Cell culture media, fetal bovine serum, supplements, and basic laboratory chemicals were obtained from Sigma (Suffolk, UK). Microtiter plates for enzyme-linked immunosorbent assay (ELISA) were obtained from Nunc (Roskilde, Denmark). Rabbit antibodies against galectin-9 (ab69630), phospho-S423/S425-Smad3 (ab52903), HIF-1α (ab51608), receptor for advanced glycation end products (RAGE) (ab3611), and granzyme B (ab134933) as well as mouse antibody against HIF-1α (ab1) were purchased from Abcam (Cambridge, UK). Mouse antibody against β-actin was purchased from Proteintech (66009-1-Ig, Manchester, UK). Goat anti-mouse and anti-rabbit fluorescently labeled dye secondary antibodies were obtained from LI-COR (Lincoln, Nebraska USA). ELISA-based assay kits for the detection of TGF-β, galectin-9, PD-L1, and Interleukin-2 (IL-2) and were purchased from Bio-Techne (R&D Systems, Abingdon, UK). All other chemicals used in this study were of the highest grade of purity and commercially available.

Cell lines used in this work were purchased from either European Collection of Cell Cultures or American Type Culture Collection (ATCC). Cell lines were accompanied by authentication test certificates. BEAS-2B normal bronchial epithelium cells and LN18 human glioblastoma cells were cultured using Dulbecco's Modified Eagle Medium (DMEM) medium supplemented with 10% fetal bovine serum, penicillin (50 IU/mL), and streptomycin sulphate (50 μg/mL). Jurkat T cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, penicillin (50 IU/mL), and streptomycin sulfate (50 μg/mL).

TALL-104 CD8-positive cytotoxic T lymphocytes, obtained from human acute lymphoblastic leukemia (TALL), were cultured in American type culture collection (ATCC)-formulated Iscove’s modified Dulbecco’s medium. To make the complete growth medium, we added recombinant human IL-2 (100 units/mL), human albumin (2.5 μg/mL), D-mannitol (0.5 μg/mL), and fetal bovine serum to a final concentration of 20% as well as penicillin (50 IU/mL) and streptomycin sulfate (50 μg/mL).

A total of five nasopharyngeal swabs collected from individuals in May to June 2020 (non-hospitalized patients) infected with SARS-CoV-2 were included into this study. Laboratory confirmation of SARS-CoV-2 was performed at the North Kent Pathology Service (Darent Valley Hospital, Dartford and Gravesham NHS Trust) using reverse transcriptase–polymerase chain reaction (RT-PCR). Specimens were kept at −80°C to ensure reliable results when used in experiments. The experimental work was performed following informed written consent of the participants under the ethical approval 21/HRA/2577 provided by the NHS Health Research Authority.

The following plasmids were used in the study: 1) the human codon optimized Wuhan-Hu-1 isolate (GenBank, MN908947.3) ancestral Spike (S) subcloned into the pCAGGS expression vector, and 2) pCAGGS expression vector without an insert was employed as a control plasmid.

Transfections were performed using FuGENE® HD transfection reagent (Promega) according to the manufacturer’s protocol.

Levels of phospho-S423/S425 Smad-3, galectin-9, and granzyme B were analyzed by Western blot and compared to the amounts of β-actin (protein loading control), as previously described (12). LI-COR goat secondary antibodies conjugated with infrared fluorescent dyes, were used as described in the manufacturer’s protocol for visualization of specific target proteins (LI-COR Odyssey imaging system was employed). Western blot data were quantitatively analyzed using Image Studio software, and values were subsequently normalized against those of β-actin named as “actin” in the figures.

