Elevated temperature inhibits SARS-CoV-2 replication in respiratory epithelium independently of IFN-mediated innate immune defenses

The pandemic spread of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the etiological agent of Coronavirus Disease 2019 (COVID-19), represents an ongoing international health crisis. A key symptom of SARS-CoV-2 infection is the onset of fever, with a hyperthermic temperature range of 38 to 41°C. Fever is an evolutionarily conserved host response to microbial infection that can influence the outcome of viral pathogenicity and regulation of host innate and adaptive immune responses. However, it remains to be determined what effect elevated temperature has on SARS-CoV-2 replication. Utilizing a three-dimensional (3D) air–liquid interface (ALI) model that closely mimics the natural tissue physiology of SARS-CoV-2 infection in the respiratory airway, we identify tissue temperature to play an important role in the regulation of SARS-CoV-2 infection. Respiratory tissue incubated at 40°C remained permissive to SARS-CoV-2 entry but refractory to viral transcription, leading to significantly reduced levels of viral RNA replication and apical shedding of infectious virus. We identify tissue temperature to play an important role in the differential regulation of epithelial host responses to SARS-CoV-2 infection that impact upon multiple pathways, including intracellular immune regulation, without disruption to general transcription or epithelium integrity. We present the first evidence that febrile temperatures associated with COVID-19 inhibit SARS-CoV-2 replication in respiratory epithelia. Our data identify an important role for tissue temperature in the epithelial restriction of SARS-CoV-2 independently of canonical interferon (IFN)-mediated antiviral immune defenses.

With respect to COVID-19, up to 90% of hospitalized patients show low (44%) to moderate (13% to 34%) grade fever [4][5][6][7][8], with ICU patients presenting a 10% higher prevalence of fever relative to non-ICU patients [6,15]. In vitro studies have shown that SARS-CoV-2 replicates more efficiently at lower temperatures associated with the upper respiratory airway (33˚C), which correlates with an overall weaker interferon (IFN)-mediated immune response to infection relative to lower respiratory airway (37˚C) infection [16]. These data suggest that tissue temperature could be a significant factor in the host immune response to SARS-CoV-2 infection and COVID-19 disease progression. However, it remains to be determined what effect elevated temperature (>37˚C) has on SARS-CoV-2 replication. We therefore set out to determine the net effect of temperature elevation on SARS-CoV-2 infection within respiratory epithelial tissue.
Utilizing a three-dimensional (3D) respiratory model that closely mimics the tissue physiology and cellular tropism of SARS-CoV-2 infection observed in the respiratory airway [16][17][18][19][20][21][22][23][24], we demonstrate temperature elevation (�39˚C) to restrict the replication and propagation of SARS-CoV-2 independently of the induction of IFN-mediated antiviral immune defenses. We show that respiratory epithelium remains permissive to SARS-CoV-2 infection at temperatures up to 40˚C, but to restrict the initiation of viral transcription leading to reduced levels of viral RNA (vRNA) replication and apical shedding of infectious virus. We identify temperature to play an important role in the differential regulation of multiple epithelial host responses to SARS-CoV-2 infection, including epigenetic, long noncoding RNA (lncRNA), and immunity-related pathways. Our data identify an important role for tissue temperature in the epithelial restriction of SARS-CoV-2 replication independently of canonical IFN-mediated antiviral immune defenses previously reported to restrict SARS-CoV-2 infection.

