All materials were purchased from Sigma-Aldrich, unless otherwise specified.

Six- to ten-week-old C57BL/6 mice from The Jackson Laboratory were used. Experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee of Northwestern University and according to National Institutes of Health guidelines for the use of rodents.

As previously described (27), mice were housed in a BioSpherix A environmental chamber (BioSpherix) with ProOx C21 O₂ and CO₂ controllers (BioSpherix), that maintain an environment of 10% CO₂/21% O₂/69% N₂ (normoxic hypercapnia). Simultaneously, age-matched mice were maintained in ambient air and used as controls. Betulin (5 mg/kg body weight) was administrated 3 days before and on the day that hypercapnia exposure began.

Primary human bronchial epithelial (HBE) cells (LONZA) were seeded as passage 1 (P1) into 100 mm dishes on PneumaCult-Ex Plus media (StemCell) and incubated at 37°C, 5% CO₂. Once cells reached 70–80% confluency, they were dissociated using Animal Component-Free Cell Dissociation Kit (StemCell) and seeded on 12 mm Transwells (Corning) coated with 0.3 mg/mL Collagen type IV from human placenta (Sigma-Aldrich). Upon reaching confluency, the apical medium was removed to obtain an air liquid interface (ALI), and the basal medium replaced with PneumaCult-ALI medium (StemCell). Medium was changed every second day and apical surfaces washed with PBS (Ca²⁺, Mg²⁺) twice per week. Cultures were used for experiments after reaching full differentiation (∼ 3-4 wk on air) as assessed by visual confirmation of beating cilia and mucus (28). BEAS-2B cells, a SV-40-transformed human bronchial epithelial cell line (ATCC CRL-9609), were maintained in DMEM:F12, 5% FBS medium while VERO (ATCC, CRL-1586) were grown on EMEM, 10% FBS. All media contained penicillin (100 U/ml) and streptomycin (100 μg/ml).

Cells were exposed to normocapnia or normoxic hypercapnia. Normocapnia is defined as humidified 5% CO₂ (PCO₂ 36 mmHg)/95% air, at 37°C (standard incubator atmosphere). Normoxic hypercapnia entailed 15% CO₂ (PCO₂ 108 mmHg)/21% O₂/64% N₂ in an environmental chamber (C-174, Biospherix) contained within the same incubator where control cultures were simultaneously exposed to normocapnia. In selected experiments, pH at 7.4 in 15% CO₂ or 7.2 in 5% CO₂ was maintained adding Tris-HCl and Tris base to the media and measured with a pHOx Plus blood gas analyzer (Nova Biomedical). Before addition to cultures, media were pre-saturated with 5% or 15% CO₂. Cells were also exposed to cigarette smoke extract (CSE, 1 μg/ml, Murty Pharmaceuticals) during culture in normocapnia or hypercapnia.

Cells were pre-exposed to normocapnia or hypercapnia in the presence or the absence of CSE for 2 days, pseudo-SARS-CoV-2 virus (p-SARS-CoV-2, Montana Molecular, 1 x 10⁸ PV/ml) was added, and cells were cultured for an additional day in normocapnia or hypercapnia, respectively. The p-SARS-CoV-2 virus is a baculovirus pseudo-typed with SARS-CoV-2 spike protein that does not replicate but delivers a genetically encoded green-fluorescent reporter to infected host cells and is safe to use with no risk of infection to laboratory personnel. Nuclei were labeled with DAPI. Cells were imaged using fluorescence microscopy (3 fields/condition from at least 4 independent experiments) and the percentage of pseudo-SARS-CoV-2 positive cells was assessed using CellProfiler™ software (29).

The presence of indicated proteins in cell homogenates were assessed by immunoblotting, as before (16), using the following antibodies: anti-ACE2 (GeneTex, SN0754) and anti-βactin (Abcam, ab8226) followed by HRP-conjugated (1:5000, CellSignal) or IRDye (1:10,000, LI-COR) secondary antibodies, respectively. Signals were captured with a LI-COR Odyssey Fc Imager and analyzed by densitometry using ImageStudio™ software (LI-COR). Images are representative of 3 or more independent experiments.

