Circadian regulation of SARS-CoV-2 infection in lung epithelial cells

The COVID-19 pandemic, caused by SARS-CoV-2 coronavirus, is a global health issue with unprecedented challenges for public health. SARS-CoV-2 primarily infects cells of the respiratory tract, via binding human angiotensin-converting enzyme (ACE2)1,2, and infection can result in pneumonia and acute respiratory dist ress syndrome. Circadian rhythms coordinate an organisms response to its environment and recent studies report a role for the circadian clock to regulate host susceptibility to virus infection3. Influenza A infection of arhythmic mice, lacking the circadian component BMAL1, results in higher viral replication4 and elevated inflammatory responses leading to more severe bronchitis5,6, highlighting the impact of circadian pathways in respiratory function. We demonstrate circadian regulation of ACE2 in lung epithelial cells and show that silencing BMAL1 or treatment with the synthetic REV-ERB agonist SR9009 reduces ACE2 expression and inhibits SARS-CoV-2 entry and RNA replication. Treating infected cells with SR9009 limits viral replication and secretion of infectious particles, showing that post-entry steps in the viral life cycle are influenced by the circadian system. Our study suggests new approaches to understand and improve therapeutic targeting of COVID-19.

Since ACE2 and transmembrane protease serine 2 (TMPRSS2) co-regulate SARS-CoV-2 internalization 1,2 we measured their expression in synchronised Calu-3 cells. ACE2 protein levels varied over 24h, with the trough (CT6) and peak (CT18) of expression associating with SARS-CoV-2pp entry (Fig.2a). In contrast, TMPRSS2 expression was similar at all time points sampled (Fig.2a). The classical model of circadian regulation is one of transcriptional control, however PCR quantification of ACE2 and TMPRSS2 transcripts showed no evidence of a rhythmic pattern in Calu-3 cells (Fig.2b), consistent with posttranscriptional regulation of ACE2. To extend these observations we quantified Ace2 and Tmprss2 transcripts in lung, liver and intestine harvested from light/dark entrained mice and observed limited evidence for a circadian pattern of expression (Supplementary Fig.3). Immunoblotting of ACE2 in the murine lung resulted in multiple bands of varying molecular weight, compromising the interpretation of these experiments. To further explore the role of circadian pathways in regulating ACE2 we shRNA silenced Bmal1, the major circadian transcriptional activator, and showed reduced ACE2 expression but a negligible effect on TMPRSS2 in Calu-3 cells (Fig.2c). To assess the impact of Bmal1 on SARS-CoV-2pp entry we infected the silenced Calu-3 cells and showed a significant reduction in pp infection, whereas VSV-Gpp infection was unaffected (Fig.2d). During the COVID-19 pandemic several Spike variants have emerged; some conferring a fitness advantage to viral entry. D614G, has become prevalent in many countries, consistent with a reported fitness advantage for infecting cells of the upper respiratory tract 7,8 . SARS-CoV-2 Spike has a unique furin cleavage site that mediates membrane fusion and deletion of this motif has been observed both in vitro 9 and in animal models of infection 10 . Importantly, pp containing either Spike variant showed reduced infection of Bmal1 silenced Calu-3 cells (Fig.2d). In summary, these data show a circadian regulation of ACE2 that significantly impacts SARS-CoV-2 entry.
The availability of a synthetic agonist (SR9009) that activates REV-ERB and modulates circadian pathways 11,12 prompted us to investigate its role in SARS-CoV-2 infection. Treating Calu-3 cells with SR9009 reduced BMAL1 promoter activity and protein expression, with no impact on cell viability (Supplementary Fig.4). SR9009 treatment reduced ACE2 in a dose-dependent manner but had no effect on TMPRSS2 expression (Fig.3a). Importantly, SR9009 inhibited SARS-CoV-2pp infection in a BMAL1dependent manner (Fig.3b). As expected from our earlier results, VSV-Gpp infection was insensitive to SR9009 treatment (Fig.3b). SR9009 was equally effective at limiting entry of pp bearing the D614G or Furin-KO Spike variants (Fig.3c). Using an independent pseudoparticle system based on VSV 1 , we showed that SR9009 treatment of Calu-3 ( Fig.3d) or VERO cells (Supplementary Fig.5) reduced particle infection. To extend our observations to a more physiologically relevant system, we showed that treating primary bronchial epithelial cells with SR9009 significantly reduced SARS-CoV-2pp infection (Fig.3e). SARS-CoV-2 Spike binding to ACE2 can induce formation of multicellular syncytia [13][14][15] and we established a realtime assay to measure cell-cell fusion 16 . Treating target cells with SR009 reduced both ACE2 expression (Fig.3f) and Spike driven cell-cell fusion (Fig.3g). In summary, these demonstrate that REV-ERB agonists repress ACE2 expression and limit SARS-CoV-2 entry and cell-cell fusion.
