key: cord-0731694-s0hs66gd
authors: Zani, Ashley; Kenney, Adam D.; Kawahara, Jeffrey; Eddy, Adrian C.; Wang, Xiao-Liang; KC, Mahesh; Lu, Mijia; Hemann, Emily A.; Li, Jianrong; Peeples, Mark E.; Hall-Stoodley, Luanne; Forero, Adriana; Cai, Chuanxi; Ma, Jianjie; Yount, Jacob S.
title: Interferon-induced transmembrane protein 3 (IFITM3) limits lethality of SARS-CoV-2 in mice
date: 2021-12-23
journal: bioRxiv
DOI: 10.1101/2021.12.22.473914
sha: 1185e6c0b4c2860be93a81d05258a74235363ea6
doc_id: 731694
cord_uid: s0hs66gd

Interferon-induced transmembrane protein 3 (IFITM3) is a host antiviral protein that alters cell membranes to block fusion of viruses. Published reports have identified conflicting pro- and antiviral effects of IFITM3 on SARS-CoV-2 in cultured cells, and its impact on viral pathogenesis in vivo remains unclear. Here, we show that IFITM3 knockout (KO) mice infected with mouse-adapted SARS-CoV-2 experienced extreme weight loss and lethality, while wild type (WT) mice lost minimal weight and recovered. KO mice had higher lung viral titers and increases in lung inflammatory cytokine levels, CD45-positive immune cell infiltration, and histopathology, compared to WT mice. Mechanistically, we observed disseminated viral antigen staining throughout the lung tissue and pulmonary vasculature in KO mice, while staining was observed in confined regions in WT lungs. Global transcriptomic analysis of infected lungs identified upregulation of gene signatures associated with interferons, inflammation, and angiogenesis in KO versus WT animals, highlighting changes in lung gene expression programs that precede severe lung pathology and fatality. Corroborating the protective effect of IFITM3 in vivo, K18-hACE2/IFITM3 KO mice infected with non-adapted SARS-CoV-2 showed enhanced, rapid weight loss and early death compared to control mice. Increased heart infection was observed in both mouse models in the absence of IFITM3, indicating that IFITM3 constrains extrapulmonary dissemination of SARS-CoV-2. Our results establish IFITM3 KO mice as a new animal model for studying severe SARS-CoV-2 infection of the lung and cardiovascular system, and overall demonstrate that IFITM3 is protective in SARS-CoV-2 infections of mice.

(IFITM3) is an antiviral restriction factor that inhibits virus fusion with host cell membranes [1] [2] [3] , and potently blocks infection by numerous enveloped viruses of human health concern, including influenza, dengue, and Zika viruses 4 . This set of viruses primarily utilizes endocytic pathways for entry into cells, and fuses with endosome membranes where IFITM3 is abundantly localized and poised to restrict infection 5, 6 . In contrast, viruses that fuse primarily at the plasma membrane, such as Sendai virus, are minimally affected by IFITM3 7, 8 . Viruses that are able to fuse at both the plasma membrane or within endosomes, such as human metapneumovirus, can be partially restricted by IFITM3, with the portion of virus that fuses at the plasma membrane largely evading restriction 9 . These patterns of restriction hold true for the vast majority of the dozens of viruses that have been tested for IFITM3 inhibition, with the notable exception of the common cold coronavirus OC43, which utilizes dual cell entry pathways similarly to metapneumovirus, but showed enhanced, rather than reduced, infection when IFITM3 was expressed in target cells 10 . This unusual finding suggests that IFITM3 may have unanticipated effects on some coronaviruses, leading several groups to examine the impact of IFITM3 and related proteins on SARS-CoV-2, which remains a critical threat to worldwide human health.

SARS-CoV-2 uses cell surface and endosomal fusion strategies for infection 11, 12 .

Experiments studying effects of IFITM3 on SARS-CoV-2 infections identified opposing activities in which IFITM3 inhibits virus entry at endosomes as expected, but enhances virus entry at the plasma membrane 13 . The overall effects of IFITM3 remain controversial, with some reports indicating that IFITM3 primarily restricts cellular SARS-CoV-2 infection 13, 14 , while others conclude that IFITM3 is a net enhancer of infection, particularly in lung cells [15] [16] [17] . Further, IFITM3 has been shown inhibit fusion of SARS-CoV-2 infected cells with neighboring cells (syncytia formation), which may be an important mechanism of viral spread within tissues 18, 19 . The overall balance and relevance of the effects of IFITM3 in SARS-CoV-2 infection of cells and pathogenesis in vivo remain unclear.

