Novel immunomodulatory properties of adenosine analogs promote their antiviral activity against SARS-CoV-2

Given the striking structural similarities between the A2AR natural ligand, adenosine, and adenosine analog drugs, we hypothesized that these compounds may interact with A2AR and exert immunomodulatory effects in addition to their intrinsic antiviral activity. To test this idea, we used computational molecular docking, to predict the interaction between adenosine analogs and A2AR. To validate the reliability of our molecular docking analysis, we first tried to dock adenosine to A2AR and compared it to the known crystal structure of A2AR in complex with adenosine, which was resolved in (Lebon et al, 2011; Lebon et al, 2011). We found that our modeling was able to reproduce the crystal structure of the A2AR-adenosine complex (Fig. EV1A), confirming that our computational modeling can reliably predict the interaction between these molecules. We then screened several adenosine analogs that are in clinical development, or clinical use, as antivirals and/or antineoplastic agents (Table 1; Appendix Fig. S1). We found that some adenosine analogs are predicted to possess a higher or similar affinity to A2AR compared to adenosine, as indicated by the ΔG value, and with similar interacting residues as adenosine (Table 1; Appendix Fig. S1). Among all the screened analogs, GS-441524 (GS), the active metabolite of the antiviral drug remdesivir (REM), had the highest predicted affinity for A2AR and very similar interacting residues as adenosine. Furthermore, GS and adenosine shared striking structural similarities (Fig. EV1B), suggesting that GS could potentially interact with A2AR. These observations prompted us to select GS to test whether it may have immunomodulatory effects through interaction with A2AR, in addition to its direct antiviral activity. To test our hypothesis, we carried out an in-depth analysis of the binding between GS and A2AR compared to the adenosine-A2AR complex. Of the 9 predicted GS-A2AR states obtained from the molecular docking simulation (Fig. EV1C), the first two states exhibit higher calculated affinities for A2AR than adenosine itself, as indicated by the ΔG values of GS (−8.4, −8.3 kcal/mol) and adenosine (−7.7 kcal/mol) in complex with A2AR (Figs. 1A–C and EV1C). In addition, these two states display similar bonds as adenosine-A2AR (Figs. 1D and  EV1C, EV2A). The first state showed a different orientation of GS compared to adenosine, while the second state has an almost identical orientation to adenosine (Fig. 1A–C). Both GS-A2AR complexes shared with adenosine-A2AR key hydrogen bonds, including the ones with SER-277, HIS-278, GLU-169, and ASN-253 (Figs. 1A–D and EV2B).

Since there are other adenosine receptors (A2BR, A1R, and A3R), apart from A2AR, and adenosine interacts with all of them, we asked if GS may interact with other adenosine receptors. We obtained docking results, using the same methodology, with A1R and A2BR crystal structures (Cai et al, 2022; Draper-Joyce et al, 2018). Docking with A3R was not attempted, because the structure of this receptor has not been resolved yet. Our modeling showed that GS has a similar affinity for A1R as adenosine and binds to the A1R active site (Fig. EV3); GS has a lower affinity than adenosine for A2BR and does not appear to bind within the active site of the receptor (Fig. EV4). Based on our molecular docking analysis, we predicted that GS interacts with A2AR with high affinity and may also interact, to a much lesser extent, with A1R.

The molecular docking analysis was useful to predict the interaction between GS and A2AR. However, we did not know whether this interaction leads to blockade or activation of the receptor. To determine whether GS has A2AR-mediated effects, we treated mice with known A2AR agonist CGS-21680 (CGS), known A2AR antagonist ZM-241385 (ZM), or REM. We used REM for in vivo studies instead of GS because REM has a superior pharmacokinetics and ensures optimal tissue distribution of its metabolite GS (Leegwater et al, 2022; Hu et al, 2021). We observed a dose-dependent increase in the neutrophils-to-lymphocytes (NLR) ratio and a reduction in the number of splenic CD69 + CD3+ activated T-cells in CGS-treated mice (Fig. 2A), in agreement with the suppressive effect of activating A2AR in T-cells; whereas treatment with the A2AR antagonist ZM or REM did not change these readouts (Fig. 2A). A2AR antagonists, which block the A2AR, do not usually show significant effects in the absence of an A2AR agonist, similar to what we observed for the known A2AR antagonist, ZM, and REM. None of the treatments changed the weight of the mice (Appendix Fig. S2). Therefore, based on these observations, we hypothesized that REM metabolite, GS, could act as an A2AR antagonist, like ZM. To address this hypothesis, we tested whether GS can antagonize the A2AR agonist CGS in a number of in vitro and in vivo studies.