Differentiation of primary human bronchial epithelial cells into ciliated respiratory epithelium supports SARS-CoV-2 replication in vitro
To establish a model suitable to study the effect of temperature on SARS-CoV-2 replication, we differentiated primary human bronchiolar epithelial (HBEp) cells isolated from 3 independent donors into stratified respiratory epithelium. Hematoxylin and eosin (H&E) and immunohistochemistry (IHC) staining of respiratory cultures demonstrated these tissues to contain a mixture of epithelial and goblet cells, with high levels of apical ciliation and angiotensin-converting enzyme 2 (ACE2) expression, the principal surface receptor of SARS-CoV-2 ( Fig 1A  and 1B) [25][26][27][28]. Infection of respiratory airway cultures with SARS-CoV-2 (England/02/2020; MOI 0.05, 10 4 plaque-forming unit (PFU)/Tissue) at 37˚C demonstrated these tissues to support infection and viral replication, with intraepithelial and apical vRNA accumulation readily detected by in situ hybridization (ISH) by 72 h postinfection (Fig 1A and 1B). Notably, we observed discrete clusters of vRNA accumulation within respiratory epithelia (Fig 1A, arrows), indicative of localized areas of intraepithelial infection, propagation, and spread [29,30]. The overall morphology of the respiratory epithelium remained largely intact, with little shedding of ciliated cells from the epithelial surface ( Fig 1A, 120 h). Measurement of genome copies by reverse transcription quantitative PCR (RT-qPCR) and infectious virus by TCID 50 in apical washes collected over time demonstrated the linear phase of virus release to occur between 48 and 120 h (Fig 1C and 1D). A substantial drop in apical infectious titers was observed at 144 h postinfection (Fig 1D), consistent with multiple waves of virus release over prolonged periods of incubation in primary airway cultures [31].

Respiratory airway cultures induce a heat stress response at elevated temperature
While the heat stress response has been extensively investigated in two-dimensional (2D) cell culture model systems [32,33], this pathway remains poorly characterized in 3D respiratory epithelia under air-liquid interface (ALI). We therefore investigated the influence of temperature on our 3D respiratory model. Mock-treated respiratory cultures were incubated at 37, 39, or 40˚C (representative of core body temperature and low to moderate grade febrile temperatures, respectively) prior to tissue fixation or RNA extraction and RNA sequencing (RNA-Seq). H&E staining and transepithelial electrical resistance (TEER) measurements demonstrated no morphological or membrane integrity changes to the respiratory epithelium upon temperature elevation (Fig 2A and 2B). RNA-Seq analysis of respiratory epithelia demonstrated no difference in the expression level of ACE2 (Fig 2C, p = 0.6753) or a reference set of 24 genes known to be constitutively expressed across a wide range of tissues and cell types (Fig 2D, p = 0.2136; [34][35][36]) upon temperature elevation. These data indicate that general transcription remains largely unperturbed at elevated temperatures up to 40˚C. Out of the 867 differentially expressed genes (DEGs) identified ( Fig 2E; Q < 0.05), DEG enrichment was observed across multiple pathways in tissues incubated at 40 relative to 37˚C, including cellular response to stress, RNA polymerase I promoter opening, and DNA methylation (Fig 2F and S1 Data). Significant DEG enrichment was also observed in cellular response to heat stress ( Fig 2F [arrow], 2G; S1 Data). We conclude that our respiratory model induces a significant heat stress response at elevated temperature without visible damage to the epithelium or induction of innate immune defenses that may otherwise influence the interpretation of SARS-CoV-2 replication studies at elevated temperature.

Elevated temperature restricts the replication of SARS-CoV-2 in respiratory airway cultures
We next examined the effect of elevated temperature on SARS-CoV-2 replication. Respiratory cultures were incubated at 37, 39, or 40˚C for 24 h prior to SARS-CoV-2 infection and continued incubation at their respective temperature. Measurement of genome copies and infectious virus from apical washes collected over time (24 to 72 h) demonstrated that extracellular viral titers were significantly decreased at both 39 and 40˚C relative to 37˚C (Fig 3A and 3B). Importantly, SARS-CoV-2 infection at 37˚C for 24 h prior to temperature upshift also led to reduced genome copy titers in apical washes (Fig 3C; p = 0.0056). These data indicate that temperature elevation, either prior to or following SARS-CoV-2 infection, is inhibitory to viral replication and/or release of viral particles. These data identify febrile temperature ranges associated with low (39˚C) to moderate (40˚C) grade fever to restrict SARS-CoV-2 propagation in respiratory epithelia.
We next investigated whether this temperature-dependent restriction was donor dependent. Respiratory cultures were incubated at 37 or 40˚C for 24 h prior to SARS-CoV-2 infection and continued incubation at their respective temperature. Measurement of genome copy titers in apical washes demonstrated viral titers were significantly reduced at 40˚C in a donorindependent manner (Fig 4A and 4B). Notably, donor-dependent profiles of restriction were observed (Fig 4B), with respiratory tissue derived from donor 1 being the most refractory to SARS-CoV-2 replication at 40˚C. We next examined if the decreased yield of infectious virus