We previously generated IFITM3 knockout (KO) mice to investigate in vivo effects of this antiviral restriction factor and confirmed a profound susceptibility of these mice to severe influenza virus infections 20 . These results were in agreement with studies linking deleterious single nucleotide polymorphisms in the human IFITM3 gene to severe influenza [21] [22] [23] . Although some studies have suggested that IFITM3 gene polymorphisms are risk factors for severe SARS-CoV-2 infection 24-28 , IFITM3 variants did not emerge as top candidates in other large genome-wide SARS-CoV-2 susceptibility studies [29] [30] [31] . In addition to possible direct effects of IFITM3 gene polymorphisms, defects in interferon responses that induce IFITM3 during infection have been linked to roughly 20% of severe COVID-19 cases, providing additional circumstances where IFITM3 insufficiency may be impacting disease 32, 33 . Given the conflicting reports of IFITM3 activity on SARS-CoV-2 in vitro, coupled with the potential, but uncertain, role of IFITM3 in modulating SARS-CoV-2 disease severity, we investigated roles of IFITM3 in SARS-CoV-2 infection in the mouse model.

To determine the impact of IFITM3 on SARS-CoV-2 disease severity, we infected wild type (WT) C57BL/6 mice and corresponding IFITM3 KO animals with the mouse-adapted SARS-CoV-2 strain MA10 34 . Importantly, our virus stock was generated with a stringent plaque purification and sequencing protocol, allowing us to utilize virus lacking commonly observed tissue culture adaptations that attenuate pathogenicity 35 . After infection, we observed a greater than 10% weight loss in WT mice with recovery to full weight by day 8 post infection (Fig 1A) . In contrast, IFITM3 KO mice lost significantly more weight starting at day 1 post infection. This accelerated weight loss continued through day 5 post infection, at which point all IFITM3 KOs had either died or met humane endpoint criteria due to severe illness and lack of movement ( Fig 1A) . We conclude that IFITM3 provides protection against severe, lethal SARS-CoV-2induced disease in mice. We next measured viral titers at days 3 and 5 post infection, and found that IFITM3 KO lung titers were significantly higher than WT titers on day 3 (Fig 1B) . Consistent with higher virus measurements, elevated levels of the inflammatory cytokine IL-6 were also detected in KO versus WT lungs at day 3 post infection (Fig 1C) . Viral titers were decreased in both groups at day 5 compared to day 3 (Fig 1B) . Nonetheless, complete virus clearance was delayed in IFITM3 KOs, with all KO mouse lungs remaining positive for virus at day 5, while only one of four WT lungs were positive for live virus at this timepoint (Fig 1B) . At day 3 post infection, we also detected a low level of live virus in the heart, spleen, and brain in a subset of the KO animals, while no virus was detected in liver or kidney tissue (Fig 1D) . Live virus was not detected in any of these extrapulmonary organs in WT or KO animals at day 5 post infection (not shown). In sum, we observed that the enhanced illness severity in IFITM3 KO mice was accompanied by increased viral titers and delayed clearance.

Mouse-adapted SARS-CoV-2 differs from the parental human isolate by only seven amino acids 34 and utilizes endogenously expressed murine ACE2 as the virus receptor for infection, making it a highly relevant model for studying viral tropism and pathogenesis in vivo.