We first performed a dose-response study by treating primary mouse CD8 + T-cells, activated with anti-CD3/CD28, with increasing doses of CGS against a fixed dose of GS (Fig. 2B). In this assay, we used CD8 + T-cells proliferation as a readout (response), since we know that CGS reduces proliferation in activated CD8 + T-cells as shown in (Monticone et al, 2022) and other works (Sorrentino et al, 2019; Leone et al, 2018; Ohta et al, 2012). As expected, we observed that CGS reduces T-cell proliferation in a dose-dependent manner (Fig. 2B). Importantly, we found that GS antagonizes CGS-mediated suppression of proliferation, reduces the efficacy (E_max) and the potency of CGS (Fig. 2B). To further test whether GS binds to A2AR and has antagonistic properties, we measured the production of cyclic AMP (cAMP), the direct secondary mediator of A2AR (Huang et al, 1997), in an A2AR-expressing cell line in response to GS, ZM, and CGS, using a luciferase assay (Fig. 2C). We found that GS antagonizes and reduces cAMP production induced by CGS, similarly to the A2AR antagonist ZM (Fig. 2C). Along the same line, we also showed that GS antagonizes the overproduction of cyclic AMP (cAMP) in response to CGS, reducing it to control (vehicle) levels, as measured by ELISA in primary mouse CD8 + T-cells (Fig. 2C).

We then asked if GS could antagonize A2AR activation in vivo and whether this translates into immunomodulatory effects. We treated uninfected, immunocompetent mice with CGS alone or in combination with REM and we analyzed immunological readouts (Fig. 2D–H). We performed the experiments in uninfected mice to exclude the possibility that any observed immunomodulatory effects may be secondary to the intrinsic antiviral activity of REM. We found that REM treatment strongly reverses CGS-induced immunological effects, including reducing NLR (Fig. 2E) and the number of circulating neutrophils, while increasing splenic T-cells and NK cells (Fig. 2F). As further proof that REM treatment modulates A2AR, we tested whether REM treatment has any effect on NICD, the active form of Notch1, which is also known to contribute to the signaling cascade downstream of A2AR (Monticone et al, 2022). We found that REM restores NICD from CGS-mediated inhibition in CD8+ and CD4+ splenic T-cells in vivo in non-infected immunocompetent mice, as well as GS restores NICD in primary mouse CD8 + T-cells in vitro (Fig. 2G,H). In all the experiments, REM or GS did not have effects as single agents, but only when used to antagonize CGS. Overall, our results provide evidence that REM active metabolite, GS, acts as an A2AR antagonist and elicits immunological effects through A2AR blockade, independently of its intrinsic antiviral activity. We have not explored the activity of GS on other adenosine receptors, and this will be the focus of future studies.

Having demonstrated an immune-stimulatory activity of GS, we tested whether stimulating CD8 + T-cell responses via A2AR blockade with immunotherapeutic A2AR antagonists could be beneficial against SARS-CoV-2 infection. We first needed to exclude that any antiviral effect observed in response to an A2AR antagonist may reflect direct antiviral activity. Therefore, we tested whether the A2AR antagonist, ZM, has any direct antiviral activity against SARS-CoV-2. We treated Vero cells with ZM or REM prior to infection with SARS-CoV-2. We did not observe a reduction in the viral burden in Vero cells in response to ZM, in contrast with REM, thus suggesting that ZM does not have intrinsic antiviral activity against SARS-CoV-2 (Fig. 3A). We then treated mice infected with SARS-CoV-2 with ZM (Fig. 3B). We observed milder weight loss, faster recovery and milder disease manifestations (clinical score) in mice receiving ZM compared to mice receiving vehicle (Fig. 3C,D). Interestingly, the histopathological analysis showed increased CD8 + T-cell infiltration in the lungs of ZM-treated mice (Fig. 3E,F) and this correlated with decreased viral load in the lungs (Fig. 3G), suggesting that ZM increases the efficiency of viral clearance by CD8 + T-cells. These findings are remarkable considering that ZM has no intrinsic antiviral activity, and suggest that systemic blockade of A2AR is beneficial for clearing SARS-CoV-2 infection, possibly by boosting CD8 + T-cell function.

