Overcoming Culture Restriction for SARS-CoV-2 in Human Cells Facilitates the Screening of Compounds Inhibiting Viral Replication

Abstract

Efforts to mitigate the coronavirus disease 2019 (COVID-19) pandemic include the screening of existing antiviral molecules that could be repurposed to treat severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections. Although SARS-CoV-2 replicates and propagates efficiently in African green monkey kidney (Vero) cells, antivirals such as nucleos(t)ide analogs (NUCs) often show decreased activity in these cells due to inefficient metabolization. SARS-CoV-2 exhibits low viability in human cells in culture. Here, serial passages of a SARS-CoV-2 isolate (original-SARS2) in the human hepatoma cell clone Huh7.5 led to the selection of a variant (adapted-SARS2) with significantly improved infectivity in human liver (Huh7 and Huh7.5) and lung cancer (unmodified Calu-1 and A549) cells. The adapted virus exhibited mutations in the spike protein, including a 9-amino-acid deletion and 3 amino acid changes (E484D, P812R, and Q954H). E484D also emerged in Vero E6-cultured viruses that became viable in A549 cells. Original and adapted viruses were susceptible to scavenger receptor class B type 1 (SR-B1) receptor blocking, and adapted-SARS2 exhibited significantly less dependence on ACE2. Both variants were similarly neutralized by COVID-19 convalescent-phase plasma, but adapted-SARS2 exhibited increased susceptibility to exogenous type I interferon. Remdesivir inhibited original- and adapted-SARS2 similarly, demonstrating the utility of the system for the screening of NUCs. Among the tested NUCs, only remdesivir, molnupiravir, and, to a limited extent, galidesivir showed antiviral effects across human cell lines, whereas sofosbuvir, ribavirin, and favipiravir had no apparent activity. Analogously to the emergence of spike mutations in vivo, the spike protein is under intense adaptive selection pressure in cell culture. Our results indicate that the emergence of spike mutations will most likely not affect the activity of remdesivir.

Keywords: A549 cells; COVID-19; Huh7.5 cells; coronavirus; galidesivir; molnupiravir; nucleotide analogs; remdesivir; sofosbuvir; virus evolution.

FIG 1

Adaptation of SARS-CoV-2 to efficient growth in Huh7.5 cells. (A) Schematic overview of the serial passages performed in Vero E6 and Huh7.5 cells. Dishes represent the culture surfaces, and green and red represent Vero E6 and Huh7.5 cells, respectively. The day of supernatant harvest used for passage is indicated above the light blue arrow that symbolizes the transfer of the culture supernatant to naive (uninfected) cells. The black arrow indicates “no passage.” (B) Comparative infectivity titers of the P2^VeroE6 (original) and P5^Huh7.5 (adapted) viruses in Huh7.5, Vero E6, and Huh7 cells. Infectivity titers (log₁₀ TCID₅₀ per milliliter) are shown on the y axis. Results are based on several independent experiments: for original and adapted viruses in Huh7.5 cells, 6 and 2 independent experiments with 4 replicates each are represented, respectively. In a third independent titration experiment, endpoint dilution for the adapted virus was not achieved (>7 log₁₀ TCID₅₀/ml), and thus, the data were not included in the graph. For the Vero E6 cells, the data presented in the graph correspond to 5 (original) and 4 (adapted) independent experiments with 4 replicates each. For the parental Huh7 cells, results are based on 3 independent experiments for each virus. Bars represent the means and standard errors of the means (SEM) from the different independent experiments. Statistical significance (P < 0.05 by an unpaired t test) is highlighted with an asterisk. (C) Comparative cytopathic effect titers of the original and adapted viruses in Huh7.5, Vero E6, and Huh7 cells. Cytopathic effect titers (log₁₀ CPE₅₀ per milliliter) are shown on the y axis. For original and adapted viruses in both Huh7.5 and Vero E6 cells, results are based on 3 independent experiments with 4 replicates each. For the original virus in Huh7.5 cells, only one experiment yielded a CPE value over the threshold, and thus, the value was not plotted (depicted with “#”). In the case of the parental Huh7 cells, results are based on 3 independent experiments as well, and none of the experiments with the original virus yielded values over the assay threshold (#). Bars represent the SEM for the different independent experiments. Statistical significance (P < 0.05 by an unpaired t test) is highlighted with an asterisk. (D) Visual comparative SARS-CoV-2 antigen staining of both original and adapted viruses after infection of Huh7.5 cells or the blank (noninfected cells), from a representative TCID₅₀ assay. Each picture represents a replicate of infections performed at the indicated dilutions (noncytopathic) for each virus and was obtained after HRP staining with an anti-spike protein antibody, using the ImmunoSpot series 5 UV analyzer as described in Materials and Methods.

FIG 2

Adaptation of SARS-CoV-2 to Huh7.5 cells permits virus culture in human lung carcinoma Calu-1 and A549 cells. (A) Comparative infectivity titers of the original and adapted viruses in Calu-1 and A549 cells. Infectivity titers (log₁₀ TCID₅₀ per milliliter) are shown on the y axis. For each cell line, results are based on 3 independent experiments with 4 replicates each. For the P2^VeroE6 (original) virus in A549 cells, none of the experiments yielded detectable titers (#). Bars represent the means and SEM for the different independent experiments. Statistical significance (P < 0.05 by an unpaired t test) is highlighted with an asterisk. (B) Comparative cytopathic effects of the original and adapted viruses in Calu-1 and A549 cells. Cytopathic effect titers (log₁₀ CPE₅₀ per milliliter) are shown on the y axis. Results are based on two independent experiments with 4 replicates each. CPE was below the detection limit in the two experiments with the original virus in both cell lines (#). (C) Visual comparative SARS-CoV-2 antigen staining of original and adapted viruses in Calu-1 cells (top) and of original, adapted, and day 42 (Vero E6) viruses in A549 cells (bottom), from representative TCID₅₀ assays.