Nonetheless, we also sought to examine the in vivo role of IFITM3 in infection with WA1, a nonadapted SARS-CoV-2 strain isolated from humans, which we also stringently propagated and sequenced. For this, we utilized mice expressing human (h)ACE2 under control of the K18 keratin promoter (K18-hACE2 mice) crossed with IFITM3 KOs. We observed that K18-hACE2/IFITM3 KO mice lost significantly more weight than K18-hACE2 control mice, and that all of the K18-hACE2/IFITM3 KO mice succumbed to infection by day 5 (Fig 1E) , at which point we ended our experiments. High virus titers at day 3 post infection were detected in both control and KO mice, with one of two experiments showing significantly higher titers in IFITM3 KOs and a second experiment showing a trend toward higher titers in the KOs (Fig 1F) . Further, cardiac dissemination of virus was detected in 8 out of 8 IFITM3 KOs compared to 1 out of 8 control mice (Fig 1G) , confirming that IFITM3 restricts cardiac dissemination of SARS-CoV-2. Overall, a protective effect of IFITM3 in SARS-CoV-2 infections was identified using both mouse-adapted and human isolates.

To mechanistically characterize roles of IFITM3 in limiting SARS-CoV-2 lung pathogenesis, we performed hematoxylin and eosin staining on lung sections from WT and IFITM3 KO animals after infection with SARS-CoV-2 MA10. All lungs from infected animals showed areas of consolidated tissue, cellular infiltration, and inflammation (Fig 1H) . We observed that larger portions of the lungs were afflicted in IFITM3 KO samples, and thus quantified cellularity scores to measure loss of open airspace for individual sections from infected WT and IFITM3 KO mouse lung sections. This unbiased method confirmed increased pathology in infected IFITM3 KO versus WT lungs at day 5 post infection (Fig 1H) , correlating with the increased illness and viral titers that we observed in these animals (Fig 1A, B) . This conclusion was supported further by immunohistochemical staining and quantification of CD45positive immune cell infiltration into the lung, which was elevated in IFITM3 KO versus WT mice at day 5 (Fig 1I) . Thus, lung pathology resulting from SARS-CoV-2 infection was exacerbated in IFITM3 KO mice.

Given that IFITM3 generally blocks infection and spread of certain viruses, we investigated the effects of IFITM3 on viral infection patterns in the lungs by staining for viral antigen (N protein) in WT and IFITM3 KO lung sections at day 2 post infection with SARS-CoV-2 MA10, representing an early timepoint at which viral titers are high 34 . We detected robust staining of the airways in both groups, but observed more disseminated staining throughout IFITM3 KO lungs (Fig 2A, B) . Indeed, large portions of WT lungs showed no staining for viral antigen, while virus was detected throughout KO lungs (Fig 2A,B) . Further, accumulation of viral antigen that likely represents shedding of necrotic, highly infected cells into the bronchioles was observed in KO lungs, even at this early timepoint (Fig 2B) . Infected cells associated with blood vessels could be seen in WT lungs primarily when adjacent to highly infected airways (Fig   2A) , while blood vessels associated with infected cells were readily apparent throughout IFITM3 KO lungs (Fig 2B) . Quantification of viral staining in lung sections from multiple mice confirmed the significant increase in viral antigen staining in IFITM3 KO versus WT mice (Fig 2C) . Overall, viral antigen imaging confirmed increased viral titers and pathology (Fig 1B, H) , but also revealed diffuse infection of cells, including the vasculature, in IFITM3 KO lungs.

We next examined the global impact of loss of IFITM3 on transcriptional programs at day 2 post infection with SARS-CoV-2 MA10 by performing lung RNA sequencing. Dimensionality reduction approaches showed divergence of gene signatures in infected WT versus KO lungs (Fig 2D) , corresponding to 1,865 differentially expressed genes between the groups (≥ |2| fold change, p-value < 0.05; a list of differentially expressed genes, expression values, fold changes, and p-values are shown in Supplementary Table 1) . GO term enrichment analysis of genes upregulated in infected IFITM3 KO versus WT lungs revealed significant increases in antiviral defense responses, angiogenesis, and inflammation (Fig 2F, Supplementary Table 2) . Among the antiviral and inflammatory genes increased in KO versus WT animals were multiple type I interferon genes, including Ifnb1 and Ifna4, and canonical interferon-stimulated genes, such as Oas1b, Mx1, and Ifit2 (Fig 2G) . Expression of chemokine genes Ccl5 and Cxcl9 was also potentiated in KO lungs relative to WT (Fig 2G) , consistent with the increased recruitment of CD45-positive immune cells that we observed (Fig 1I) . Classic angiogenesis-driving genes with significantly elevated expression in KO versus WT animals included Vegfa, Hif3a, Wnt7a, and Pdgfa (Fig 2G) . These statistically significant increases in activation of inflammatory antiviral and angiogenesis pathways in KO lungs were further confirmed by Molecular Signature Database Hallmark gene set enrichment analysis (Fig 2F) . Likewise, analysis with the PanglaoDB database of single cell RNA sequencing experiments identified significant associations with gene signatures specific to pneumocytes and endothelial cells within the KOspecific upregulated genes (Fig 2F) . Furthermore, TRRUST Transcriptional Regulatory Network database analysis identified interferon regulatory factors, IRF3 and IRF7, as well as the proangiogenic protein ETS1, as transcriptional factors likely involved in upregulation of genes in infected IFITM3 KO versus WT lungs (Fig 2F) . Thus, multiple independent analyses of our data converged on inflammatory interferon and angiogenic/endothelial nodes as key pathways that Table 2 ). Overall, our results demonstrate that IFITM3 restrains SARS-CoV-2 replication and spread through the lungs and cardiovascular system, thus limiting induction of pathological interferon and angiogenesis gene expression programs.