To detect PD-L1, CD3ϵ, interferon-γ (IFN-γ), HIF-1α, and vascular endothelial growth factor (VEGF) mRNA levels, we used quantitative real-time PCR (qRT-PCR) (6, 12). Total RNA was isolated using a GenElute™ mammalian total RNA preparation kit (Sigma-Aldrich) according to the manufacturer’s instructions, followed by RT-PCR of a target protein mRNA (also performed according to the manufacturer’s protocol). This was followed by qRT-PCR. The following primers were used:

PD-L1: (forward) 5′-AAATGGAACCTGGCGAAAGC-3′ and (backward) 5′-GATGAGCCCCTCAGGCATTT-3′; CD3ϵ: (forward) 5′-TCCCAACCCAGACTATGAGC-3′ and (backward) 5′-CAAGACTAGCCCAGGAAACAG-3′; IFN-γ: (forward) 5′- TCCCATGGGTTGTGTGTTTA-3′ and (backward) 5′-AAGCACCAGGCATGAAATCT-3′; HIF-1α: (forward) 5′-CTCAAAGTCGGACAGCCTCA-3′ and (backward) 5′-CCCTGCAGTAGGTTTCTGCT-3′; VEGF: (forward) 5′-GTATAAGTCCTGGAGCGT-3′ and (backward) 5′-CTCGGAGGGAGTCCCAAA-3′; and β-actin: (forward) 5′-TGACGGGGTCACCCACACTGTGCCCATCTA-3′ and (backward) 5′-CTAGAAGCATTTGCGGTCGACGATGGAGGG-3′. Reactions were performed using a LightCycler^® 480 qRT-PCR machine and SYBR Green I Master kit (Roche, Burgess Hill, UK). The assay was performed according to the manufacturer’s protocol. Values of investigated protein mRNA levels were normalized against those of β-actin.

Chromatin immunoprecipitation (ChIP) was performed as described recently (12, 13). In brief, 5 × 10⁶ cells were subjected to immunoprecipitation. Cross-linking was then performed using 1.42% formaldehyde, followed by quenching for 5 min with 125 mM glycine. Cells were then washed twice with Phosphate-buffered saline (PBS) and subjected to ChIP according to ChIP-IT high-sensitivity kit (Active Motif) protocol. Immunoprecipitation was performed using anti–HIF-1α antibody. IgG isotype control antibody was used for a negative control IP. The epitopes recognized by these antibodies do not overlap with DNA and co-activator binding sites of these proteins. Immunoprecipitated DNA was then purified and subjected to qRT-PCR which was performed as outlined above. The following primers were designed using NCBI Primer-Blast primer designing tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/): TGF-β: (forward) 5′-ATCGCCATCATCATCATCCACTGAGC-3′ and (backward) 5′-TTGCAGTCCATGGCATAGGG-3′. These primers allow amplification of the fragments of promoter regions of corresponding genes, which surround HIF-1-binding sites [so-called hypoxia response elements (HREs)].

Levels of secreted/soluble galectin-9, TGF-β, and IL-2 were measured in nasal secretion from the back of the nose and throat (nasopharyngeal swab) or cell culture conditioned media, and PD-L1 quantities were also analyzed in cell lysates by ELISA using R&D Systems kits (DY2045 for galectin-9, DY240 for TGF-β, DY202 for IL-2, and DY156 for PD-L1) according to the manufacturer’s protocols.

Cell surface levels of RAGE protein were measured using on-cell Western analysis performed using a LI-COR Odyssey imager, and the assay was performed in line with the manufacturer’s recommendations as previously described (14).

HIF-1α prolyl hydroxylase (PHD) activity was analyzed as described before using the method based on ability of HIF-1α PHD to hydroxylate specific peptide (15). HIF-1 DNA-binding activity was measured by the method, which was described recently (7). Briefly, 96-well Maxisorp™ microtiter plates were coated with streptavidin and blocked with Bovine serum albumin (BSA). A volume of 2 pmol/well biotinylated 2HRE (HRE) containing oligonucleotide was immobilized by 1 h of incubation at room temperature. The plates were then washed with TBST buffer [10 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 0.05% Tween-20), followed by 1 h of incubation with cell lysate at room temperature. The plate was again washed with Tris-buffered saline with Tween 20 (TBST) buffer and mouse anti–HIF-1α antibody (1:1,000 in TBS with 2% BSA) was added. After 1 h of incubation at room temperature, the plate was washed with TBST buffer and then incubated with LI-COR goat anti-mouse secondary antibody labeled with infrared fluorescent dye. After extensive washing with TBST, the plate was scanned using a LI-COR Odyssey fluoroimager.