After euthanasia, mouse lungs were perfused with HBSS via the right ventricle and a 20-gauge angiocath was sutured into the trachea. Lungs and heart were removed en bloc, and lungs were inflated at a pressure of 16 cm H₂O with 0.8 ml of formalin. Tissue was embedded in paraffin and 5-μm sections were then deparaffinized with xylene, rehydrated by using graded ethanol, and subjected to antigen retrieval using sodium citrate buffer (10 mM, pH 6.0), as before (16). Tissues were labeled with anti-ACE2 (GeneTex, SN0754), anti ABCA1 (Thermo, MA5-16026) or anti-SREBP2 (Novus, NBP1-54446SS) antibodies followed by Alexa-conjugated secondary antibodies (1 μg/ml). Cells were fixed with 4% PFA for 15 min or methanol (-20°C for 5 min), permeabilized with 0.1% Triton X100 for 5 min, and labeled with anti-ACE2 (Abcam, ab15348 or GeneTex, SN0754) or anti-SREBP2 (Novus, NBP1-54446SS) antibodies followed by Alexa-conjugated secondary antibodies. Cells were co-labeled with acetylated tubulin (cilliated cell marker) using anti-acetylated-tubulin (T7451) antibody, followed by Alexa-conjugated antibody (1 μg/ml). In all cases, DAPI was used to visualize nuclei, and Gel/Mount (Biomeda) was used to mount the slides. Of note, mouse or rabbit IgGs were used as a nonimmune staining control that was negative for all protocols. Images were obtained using the same exposure time for all samples from a given experimental set using Axiovert 200M Fluorescence Microscope (Zeiss). Exposure time was selected based on the brightest stained sample to avoid saturation, and used for all other samples in the set, resulting in equal subtraction of background autofluorescence from all sets. Tissue images are representative of at least 3 independent experiments. Cells were also imaged using fluorescence microscopy (3 fields/condition from at least 4 independent experiments), and fluorescence intensity was quantified using NIH ImageJ software. Nuclei/field were assessed with CellProfiler cell image analysis software. ACE2 fluorescence intensity was expressed in arbitrary units (AU)/cell.

We included genes obtained by Gene Ontology (GO) analysis of biological processes for cholesterol efflux and biosynthesis from our previous published study (18) investigating the transcriptomic response to hypercapnia in human bronchial epithelial cells (GSE110362). We determined transcriptional changes induced by elevated CO₂ in HBE cells differentiated at ALI were exposed to hypercapnia for 24 h or maintained in normocapnia as a control using Affymetrix GeneChip Hybridization. Relative gene expression was represented as a heat map.

Cells were lysed using RIPA buffer and cholesterol was measured using Amplex red assay following manufacturer’s protocol. Cholesterol concentration was expressed as µg/mg protein. In addition, cells were incubated with filipin III (1 µg/ml), cholera toxin B, or LipidSpot 488 (Biotium) at 37°C for 30 min, to stain total cholesterol, ganglioside GM1 in lipid rafts, or lipid droplets (LD), respectively. Nuclei were stained with DAPI, then imaged by fluorescence microscopy.

RNeasy Mini Kit (Qiagen) was used to extract RNA that was reverse transcribed with an iScript cDNA Synthesis Kit (Bio-Rad), as before (16). PrimeTime™ Predesigned qPCR Assays FAM-labeled primer/probes sets Hs.PT.58.27645939 for ACE2 and Hs.PT.58v.18759587 for the housekeeping gene B2M were used for amplification using the CFX Connect Real-Time System (Bio-Rad). ACE2 gene expression was normalized to B2M. Comparative CT method (ΔΔCT) was used to assess mRNA relative expression, as before (16).

