factors and nuclear factor kappa B, and launch two programs, interferon-driven cellular defence and chemokine-driven leukocyte recruitment. It was also known that respiratory viruses encode antagonists of this system, with SARS-CoV-1 antagonism attributed to ORF3B, ORF6 and the nucleocapsid protein plus the host shutoff nuclease nsp1, and influenza A virus antagonism attributed to NS1. Work on SARS-CoV-1 had proposed that clinical severity stems from a dysregulated immune response with delayed type I interferon and improper recruitment of inflammatory monocyte and macrophage populations. What was missing for SARS-CoV-2 was a direct, matched comparison of the host response across systems and against other respiratory viruses.
Central question
How does the host transcriptional response to SARS-CoV-2 differ from the response to other respiratory viruses, and does that difference account for the inflammatory character of COVID-19?
Experimental strategy
The design is comparative at every level. Rather than characterising SARS-CoV-2 alone, the authors place it alongside two other pathogenic coronaviruses and three common respiratory viruses, using the same cell systems, the same sequencing pipeline and the same differential expression framework, so that the response can be positioned rather than merely described. Because A549 cells express little ACE2 and support poor replication, adenoviral delivery of ACE2 is used to separate the question of permissiveness from the question of response, with an mCherry vector as the matched control, and infections are run at both low and high multiplicity because the two conditions distinguish a virus that fails to trigger detection from one whose antagonist is saturated. Primary bronchial epithelial cells provide a physiological comparison, and the inclusion of an influenza A virus lacking NS1 supplies a positive control for what a full interferon response in those cells looks like when antagonism is removed. Ruxolitinib is used to ask whether the induced genes depend on interferon signalling at all. The ferret study adds time and an intact host with an immune system, sampling the upper respiratory tract repeatedly and the trachea at a fixed day. Human material closes the loop in two ways, transcriptionally in lung tissue and at the protein level in serum, where circulating cytokines can be measured in a larger group than the tissue analysis allows.
Key findings
- Viral read fraction across infections ranged from 0.1 percent to over 50 percent of total reads. A549 cells were relatively non-permissive for SARS-CoV-2 while Calu-3 cells supported roughly 15 percent viral reads, and adenoviral ACE2 raised SARS-CoV-2 in A549 cells to about 54 percent of reads at low multiplicity, corroborated by nucleocapsid immunoblot and by more than three orders of magnitude higher Envelope and nsp14 transcript (Figure 1, A through D).
- Despite that high viral load at low multiplicity, there was no TBK1 activation and no induction of STAT1 or MX1. Raising the multiplicity tenfold did engage interferon even though total viral reads at 24 hours were comparable. The authors offer two readings, that an antagonist is overcome at high multiplicity or that high multiplicity generates pathogen patterns that may not form physiologically, and they do not adjudicate between them.
- SARS-CoV-2 was sensitive to interferon beta pretreatment. Blocking interferon signalling with ruxolitinib prevented interferon-stimulated gene induction without increasing viral Spike or viral reads appreciably, and had minimal effect on cytokine and chemokine induction. The last point is the direct evidence that the chemokine response is independent of type I and type III interferon signalling.
- Principal component analysis placed the response to SARS-CoV-2 under conditions of high replication apart from all other viruses tested, while MERS-CoV clustered with SARS-CoV-1 and influenza A virus in a pattern of overall antiviral repression, and parainfluenza virus 3 and respiratory syncytial virus formed a separate cluster marked by high interferon and interferon-stimulated gene expression (Figure 1, E and F).
- In primary bronchial epithelial cells, interferon beta induced 381 genes, most of which were also induced by the NS1-deficient influenza A virus. SARS-CoV-2 and wild-type influenza A virus produced responses similar in magnitude but sharing only eight significantly induced genes, among them IL-6, IRF9, ICAM1 and TNF (Figure 2A).
- In those cells type I and type III interferons were undetectable for both SARS-CoV-2 and wild-type influenza A virus with only a small subset of interferon-stimulated genes induced, whereas the NS1-deficient virus induced IFNB and IFNL1 through 3 robustly (Figure 2, C and D). For influenza A virus the muting is attributable to NS1. For SARS-CoV-2 the paper does not identify the responsible viral product.
- Despite absent interferon expression, SARS-CoV-2 in primary bronchial cells induced CCL20, CXCL1, IL-1B, IL-6, CXCL3, CXCL5, CXCL6, CXCL2, CXCL16 and TNF, together with enrichment for chemokine signalling and a response to type II interferon (Figure 2, C and E).
- In ferrets, SARS-CoV-2 reached 1.2 percent of nasal wash reads at day three, fell to 0.05 percent by day seven and was undetectable at day 14. Transcriptional change was negligible at day one, began at day three with CCL8 and CXCL9, and expanded by day seven to include CCL2 and CXCL9 even as virus waned, accompanied by upregulation of leukocyte markers including CD163, CD226, CCR5, CCR6, CXCR1, CXCR2 and CXCR7 (Figure 3, B and C).
