ABSTRACT
There are, besides Remdesivir (RDV), no approved antivirals for the treatment and/or prophylaxis of SARS-CoV-2 infections. To aid in the search for antivirals against this virus, we explored the use of human tracheal airway epithelial cells (HAEC) and human small airway epithelial cells (HsAEC) grown at the air/liquid interface (ALI) and infected at the apical side with either one of two different SARS-CoV-2 isolates. The virus was shown to replicate to high titers for extended periods of time (at least 8 days) and, in particular an isolate with the D614G in the spike (S) protein did so more efficiently at 35°C than at 37°C. The effect of a selected panel of reference drugs that were added to the culture medium at the basolateral side of the system was explored. GS-441524 (the parent nucleoside of Remdesivir), EIDD-1931 (the active metabolite of Molnupiravir) and IFN (β1 and λ1) all resulted in a dose-dependent inhibition of viral RNA and infectious virus titers at the apical side. However, AT-511 (a guanosine nucleotide previously reported to inhibit SARS-CoV-2) failed to inhibit viral replication. Together, these results provide a reference for further studies aimed at selecting SARS-CoV-2 inhibitors for further preclinical and clinical development.
INTRODUCTION
Besides Remdesivir (RDV), there are no approved antivirals for the treatment and/or prophylaxis of SARS-CoV-2 infections, although the clinical benefit of RDV is still a matter of debate1. Major efforts are ongoing to develop novel antiviral drugs. To aid in their development physiological relevant models are needed, in particular because typically immortal cell lines also originating from non-respiratory (and often non-human) tissue are being used in early preclinical studies. For example, VeroE6, a widely used cell line in SARS-CoV-2 studies is defective in the expression of main the SARS-CoV-2 receptors (angiotensin-converting enzyme 2 (ACE2) and transmembrane protease serine 2 (TMPRSS2)). Hence, screenings campaigns often result in the discovery of antiviral agents that regulate autophagy pathways and endosomal-lysosomal maturation which may not be pertinent or translatable as SARS-CoV-2 therapies2. Meanwhile, air-liquid interface of differentiated primary human airway epithelial cells (HAEC) possess the architecture and cellular complexity of human lung tissue and are permissive to variety of respiratory viral infections3,4. Containing all relevant cell types of the lower respiratory tract (ciliated, goblet and basal cells) which includes ACE2 and TMPRSS2 expressing cells, this system allows to dissect the host-pathogen interactions at the molecular and cellular levels and provides a platform for the profiling of antiviral drugs.
In this study, we explored the effect of a selected number of reported SARS-CoV-2 inhibitors in HAEC ALI cultures on the replication of different SARS-CoV-2 isolates. Our results provide a reference set of data for the preclinical development of SARS-CoV-2 inhibitors.
MATERIALS AND METHODS
Cells and virus isolates
The African monkey kidney cell line VeroE6 tagged green fluorescent protein (VeroE6-GFP, kindly provided by M. van Loock, Janssen Pharmaceutica, Beerse, Belgium) and VeroE6 were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, catalogue no. 41965-039) supplemented with 10% v/v heat-inactivated foetal bovine serum (HI-FBS; HyClone, catalogue no. SV03160.03), 1% v/v sodium bicarbonate 7.5% w/v (NaHCO3; Gibco, catalogue no. 25080-060), and 1% v/v Penicillin-Streptomycin 10000 U/mL (P/S; Gibco, catalogue no. 15140148) at 37°C and 5% CO2. The hepatocellular carcinoma cell line Huh7 (kindly provided by Ralf Bartenschlager, University of Heidelberg, Germany) was propagated in DMEM supplemented with 10% HI-FBS, 1% NaHCO3, 1% P/S, 1% non-essential amino acids (NEAA; Gibco, catalogue no. 11140050), and 2% HEPES 1M (Gibco, catalogue no. 15630106) at 37°C and 5% CO2. All assays involving virus growth were performed in the respective cell growth medium containing 2% (VeroE6-GFP) or 4% (Huh7) instead of 10% FBS.