Data were analyzed using Prism 9.0 (GraphPad). Student’s t test and Levene’s test were used to determine differences between two groups and homogeneity of variances, respectively. ANOVA followed by Sidak’s multiple comparisons test was used to analyze differences between multiple groups. In all analyses, p < 0.05 was considered significant.

To investigate the effects of hypercapnia on ACE2 protein expression in vivo, mice were exposed to 10% CO₂/21% O₂/69% N₂ (normoxic hypercapnia) for 7, 14 and 21 days, or ambient air as control. We and our colleagues have previously shown that exposure of mice 10% CO₂ for up to 21 days causes no obvious signs of distress or overt changes in the animals’ appearance, and that they gain weight normally in comparison with air-exposed control mice (16, 27, 30, 31). In addition, we showed that exposure to normoxic hypercapnia for up to 7 days caused no histopathologic lung injury (16). In the current study, hypercapnia increased ACE2 expression in the bronchi of mice exposed to 10% CO₂ for 7, 14 and 21 days ( Figure 1A and Supplementary Figure 1 ). SARS-CoV-2 does not replicate in wild-type mice, due to poor binding to mouse ACE2 (32). Therefore, to study the impact of hypercapnia on SARS-CoV-2 infection, and the mechanisms involved, we used polarized HBE cells differentiated by culture at ALI, BEAS-2B cells and VERO cells, all of which support the replication of SARS-CoV-2 (33–35). We found that culture under hypercapnic conditions (15% CO₂, PCO₂ 108) for 2 to 4 days, as compared to culture in normocapnia (5% CO₂, PCO₂ 36), increased ACE2 protein expression in HBE ( Figure 1B ), BEAS-2B ( Figure 1C ) and VERO cells ( Figures 1D, E ). Of note, ACE2 mRNA expression was increased after 1 day of culture in hypercapnia in BEAS-2B cells (data not shown).

To determine the effect of hypercapnia on SARS-CoV-2 pseudovirus infection, epithelial cells were cultured in normocapnia or hypercapnia for 2 days, incubated with p-SARS-CoV-2 in normocapnia or hypercapnia for an additional day, then fixed and assessed for GFP expression, i.e. PV positivity, by fluorescence microscopy. We found that culture in elevated CO₂ increased the percentage of PV positive cells in HBE ( Figure 2A ), BEAS-2B ( Figures 2B, C ), and VERO ( Figures 2D, E ) cell cultures, indicating that hypercapnia increases p-SARS-CoV-2 entry into epithelial cells. Culture of BEAS-2B and VERO cells in hypercapnia after infection with p-SARS-CoV-2 under normocapnic conditions did not increase the percentage of GFP positive cells ( Supplementary Figure 2 ), excluding the possibility that hypercapnia increases the apparent percentage of PV positive cells by increasing GFP transcription, translation or fluorescence intensity after viral uptake. The increase in viral entry occurred in the same time frame as the hypercapnia-induced increase in ACE2 protein expression in these cells ( Figures 1A–E ). Of note, when epithelial cells cultured in hypercapnia for 2 days were returned to culture in normocapnia for an additional day, ACE2 protein expression ( Supplementary Figure 3A ) and p-SARS-CoV-2 entry ( Supplementary Figures 3B, C ) were reduced in comparison with cells continuously cultured in elevated CO₂. Thus, the increases in ACE2 expression and viral entry caused by hypercapnia are reversible.