PLOS BIOLOGY
Temperature elevation restricts SARS-CoV-2 replication observed at elevated temperature occurred due to viral genetic mutation. Amplicon sequencing of vRNA isolated from donor 1 apical washes at 72 h identified a total of 7 unique singlenucleotide polymorphisms (SNPs) relative to input sequence with no indels being identified (Table 1 and S2 Data). SNPs were only identified in genomic sequences isolated from apical washes incubated at 37˚C (Table 1). Thus, the temperature-dependent restriction observed in SARS-CoV-2 replication at 40˚C is not a consequence of viral mutation. As the generation of SNPs requires vRNA replication, the absence of genomic SNPs at 40˚C may reflect a temperature-dependent block in vRNA replication. To investigate this, we quantified the intracellular levels of vRNA from within infected tissues. RT-qPCR analysis using 2 independent primerprobe sets (N and ORF1a) demonstrated significantly lower levels of intracellular vRNA in tissues incubated at 40˚C relative to control samples ( Fig 4C). ISH staining of tissue sections confirmed abundant levels of intraepithelial vRNA accumulation within all infected tissues incubated at 37˚C, with little to no staining observed at 40˚C (Fig 4D). Together, these data indicate that respiratory tissue remains permissive to SARS-CoV-2 infection (viral entry) but refractory to vRNA replication to levels sufficient for robust ISH detection.

Elevated temperature restricts the onset of SARS-CoV-2 transcription independently of the IFN-mediated innate immune defenses
As we could detect significant levels of intracellular SARS-CoV-2 RNA within infected tissues at elevated temperature (Fig 4C), the temperature-dependent restriction observed could reflect enhanced levels of immune activation following pattern recognition receptor (PRR) detection of viral PAMPs [37,38]. To investigate this, we compared the relative transcriptome profiles from mock-treated or SARS-CoV-2-infected tissues at 37 or 40˚C derived from 3 independent donors (Figs 5 and S1-S4, S1 and S3-S5 Data). Principle component analysis (PCA) demonstrated host transcriptome signatures to be donor specific (S1 Fig

PLOS BIOLOGY
Temperature elevation restricts SARS-CoV-2 replication (donor 1, Caucasian male aged 63 years; donor 2, Hispanic male aged 62 years). Analysis of pooled data identified distinct profiles of DEG (Q < 0.05) expression for each paired experimental condition analyzed (Fig 5A and 5B and S6 Data), with clusters of unique and shared DEG expression (Fig 5B, purple lines). Significant differences in lncRNA expression were also identified between each paired condition (S5 Fig and S7 Data). Pathway analysis identified distinct profiles of DEG enrichment, with infection at 37˚C inducing an elevated immune signature (immune system process, GO:0002376) relative to all other conditions (Fig 5C, arrow). Notably, SARS-CoV-2 infection at 40˚C failed to register a significant immune response above that observed for mock treatment ( Fig 5C [arrow], gray box; p > 0.05). Collectively, these data identify tissue temperature to play an important role in the differential regulation of transcriptional host responses to SARS-CoV-2 infection, with enriched levels of immune gene expression observed during infection at 37˚C. Investigating further, we compared the transcriptome profiles from SARS-CoV-2-infected tissues at 37 and 40˚C. Out of the 1,934 DEGs (Q < 0.05) identified (Fig 6A), pathway analysis identified immune system and cytokine-related pathways to be suppressed during SARS-CoV-2 infection at 40 relative to 37˚C (Fig 6B and 6C and