SARS-CoV-2 isolate BetaCoV/Germany/BavPat1/2020 (EPI_ISL_406862|2020-01-28, kindly provided by C. Drosten, Charité, Berlin, Germany) and BetaCov/Belgium/GHB-03021/2020 (EPI_ISL_407976|2020-02-03) retrieved from RT-qPCR-confirmed COVID-19 positive patients in January and February 2020 were described previously5,6. The generation of virus stocks by serial passaging in Huh-7 and VeroE6 cells were fully reported7,8. BavPat1 isolate (passage 2 (P2)) and GHB-03021 isolate (P6 and P7) were used for the air liquid-interface experiment while only the latter was used for standard in vitro assays in VeroE6-GFP cells (P6 and P7) and in Huh7 cells (P9). The genomic sequence of both isolates is highly similar. BavPat1 carries the D614G amino acid change in the spike-protein while the GHB-03021 has a ΔTQTNS deletion at 676-680 residues that is typical for SARS2 strains that have been passaged several times on VeroE6 cells. All infectious virus-containing works were conducted in biosafety level 3 (BSL-3) and 3+ (CAPs-IT) facilities at the Rega Institute for Medical Research, KU Leuven, according to institutional guidelines.
Compounds
GS-441524 and EIDD-1931 were purchased from Carbosynth (United Kingdom) and R&D Systems (USA) respectively. Stock solutions (10 mM) were prepared using analytical grade dimethyl sulfoxide (DMSO). AT-511 was synthesized and chemically validated at the California Institute for Biochemical Research (Calibr) (La Jolla, CA) and used as a 10 mM DMSO solution. The biological activity of AT-511 was confirmed in an antiviral assay with hepatitis C (data not shown). IFN λ1 was purchased from R&D Systems and IFN β-1a was a kind gift from the laboratory of Immunobiology (Rega Institute, KU Leuven, Belgium), which were reconstituted in sterile phosphate buffered saline (PBS, Life Technologies) containing at least 0.1% FBS.
In vitro standard antiviral and toxicity assays
VeroE6-GFP cells were seeded at a density of 25000 cells/well in 96-well plates (Greiner Bio One, catalogue no. 655090) and pre-treated with three-fold serial dilutions of the compounds overnight. On the next day (day 0), cells were infected with the SARS-CoV-2 inoculum at a multiplicity of infection (MOI) of 0.001 median tissue infectious dose (TCID50) per cell. The number of fluorescent pixels of GFP signal determined by High-Content Imaging (HCI) on day 4 post-infection (p.i.) was used as a read-out. Percentage of inhibition was calculated by subtracting background (number of fluorescent pixels in the untreated-infected control wells) and normalizing to the untreated-uninfected control wells (also background subtracted). The 50% effective concentration (EC50, the concentration of compound required for fifty percent recovery of cell-induced fluorescence) was determined using logarithmic interpolation. Potential toxicity of compounds was assessed in a similar set-up in treated-uninfected cultures where metabolic activity was quantified at day 5 using the MTS assay as described earlier9. The 50% cytotoxic concentration (CC50, the concentration at which cell viability reduced to 50%) was calculated by logarithmic interpolation.
Huh7 cells were pre-seeded at 6000 cells/well in 96 well-plates (Corning, catalogue no.3300) and incubated overnight at 37°C and 5% CO2. On day 0, cells were firstly treated with the three-fold serial dilution of a potential antiviral, followed by either the inoculation of SARS-CoV-2 at MOI of 0.0037 TCID50/cell or addition of fresh medium. After 4 days, differences in cell viability caused by virus-induced cytopathic effect (CPE) or by compound-specific toxicity were evaluated using MTS assays. The EC50 and CC50 were calculated as above-mentioned.
Viral infection of reconstituted human airway epithelium cells
Tracheal HAEC (catalogue no. EP01MD) and human small airway epithelium cells (HsAEC) (catalogue no. EP21SA) from healthy donors were obtained from Epithelix (Geneva, Switzerland) in an air-liquid interphase set-up. After arrival, the insert was washed with pre-warmed 1x PBS (Gibco, catalogue no. 14190-094) and maintained in corresponding MucilAir medium (Epithelix, catalogue no. EP04MM) or SmallAir medium (Epithelix, catalogue no. EP64SA) at 37°C and 5% CO2 for at least 4 days before use. On the day of the experiment, the H(s)AEC were first pre-treated with basal medium containing compounds at different concentrations for indicated hours, followed by exposing to 100 μL of SARS-CoV-2 inoculum from apical side for 1.5 hours. Then the cultures were incubated at the indicated temperatures. The first apical wash with PBS was collected either right after the removal of viral inoculum (day 0) or 24 hours later (day 1 post-infection (p.i.)). Every other day from day 0, subsequent apical washes were collected whereas compound-containing medium in the basolateral side of the H(s)AEC culture was refreshed. Wash fluid was stored at −80°C for following experiments.