Increased induction of interferon and interferon-responsive genes may be the result of heightened levels of virus replication seen in IFITM3 KO animals (Fig 1B and Fig 2A, B) .

Alternatively, these results are consistent with feedback inhibition of type I interferon induction and inflammation by IFITM3, as has been previously suggested 36, 37 . Type I interferons provide a double-edged sword in COVID-19, limiting virus replication early in infection while also contributing to pathological inflammation 38, 39 . Furthermore, angiogenesis and thrombosis rapidly emerged as hallmarks of SARS-CoV-2-associated disease 40 . The observed upregulation of angiogenic gene programs may result from direct infection of blood vessel cells, or from infected cells mediating bystander signaling. We also noted that several coagulation-associated genes had enhanced upregulation in the KO animals, such as Vwf, F2r, F2rl3, Plat, Fgg, F3, and Thbd Table 2 The sequencing libraries were multiplexed and clustered onto a flowcell. After clustering, the flowcell was loaded onto the Illumina HiSeq instrument according to manufacturer's instructions.

The samples were sequenced using a 2x150bp Paired End (PE) configuration. Image analysis and base calling were conducted by the HiSeq Control Software (HCS). Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into fastq files and de-multiplexed using Illumina bcl2fastq 2.17 software. One mis-match was allowed for index sequence identification."

Resulting fastq files were analyzed by ROSALIND® (https://rosalind.bio/), with a HyperScale architecture developed by ROSALIND, Inc. (San Diego, CA). Read Distribution percentages, violin plots, identity heatmaps, and sample MDS plots were generated as part of the QC step. Statistical analysis for differential gene expression was done using the "limma" R library 41 . The principal components analysis and volcano plots were re-formatted in PRISM graphing software using values downloaded from ROSALIND. Heatmaps were generated using gene expression values downloaded from ROSALIND and Morpheus software (https://software.broadinstitute.org/morpheus). Hypergeometric distribution was used to analyze the enrichment of pathways, gene ontology, and other ontologies. The topGO R library was used to determine local similarities and dependencies between GO terms in order to perform Elim pruning correction. Several database sources were referenced for enrichment analysis, including MSigDB 42,43 , TRRUST 44 , and PanglaoDB 45 . Enrichment was calculated relative to a set of background genes relevant for the experiment. 

IFITM proteins restrict viral membrane hemifusion

IFITM3 restricts influenza A virus entry by blocking the formation of fusion pores following virus-endosome hemifusion

IFITM3 requires an amphipathic helix for antiviral activity

Lessons in self-defence: inhibition of virus entry by intrinsic immunity

Phosphorylation of the antiviral protein interferon-inducible transmembrane protein 3 (IFITM3) dually regulates its endocytosis and ubiquitination

Identification of an endocytic signal essential for the antiviral action of IFITM3

Ubiquitin Ligase NEDD4 Promotes Influenza Virus Infection by Decreasing Levels of the Antiviral Protein IFITM3

Palmitoylation on conserved and nonconserved cysteines of murine IFITM1 regulates its stability and anti-influenza A virus activity