Lactate dehydrogenase A (LDHA) activity was analysed using Sigma-Aldrich assay kit and reagents (with modification - 2-OG was employed as a substrate) according to the standard protocol based on oxidation of NADH+H+. FX11 LDHA inhibitor was used to confirm specificity of the reaction. 2-Hydroxy-glutarate (2-OH-G, D form) was measured using Abcam colorimetric assay kit (ab211070) according to the manufacturer’s protocol.

Granzyme B activity in cell lysates was measured using a fluorometric assay based on the ability of this enzyme to cleave the fluorogenic substrate Ac-IEPD-AFC (Sigma-Aldrich) as described before (16).

L-kynurenine (LKU) levels were analyzed on the basis of ability of this amino acid to react with 4-dimethylamino)-benzaldehyde (17). Briefly, we took 160 µL of cell culture conditioned medium or nasal secretion from the back of the nose and throat (nasopharyngeal swab) and added 10 µL of 30% (v/v) trichloroacetic acid to each sample. Samples were incubated for 30 min at 50°C in order to hydrolyze N-formylkynurenine to L-kynurenine (LKU). The samples were then centrifuged at 3,000 g for 10 min. Supernatants (100 µL) were transferred to wells of a 96-well flat-bottom plate and mixed with 100 µL of freshly prepared Ehrlich’s reagent [1.2% w/v 4-(dimethylamino)-benzaldehyde in glacial acetic acid] followed for 10 min of incubation at room temperature. Absorbance was measured using a microplate reader at 492 nm.

IDO-1 activity (or enzymatic conversion of L-Trp into N-formylkynurenine), which was then further converted into LKU, was measured using a previously described method (17) with minor modifications. Briefly, the cell or tissue lysate was added to the reaction mixture containing 50 mM potassium phosphate buffer (pH 6.5), 20 mM ascorbate, catalase (100μg/mL), and 2 mM L-Trp. The reaction was carried out at 37°C for 60 min and terminated by adding 10 μL of 30% (v/v) trichloroacetic acid to 160 μL of sample. Further operations were performed as previously described.

Glycolytic degradation of glucose was analyzed using a colorimetric assay as described previously (17, 18). Briefly, the assay was performed using cell lysates and the conversion of glucose into lactate in the absence of oxygen. Cell lysates were incubated for 1 h at 37°C with 1% glucose solution in an anaerobic chamber. Two percent trichloroacetic acid solution was then used to precipitate proteins. This was followed by carbohydrate precipitation using saturated CuSO₄ solution in combination with Ca(OH)₂ (at a final concentration of 60 mg/mL). Lactate was then converted into acetaldehyde using concentrated H₂SO₄ at 90°C for 1 min and cooled on ice. Acetaldehyde was detected using the veratrole (1,2-dimethoxybenzene) test. Methyl-glyoxal (MGO) was also detected colorimetrically in the cell culture medium following biochemical modifications. In brief, MGO was condensed with reduced Glutathione (GSH) (1 mM) for 10 min at 37°C. The complex was then converted into lactate by glyoxalases I and II at pH 8.0 (glyoxalase I converted the complex into D-lactoyl-glutathione, which was then converted into lactate by glyoxalase II). Lactate was then detected colorimetrically, as described above.

Activity of DiFMUP (6,8-difluoro-4-methylumbelliferyl phosphate)–specific phosphatases (here, we call them “SHP-2–like phosphatases”) was analyzed using a flurometric assay kit (E12020, Thermo Fisher Scientific) according to the manufacturer’s instructions.