It has been reported that during SARS-CoV-2 infection, IFN-γ production by CD8+ and Th1 CD4 + T-cells promotes viral clearance, whereas IL-10 and IL-17 released by Th2 and Th17 CD4 + T-cells are associated with severe disease outcomes (Huang et al, 1997; Tan et al, 2021; Pavel et al, 2021). Therefore, we asked how manipulation of A2AR changes cytokine production in different lymphocyte populations. In agreement with our previous work (Monticone et al, 2022), we found that activation of A2AR by the A2AR agonist, CGS, downregulates the active form of Notch1, NICD—a key regulator of T-cell functions, including cytokine production (Monticone et al, 2022; Cho et al, 2009; Palaga et al, 2003)—and suppresses proliferation, whereas treatment with ZM rescues NICD and proliferation in primary mouse CD8 + T-cells and CD4 + T-cells (Fig. 4A–D). Consistently, we found that CGS and ZM antagonize each other’s effects on cytokine production: CGS suppresses IFN-γ, whereas ZM rescues IFN-γ production from CGS in CD8 + T-cells and Th1 CD4 + T-cells (Fig. 4E,F); CGS does not change IL-10 in Th2 CD4 + T-cells, while ZM reduces IL-10 when used as a single treatment or against CGS (Fig. 4G); CGS increases IL-17 in Th17 CD4 + T-cells, whereas ZM restores it to control (vehicle) level, antagonizing CGS (Fig. 4G,H). Overall, our results indicate that A2AR antagonism could stimulate protective adaptive immune responses, and this could promote the clearance of SARS-CoV-2 infection.

Having found previously unknown immunomodulatory properties of REM, through the adenosine analog GS, and that a non-antiviral A2AR antagonist promotes clearing of SARS-CoV-2 in mice, we then asked if REM has similar immunomodulatory effects in SARS-CoV-2-infected animals. To test our hypothesis, we analyzed immunological readouts in mice infected with SARS-CoV-2 and treated with REM or vehicle (Fig. 5A). As expected, we observed faster recovery in mouse weight (Fig. 5B), faster viral clearance as measured by RT-PCR and IHC and milder disease manifestations (Figs. 5C and EV5A,B) in mice receiving REM compared to mice receiving vehicle. REM-treated mice started to regain weight at day 4 post-infection (Fig. 5B) and showed immune infiltration in the lungs with no significant damage (Fig. 5D), whereas we observed increased inflammation, hemorrhage, and significant damage, from bronchiolar shedding, in the lungs of mice receiving vehicle (Fig. 5D). Importantly, we found that REM treatment results in a remarkable increase of CD8 + T-cells in the lungs of infected mice (Fig. 5E,I), similar to what was observed with ZM treatment. Consistently, we found transcriptional upregulation of genes involved in adaptive immune responses and A2AR antagonism, including IFN-γ response and IL-2 signaling (Fig. 5H), in the lungs of mice treated with REM. These findings may suggest that the treatment is promoting T-cell immune responses in the lungs which contribute to viral clearance. Interestingly, we also observed a significant reduction of neutrophils (Figs. 5F and  EV5C) and a drastic reduction of Neutrophils Elastase (NE) in the lungs of REM-treated mice (Figs. 5G and EV5C), which may indicate suppression of neutrophil recruitment and neutrophil extracellular traps secretion (NET) by neutrophils, key factors contributing to severe outcomes in COVID-19 (Barnes et al, 2020). In agreement with the observed increase in CD8 + T-cells and decrease in neutrophils in the lungs, we also observed the remarkable reversal of NLR in response to REM (Fig. 5J), which was also accompanied by the reduction of neutrophils and an increase in CD3 + T-cells in circulation (Fig. EV5D,E). These effects are in agreement with our results in uninfected mice treated with REM or infected mice treated with ZM, thus suggesting that these effects are largely independent of REM intrinsic antiviral activity and dependent on its immunomodulatory properties. Furthermore, these immunological effects were observed at 4-days post-infection when mice treated with REM started to regain weight, whereas mice treated with vehicle experienced the worst decrease in weight (Fig. 5B), suggesting that the immunological effects of REM may contribute to ameliorating disease manifestations and outcomes. Finally, we found decreased cAMP expression (the second mediator of A2AR) by immunohistochemistry in correspondence to CD8 + T-cell infiltration in the lungs of REM-treated mice (Fig. 6A), which supports the hypothesis that REM elicits immunological effects through A2AR antagonism.