In a previous transcriptional profiling study (18) we found that culture in hypercapnia increased expression of cholesterol biosynthesis genes including HMGCS1 and decreased expression of ABC transporters which promote cholesterol efflux ( Figures 3A ). These results suggested that culture in hypercapnia might increase cellular cholesterol by upregulating the pathway leading to its synthesis ( Figure 3B ) and decreasing export of cholesterol from the cell. Interestingly, high cellular cholesterol content has been shown to increase SARS-CoV-2 infection (20, 22, 36). Here we found that exposure of mice to normoxic hypercapnia for 7 d increased protein expression of HMGCS1 ( Figure 3C ) and decreased expression of ABCA1 protein expression in bronchial epithelium ( Figure 3D ). Simultaneously, hypercapnia increased protein expression of SREBP2, the master transcription factor that upregulates transcription of cholesterol biosynthesis genes and downregulates transcription of ABC transporters (19), both in mouse bronchial epithelium ( Figure 3E ) and in cultured BEAS-2B cells ( Figures 3F, G ). Furthermore, hypercapnia activated SREBP2, indicated by the increase in expression of its cleaved isoform (a) ( Figure 3G ), which is released from the endoplasmic reticulum and translocated to the nucleus. In sum, these effects of hypercapnia would be expected to increase cholesterol accumulation in epithelial cells. Indeed, culture in hypercapnia increased overall cellular cholesterol in BEAS-2B cells ( Figure 3H ) and cholesterol localization specifically in lipid rafts in BEAS-2B and VERO cells ( Figures 3I, J ).

Since increasing CO₂ exposure lowers the pH of the culture media (37), we assessed whether the increases in ACE2 expression, p-SARS-CoV-2 entry, and cellular cholesterol accumulation could be due to extracellular acidosis, as opposed to the increase in CO₂, per se. In these studies, we buffered culture media with Tris-HCl and Tris base to reach pH 7.4 with 15% CO₂, and pH 7.2 with 5% CO₂ (non–Tris-buffered media in 5 and 15% CO₂, have pH 7.2 and pH 7.4, respectively). When epithelial cells were cultured in these media, we observed greater ACE2 protein expression ( Supplementary Figures 4A, B ) and a higher percentage of cells staining positively for p-SARS-CoV-2 ( Supplementary Figures 4C, D ) in hypercapnia (15% CO₂) than in normocapnia (5% CO₂), regardless of whether the pH was 7.2 or 7.4. Likewise, hypercapnia increased cholesterol accumulation ( Supplementary Figures 4E–G ) at both pH 7.2 and pH 7.4. Thus, the effects of hypercapnia on ACE2 expression, p-SARS-CoV-2 entry and cholesterol accumulation in epithelial cells are not attributable to extracellular acidosis, but instead result from the higher PCO₂, independent of pH.