RNA extraction and quantitative reverse transcription-PCR (RT-qPCR)
Viral RNA in the apical wash was isolated using the Cells-to-cDNA™ II cell lysis buffer kit (Thermo Fisher Scientific, catalogue no. AM8723). Briefly, 5 μL wash fluid was added to 50 μL lysis buffer, incubated at room temperature (RT) for 10 min and then at 75°C for 15 min. 150 μL nuclease-free water was additionally added to the mixture prior to RT-qPCR. In parallel, a ten-fold serial dilution of corresponding virus stock was extracted. The amount of viral RNA expressed as TCID50 equivalent per insert (TCID50e/insert) was quantified by RT-qPCR using iTaq universal probes one-step kit (Bio-Rad, catalogue no. 1725141), and a commercial mix of primers for N gene (forward primer 5’-GACCCCAAAATCAGCGAAAT-3’, reverse primer 5’-TCTGGTTACTGCCAGTTGAATCTG-3’) and probes (5’-FAM-ACCCCGCATTACGTTTGGTGGACC-BHQ1-3’) manufactured at IDT Technologies (catalogue no. 10006606). The reaction (final volume: 20 μL) consisted of 10 μL one-step reaction mix 2X, 0.5 μL reverse transcriptase, 1.5 μL of primers and probes mix, 4 μL nuclease-free water, and 4 μL viral RNA. The RT-qPCR was executed on a Lightcycler 96 thermocycler (Roche), starting at 50°C for 15 min and 95°C for 2 min, followed by 45 cycles of 3 sec at 95°C and 30 sec at 55°C.
Titration using a 50% tissue culture infectious dose (TCID50) assay
VeroE6 cells were seeded in 96-well tissue culture plates at a density of 1×104 cells/180 μL/well. After 24 hours, serial 10-fold dilutions of ALI wash fluid were prepared in the plates. Cells were incubated for 3 days at 37°C and evaluated microscopically for the absence or presence of virus induced cytopathic effect (CPE). The infectious viral titer was determined by end-point titration, expressed as TCID50/ml. Virus titers were calculated by using the Spearman and Karber method as previously reported10,11.
Statistical analysis
All statistical comparisons in the study were performed in GraphPad Prism (GraphPad Software, Inc.). Statistical significance was determined using the ordinary one-way ANOVA with Dunnett’s multiple comparison test. P-values of ≤0.05 were considered significant.
RESULTS
Replication kinetics of SARS-CoV-2 in reconstituted human airway epithelia
We first compared the replication kinetics of the Belgian isolate GHB-03021 and the German isolate BavPat1. The main differences in the genomes of these viruses is the D614G amino acid change in the spike-protein of BavPat1 and the deletion of several amino acids near the furin-cleavage site in the GHB-03021 isolate (because of extensive passaging in VeroE6 cells). The replication kinetics was investigated at respectively 35 and 37°C in both cultures from tracheal cells (HAEC) or from small airway cells (HsAEC). In preliminary experiments, it was observed that an input of 102, 103 or 104 TCID50/insert resulted in comparable levels of virus production (data not shown). We therefore selected 2×103 TCID50/insert as the viral input for this experiment. Overall, BavPat1 infected the cultures more efficiently than the GHB-03021 isolate did (Fig. 1). For example, at 37°C not all bronchiole-airway derived inserts infected with GHB-03021 resulted in a productive infection whereas all cultures infected with BavPat1 showed productive infection under all conditions. Virus replication was for both isolates higher at 35°C and more reproducible as compared with a temperature of 37°C. This difference was more pronounced for the HsAEC than for the HAEC cultures.
