key: cord-295560-0rpguepv
authors: Yan, Kexin; Rawle, Daniel J.; Le, Thuy T.T.; Suhrbier, Andreas
title: Simple rapid in vitro screening method for SARS-CoV-2 anti-virals that identifies potential cytomorbidity-associated false positives
date: 2020-10-14
journal: bioRxiv
DOI: 10.1101/2020.10.13.338541
sha: 
doc_id: 295560
cord_uid: 0rpguepv

The international SARS-CoV-2 pandemic has resulted in an urgent need to identify new anti-viral drugs for treatment of COVID-19 patients. The initial step to identifying potential candidates usually involves in vitro screening. Here we describe a simple rapid bioassay for drug screening using Vero E6 cells and inhibition of cytopathic effects (CPE) measured using crystal violet staining. The assay clearly illustrated the anti-viral activity of remdesivir, a drug known to inhibit SARS-CoV-2 replication. A key refinement involves a simple growth assay to identify drug concentrations that cause cellular stress or “cytomorbidity”, as distinct from cytotoxicity or loss of viability. For instance, hydroxychloroquine shows anti-viral activity at concentrations that slow cell growth, arguing that its purported in vitro anti-viral activity arises from non-specific impairment of cellular activities.

The international SARS-CoV-2 pandemic has resulted in an urgent need to identify new anti-viral 18 drugs for treatment of COVID-19 patients. The initial step to identifying potential candidates usually 19

involves in vitro screening. Here we describe a simple rapid bioassay for drug screening using Vero 20 E6 cells and inhibition of cytopathic effects (CPE) measured using crystal violet staining. The assay 21

clearly illustrated the anti-viral activity of remdesivir, a drug known to inhibit SARS-CoV-2 22

replication. A key refinement involves a simple growth assay to identify drug concentrations that 23 cause cellular stress or "cytomorbidity", as distinct from cytotoxicity or loss of viability. For instance, 24

hydroxychloroquine shows anti-viral activity at concentrations that slow cell growth, arguing that its 25 purported in vitro anti-viral activity arises from non-specific impairment of cellular activities.

The global SARS-CoV-2 pandemic has resulted in widespread activities seeking to identify new 28

anti-viral drugs that might be used to treat COVID-19 patients [1] [2] [3] [4] . Remdesivir has emerged as a lead 29 candidate with clear anti-viral activity in vitro 5 and non-human primates 6 , with some promising 30 early results in human trials 7, 8 . The quest for new anti-viral drugs for SARS-CoV-2 (as for other 31 viruses) usually begins with in vitro screening to identify potential candidates 9-11 . Initial screening 32 usually involves assessing whether drugs can inhibit virus replication in a permissive cell line, with 33

Vero E6 cells widely used for SARS-CoV-2. Such in vitro screening approaches often identify drugs 34 that work well in vitro, but ultimately fail to have anti-viral activity in vivo. For Translational inhibition is also a key anti-viral response, which is able to inhibit replication of many 51 viruses 22, 25, 26 including coronaviruses 27 . Thus a drug that has no specific anti-viral activity, but that 52 is able to induce cellular stress, may therefore inhibit virus replication non-specifically and generate 53 a potential false positive in the screening assay.

A key outcome of stress responses is usually to slow cell growth, allowing the cell to either 55 recover, or if stress and/or damage is excessive, to induce cell death [28] [29] [30] . Cells that are slightly 56 poisoned or otherwise compromised (without induction of stress responses) would likely also show 57 reduced growth rates. Cell growth of Vero E6 cells can be very simply measured by seeding 400 58 cells per well in triplicate into a 96 well flat bottom plate and culturing with a range of drug 59 concentrations for 4 days followed by crystal violet staining. The percentage of protein staining 60 relative to a no-drug control is then calculated and provides a simple measure of the drug 61 concentration that slows cell growth. Perhaps not surprisingly the drug concentrations that caused 62

inhibition of cell growth were usually lower than the drug concentrations that caused cytotoxicity 63 (Fig. 1 , compare black circles with green squares). For some drugs the concentration differences for 64 these two activities were >10 fold ( Fig. 1 , ribavirin, cycloheximide, oleuropein, didemnin B).

Inhibition of cell growth is not really cytostasis, which generally means no growth, and not really 66 cytotoxicity, which is generally viewed as cell death. The reason(s) for reduced cell growth induced 67 by any given drug may not be clear, and may be related to stress responses or some other phenomena 68

that compromises the cells normal metabolic activities. Hence we suggest the term "cytomorbidity" 69

to infer a level of cytotoxicity insufficient to kill the cells or induce cytostasis, but sufficient to stress 70 or compromise the cells, with a simple growth bioassay used to indicate cytomorbidity.

A simple rapid bioassay for screening drugs for potential antiviral activity against SARS-CoV-2 72 is to determine whether the drug can inhibit virus-induced cytopathic effects (CPE) in Vero E6 cells.