Lastly, we used transcriptomics of lungs at 4 and 7 days post-infection to explore the effects observed on the functional modulation of immune cells and responses elicited by REM. Using KEGG pathway analysis, we found that REM significantly changes genes of pathways associated with A2AR, including the cAMP signaling pathway, purine metabolism, and HIF-1 signaling pathway (Huang et al, 1997; Poth et al, 2013) (Fig. 6B), thus further suggesting that REM acts on A2AR. In agreement with the different disease course observed in REM-treated mice vs. vehicle, we identified a significant number of differently expressed genes (DEGs) when comparing the lungs of REM-treated mice vs. vehicle at 4 and 7 days post-infection (Fig. 6C). To interrogate the function of the identified DEGs we performed Gene Ontology (GO) analysis of Biological Processes (BP). These analyses showed that REM significantly changes the expression of genes involved in immune processes at 4-days post-infection, including defense response to other organisms, immune response, T-cell chemotaxis, and cytokine secretion (Fig. 6D), and genes involved in tissue repair processes at 7-days post-infection, including cellular component organization and biogenesis, biological adhesion, and tissue morphogenesis (Fig. 6E). Consistently, we found upregulation of pathways associated with tissue damage in COVID-19, including Reactive Oxygen Species (ROS) pathway, EF2 targets, G2M checkpoint, mitotic spindle, and MYC targets (Ahern et al, 2022), in lung samples from mice receiving vehicle, both at 4 and 7 days post-infection (Fig. EV5F). These results are consistent with the hypothesis that REM elicits protective antiviral immune responses that lead to faster recovery and milder disease course. Collectively, our findings support the hypothesis that REM may have immunomodulatory effects which are achieved, in a significant part, through A2AR antagonism, and may be independent of its intrinsic antiviral activity.

Twelve-week-old female C57Bl/6 J mice were purchased from The Jackson Laboratory. SARS-CoV-2-infected mice were housed in the Animal Biosafety Level 3 Facility at the University of Pennsylvania under the standard dark/light cycle, ambient temperature, humidity, and within the Allentown individual ventilated BCU2 caging system. Non-infected mice were housed in the LSUHSC Animal facility and kept in specific pathogen-free conditions under standard dark/light cycles and ambient conditions. All work was in adherence to the University of Pennsylvania’s and Louisiana Health Sciences Center—New Orleans (LSUHSC-NO) approved ABSL-3 IBC and IACUC protocols.

Mice were randomly assigned to one of the treatment groups: mock, ZM (ZM-241385, Tocris) or REM (Remdesivir, Tocris). Mice were intraperitoneally injected with 10 mg/kg ZM or REM, 2 h prior to infection and the treatment was continued daily for the duration of the experiment. Two-hours post-treatment, mice were anesthetized with ketamine-xylazine and infected with 10⁵ PFU’s of mouse-adapted SARS-CoV-2 (MA30) (Dinnon et al, 2020) intranasally in 40 uL DMEM. The virus was propagated from Stanley Perlman’s MA30 stock virus in TMPRSS2 expressing Vero cells. The TMPRSS2 Vero cells were grown in Dulbecco’s modified Eagle’s medium DMEM, supplemented with 10% FBS. The virus was sequenced after propagation and found to match the input strain. Non-infected mice received 0.2 or 2 mg/kg CGS (CGS-21680, Tocris) or, 10 or 20 mg/kg ZM or REM intraperitoneally daily (100 ul in PBS-DMSO) for the duration of the experiment.

Mice were monitored and scored for clinical disease and body weight daily. Numerical clinical scores were assigned based on disease manifestations and weight loss rate as shown in Fig. EV5B.

At the clinical or experimental endpoint (Day 4 or 7 post-infection), mice were anesthetized by isoflurane and necropsied. Blood was collected through cardiac puncture and placed in Sarstedt EDTA tubes and allowed to sit for 10 min. Mice were perfused transcardially with 1X PBS. Followed by lungs and spleen collected and placed in 10% neutral buffered formalin separately. All tissues were inactivated and removed from the Animal Biosafety Level 3 facility in accordance with the University of Pennsylvania IBC protocol.