IFITM3 Restricts Human Metapneumovirus Infection

Interferon induction of IFITM proteins promotes infection by human coronavirus OC43

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

Cell entry mechanisms of SARS-CoV-2

Opposing activities of IFITM proteins in SARS-CoV-2 infection

Rapalogs downmodulate intrinsic immunity and promote cell entry of SARS-CoV-2. bioRxiv

IFITM proteins promote SARS-CoV-2 infection and are targets for virus inhibition in vitro

IFITM dependency of SARS-CoV-2 variants of concern. bioRxiv

The P681H mutation in the Spike glycoprotein confers Type I interferon resistance in the SARS-CoV-2 alpha (B.1.1.7) variant. bioRxiv

Syncytia formation by SARS-CoV-2-infected cells

SARS-CoV-2 Alpha, Beta, and Delta variants display enhanced Spike-mediated syncytia formation

IFITM3 protects the heart during influenza virus infection

IFITM3 restricts the morbidity and mortality associated with influenza

SNP-mediated disruption of CTCF binding at the IFITM3 promoter is associated with risk of severe influenza in humans

Antiviral protection by IFITM3 in vivo

The frequency of combined IFITM3 haplotype involving the reference alleles of both rs12252 and rs34481144 is in line with COVID-19 standardized mortality ratio of ethnic groups in England

Interferon-Induced Transmembrane Protein 3 Genetic Variant rs12252-C Associated With Disease Severity in Coronavirus Disease

Interferon-induced transmembrane protein-3 genetic variant rs12252 is associated with COVID-19 mortality

The Interferon-induced transmembrane protein 3 gene (IFITM3) rs12252 C variant is associated with COVID-19

Association between the interferon-induced transmembrane protein 3 gene (IFITM3) rs34481144 / rs12252 haplotypes and COVID-19

Integrative genomic analyses identify susceptibility genes underlying COVID-19 hospitalization

Genetic mechanisms of critical illness in COVID-19

Genomewide Association Study of Severe Covid-19 with Respiratory Failure

Inborn errors of type I IFN immunity in patients with life-threatening COVID-19

Autoantibodies against type I IFNs in patients with life-threatening COVID-19

A Mouse-Adapted SARS-CoV-2 Induces Acute Lung Injury and Mortality in Standard Laboratory Mice

Loss of furin cleavage site attenuates SARS-CoV-2 pathogenesis

IFITM3 inhibits virus-triggered induction of type I interferon by mediating autophagosome-dependent degradation of IRF3

The antiviral restriction factor IFN-induced transmembrane protein 3 prevents cytokine-driven CMV pathogenesis

Mouse model of SARS-CoV-2 reveals inflammatory role of type I interferon signaling

The interferon landscape along the respiratory tract impacts the severity of COVID-19

Pulmonary Vascular Endothelialitis, Thrombosis, and Angiogenesis in Covid-19

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

Molecular signatures database (MSigDB) 3.0

Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles

TRRUST v2: an expanded reference database of human and mouse transcriptional regulatory interactions

PanglaoDB: a web server for exploration of mouse and human single-cell RNA sequencing data

The authors thank Dr. Kara Corps (OSU Comparative Pathology and Mouse Phenotyping Core 

The authors do not have any conflicts of interest to disclose relating to this work.

WT and IFITM3 KO mice were intranasally infected with 10 5 TCID50 SARS-CoV-2 MA10. A) Weight loss (*p < 0.05, ** p < 0.01, ANOVA with Bonferoni's multiple comparison's test, data from three independent experiments, scull and crossbones indicates that KO mice did not survive beyond this timepoint), B) viral titers in lung homogenates at the indicated days post infection (**p < 0.01, Mann-Whitney test, data from two independent experiments), C) IL-6 levels in lung homogenates on day 3 post infection (***p < 0.001, t-test, data from two independent experiments), and D) viral titers in extrapulmonary organs on day 3 post infection were measured. E-G) K18-hACE2 (hACE2) and K18-hACE2/IFITM3 KO mice were infected with 10 4 pruning value is graphed), the top five most significant Hallmark gene sets, the significant associations from PanglaoDB analysis, and the significant associations from TRRUST transcription factor analysis are shown. G) Heat maps for genes from the top two most significant GO Biological Processes as shown in F.