PLOS BIOLOGY
Temperature elevation restricts SARS-CoV-2 replication S8 Data). DEG analysis identified IFNB1 (IFNβ) and IFNL1-3 (IFNλ) transcript levels to only be up-regulated above 1 CPM during SARS-CoV-2 infection at 37˚C (Fig 6D and S8 Data). While a small number of IFN-stimulated genes (ISGs) were identified to be up-regulated within infected tissues at 40˚C relative to mock treatment (9 ISGs in total; S3 Fig), no net increase in immune DEG expression was observed between mock or SARS-CoV-2-infected tissues at 40˚C relative to mock treatment at 37˚C (Fig 6C). Together, these data suggest that SARS-CoV-2 infection of respiratory epithelia at 40˚C fails to induce a robust IFN response. To corroborate this result, we performed RT-qPCR on RNA extracted from mock or SARS-CoV-2-infected tissues derived from multiple donors. The induction of IFNB1 and IFNL1 expression was readily observed in SARS-CoV-2-infected tissues at 37˚C, with no significant change observed during infection at 40˚C relative to mock-treated samples (Figs 6E and S6). Similarly, significant levels of ISG induction (Mx1, ISG15, and IFIT2) were only observed in SARS-CoV-2-infected tissues at 37˚C (Figs 6F and S6). Notably, the levels of ISG induction at 37˚C were relatively weak (3-to 6-fold relative to mock treatment; Fig 6F), indicative of low levels of innate immune stimulation at 72 h postinfection [20,39]. Infection of respiratory cultures treated with the Janus kinase (JAK) inhibitor Ruxolitinib (Ruxo; 5 μM), which blocks IFN-mediated JAK-STAT signaling during virus infection [40,41], demonstrated no significant change in SARS-CoV-2 apical titers relative to carrier control (DMSO) treatment ( Fig  6G). ThusAU : PleasecheckandconfirmthattheeditstothesentenceThus; incontrasttotemperatureelevation , in contrast to temperature elevation (Figs 3 and 4), these data demonstrate IFNmediated innate immune defenses to have a negligible effect on viral yield up to 72 h postinfection in our model system. We next examined the influence of temperature on SARS-CoV-2 replication in VeroE6 cells, a Chlorocebus sabaeus (African green monkey) cell line known to be permissive to SARS-CoV-2 infection but defective in type-I IFN-mediated host defenses [26,42,43]. Temperature elevation to 40˚C restricted SARS-CoV-2 replication, with or without preincubation at elevated temperature, relative to infection at 37˚C (S7 Fig). Collectively, these data demonstrate the temperature-dependent restriction of SARS-CoV-2 replication to occur independently of IFN-mediated antiviral immune defenses.
We next investigated at which stage in the replication cycle SARS-CoV-2 became restricted at elevated temperature. RNA-Seq analysis of SARS-CoV-2-infected tissues demonstrated no Table 1. SARS-CoV-2 sequence summary. SARS-CoV-2 RNA isolated from apical washes collected from donor 1 infected tissues incubated at 37 or 40˚C at 72 h were subjected to amplicon sequencing. Reference (ref) sequences (Wuhan, MN908947.3; England 2, EPI_ISL_407073) were used for consensus alignment and SNP identification. Unique SNPs (not detected in input and >5% base count frequency) and frequency (%base counts) shown. Full SNP analysis presented in S2 Data.

SNP frequency (%)
Amino acid change (relative to Wuhan)  (Fig 7B-7D). Importantly, a substantial decrease in the expression of viral subgenomic RNAs (sgRNAs) was also detected at elevated temperature ( Fig 7E, p = 0.0018). As the expression of SARS-CoV-2 sgRNAs requires the onset of viral transcription [44], these data identify tissue temperature to play an important role in the regulation of viral transcription. Correspondingly, indirect immunofluorescence staining of tissue sections demonstrated significantly fewer SARS-CoV-2 nucleocapsid (N) positive foci in tissues infected at 40 relative to 37˚C, confirming reduced levels of viral gene expression (Fig 7F and 7G, p < 0.0001). Costaining of tissue sections for Mx1 expression, a well-established ISG product (Fig 6D and 6F), demonstrated Mx1 expression to be localized in proximity to SARS-CoV-2 infectious foci at 37˚C, with little to no staining observed within infected  (Fig 7F). We conclude that infection of respiratory tissue at elevated temperature restricts SARS-CoV-2 replication through a mechanism that inhibits viral transcription independently of IFN-mediated antiviral immune defenses previously shown to restrict SARS-CoV-2 replication [37,38].