Effect of selected antivirals on SARS-CoV-2 replication in HAEC cultures
Three nucleoside analogues that are known as inhibitors of SARS-CoV-2 replication were selected as reference for studies in HAEC cultures. These included GS-44152412–16 (the parent nucleoside of RDV), EIDD-193116–18 (the active metabolite of Molnupiravir) and AT-51117 [a guanosine nucleotide analogue with activity against hepatitis C virus (HCV)]. In order to select a suitable concentration range of these molecules to be used in the HAEC cultures, we first explored their effect in VeroE6 and Huh7 cell lines. Both GS-441524 and EIDD-1931 selectively inhibited SARS-CoV-2 replication. On the other hand, AT-511 was surprisingly entirely devoid of antiviral activity (Table 1).
At 10 μM, GS-441524 sterilized the HAEC cultures from the GHB-03021 virus. Indeed, no virus production was detected during the first 9 days of treatment and when treatment was stopped, no rebound was observed over the next 5 days of culturing. When evaluated at a concentration of 1 μM, GS-441524 reduced virus yield by ~1 log10 during the time of treatment, but lost activity once the compound was removed from the culture. In a separate experiment, GS-441524 at 3 μM resulted in complete inhibition of virus production upon infection with BavPat1 (Fig. 2F-H). Also, 10 μM of EIDD-1931 resulted in a pronounced antiviral effect (Fig. 3F-J). AT-511, however, at the various concentrations tested (1 and 10 μM) was devoid of an antiviral effect (Fig. 3A-E).
Prophylactic interferon type I and type III reduce SARS-CoV-2 production
Human IFN has been used to treat several viral infections19,20 and recently clinical trialsagainst SARS-CoV-2 are ongoing (ClinicalTrials.gov number: NCT04315948, NCT04385095, and NCT04492475). Therefore, we investigated whether IFN β-1a and IFN λ1 exert activity when used as a prophylactic monotherapy. Tracheal cultures were pre-treated with either 5 and 50 ng/mL IFN λ1 (5 ng/mL is the average concentration secreted in the basal medium of infected HAEC cultures21) or 1 and 100 IU/mL IFN β-1a for 24 hours, and subsequently infected with BavPat1. Both drugs were able to reduce viral titers in a dose-dependent manner (Fig. 4A, 4F). Viral loads were reduced by 100 IU/mL IFN β-1a (3.3 log10 vRNA reduction, 3.6 log10 titer reduction) and 50 ng/mL IFN λ1 (4.2 logs vRNA reduction, 5.0 log10 titer reduction) on day 4 p.i. (Fig. 4B, 4D, 4G, 4I respectively). At later time points viral load in the treated samples increased again.
DISCUSSION
We demonstrate that ex vivo models reconstituted from human tracheal or small airway epithelium are permissive for SARS-CoV-2 infection and robustly produces viral progenies from the apical side in long-term experiments (up to 14 days p.i.). Recent studies report on the effect of different SARS-CoV-2 isolates and incubation temperatures on virus replication kinetics22–25. We used two isolates, BavPat1 and GHB-03021, whereby the BavPat1 provedto be more readily infectious. The BavPat1 isolate carries the p.D614G substitution in the spike (S) protein while the GHB-03021 has a deletion of several amino-acids in the S1/S2 boundary that is typically found in a VeroE6-adapted isolates7. The spike substitution D614G has been reported to increase the stability and infectivity of virions in HAEC culture by enhancing the ACE2-receptor-binding26,27. Isolates with this substitution have become globally dominant26–29. Meanwhile, it has been noted that the continued propagation of SARS-CoV-2 in Vero cells causes several substitutions or deletions in the S1/S2 boundary25,30–33, which are only rarely observed in clinical samples31,32. We speculate that the adaptation to Vero cells results in a phenotype that allows more efficient entry through an ACE2-independent pathway. This entry mechanism would enhance entry in VeroE6 cells but would limit entry in primary lung epithelial cells. Further mechanistic studies are required to elucidate this hypothesis.
The anatomical distance and ambient temperature between upper and lower human respiratory tracts have a profound influence on the replication kinetics of respiratory viruses22,34–36. In agreement with other studies, we observed SARS-CoV-2 growth in favour of lower temperature (35°C) which can be attributed to the temperature preference of SARS-CoV-2 S protein for its folding and transport24,37. Altogether, both primary HAEC and HsAEC cultures are shown to be a robust model for SARS-CoV-2 replication that can be used for antiviral drug profiling.