Hypercapnia increases ACE2 expression ( Figures 1A–E ), p-SARS-CoV-2 internalization ( Figures 2A–E ) and cellular cholesterol ( Figures 3H–J ) in epithelial cells. Thus, we investigated the impact of blocking cholesterol synthesis on ACE2 protein and virus entry. Betulin, a plant-derived pentacyclic triterpene (38) that inhibits activation of SREBP2, thereby blocking its ability to regulate cholesterol synthesis and efflux genes (39), prevented the hypercapnia-induced increase in ACE2 expression in mouse bronchial epithelium in vivo ( Figure 4A ) and VERO cells in vitro ( Figure 4B ). AM580, a synthetic retinoid derivative that inhibits SREBP1/2 binding to promoter/enhancer regions of multiple lipogenic genes (40), similarly blocked the effect of hypercapnia in VERO cells ( Figure 4B ). In addition, both betulin and AM580 blocked p-SARS-CoV-2 entry in BEAS-B ( Figure 4C ) and VERO ( Figure 4D ) cells. Betulin and AM580 are not approved for use in humans, while statins are widely used lipid-lowering agents in the clinic. Statins inhibit 3-hydroxy-3-methyl glutaryl (HMG)-CoA reductase (HMGCR), a rate-limiting enzyme that catalyzes the conversion of HMG-CoA into L-mevalonate ( Figure 3B ), thereby inhibiting cholesterol biosynthesis (41, 42). Given their safety record, efficacy and affordability, statins are attractive candidates for host-directed therapy against infectious diseases (43). Moreover, several studies suggest a significant protective effect of statins on patients with COVID-19 (23–26). Therefore, we tested fluvastatin and rosuvastatin, and found that both statins block ACE2 expression in VERO cells ( Figure 4E ) and p-SARS-CoV-2 entry in BEAS-2B and VERO cells ( Figures 4F, G ). Of note, none of the inhibitors used in the present work reduced p-SARS-CoV-2 entry under normocapnic culture conditions ( Supplementary Figure 5 ), suggesting that this effect occurs only when cholesterol accumulation is increased and not in the basal state. Fluvastatin and rosuvastatin also inhibited the hypercapnia-induced increase in cholesterol in lipid rafts in VERO ( Figure 4H ) and in BEAS-2B ( Supplementary Figure 6A ) cells. Of note, uptake of statins into cells is mediated by organic anion transporting polypeptide (OATP)1B1 (encoded by SLCO1B1), while efflux of statins from cells is mediated by the breast cancer resistance protein (BCRP, encoded by ABCG2) (44). In previously published work, we confirmed that SLCO1B1 and ABCG2 are transcribed in HBE cells (18). We show here that culture in hypercapnia did not alter mRNA expression of either of these statin transport genes ( Supplementary Figure 7 ), excluding the possibility that statins selectively block hypercapnia-induced increases in cholesterol and ACE2 expression by altering drug transport into or out of epithelial cells. Cholesterol depletion from lipid rafts using methyl-β-cyclodextrin (MβCD) (45) decreases the interaction between the coronavirus spike protein and the ACE2 receptor (20, 46). Thus, we assessed the impact of MβCD on the effect of elevated CO₂ in VERO and BEAS-2B cells, and found that it blocked the hypercapnia-induced increases in ACE2 expression ( Figure 4I ) and p-SARS-CoV-2 entry ( Figures 4J, K ). These results indicate that blocking cholesterol synthesis and increasing its removal from the cell both prevent the increases in ACE2 protein expression and viral entry caused by elevated CO₂.

Active tobacco smoking is associated with increased mortality in patients with chronic hypercapnia (47) and COVID-19 (48). In addition, exposure to cigarette smoke increases expression of ACE2 in the respiratory tract (49), promotes cholesterol accumulation and decreases expression of ABC transporters in human bronchial epithelial cells (50). Since patients with advanced lung disease and hypercapnia may continue to smoke, we examined the impact of exposure to CSE, alone and in combination with hypercapnia, on ACE2 expression, p-SARS-CoV-2 entry, and cholesterol accumulation in epithelial cells. We found that CSE increases ACE2 protein at day 3 ( Figures 5A, B ) and mRNA expression at day 1 ( Figure 5C ), p-SARS-CoV-2 entry ( Figures 5D, E ) into epithelial cells in conjunction with increased cholesterol accumulation ( Figures 5F–H ). The effects of CSE and hypercapnia were similar in magnitude, but they were not additive ( Figures 5A–H ). These results suggest that hypercapnia and cigarette smoking independently augment cholesterol accumulation by similar mechanisms, both increasing ACE2 expression and p-SARS-CoV-2 entry, but in a non-additive, non-synergistic manner.

As noted above, fluvastatin and rosuvastatin blocked hypercapnia-induced increases in cholesterol accumulation, ACE2 protein expression, and p-SARS-CoV-2 entry into epithelial cells ( Figures 4E–H ). Both statins also blocked the increase in cholesterol in lipid rafts stimulated by CSE in VERO ( Figure 6A ) and BEAS-2B cells ( Supplementary Figure 6B ). Likewise, the statins blocked the CSE-induced increases in ACE2 expression ( Figure 6B ) and p-SARS-CoV-2 internalization ( Figures 6C, D ). Thus, inhibiting cholesterol synthesis with statins blocks cellular cholesterol accumulation caused by cigarette smoke, resulting in decreased expression of ACE2 and reduced entry of p-SARS-CoV-2 into epithelial cells.