FIG 3

Correlation between ACE2 expression and the viability of original and adapted SARS-CoV-2 in different cell lines. (A) Graph showing intracellular ACE2 expression in the indicated cell lines assessed by Western blotting of ACE2. Values represent the fold amounts of ACE2 in each cell type relative to the levels detected in BHK-21 (baby hamster kidney) cells (black bar). Each gray bar represents the mean and standard deviation (error bars) from three independent experiments. An image of one of the three independent Western blots showing the bands corresponding to ACE2 (green) and the common housekeeping protein β-tubulin (red) is shown below the graph. (B) Representative image of cells infected with 100 μl of an undiluted stock of the original virus (MOI of 4 according to Vero E6). SARS-CoV-2 antigen staining was performed as indicated in the legends of Fig. 1D and Fig. 2C. No infected cells were detected in A549 cells. (C) Effect of ACE2 blocking using an anti-hACE2 antibody on SARS-CoV-2 infection for the different cell types. The y axis represents the percentage of the number of virus-positive cells normalized to nontreated controls, and bars show means and standard errors of the means from 7 and 4 replicates for 10 μg and 40 μg of ACE2-blocking antibody, respectively.

FIG 4

Dependence on SR-B1 and sensitivity to interferon of original and adapted SARS-CoV-2 in different cell lines. (A) Effect of treatment with the SR-B1 antagonist ITX5061 on SARS-CoV-2 infection. The y axis represents the percentage of the number of virus-positive cells normalized to nontreated controls, and bars show means and standard errors of the means from seven replicates. “#” indicates a lack of data as the corresponding concentrations of ITX5061 were cytotoxic in Huh7.5 cells. (B) Treatment with interferon alpha 2b. The y axis represents the percentage of the number of virus-positive cells normalized to nontreated controls, and bars show means and standard errors of the means from 6 replicates for Vero E6, Calu-1, and A549 cells and 3 replicates for Huh7.5 cells. In cases where the bar is not visible, the values were close to zero.

FIG 5

Neutralization of original and adapted SARS-CoV-2 variants. (A) Neutralization was analyzed for plasma samples obtained following recovery from COVID-19 in six individuals. Each graph shows the data obtained from a single experiment for each plasma sample against original and adapted viruses. The lines represent the nonlinear regression of virus inhibition compared to nontreated controls (y axis) in the different plasma dilutions (x axis). Bars represent the standard deviations from at least 3 replicates. The calculated 50% inhibitory dilution (ID₅₀) value for each virus variant is shown in parentheses. The dotted line highlights the 50% neutralization level. All infections were performed at an MOI of 0.01 except for infection with the original virus in the neutralization of plasma sample M57, which was performed at an MOI of 0.02. (B) Comparison of the ID₅₀ values between the two groups (original and adapted variants). No significant differences were found using a Wilcoxon-Mann-Whitney test (P > 0.05).

FIG 6

Structural overview of changes found in the spike protein of SARS-CoV-2 after culture adaptation in Huh7.5 cells. A multiple-sequence alignment of the partial spike protein sequences of SARS-CoV-2 (GenBank accession number MN908947), P2^VeroE6 (original) virus, P5^Huh7.5 (adapted) virus, SARS-CoV (GenBank accession number AY278741), and MERS-CoV (GenBank accession number JX869059) was carried out using MUSCLE software (78). (A) Alignment of the area containing the N-terminal 9-amino-acid deletion in adapted viruses, which corresponds to the protein-specific positions indicated with numbers. The SARS-CoV-2 sequences are shown in green, with the N74 glycosylation site highlighted in yellow; the SARS-CoV sequence is in gray; and the MERS-CoV sequence is in black. (B) Structural alignments of SARS-CoV and MERS-CoV spike proteins (PDB accession numbers 5X58 [79] and 6Q04 [80], respectively) to the SARS-CoV-2 spike protein (PDB accession number 7JJI [81]) using PyMOL (82) with the same parts of the sequences as the ones shown in panel A. The same color coding as in panel A is used. N74 is shown as sticks in SARS-CoV-2, and an attached glycan is illustrated schematically. The structure for the adapted virus was generated from the SARS-CoV-2 structure by introducing the deletion and modeling the loop closure (“HAKRFD”) with ModLoop (83). (C) Alignment of the area around the S2′ cleavage site (indicated with a red arrow). The observed P812R mutation is highlighted in orange. (D) Structural alignments with the same parts of the sequences as the ones shown in panel C. Structures were generated as explained above for panel B using the same PDB entries except for the SARS-CoV spike protein (PDB accession number 5XLR [84]). Residues that align with P812 and R815 in SARS-CoV-2 are represented as sticks. A part of the MERS-CoV sequence (“SISTGSRS”) was modeled using ModLoop (83), as the residues were missing in the experimental structure. The S2′ cleavage site is indicated with a red arrow.