A promising target for the development of novel antiviral agents active against coronaviruses is the viral RNA-dependent RNA polymerase (RdRp)38. Remdesivir, a phosphoramidate prodrug of an adenosine C-nucleoside, has been approved as the first COVID-19 therapy. However, its effectiveness is still a matter of debate1. In addition, it has a challenging pharmacological profile allowing intravenous administration only39–41. We demonstrate that the parent nucleoside GS-44152412–16 can “sterilize” H(s)AEC cultures from SARS-CoV2 as no rebound of the virus was noted several days after removal of the molecule. Differences in antiviral potencies of RDV and GS-441524 have been reported depending on the cell lines used, which correlates with the formation of the biologically active (5’-triphosphate) metabolite12,13. Data from a pharmacokinetic study in mice suggests that GS-441524 could possibly be considered as an oral drug12.
AT-527 is currently being evaluated in phase II clinical trials for COVID-19 (ClinicalTrials.gov). Surprisingly, we did not observe anti-SARS-CoV-2 activity of AT-511, the free base form of AT-527, in VeroE6 and Huh7 cells, nor did we observe antiviral activity in the HAEC cultures. This is in contrast with a recent publication where sub-micromolar activities of AT-511 were observed in very similar assay systems17. At this moment we have no explanation for this discrepancy. One possibility is that small differences in the assay conditions may influence the metabolization of AT-511 to its active form and thus influence its antiviral activity. As AT-511 is a double pro-drug it may be more susceptible to these nuances.
In addition to RDV and AT-511, we also investigated the effect of the nucleoside analogue EIDD-1931 which is the active metabolite of the ester prodrug Molnupiravir (EIDD-2801). EIDD-1931 has been reported to exert antiviral activity against various human coronaviruses and Molnupiravir is currently in clinical trials for SARS-CoV-217,18. Initial interim data from a phase II study provides first evidence for antiviral activity in COVID patients (https://www.croiconference.org/abstract/reduction-in-infectious-sars-cov-2-in-treatment-study-of-covid-19-with-molnupiravir/). Like GS-441524, EIDD-1931 also results in a pronounced antiviral effect in the human airway epithelium cell cultures, which is consistent with another report18.
Also, a significant inhibitory effect of IFN β-1a and IFN λ1 was noted, although at high concentrations and in particular during the first days of the treatment. At later time-points, viral replication increased in the treated cultures, suggesting that the virus can escape the effect of IFN. The effective concentration of IFN β-1a used in this study is comparable with the clinically achievable concentration and is in line with other reports19–20–42.
In conclusion, we assessed (i) the replication of two SARS-CoV-2 isolates in H(s)AEC cultures and (ii) the antiviral effect of a selected list of inhibitors. These data provide a reference when developing yet other inhibitors of SARS-CoV2 replication.
Conflict of interest
All authors declare that there is no conflict of interest.
Author contributions
J.N. and D.J. conceptualized and supervised the project. T.N.D.D and D.J. designed the research. T.N.D.D. performed the ALI-related experiments. T.N.D.D. analysed data. A.J.C, P.A.G, and M.D.B. characterized AT-511 structure and tested its activity against HCV. T.N.D.D. wrote the first draft of the manuscript. D.J., S.D.J., L.V., and J.N. edited the manuscript. L.V., D.J. and J.N. acquired the funding.
Acknowledgements
We thank Birgit Voeten, Niels Cremers, Tina van Buyten and Thibault Francken for their excellent technical assistance. We also thank Piet Maes for kindly providing the SARS-CoV-2 GHB-03021/2020 isolateused in this study. This research was supported by Bill & Melinda Gates Foundation (BMGF) under grant number INV-006366. T.N.D.D received the fellowship from European Union’s Horizon 2020 research and innovation programme under Marie Sklodowska-Curie grant agreement No. 812673 (OrganoVIR project). Part of this research work was performed using the ‘Caps-It’ research infrastructure (project ZW13-02) that was financially supported by the Hercules Foundation and Rega Foundation, KU Leuven.