The percentage of apoptotic/necrotic [propidium iodide (PI)–positive)] cells was measured using an Annexin V/PI assay kit (APOAF-20TST, Sigma-Merck). We measured Annexin V and PI-positive cells using flow cytometry according to the manufacturer’s protocol. Quadrant gates were used to distinguish the populations by fluorescence: Annexin V+ PI−, Annexin V+ PI+, and Annexin V− PI+ populations from the double-negative population (Annexin V− PI−). The gate was set according to the unstained control. We also measured Annexin V and PI-positive cells by fluorometric assay using 96-well plates, where the cells could stay adherent in order to rule out false-positive and false-negative results. The results were calculated as % of cells stimulated with 1mM H₂O₂ for 6 h. Cell proliferative activity was assessed using an MTS assay kit (Promega, G3580) according to the manufacturer’s protocol.

Each experiment was performed at least four times, and statistical analysis was conducted using a two-tailed Student's t-test or the Kruskal-Wallis test. Statistical probabilities (p) were expressed as * when p < 0.05, ** when p < 0.01, and *** when p < 0.001.

First of all, we investigated nasopharyngeal swabs (taken in the active phase of disease at diagnosis and before treatment) obtained from five non-hospitalized individuals affected by Wuhan strain of SARS-CoV-2 virus versus five healthy individuals. We found that, in swabs of COVID-19 individuals, the levels of galectin-9, TGF-β, and LKU were highly upregulated compared to those of healthy individuals ( Figures 1A–C ). We also observed significant upregulation of PD-L1 mRNA levels in COVID-19 individuals ( Figure 1D ) as well as significant upregulation of ϵCD3 mRNA levels ( Figure 1E ). This means increased levels of infiltrated CD3-positive T lymphocytes, because T cells express and use this protein, and, respectively, a significant upregulation of IFN-γ mRNA levels was observed. Interestingly, mRNA levels of both HIF-1α and VEGF (encoded by HIF-1 downstream gene and, thus, used to analyze HIF-1 activity) were significantly increased in the swabs of COVID-19 individuals (Figures 1G, H). This means that HIF-1 transcription complex was more active in COVID-19 individuals because it is known to specifically upregulate VEGF gene transcription (19). However, we were keen to understand the cause of this effect because the tissues subjected to taking the swabs cannot have highly decreased oxygen availability. On the other hand, the S protein of SARS-CoV-2 was reported to upregulate aerobic glycolysis leading to translocation of all the pyruvate into mitochondria (20, 21). This can leave existing lactate dehydrogenase (LDH) A enzyme molecules “unemployed.” In addition, in this case, LDH A can in theory convert cytosolic 2-OG into L-2-OH-G (22). The latter cannot be used as a cofactor in HIF-1α PHD reaction, leading to attenuation of this reaction and stabilization/accumulation of HIF-1α protein.

To investigate this hypotheis of 2-OG conversion/removal, we measured D-2-OH-G in swabs of both SARS-CoV-2–infected and healthy individuals. We found that 2-OH-G was not detectable at all in the swabs of all healthy individuals and was clearly detectable in swabs of COVID-19 individuals ( Figure 1I ). As such, we hypothesized that this process triggers activation of HIF-1, which was shown to upregulate TGF-β production (7), and TGF-β is known to induce expression of galectin-9, PD-L1, and IDO1 (converts amino acid L-tryptophan into formyl-kynurenine, which is then turned into LKU) via the Smad3 pathway (6, 17). Thus, galectin-9, PD-L1, and LKU are suppressing T-cell function (6, 17).