Discussion
A defining symptom of COVID-19 is the onset of fever with a febrile temperature range of 38 to 41˚C [4][5][6][7][8]. However, the effect of elevated temperature on SARS-CoV-2 tissue tropism and replication has remained to be determined. Here, we identify a temperature-sensitive

PLOS BIOLOGY
Temperature elevation restricts SARS-CoV-2 replication phenotype in SARS-CoV-2 replication in respiratory epithelia that occurs independently of the robust induction of IFN-mediated innate immune defenses known to restrict SARS-CoV-2 replication [37,38]. The differentiation of stratified respiratory epithelium has proven to be a valuable research tool to investigate the cellular tropism, replication kinetics, and immune regulation of SARS-CoV-2, as these tissue models mimic many aspects of infection observed in animal models and COVID-19 patients [16][17][18][19][20][21][22][23][24]29,[45][46][47][48][49]. While the use of such 3D models represents an important advancement over traditional two-dimensional cell culture systems, the absence of circulating immune cells (e.g., macrophages, natural killer cells, dendritic cells, and neutrophils), which modulate fever and proinflammatory immune responses to infection is an important limiting factor [9,10,50]. Thus, we limit our conclusions to the effect of elevated temperature on SARS-CoV-2 infection and replication within respiratory epithelium. Consistent with previous reports [16,17,[19][20][21]30,38,49], SARS-CoV-2 infection of respiratory epithelium at 37˚C induced a proinflammatory immune signature (S2D Fig; IFNβ and IFNλ1-3). While SARS-CoV-2 is known to induce a comparatively weak immune response relative to other respiratory viruses (e.g., IAV) [20,39], we observed localized areas of epithelial infection (Fig 1A, arrows) that were coincident with elevated levels of ISG expression by 72 h postinfection (Fig 7F, Mx1). These findings are consistent with reports that have shown localized areas of infection in the respiratory airway of infected ferrets and surface epithelium of organoid cultures [29][30][31]49]. Thus, the proportion of epithelial infection and/or rate of intraepithelial spread may account for the differences observed in immune signature reported between different respiratory pathogens [20,51]. As we observed localized areas of ISG induction, our data support a model of infection where cytokine (e.g., IFN) receptor-binding consumption kinetics may be a rate limiting factor in the protection of respiratory epithelia to SARS-CoV-2 infection [52,53]. Such observations warrant further investigation to determine whether these localized areas of epithelial immune protection are induced by specific cytokines (e.g., IFNβ or IFNλ1-3) and/or correlate with COVID-19 disease progression.
We demonstrate 3D respiratory epithelial cultures under ALI to induce a robust heat stress response upon temperature elevation without loss of epithelium integrity, disruption to general cellular transcription, or damage-associated molecular pattern (DAMP) activation of innate immune defenses (Figs 2 and S4). Temperature elevation alone was not sufficient to block SARS-CoV-2 entry (Figs 3, 4, and S7), but refractory to SARS-CoV-2 transcription leading to reduced levels of vRNA accumulation and apical shedding of infectious virus (Figs 3, 4, and 7). Thus, we identify febrile temperatures associated with COVID-19 to play an important role in limiting the epithelial replication of SARS-CoV-2 infection. Importantly, SARS-CoV-2 restriction occurred independently of the robust induction of type-I (IFNβ) or type-III (IFNλ) IFN-mediated immune defenses (Figs 5, 6, and S7), despite infected tissues having abundant levels of intracellular vRNA (Fig 4C). We posit that the lack of immune induction observed at 40˚C is likely a consequence of diminished levels of vRNA replication, which has been shown to play an important role in the production of PAMPs (e.g., dsRNA intermediates) required for the activation of innate immune defenses to other respiratory pathogens [54][55][56][57]. However, PRR detection of PAMPs plays an important role in the induction of a proinflammatory cytokines (including IL-6) required to mount a fever response to microbial challenge in vivo [9,10]. Thus, additional animal studies will be required to determine whether the temperaturedependent restriction of SARS-CoV-2 occurs independently of an IFN response in vivo. While speculative, we posit that low to moderate grade fever may confer protection to respiratory tissue within SARS-CoV-2-infected individuals as a component of a homeostatically controlled non-hyperinflammatory immune response to infection.