BMAL1 and REV-ERB regulate gene expression by binding E-box or ROR response elements (RORE), respectively, in the promoter and enhancer regions of their target genes 17,18 . A genome-wide CRISPR screen identified 153 host factors that are important in SARS-CoV-2 infection 19 and bio-informatic analysis 20 identified 144 canonical E-box motifs 'CANNTG' and 80 ROR response elements 'RGGTCA' in these genes (Fig.4a). Furthermore, analysing murine transcript data from BMAL1 ChIP-seq 21 and REV-ERB knock-out animals 22 identified putative target genes and suggested that ~30% of SARS-CoV-2 host factors are BMAL1 or REV-ERB targets (Fig.4a), suggesting a role for circadian pathways in SARS-CoV-2 RNA replication. To test this hypothesis we validated our earlier pseudoparticle data with an authentic viral replication system and demonstrate a significant reduction of SARS-CoV-2 (Victoria 01/20 strain) replication in Bmal1 silenced Calu-3 compared to parental cells (Fig.4b). Furthermore, SR9009 treatment significantly reduced SARS-CoV-2 replication, assessed by intracellular RNA and the secretion of infectious particles (Fig.4c). These observations were recapitulated with VERO cells (Supplementary  Fig.6). To define whether circadian pathways regulate post-entry steps in the SARS-CoV-2 life cycle, we evaluated the effect of SR9009 on viral replication when added before or after virus inoculation. The agonist reduced viral RNA levels under both conditions (Fig.4c), leading us to conclude that both SARS-CoV-2 entry and replication are circadian regulated.
Our studies show circadian regulation of ACE2 in lung epithelial cells and show striking inhibitory effects of BMAL1 silencing and the synthetic REV-ERB agonist SR9009 treatment on ACE2 expression, SARS-CoV-2 entry and virus replication in lung epithelial cells. Of note, SARS-CoV-1 and alpha NL63 also require ACE2 to enter cells 23 and our data support a role for circadian factors in regulating the infection of these related coronaviruses. The human Ace2 promoter encodes putative binding sites for BMAL1/CLOCK and REV-ERB, however, we did not observe binding of these factors by chromatin immunoprecipitation-qPCR in Calu-3 cells (Supplementary Fig.7). In addition, there was no evidence for rhythmic expression of Ace2 transcripts in Calu-3 or in the lung, liver or intestine from entrained mice, consistent with bio-informatic analyses of published diurnal/circadian datasets from baboons 24 and mice 25 . These results are consistent with recent studies showing a role for post-transcriptional and post-translational mechanisms in regulating the circadian clock [26][27][28] . There is an emerging role of microRNAs (miRNAs) in modulating the circadian clock 29,30 and miRNAs have been reported to regulate ACE2 expression 31 , providing a possible post-transcriptional mechanism. Our studies show ACE2 to be a rhythmically 'moving' target and this could be relevant for evaluating new treatments targeting this step in the viral life cycle.
In addition to the circadian regulation of ACE2-mediated viral entry, we observed a marked suppression of SARS-CoV-2 RNA and genesis of infectious particles following SR9009 treatment and in Bmal1 silenced cells. Our bio-informatic analysis suggests that 30% of SARS-CoV-2 host factors are potentially BMAL1/REV-ERB regulated, highlighting a role for circadian-signalling to influence multiple steps in the SARS-CoV-2 life cycle. Further work is needed to characterise the circadian-dependent mechanisms of SARS-CoV-2 repression. A key finding from our study is the potential application of chronomodifying drugs for the treatment of COVID-19 32 . Dexamethasone is one of the few drugs that can reduce the severity of COVID-19 33 and is known to synchronise circadian pathways [34][35][36] . Over the last decade, a number of compounds that target core clock proteins have been developed 37 , including REV-ERB 38,39 and RORs 40,41 agonists that have been shown to inhibit hepatitis C virus and HIV replication 42,43 . A report demonstrating REV-ERB dependent and independent effects of SR9009 44 suggests some additional offtarget effects. We cannot exclude the possibility of additional pathways contributing to SR9009 antiviral activity; however, our use of genetic targeting approaches confirms a role for BMAL1 in regulating SARS-CoV-2 replication. REV-ERBa agonists can also impact the host immune response by suppressing IL-6 45 and so may offer a 'two pronged' approach to reduce viral replication and adverse host immune responses.