Vero cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% FBS and 1% Pen Strep and plated in six-well tissue culture plates at 1-million cells per well. Cells were allowed to adhere overnight in a CO₂ incubator at 37 °C. The following day, the culture media was aspirated from each well and 2 mL of fresh DMEM + drug were added. After an hour of incubation, cells were infected with a multiplicity of infection (MOI) of 0.05 in DMEM + drug. Two hours later, the virus was removed and washed with 1X phosphate-buffered saline (PBS). Fresh DMEM + drug was added to mock or virus-exposed cells and incubated for 24 h. Twenty-four hours post viral exposure, the cellular layer within each well was collected in TRIzol and homogenized. The cellular homogenate was added to Trizol at a 1:3 ratio and extracted using a Zymo Direct-zol RNA kit following the manufacturer's protocol. RNA was reverse transcribed and amplified using iScript cDNA Synthesis Kit (Biorad). Gene-specific primers for SARS-CoV-2 N gene (F:TTACAAACATTGGCCGCAAA and R:GCGCGACATTCCGAAGAA) and human HPRT1 (F: CATTATGCTGAGGATTTGGAAAGG and R: CTTGAGCACACAGAGGGCTACA) with Power SYBR Green PCR Master Mix (Applied Biosystems) were used to amplify viral and cellular RNA by QuantStudio 3 Flex Real-Time PCR Systems (Applied Biosystems). The relative expression levels of target genes were calculated using HPRT1 RNA as an internal control.

CD8+ and CD4 + T-cells were aseptically isolated from the spleens and lymph nodes of C57BL/6 (6–12 weeks old) mice using the negative selection Easysep mouse CD8 + T-cell isolation kit or Easysep mouse CD4 + T-cell isolation kit (StemCell Technologies) according to the manufacturer’s instructions. CD8+ and CD4 + T-cells were cultured in RPMI supplemented with 10% fetal bovine serum, 4 mM l-Glutamine, 50 U/ml penicillin, 50 µg streptomycin, and 50 µM 2-mercaptoethanol. The cells were activated in plates coated with anti-mouse CD3ε and anti-mouse CD28 antibodies (1 µg/ml, BD Biosciences) for up to 72 h. Th1, Th2, and Th17 CD4 + T-cells were differentiated and cultured using the CellXVivo Mouse Th1 Cell differentiation Kit, CellXVivo Mouse Th2 Cell differentiation Kit, or the CellXVivo Mouse Th17 Cell differentiation Kit (Biotechne, R&D Systems) according to the manufacturer’s instructions. T-cell proliferation was measured by labeling primary CD8+ or CD4 + T-cells with 1 μM carboxyfluorescein diacetate succinimidyl ester (CFSE, Thermo Fisher) for 10 min at 37 °C before activation. T-cell proliferation was measured by flow cytometry (Beckman Coulter) as indicated by CFSE dilution, and data were analyzed using Kaluza software (Beckman Coulter). Neutrophils and macrophages were isolated from the peritoneal cavity of C57BL/6 (6–12 weeks old) mice at 24 and 96 h post-injection with 3% Brewer thioglycolate medium (Zhang et al, 2008). After euthanizing the mice, the collection of peritoneal cells was performed by injecting 3 ml of 1X PBS into the peritoneal cavity and carefully aspirating the fluid at least 5 times.

Collected mouse blood was transferred into the ACK lysis buffer and incubated at room temperature for 5 min. 1X PBS was added to stop lysis and then centrifuged at 300 × g for 5 min at 4 °C. This was completed twice. The supernatant was removed, and the pellet was resuspended in 5 mL of 4% paraformaldehyde (PFA). Mouse spleens were collected, minced by gently crushing the spleen using the flat end of the 3 cc syringe plunger, and then filtered through a 70 µm filter. The cell suspension was centrifuged at 400 × g for 5 min and the supernatant was removed. ACK lysing buffer (1 ml) was added and incubated at RT for 2 min. 1X PBS was added to stop lysis and centrifuged at 300 × g for 5 min at 4 °C, the supernatant was removed, and the pellet was resuspended in 1X PBS. Prior to staining, Fc block (1 µg/10⁶ cells) was added to samples and incubated for 5 min at RT. Samples were then incubated with antibodies for 30 min at 4 °C, washed with PBS, and centrifuged at 400 × g for 5 min. Samples were fixed with 4% PFA for 20 min at 4 °C. For intracellular staining, 250 µl Permeablization buffer (BD Transcription Factor Perm Set) was added and incubated for 20 min at 4 °C, washed with PBS, and centrifuged at 400 × g for 5 min. Samples were then Incubated with antibodies for 30 min at 4 °C. Samples were washed, resuspended in PBS and run on a flow cytometer. Antibody used in this study: anti-CD45 (BD 553079), anti-NICD (BD 552768), anti-IFN-γ (BD 562303), anti-NK1.1 (BD 566502), anti-CD4 (BD 563933), anti-CD69 (BD 560689), FVS780 (BD 565388), anti-CD3 (BD 562600), anti-CD8 (BD 563046), anti-Ly6G (BD 551461), anti-CD11b (BD 557686), anti-IL-10 (BD 554468), and anti-IL-17 (BD 559502). T-cell proliferation was measured by labeling primary CD8+ or CD4 + T-cells with 1 μM carboxyfluorescein diacetate succinimidyl ester (CFSE, Thermo Fisher) for 10 min at 37 °C before activation and treatments. Samples were run on a Gallios cytometer (Beckman Coulter) and data analyzed using Kaluza software (Beckman Coulter). A diagram of the gating strategy is provided in Appendix Fig. S3.