While we identify tissue temperature to play an important role in the regulation of SARS-CoV-2 transcription and replication in respiratory epithelia (Fig 7D and 7E), the precise molecular mechanism(s) of restriction remain to be determined. The heat stress response has been shown to play both a positive (proviral) and negative (antiviral) role during virus replication [58][59][60]. Our transcriptomic analysis identified respiratory epithelial host responses to be differentially regulated in response to both temperature elevation and SARS-CoV-2 infection (Figs 5, 6, and S1-S5). Thus, SARS-CoV-2 infection of respiratory epithelia at elevated temperature elicits a distinct host response that may contribute to the restriction of SARS-CoV-2 replication at multiple stages of infection. For example, we identify lncRNAs to be differentially expressed in response to both SARS-CoV-2 infection and temperature (S5 Fig), which are known to influence the outcome of virus infection independently of IFN-mediated immune defenses [61]. Thus, multiple gene products and/or pathways may contribute to the sequential or accumulative restriction of SARS-CoV-2 replication within respiratory epithelia at elevated temperature.
We identify temperature elevation to correlate with lower levels of SARS-CoV-2 sgRNA expression (Fig 7E), identifying a temperature-dependent block in SARS-CoV-2 transcription. We hypothesize that this restriction may relate to an inhibition in SARS-CoV-2 RNA-dependent RNA polymerase (RdRp) activity and/or binding affinity to vRNA, as the genomic replication activity of the IAV RdRp polymerase is known to be restricted at elevated temperature (41˚C) [62]. Notably, avian IAV cold adaptation of RdRp is a host determinant of IAV zoonosis, which naturally replicates in the intestinal tract of birds at higher temperatures [63]. As circulating strains of SARS-CoV-2 are susceptible to nonsynonymous mutation within RdRp coding sequences (http://cov-glue.cvr.gla.ac.uk/#/home), molecular studies are warranted to determine if such amino acid substitutions might influence the thermal restriction of SARS-CoV-2 replication. While SARS-CoV-2 has been reported to replicate to higher viral titers at temperatures associated with the upper respiratory airway (33˚C) relative to core body temperature (37˚C) [16], the replication of coronaviruses at febrile body temperatures has remained poorly characterized. While temperature-sensitive (ts) coronavirus mutants have been identified for feline infectious peritonitis virus (tsFIPV) and murine hepatitis virus (MHV tsNC11), their parental wild-type derivatives were able to replicate at temperatures �39˚C [64,65]. The replication defect for tsFIPV was attributed to a block in viral maturation that limited its replication in cats to the upper respiratory airway [64], whereas the growth defect in tsNC11 was attributed to a coding mutation within the macrodomain and papain-like protease 2 domain of the nonstructural protein 3 [65]. Thus, we present the first evidence demonstrating a circulating strain of coronavirus to be sensitive to temperature thermoregulation. Genomic analysis of SARS-CoV-2 isolated from apical washes failed to identify any unique SNPs from infected tissues at 40˚C relative to input sequence. These data support the temperature-dependent block in SARS-CoV-2 propagation to occur prior to vRNA replication, a prerequisite requirement for genomic mutation. As such, heat adaptation gain of function experiments through serial passage of SARS-CoV-2 at elevated temperature (�39˚C) may shed light on whether the restriction observed is related to viral and/or cellular host factors. Importantly, however, appropriate levels of biosafety (ethical and genetic modification) should be considered prior to such experimentation.
In summary, we identify an important role for tissue temperature in the cellular restriction of SARS-CoV-2 during infection of respiratory epithelia that occurs independently of the robust induction of canonical IFN-mediated antiviral immune defenses known to restrict SARS-CoV-2. We demonstrate tissue temperature to significantly influence the differential regulation of epithelial host responses to SARS-CoV-2 infection and the progression of viral transcription upon tissue infection. Future investigation is warranted to determine the precise mechanism(s) of restriction, as this may uncover novel avenues for therapeutic intervention in the treatment of COVID-19.