Circadian clocks can also impact on the pharmacokinetics and pharmacodynamics of drug responses 46 . Epidemiological [47][48][49] and molecular evidence 50 shows that night shift workers suffer circadian disruption and are at an increased risk of developing chronic inflammatory diseases. Identifying whether shift work is a risk factor for acquiring SARS-CoV-2 infection or developing moe severe disease could inform public health policy. Our observations raise questions as to how our in vitro studies translate to humans, where the time of exposure to SARS-CoV-2 may impact on the likelihood of infection, the host response, virus shedding, transmission and disease severity and are worthy of further investigation.   (a) Calu-3 cells were treated with SR9009 (5 or 10 µM) for 24h and assessed for ACE2 and TMPRSS2 expression together with housekeeping β-actin by western blotting. (b) Control or shBmal1 silenced Calu-3 cells were treated with SR9009 for 24h followed by infection with SARS-Cov-2pp or VSV-Gpp. After 24h viral entry was assessed by measuring luciferase activity and expressed relative to untreated cells (mean ± S.E.M., n = 3, Kruskal-Wallis ANOVA with Dunn's test). (c) Control or SR9009 treated Calu-3 cells were infected with SARS-CoV-2 wild type (WT), D614G or furin mutant pp. Viral entry was expressed relative to untreated cells 24h later (mean ± S.E.M., n = 4-6, Kruskal-Wallis ANOVA with Dunn's test). (d) Control or SR9009 treated Calu-3 cells were infected with SARS2-S-VSVpp and 24h later luciferase activity measured and data expressed relative to untreated cells (mean ± S.E.M., n = 3). Stats not needed for dose curves (e) Primary Bronchial Epithelial Cells (PBECs) were treated with SR9009 (20 µM) for 24h prior to SARS-CoV-2pp infection. Viral entry was measured as luciferase activity and data presented as mean ± S.E.M of triplicate wells from three independent donors relative to untreated control cells, Paired two-tailed t test. (f) SR9009 treated Huh-7 cells were assessed for ACE2 expression together with the housekeeping gene β-actin. (g) Huh-7 target cells expressing a split rLuc-GFP reporter (8)(9)(10)(11) were treated overnight with SR9009 at the indicated concentrations or with DMSO vehicle (Ctrl.) before culturing with Spike expressing HEK-293T cells with the split rLuc-GFP (1-7) and SR9009 added for the indicated times. A representative image of SARS-CoV-2 Spike induced syncytia at 3 days post coculture is shown (left panel). GFP-positive syncytia were quantified every 4h using an IncuCyte real-time imaging platform (right panel). Five fields of view were obtained per well at 10x magnification and GFP expression quantified by calculating the total GFP area using the IncuCyte analysis software.
Cell culture. Calu-3, Huh-7, HEK293T and VERO E6 cells were cultured in DMEM supplemented with 10% fetal bovine serum (FBS), 2mM L-glutamine, 100 U/mL penicillin and 10μg/mL streptomycin (all reagents from Life Technologies/Thermo Fisher). VERO E6/TMPRSS2 cells (a VERO E6 cell line stably overexpressing the TMPRSS2 gene, kindly provided by Dr Makoto Takeda at Department of Virology III, National Institute of Infectious Diseases 52 ) were cultured in DMEM supplemented with 10% fetal bovine serum, 10 units/mL penicillin, 10 mg/mL streptomycin, 10 mM HEPES (pH 7.4), and 1 mg/mL G418 (Life Technologies, UK). All cell lines were maintained at 37 o C and 5% CO2 in a standard culture incubator. Human PBECs were obtained using flexible fibreoptic bronchoscopy from healthy control volunteers under light sedation with fentanyl and midazolam. Participants provided written informed consent. The study was reviewed by the Oxford Research Ethics Committee B (18/SC/0361). Airway epithelial cells were taken using 2mm diameter cytology brushes from 3rd to 5th order bronchi and cultured in Airway Epithelial Cell medium (PromoCell, Heidelberg, Germany) in submerged culture. SARS-CoV-2 VSV pp were generated as previously reported 1 using reagents provided by Stefan Pöhlmann (Infection biology unit, German Primate Center, Göttingen, Germany). Briefly, HEK-293T cells were transfected with an expression construct encoding a c-terminal truncated version of SARS-CoV-2-S for assembly into VSV pp (pCG1-SARS-CoV-2-S-DC) 55 . After 24h, the transfected cells were infected with a replication-incompetent VSV (VSV*DG 56 ) containing GFP and firefly luciferase reporter genes, provided by Gert Zimmer (Institute of Virology and Immunology, Mittelhäusern, Switzerland). After 1h, the cells were washed with PBS before medium containing a VSV-G antibody (I1, mouse hybridoma supernatant from CRL-2700, ATCC) and supernatants harvested after 24h. The VSV*DG used for generating the pps was propagated in BHK-21 G43 cells stably expressing VSV-G. Viral titres were determined by infecting Calu-3 cells and measuring cellular luciferase after 48h.