Mouse lungs were inflated and fixed in 10% buffered formalin for tissue processing. After paraffin embedding, tissues were sectioned at a thickness of 4 μm, placed on electromagnetically charged slides (Fisher Scientific; Waltham, MA), and stained with Hematoxylin & Eosin (H&E) for routine histologic analysis. Immunohistochemistry was performed using the Avidin-Biotin-Peroxidase complex system, according to the manufacturer’s instructions (Vectastain Elite ABC Peroxidase Kit; Vector Laboratories, Burlingame, CA). Our modified protocol includes meting the paraffin at 56 °C for 15 min, clearing in xylenes three times for 15 min each, rehydration through descending grades of alcohol up to water, non-enzymatic antigen retrieval with 0.01 M sodium citrate buffer pH 6.0 at 95 °C for 25 min in a vacuum oven, and endogenous peroxidase quenching with 3% H₂O₂ in methanol, Following these steps, slides were washed with PBS and blocked in PBS/0.1% BSA containing 5% normal horse serum (for mouse monoclonal antibodies), or normal goat serum (for rabbit generated antibodies) for 2 h at room temperature, then incubated overnight with primary antibodies. Primary antibodies included a rabbit polyclonal anti-nucleocapsid of SARS-CoV-2 (Novus Biologicals, NB100-56576, 1:200 dilution), a rabbit recombinant monoclonal anti-CD8a (Abcam, EPR21769, 1:2000 dilution), a mouse monoclonal anti-cyclic adenosine monophosphate (cAMP) (AbboMax, Clone ABM486, 1:200 dilution), a rabbit polyclonal anti-Ly-6G (Invitrogen, PA5-141170, 1:250 dilution) and a rabbit recombinant monoclonal anti-Neutrophil Elastase (Invitrogen, Clone 8X4E6, 1:100 dilution). After overnight incubation, slides were rinsed thoroughly with PBS and incubated with biotinylated secondary anti-rabbit or anti-mouse antibodies for 1 h at room temperature. Then, Avidin-Biotin complexes were incubated for 1 h at room temperature, and finally, slides were developed using a diaminobenzidine substrate (DAB, Roche Laboratories), counterstained with Hematoxylin, and mounted with Permount. Images were collected using a BX61 Olympus microscope equipped with a high-resolution Olympus DP80 digital camera and CellSense image capture software.

Total RNA was extracted from FFPE tissue samples using the AllPrep DNA/RNA FFPE kit (Qiagen). Directions from the manufacturer were followed. Tissue sections of 20 μm were first deparaffined using Sigma’s xylene substitute, followed by two ethanol washes and air dried for 10 min at 37 °C. After deparaffinization, tissues were lysed in a lysis buffer containing proteinase K for 15 min at 56 °C. RNA containing supernatant was then de-crosslinked at 80 °C for 15 min and purified using the RNA purification columns as indicated by the manufacturer’s instructions. A treatment with DNase I (Qiagen) was included. RNA was eluted in 20 μl of H₂O after 5 min of incubation at room temperature. RNA quantification was performed using the Qubit RNA HS Assay kit (Invitrogen) and RNA quality was assessed with the Agilent 2100 bioanalyzer (Agilent Technologies). Libraries were generated using Illumina’s TruSeq RNA exome library preparation kit and according to the manufacturer’s instructions, with the following modifications (Wen et al, 2017): 300 ng of input RNA was used, PCR cycles were reduced to nine cycles, and hybridization time was extended to 16 h. After library preparation, free adapters were blocked using Illumina’s free adapter-blocking reagent. Libraries were validated on Agilent’s 2100 bioanalyzer (Agilent Technologies) using a DNA 1000 kit and quantified using the Qubit dsDNA HS Assay kit (Invitrogen). Sequencing was performed on Illumina’s NextSeq 2000 instrument using a NextSeq 2000 P2 200 cycles kit with pair-end 76 bp reads.