Materials and methods
Virus SARS-CoV-2 (SCV2) strain BetaCoV/England/02/2020/EPI_ISL_407073 was isolated from a COVID-19 patient (Public Health England). The virus was passaged 3 times in Vero E6 cells and genotype sequence confirmed by Illumina sequencing. All experiments were performed in a Biosafety level 3-laboratory at the MRC-University of Glasgow Centre for Virus Research (SAPO/223/2017/1a).

SARS-CoV-2 replication kinetics in Vero E6 cells
Vero E6 cells (a gift from Michelle Bouloy, Institute Pasteur, France) were propagated in Dulbecco's Modified Eagle Medium (DMEM GlutaMAX; ThermoFisher, 31966-021) supplemented with 10% fetal calf serum (FCS; ThermoFisher, 10499044 [Lot Number 08G8293K]) at 37˚C with 5% CO 2 . Cells were seeded the day before infection at a density of 2 × 10 5 cells per well in a 12-well plate and infected with 10 4 PFU of SARS-CoV-2 diluted in serum-free DMEM for 120 min with occasional rocking. The inoculum was removed and replaced by DMEM + 10% FCS. Supernatant was collected every 24 h and snap frozen prior to analysis. For temperature inhibition experiments, cells were either preincubated at 37 or 40˚C for 24 h prior to infection with continued incubation at their respective temperatures or incubated at the indicated temperatures after infection at 37˚C after the addition of overlay medium (as indicated).

Differentiation and infection of respiratory epithelium
Three independent donors of primary HBEp cells were used for this study. Donor 1, 63-yearold Caucasian male (EpithelixSarl, 02AB077201F2); donor 2, 62-year-old Hispanic male (EpithelixSarl, AB079301); donor 3, 16-year-old Caucasian female (Lonza, CC-2540S). All donors were healthy and nonsmokers. For expansion and amplification, cells were propagated at 37˚C with 5% CO 2 in hAEC culture medium (Epithelix, EP09AM). All differentiation experiments were performed on cells that had been passaged a total of 4 times. Cells were seeded onto transwells (Falcon, 734-0036) at a cell density of 3 × 10 4 cells per transwell and grown to confluency. Cells were differentiated under ALI in PneumaCult-ALI medium containing hydrocortisone and heparin supplements (STEMCELL Technologies, 05001) as specified by the manufacturer's guidelines. Cells were differentiated for a minimum of 35 days, with media changed every 48 h and washed twice weekly in serum-free DMEM after day 20 post-ALI to remove mucus build up. ALI cultures were apically washed twice before infection, 24 h prior to infection and immediately preceding infection. Cultures were inoculated apically with 100 μl of serum-free DMEM containing 10 4 PFU (MOI of 0.05) of SARS-CoV-2 (based on Vero E6 titres) for 120 min. The inoculum was removed, and the apical surface washed once in serum-free DMEM (0 h time point). Unless stated otherwise, ALI cultures were incubated at 37˚C with 5% CO 2 . For temperature inhibition experiments, ALI cultures were preincubated at 37 or 40˚C for 24 h prior to infection with continued incubation at their respective temperatures. For temperature upshift experiments, ALI cultures were infected at 37˚C and incubated for 24 h prior to temperature upshift to 40˚C or continued incubation at 37˚C. For Ruxolitinib (Ruxo) experiments, DMSO (carrier control) or Ruxo (Selleckchem, S1378) was added to basal medium at a final concentration of 5 μM for 16 h prior to infection at 37˚C. Cultures were incubated at 37˚C and basal media changed daily containing fresh inhibitor. Apical washes were collected every 24 h in 200 μl of serum-free DMEM incubated for 30 min to collect infectious virus. Samples were stored at −80˚C until required. Tissues were fixed at the indicated time points in 8% formaldehyde (Fisher Scientific, F/1501/PB17) for 16 to 24 h prior to paraffin embedding and processing. TEER measurements were conducted using an ERS-2 Voltohmmeter (Millicell; MERS00002). PBS was added to apical and basal chambers of ALI cultures or empty transwells and allowed to stand at room temperature (RT) for 15 min. Readings were taken prior to temperature shift (day 0) and every 24 h for 3 days. TEER measurements were calculated by subtracting the average of empty transwells from ALI epithelial readings and reported as Ωcm 2 .

Plaque assay
Vero E6 cells (subclone 6F5, MESO) were seeded at a density of 2.5 × 10 5 cells/well in 12-well plates the day before infection. Ten-fold serial dilutions of virus were prepared in DMEM + 2% FCS in a final volume of 500 μl. Cells were infected for 60 min with occasional rocking at 37˚C in 5% CO 2 . The inoculum was removed and 1 ml of overlay containing 0.6% Avicel (Avicel microcrystalline cellulose, RC-591) made up in DMEM containing 2% FCS was added to each well. Cells were incubated for 3 days at 37˚C in 5% CO 2 prior to fixation in 8% formaldehyde and Coomassie brilliant blue staining (as described above). Plaques were counted manually and the PFU per ml calculated.