- The overall magnitude of the ferret upper respiratory response was lower than for a comparable influenza A virus infection, and influenza A virus produced the far larger antiviral signature including MX1, ISG20, OASL and Tetherin. SARS-CoV-2 produced a distinct signature enriched for cell death and leukocyte activation including IL1A and CXCL8. At day 14, with no viral reads, IL-6 and IL1RN remained elevated.
- In ferret trachea at day three, both viruses induced comparable monocyte and lymphocyte marker signatures, while SARS-CoV-2 uniquely induced genes aligning with haematopoietic progenitor cells. The authors state explicitly that whether this reflects induced haematopoiesis and whether it contributes to COVID-19 requires further work.
- Post-mortem lung from two COVID-19 patients compared with two healthy lungs showed about 2,000 differentially expressed genes, with a subset of interferon-stimulated genes induced, no type I or type III interferon detected by sequencing or semiquantitative PCR, and robust chemokine induction including CCL2, CCL8 and CCL11 (Figure 4A).
- Serum from 24 patients who tested positive and 24 controls showed no detectable interferon beta or lambda family interferons and significantly elevated IL-6, IL1RA, CCL2, CCL8, CXCL2, CXCL8, CXCL9 and CXCL16 (Figure 4, B and C). The authors state the sample size is not necessarily representative and that additional sampling is required.
Mechanistic model
The study does not establish a mechanism and the authors do not claim one. It defines a reproducible response phenotype, low type I and type III interferon with partial interferon-stimulated gene induction alongside strong chemokine and interleukin expression, and shows that this phenotype recurs across cell lines, primary epithelium, an animal model and human samples. The ruxolitinib experiment constrains the wiring by showing the chemokine arm does not require interferon signalling, and the interferon beta pretreatment shows the virus is sensitive to interferon when it is present, so the low interferon output is not a matter of resistance. What is not determined is why interferon induction is low. The paper proposes that an antagonist prevents engagement and is overcome at high multiplicity, offers the alternative that high multiplicity generates pathogen patterns that would not form physiologically, and notes a further possibility that a subset of cells refractory to the antagonist produces the limited interferon seen in vivo. No viral gene product is assigned this role here. The central interpretive claim, that reduced innate antiviral defence coupled with exuberant inflammatory cytokine production are the defining and driving features of COVID-19, is presented by the authors as a proposal drawn from correlated observations across systems, not as a demonstrated causal chain from response pattern to disease. The suggested parallel with cytokine release syndrome and the consequent suggestion that tocilizumab or anakinra might help are explicitly flagged by the authors as requiring formal testing, and the speculation about a restricted immune response in older people is labelled as such.
Conceptual or technical advance
Assembling one matched comparison across six respiratory viruses and four levels of biological system, produced within weeks of the pathogen's emergence, gave a reference description of the SARS-CoV-2 host response and a public dataset. The comparative framing is what makes the result interpretable, since a low interferon response only means something against viruses that produce a high one in the same cells. The demonstration that chemokine induction proceeds without interferon signalling separates two arms of the response that are usually considered together and identifies the inflammatory arm as the one that is intact. Practically, the work reoriented attention for treatment from boosting or mimicking interferon toward controlling inflammation, and it supplied specific circulating mediators to track. The observation that interferon induction depends strongly on multiplicity of infection is also a methodological caution for any cell culture study of this virus.
Relationship to the broader research program
The study applies to a newly emerged pathogen the comparative transcriptional profiling approach the laboratory had developed for influenza A virus and other RNA viruses, using the same reliance on sequencing viral and host reads from the same libraries and the same use of an NS1-deficient influenza A virus as the reference for an unantagonised response. Its reference list includes the laboratory's own review on the evolution of antiviral defence systems, its work on influenza A virus transmission bottlenecks in ferrets, its microRNA-based strategy for mitigating risk in gain-of-function influenza studies, and a companion preprint on host detection of negative-sense RNA viruses. It also cites Sharma and colleagues, 2003, for TBK1 as the kinase responsible for type I and type III interferon expression, which connects this work to the senior author's doctoral training period. Reading that connection as a continuous research arc would be category 3 synthesis.
Related publications
- Sharma and colleagues, 2003, methodological foundation. Cited here for the identification of TBK1 as the kinase driving type I and type III interferon expression, the node whose activation is assayed as a readout of detection.
- Blanco-Melo and colleagues, 2020 preprint on ribonucleoprotein genomic structure and host detection of negative-sense RNA viruses, companion. Cited as the source of the recombinant parainfluenza virus 3 used here.
- Varble and colleagues, 2014, predecessor. Cited for influenza A virus transmission bottlenecks defined by infection route and recipient host, supporting the inference about transmission potential from nasal virus.
- Langlois and colleagues, 2013, methodological foundation. Cited as the source of the influenza A virus strains used.
- tenOever, 2016, review or synthesis. The laboratory's review on the evolution of antiviral defence systems, cited for the framing of the cellular response.