FASTQ files from the NextSeq2000 run were downloaded from the Illumina BaseSpace hub and uploaded to Partek Flow. Contaminants (rDNA, tRNA, and mtrDNA) were removed using Bowtie v2.2.5. Sequence reads were then aligned to STAR v2.7.8a to the mm10 version of the mouse genome. The reads were quantified with RefSeq Transcripts 96. For differential analysis, the samples were split based on days of treatment (4 and 7 days), and filters were applied to remove features with a maximum of five reads across all samples. The remaining features (>15,000) were normalized with TMM and transformed by log2 (+0.0001/TMM/log2). Heatmaps, pathways maps and gene ontology (GO) were all obtained by the tools embedded in Partek Flow. The sequencing data have been deposited to the Gene Expression Omnibus (GEO) under the accession number GSE236563.

Prepared viral RNA was assayed using the Seegene Allplex^TM 2019-nCoV2 assay (RP10250X) per manufacturer’s instructions, run on a BioRad CFX96 Touch, and visualized in the Seegene Viewer software. A standard curve using a genomic RNA from SARS-CoV-2/USA-WA1/2020 (ATCC; VR-1986D) was used to quantitate the number of viral genomes (copy number) present in each sample. Viral clearance rates were measured by comparing viral copy numbers at different days post-infection.

Single docking experiments of adenosine receptors (A2A: Protein Data Bank ID 2YDO, A1: Protein Data Bank ID 6D9H, A2B: Protein Data Bank ID 8HDP) to adenosine (PubChem ID 60961) and GS-441524 (PubChem ID 44468216) were performed using AutoDock Vina (Trott and Olson, 2010). AutoDock Tools was used to prepare adenosine receptor 2A and ligands for docking and defining the binding site coordinates (Forli et al, 2016). Predicted structures were visualized by PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.) and BIOVIA Discovery Studio (BIOVIA, Dassault Systèmes, Discovery Studio, San Diego: Dassault Systèmes, 2021).

IFN-γ production was measured in supernatants of primary immune cell cultures using an IFN-γ mouse ELISA kit (Invitrogen). Cytokines and chemokines from macrophage and neutrophil culture supernatants were measured using the antibody-based multiplex Proteome Profiler Mouse Cytokine Array Kit (Biotechne, R&D Systems), according to the manufacturer’s instructions.

The production of cAMP was measured in primary mouse CD8 + T-cells or in an A2AR-expressing cell line (Abeomics, 14-522ACL, A2AR-expressing Cho cell line) using a cAMP competitive ELISA kit (Abcam) or a luciferase reporter assay (cAMP-Glow Assay, V1501, Promega) according to the manufacturer’s protocol.

Protein lysates were loaded on 4–15% SDS-PAGE gels (BioRad) and transferred to PVDF membranes (Millipore). Blots were incubated at 4^oC overnight with primary antibody diluted in Intercept blocking buffer (Licor). The next day, blots were incubated with IRDye fluorescent secondary antibodies (Licor) for fluorescence detection, for 1 h at RT. Proteins were visualized by imaging the blots on an Odyssey scanner (Licor). The following primary antibodies were used for Western blot in this study: anti-Notch1 (D1E11, Cell Signaling) for Notch1 full length and cleaved forms; anti-cleaved Notch1 Val1744 (D3B8, Cell Signaling or PA5-99448, Invitrogen-Thermo Fisher) for NICD; anti-β-actin (AC-15, SCBT).

Statistical significance was measured using a two-tailed unpaired Student’s t-test for comparisons between two groups and one-way ANOVA with Bonferroni correction for multiple comparisons. Non-linear regression was used for comparisons of mouse weight between two sample groups over time. P values ≤0.05 were considered significant. Limma-trend was used to identify genes differentially expressed between treatment (REM) vs. vehicle, for both treatment days. Genes with p value <0.05 and fold change (FC) >2.0 were considered significant.