In situ hybridization and immunostaining of respiratory tissue sections
Paraffin-embedded tissues were sectioned using a microtome (approximately 2 to 3 μm thick) and mounted on glass slides. Tissue sections were stained with H&E. For immunofluorescence staining, tissue sections underwent antigen retrieval by citrate (pH 6) or EDTA (pH 8) pressure cooking (as indicated below). RNAscope was used for the detection of SARS-CoV-2 RNA using a Spike-specific probe set (Advanced Cell Diagnostics, 848561 and 322372) following the manufacturer's protocol, which included a pretreatment with boiling in target solution and proteinase K treatment. ACE2 was detected using a rabbit polyclonal anti-hACE2 antibody (Cell Signaling, 4355S; citrate antigen retrieval) and EnVision+ anti-rabbit HRP (Agilent, K4003). Slides were scanned with a bright field slide scanner (Leica, Aperio Versa 8). Mx1 was detected using a mouse monoclonal anti-Mx1 antibody ( [67]; EDTA antigen retrieval). Nucleocapsid protein was detected using a sheep polyclonal anti-SARS-CoV-2 N protein antibody (University of Dundee, DA114; EDTA antigen retrieval). Secondary antibodies for detection were rabbit anti-sheep AlexaFluor 555 (abcam, ab150182) and rabbit anti-mouse AlexaFluor 488 (Sigma-Aldrich, SAB4600056). Nuclei were stained with ProLong Gold (Life Technologies, P36941). Images were collected using a Zeiss LSM 710 confocal microscope using 40× Plan-Apochromat oil objective lens (numerical aperture 1.4) using 405, 488, and 543 nm laser lines. Zen black software (Zeiss) was used for image capture and exporting images, with minimal adjustment (image rotation) in Adobe Photoshop for presentation.

Viral genomic amplicon sequencing
vRNA was extracted from input or extracellular released virus collected from apical washes (72 h postinfection) using TRIzol Reagent. RNA quality control was performed using Qubit (Invitrogen) and Agilent 4200 Tapestation System (Agilent). For amplicon-based sequencing, amplicons were prepared according to the protocol developed by the ARTIC Network using primer version 3 (https://artic.network/ncov-2019). DNA fragments were cleaned using AMPure beads (Beckman Coulter). Illumina sequencing libraries were prepared using 40 ng of amplicon and the KAPA HyperPrep kit (Roche, KK8502), as per manufacturer's protocol. Samples were indexed using the NEBNext Multiplex Oligos for Illumina Dual Index Primers Set 1 (New England Biolabs, E6440S) and libraries amplified with 8 cycles of PCR. Amplified libraries were cleaned using AMPure beads (Beckman Coulter) and quantified using Qubit dsDNA HS kit (Invitrogen) and run on an Agilent 4200 Tapestation System (Agilent). Equimolar amounts of libraries were pooled, loaded at a final concentration of 11 pM, and sequenced in an Illumina MiSeq System (Illumina) using a MiSeq v2 cartridge 500 cycle kit. Sequences have been deposited in the European Nucleotide Archive (https://www.ebi.ac.uk/ ena/browser/home), accession number PRJEB41332. Reads were aligned against Wuhan-Hu-1 reference strain (GenBank accession MN908947.3) using BWA [71] followed by primer trimming and consensus calling with iVar [76]. The assembled data were parsed using Diversi-Tools (http://josephhughes.github.io/btctools/) to determine the frequency of nucleotides at each site and observe the mutations. COV-GLUE (http://cov-glue.cvr.gla.ac.uk/#/home) was used for identifying amino acid replacement following by mutation.

Statistical analysis
The number (N) of tissues sampled per experimental condition is shown throughout and derived from a minimum of 3 independent biological experiments. GraphPad Prism (version 9.2.0) was used for statistical analysis on all data points per sample condition. Statistical tests and p-values are shown throughout. Statistically significant differences were accepted at p < 0.05.