Remdesivir is known to inhibit SARS-CoV-2 replication 5 and is used herein to illustrate the behavior 74

of an effective drug in this bioassay. Remdesivir was able to inhibit virus-induced CPE by 50% at 75 ≈1 µg/ml and the drug caused 50% cytotoxicity at ≈100 µg/ml, providing a selectivity index of ≈100. 76

Importantly, remdesivir showed cytomorbidity at ≈80 µg/ml, which still leaves a selectivity index of 77 ≈80 ( Fig. 1 and 2 , Remdesivir). Hydroxychloroquine was able to inhibit viral CPE by 50% at ≈20 78 µg/ml and showed a 50% loss of viability using the MTS assay at ≈100 µg/ml, suggesting a selectivity 79

index of ≈5. However, cytomorbidity was clearly evident at ≈20 µg/ml, so the anti-viral activity 80 occurred at similar concentrations to those that caused cytomorbidity ( Fig. 1, Hydroxychloroquine) ; 81

indicating a potential false positive. The overlapping activities are clearly evident when the crystal 82 violet stained plates are viewed (Fig. 2) . 83

The close relationship between anti-viral activity and translation inhibition (inherent in the stress 84 responses described above) can be seen with the use of the translation inhibitors, cycloheximide and 85 didemnin B. These drugs provide selectivity indices of >10, when comparing viral CPE inhibition 86 and cytotoxicity. However, concentrations that inhibited viral CPE again overlapped with those that 87 caused cytomorbidity ( Fig. 1 , Cycloheximide, Didemnin B). The drug γ-mangostin would appear to 88

have a small level of anti-viral activity with a low selectivity index, but again this activity overlapped 89

with the cytomorbidity (Fig. 1 , γ-mangostin). Thus, as for hydroxychloroquine, the assay results for 90 these latter drugs provide no supportive data for anti-viral activity, instead they suggest these drugs 91

inhibit viral replication non-specifically by impairing cellular activities. Nitazoxanide showed some 92

anti-viral activity, but this coincided with cytotoxicity, providing an example of the conventional 93 cytotoxicity control that would be used to argue that the drug has no specific anti-viral activity and 94

has a selectivity index of 1 (Fig. 1, Nitazoxanide) . Curiously, higher concentrations of nitazoxanide 95

were needed to inhibit cell growth than were needed to induce cytotoxicity; likely an example cell 96 density associated toxicity. 97

The frequently used MTS assay, as expected, often gave results similar to those provided by the 98 cytotoxicity assay. Importantly the MTS assay did not provide a measure of cytomorbidity, 99

presumably because mitochondria largely remain active even in stressed cells and/or cells in G0 100

(cytostasis). For oleuropein, cyclosporine A and γ-mangostin, cytomorbidity was associated with an 101 increase in MTS activity (Fig. 1 ). The MTS bioassay may thus provide slightly misleading 102 information in this context; i.e. increased mitochondrial activity, rather than indicating increased cell 103 numbers, can sometimes be associated with stress or mild toxicity. 104

The CPE based assay described herein is really only useful for screening drugs that target the virus 105

directly. For instance, drugs whose mechanism of action requires induction of type I interferons, 106

would be ineffective as Vero E6 cells do not make type I interferons. Another limitation of using 107

virus-induced CPE as a read-out for anti-viral drugs is sensitivity. Higher drug concentrations are 108 likely needed to prevent viral CPE (overwhelming infection resulting in cell death) than would be 109 needed to inhibit viral replication as measured (for instance) by qRT-PCR of virus released into 110 culture supernatants 31 . Nevertheless, the CPE-based assay represents a screening tool able rapidly 111

to identify promising anti-viral candidates. More sensitive assays could be also envisaged for 112

assessing cytomorbidity, such as measuring activation of stress factors such as ATF3 32 , analyzing 113 cell cycle perturbations by flow cytometry or cell growth kinetics using the IncuCyte live-cell analysis 114

system. However, the simple growth assay proposed herein allows rapid identification of drug 115

concentrations that disrupt cellular activities/functions, which are often sufficient to inhibit viral 116

replication non-specifically. The cytomorbidity assay thereby flags potential false positives. 117 MTS assay; at 100% same OD as control cells with no drug or virus, at 0% background OD.

Cytomorbidity; at 100% cells are growing at their normal rate, 0% no cell growth. Viral CPE; at 100% the drug has completely prevented viral CPE, 0% represents full viral CPE (no antiviral activity). cell debris removed by centrifugation at 3000 x g for 15 min at 4°C, and virus aliquoted and 23 stored at -80 °C. Virus titers were determined using standard CCID50 assays (see below).

The virus was determined to be mycoplasma free using co-culture with a non-permissive cell 25 line (i.e. HeLa) and Hoechst staining as described 2 . 

Cycloheximide, nitazoxanide, ribavirin, hydroxychloroquine 47 sulfate, γ-mangostin and oleuropein were all purchased from Sigma Aldrich

Ribavirin and hydroxychloroquine sulfate was dissolved in Ultrapure Distilled 50

All other drugs were dissolved in DMSO (Sigma Aldrich)

Vero E6 cells were plated as above, 10 4 /well in 54 triplicate in 100 µl medium and cultured overnight. The drug was diluted in 2 fold serial 55 dilutions in RPMI 1640 supplemented with 2% FCS, and 50 µl was then added per well

The plates were cultured for 4 days, after which they were 58 fixed and stained with crystal violet as above. A MTS assay was performed where indicated 59 (before fixation and crystal violet staining) using CellTiter 96 AQueous One Solution Cell 60 Proliferation Assay (MTS) (Promega) as per manufacturer's instructions

Vero E6 cells were plated at 400 cells per well in 100 µl medium

All other steps were performed as described above for cytotoxicity testing

Vero E6 cells were plated as above, 10 4 /well in triplicate in 100 66 µl medium and cultured overnight. The drug was added at 4 times the indicated final 67 concentration in 50 µl RPMI 1640 supplemented with 2% FCS

After 4 days of 69 culture the cells were fixed and stained with crystal violet as above

Heat shock protein 10 inhibits lipopolysaccharide-induced 76 inflammatory mediator production

Complete removal of 78 mycoplasma from viral preparations using solvent extraction

 82 We thank Dr I Anraku for his assistance in managing the PC3 (BSL3) facility at QIMR