SARS-CoV-2 propagation and infection.
For infection experiments, the SARS-CoV-2 (Victoria 01/20 isolate) obtained from Public Health England was propagated in Vero E6 cells. Naïve Vero E6 cells were infected with SARS-CoV-2 at an MOI of 0.003 and incubated for 48-72h until visible cytopathic effects were observed. Culture supernatants were harvested, clarified by centrifugation to remove residual cell debris and stored at -80 o C before measuring the infectious titre by plaque assay. Briefly, Vero E6 cells were inoculated with serial dilutions of SARS-CoV-2 stocks for 2h followed by addition of a semi-solid overlay consisting of 3% carboxymethyl cellulose (SIGMA). Cells were incubated for 72h, plaques enumerated by fixing cells using amido black stain and plaque-forming units (PFU) per mL calculated. For infection of Calu-3 cells, cells were plated 24h before infecting with the stated MOI. Cells were inoculated with virus for 2h after which the unbound virus was removed by washing three times with with phosphate buffered saline (PBS). Unless otherwise stated, infected cells were maintained for 24h before harvesting for downstream applications.

SARS-CoV-2 cell-cell fusion assay.
The SARS-CoV-2 cell-cell fusion assay was performed as described previously 16 . Briefly, HEK-293T Lenti rLuc-GFP 1-7 (effector cells) and Huh-7.5 Lenti rLuc-GFP 8-11 (target cells) cells were seeded separately at 7.5x10 5 per well in a 6 well dish in 3mL of phenol-red free DMEM, supplemented with 10% FBS, 1% sodium pyruvate and 1% penicillin/streptomycin (10,000 U/mL) before culturing overnight at 37 o C, 5% CO2. The effector cells were transfected with a plasmid expressing SARS-CoV-2 Spike or a blank vector. Meanwhile, target cells were diluted to 2x10 5 /mL and 100µL seeded into a clear, flat-bottomed 96 well plate and mock-treated or treated with SR9009 (3µM or 7µM). The following day, the effector cells were harvested, diluted to 2x10 5 /mL and 100µL cultured with the drug/target cell mix, with SR9009 concentrations maintained at 3µM or 7µM. To quantify GFP expression, cells were imaged every 4h using an IncuCyte S3 live cell imaging system (Essen BioScience). Five fields of view were obtained per well at 10x magnification and GFP expression determined by calculating the total GFP area using the IncuCyte analysis software.
Bioinformatics. The published 153 SARS-CoV-2 host factors were converted to Entrez gene names 19 . BMAL1 regulated genes were obtained from the published BMAL1 ChIP-seq in the mouse liver 21 . REV-ERB regulated genes were defined from published liver-specific loss of the REV-ERB paralogues 22 . Promoters (-1kb from TSS) of genes encoding SARS-CoV-2 host factors were analyzed with the HOMER (Hypergeometric Optimization of Motif EnRichment) tool for motif discovery (E-box motif CANNTG; RORE motif RGGTCA).

Animals.
Mouse experiments were carried out at the Institute of Viral and Liver Disease animal facility (approval number E-67-482-7). C57BL/6J male mice were purchased from Charles River and housed in individually ventilated cages under a 12/12 dark/light cycle with a ZT0 corresponding to 7am. After two weeks of acclimatisation, nine-week-old mice were sacrificed at different time points (ZT0, ZT4, ZT8, ZT12, ZT16, ZT20; n = 5/time point). Organs were harvested after exsanguination by intracardiac puncture, frozen in liquid nitrogen and kept at -80°C until further processing.

DATA AVAILABILITY
The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files or are available from the authors upon request.