To verify the finding and hypotheses described above in the in vitro cell culture model, we used human bronchial epithelium BEAS-2B cells. Cells were transfected with Wuhan SARS-CoV-2 S protein expression plasmid (S-SC2W) vs. empty vector (EV; control). After transfection cells were cultured for 40 h followed by investigation of the effects caused by S protein presence in the cells. We found that transfection of EV or S-SC2W containing vector reduced cell proliferative activity of BEAS-2B cells but did not affect their viability ( Supplementary Figures 1A–C ). We also found that 2-OH-G was not detectable in wild-type BEAS-2B cells as well as in the cells transfected with EV. But in the cells transfected with S-SC2W, it was clearly detectable ( Figure 2A ). At the same time, we were measuring LDHA activity using 2-OG as a substrate. It was obviously higher in S protein-containing BEAS-2B cells. At the same time, when LDHA inhibitor was added to the system containing S-SC2W–transfected cell lysate as enzyme preparation, conversion of 2-OG was significantly downregulated confirming the ability of LDHA to act on 2-OG ( Figure 2B ). This led to attenuation of HIF-1α PHD activity as well as high upregulation of HIF-1 DNA-binding activity and TGF-β production by BEAS-2B cells ( Figure 2C ). As a result, glycolysis was significantly upregulated in S-containing BEAS-2B cells compared to that in controls. As a side effect of glycolysis upregulation, cells produced increased amounts of MGO, which is known to trigger generation of advanced glycation end products (AGE; act through RAGE, receptor of AGE, present in all the studied cells), which supports the activity of mitogen-activated protein kinase cascades (14, 18), thus supporting expression of viral proteins/capsids (if we are dealing with functional virus). These results are shown in Supplementary Figure 2 . Phospho-Smad3 was highly upregulated, which was in line with increased expression of galectin-9 and PD-L1 in S-containing BEAS-2B cells ( Figure 2D ). IDO1 activity and respectively LKU release were highly increased in S-SC2W–containing BEAS-2B cells ( Figure 2E ). We also verified the role of HIF-1 in TGF-β expression by performing ChIP using anti–HIF-1α antibody followed by RT-PCR in resting BEAS-2B cells and those exposed for 4 h to 50 µM CoCl₂ (to induce HIF-1 activity). We confirmed the direct interaction of HIF-1 with TGF-β gene promoter region ( Figure 2F ). As such, we can confirm that Wuhan SARS-CoV-2 S protein upregulates glycolysis, and we uncovered the mechanism underlying this process. This mechanism also leads to activation of immune evasion machinery similar to the one operated by cancer cells. Schemes outlining the details of these events are shown in Figures 3A, B .

We asked whether BEAS-2B cells transfected with SARS-CoV-2 S protein (Wuhan strain) display stronger immune evasion activity compared to wild-type BEAS-2B cells. We co-cultured wild-type BEAS-2B cells and those transfected with EV or S-SC2W with Jurkat T cells (as T helpers) at the ratio 1:1 for 16 h. BEAS-2B cells are expressing MHC II type proteins and, thus, can engage in interactions with CD4-positive T helpers (23). We found that, in the co-culture containing S-SC2W–transfected BEAS-2B cells, levels of released galectin-9 were significantly higher and levels of IL-2 were significantly lower compared to those in other cell types ( Figures 4A, B ). At the same time, in Jurkat T cells co-cultured with BEAS-2B containing S protein, activities of SHP-2–like phosphatases were clearly upregulated, altogether meaning activation of PD-1-type signaling events (24). BEAS-2B cells (wild type and those transfected with S-SC2W) were co-cultured with LN-18 high-grade glioblastoma cells in order to model co-habitation of cancer and SARS-CoV-2–affected cells. Then, TALL-104 cytotoxic T cells were added in the approximate ratio 1:1 ( Figure 4C ) for 24 h. We found that, in the case where we had co-habitation of LN-18 and S-SC2W–transfected BEAS-2B cells, injection of granzyme B into adherent cells was completely attenuated compared to other cases ( Figures 4C, D ). At the same time, in this co-culture, there was the highest level of T cell killing and the highest level of galectin-9 secretion. These results suggest that S-SC2W–transfected cells support cancer cells in their fight with cytotoxic T cells in vitro. To further verify the immunosuppressive activity of BEAS-2B cells transfected with S-SC2W, we performed the same experiment (the outline is shown in Supplementary Figure 3A ) but co-culturing just BEAS-2B cells (wild type, BEAS-2B transfected with EV, or BEAS-2B transfected with S-SC2W) with TALL-104 cytotoxic T cells as described above. BEAS-2B transfected with S-SC2W showed similar immune evasion potential as reported in Figure 4 (see Supplementary Figure 3B for details). In addition, it was much higher compared to wild-type BEAS-2B cells and those transfected with EV ( Supplementary Figure 3B ).