Limitations and boundaries
The human tissue transcriptional analysis rests on two post-mortem COVID-19 lungs and two healthy lungs, all from males over 60, which the authors state plainly, and the serum study covers 24 cases and 24 controls, which they also describe as not necessarily representative. The comparison viruses were run at multiplicities and durations that differ between viruses, for example influenza A virus at multiplicity five for nine hours against SARS-CoV-2 at multiplicity 0.2 or two for 24 hours, so magnitude comparisons between viruses carry that caveat. SARS-CoV-1 and MERS-CoV data were taken from a previously published dataset rather than generated alongside. The A549 system depends on adenoviral ACE2 delivery, an artificial arrangement that also introduces a second virus vector. Interferon induction proved strongly dependent on multiplicity of infection, and the authors themselves note that high multiplicity conditions may not reflect physiological infection, which limits how firmly the low interferon phenotype can be stated. The ferret study used small groups, with two naive animals, six infected with SARS-CoV-2 and two per influenza comparison, and it samples the upper respiratory tract and trachea rather than the lower airway where severe disease occurs. No mechanism or viral antagonist is identified, no severity stratification of patients is available, and the inference that the observed response pattern drives COVID-19 pathology remains an interpretation of correlated observations. Therapeutic suggestions are raised by the authors as untested possibilities.
Audience summaries
25 words
SARS-CoV-2 provokes unusually little interferon while driving strong chemokine and IL-6 output, a combination seen in cells, primary airway tissue, ferrets and patients, and linked to inflammatory disease.
75 words
Comparing SARS-CoV-2 with five other respiratory viruses in the same systems showed a distinctive host response. Infected cells made little type I or type III interferon and only some interferon-stimulated genes, yet produced abundant chemokines and IL-6. The same pattern appeared in primary airway cells, in infected ferrets and in patient lung and serum. Blocking interferon signalling did not reduce the chemokine output, indicating the inflammatory arm operates independently of interferon.
150 words
Matched transcriptional profiling across respiratory cell lines, primary bronchial epithelium, ferrets and COVID-19 patient material positions the SARS-CoV-2 host response against SARS-CoV-1, MERS-CoV, influenza A virus, parainfluenza virus 3 and respiratory syncytial virus. In A549 cells made permissive by adenoviral ACE2, SARS-CoV-2 replicated to high levels at low multiplicity without activating TBK1 or inducing STAT1 and MX1, while high multiplicity did engage interferon, an observation the authors attribute either to saturation of an unidentified antagonist or to non-physiological pattern formation. The virus was sensitive to interferon beta, and ruxolitinib abolished interferon-stimulated gene induction while leaving chemokine induction largely intact. Primary bronchial cells infected with SARS-CoV-2 shared only eight induced genes with wild-type influenza A virus yet produced a broad chemokine program. Ferrets showed a cytokine response that outlasted viral clearance. Patient lung showed chemokines without detectable interferon, and patient serum showed raised IL-6, IL1RA and multiple chemokines.
Discoveries supported by this paper
Documented publication relationships
- Triggering the Interferon Antiviral Response Through an IKK-Related Pathway — conceptual extension.
- Triggering the Interferon Antiviral Response Through an IKK-Related Pathway — methodological foundation.
- The Global Phosphorylation Landscape of SARS-CoV-2 Infection — methodological foundation.
- A Human Pluripotent Stem Cell-based Platform to Study SARS-CoV-2 Tropism and Model Virus Infection in Human Cells and Organoids — methodological foundation.
- SARS-CoV-2 infects human adult donor eyes and hESC-derived ocular epithelium — predecessor.
- Ancient viral genomes reveal introduction of human pathogenic viruses into Mexico during the transatlantic slave trade — related.
- Leveraging the antiviral type I interferon system as a first line of defense against SARS-CoV-2 pathogenicity — predecessor.
- Immune memory from SARS-CoV-2 infection in hamsters provides variant-independent protection but still allows virus transmission — predecessor.
- Reduced Nucleoprotein Availability Impairs Negative-Sense RNA Virus Replication and Promotes Host Recognition — companion.
- The NF-κB Transcriptional Footprint Is Essential for SARS-CoV-2 Replication — predecessor.
- SARS-CoV-2 infection in hamsters and humans results in lasting and unique systemic perturbations after recovery — predecessor.
- A diminished immune response underlies age-related SARS-CoV-2 pathologies — predecessor.
- The Host Response to Influenza A Virus Interferes with SARS-CoV-2 Replication during Coinfection — predecessor.
- Host protein kinases required for SARS-CoV-2 nucleocapsid phosphorylation and viral replication — methodological foundation.
- Non-cell-autonomous disruption of nuclear architecture as a potential cause of COVID-19-induced anosmia — predecessor.
- Delayed engagement of host defenses enables SARS-CoV-2 viremia and productive infection of distal organs in the hamster model of COVID-19 — predecessor.