Cell lines and viruses used
The cell lines used to develop the image cytometry method for SARS-CoV-2 antiviral testing have been published previously: Human lung adenocarcinoma cells stably transfected to express ACE2 (ACE2-A549s) (15), Human lung adenocarcinoma (Calu3) cells (ATCC HTB-55), two human pharyngeal carcinoma (Fadu, Detroit 562) cell lines (ATCC HTB-43 and ATCC CCL-138), human hepatoma (Huh7) cells (unknown sex, a gift from Dr. Charles Rice) (16), and African green monkey kidney epithelial (Vero E6) cells (ATCC CRL-1586).
ACE2-A549s, Huh7, and Vero E6 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco/Life Technologies, Carlsbad, CA) and supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Atlas Biologicals, Fort Collins, CO). Calu3 cells were maintained in DMEM and supplemented with 15% non-heat inactivated FBS. Fadu and Detriot 562 cells were maintained in Minimum Essential Media (MEM) (Gibco/Life Technologies) supplemented with 10% non-heat inactivated FBS. All media was also supplemented with 2 mM nonessential amino acids (HyClone, Logan, UT), 2 mM L-glutamine (HyClone), and 25 mM HEPES buffer. All cell lines were maintained at 37°C with 5% CO2.
The mNeonGreen SARS-CoV-2 virus strain was provided by the World Reference Center for Emerging Viruses and Arboviruses at the University of Texas Medical Branch, Galveston, TX (17). This infectious clone was constructed by Xie et al. based on the virus strain (2019-nCOV/USA_WA01/2020) isolated from the first reported SARS-CoV-2 case in the US (18–20). Viral stocks were amplified in Vero E6 cells to passage 1 (P1) with a titer of 9.67 x 105 PFU/mL and were stored at -80°C. Viral infections were carried out at specific multiplicity of infection (MOI) and incubation time for assay development. During all infections, cells were overlaid with either MEM or DMEM supplemented with 2% heat-inactivated FBS, 2 mM non-essential amino acids, 2 mM L-glutamine, and 25 mM HEPES buffer.
Celigo Image Cytometer and image analysis software
The Celigo Image Cytometer has been described previously, which utilizes one bright field (BF) and four fluorescence (FL) imaging channels: Blue (Ex/Em: 377/470 nm), Green (Ex/Em: 483/536 nm), Red (Ex/Em: 531/629 nm), and Far Red (Ex/Em: 628/688 nm) in combination with high-power light-emitting diodes (21–24).
The Celigo software application “Confluence 1” was used to measure the host cell confluence percentages using the acquired bright field images. The preset ANALYZE parameters were optimized to automatically measure confluence percentages. The confluence percentages were calculated as the ratio of cell surface area coverage divided by total surface area in the well measured directly from the image cytometer. The results were used to optimize Calu3 seeding density.
The Celigo software application “Target 1 + 2 + Mask” was used to identify the total number of Hoechst-positive host cells (Ex/Em: 352/461 nm) in the Blue channel and the number of mNeonGreen fluorescent infected cells (Ex/Em: 506/517 nm) in the Green channel. The Celigo instrument was set up to acquire images in the Target 1 (BF), Target 2 (Green), and Mask (Blue) channels, where the exposure times for mNeonGreen and Hoechst were 60,000 and 145,000 µs, respectively. Image-based autofocus was used to focus in the BF channel, and the focus offsets were applied for the BF (-2 µm), Green (0 µm) and Blue (-15 µm) channels. The preset ANALYZE parameters were optimized to identify the Hoechst-positive host cells above an intensity threshold of 2, then the gating feature was used to determine the mNeonGreen-positive infected cell number.
The Celigo software application “Target 1 + 2 + 3 + 4” was used to identify the total number of Hoechst-positive cells in the Blue channel, the number of live/metabolically active Calcein AM-positive cells (Ex/Em: 488/520 nm) in the Green channel, and dead propidium iodide-positive cells (PI, Ex/Em: 496/636 nm) in the Red channel. The Celigo instrument was set up to acquire images in the Target 1 (BF), Target 2 (Blue), Target 3 (Green), and Target 4 (Red) channels, where the exposure times for Hoechst, Calcein AM, and PI were 23,000, 10,000, and 10,000 µs, respectively. Image-based autofocus was used to focus in the BF channel, and the focus offsets were applied for the BF (-8 µm), Blue (-13 µm), Green (+ 7 µm) and Red (0 µm) channels. The preset ANALYZE parameters were optimized to identify the fluorescent positive cells above an intensity threshold of 4.
Host cell line and MOI selection
Multiple host cell lines were infected with different MOIs of the mNeonGreen SARS-CoV-2 virus to select the most appropriate cell type for developing the antiviral testing method. First, the ACE2-A549, Calu3, Huh7, Fadu, Detroit 562, and Vero E6 cell types were seeded in a 96-well plate (Greiner 655180) at 20,000 cells/well and allowed to adhere overnight. After 24 hours post-seeding, the ACE2-A549, Calu3, and Huh7 were infected with MOIs of 0.01, 0.1, and 1. The Fadu and Detroit 562 were infected with MOIs of 0.001, 0.01, 0.1, and 1. After 72 hours post-infection (hpi), the cells were stained with ViaStain™ Hoechst 33342 (Nexcelom, Lawrence, MA) following manufacturer’s instructions. Subsequently, the wells were scanned using the image cytometer to determine the total cell count and infected cell count for selecting the most optimal cell type. The infected cell count results were normalized to the average total and infected cell count for the control. Each cell line and MOI was tested at n = 3 with the equation \({Log}_{10}\left(mNeonGreen count\right)= \frac{\frac{{Hoechst}_{Sample}^{+}}{AVE {Hoechst}_{Control}^{+} }\times {GFP}_{Sample}^{+}}{AVE {GFP}_{Control}^{+}}\).
Cell seeding density selection
A cell proliferation experiment was conducted to optimize the seeding density for the antiviral testing method using Calu3 that was selected from the experiment described previously. The Calu3 cells were seeded at densities of 1 x 105, 8 x 104, 6 x 104, 4 x 104, 2 x 104, and 1 x 104 cells/well in a 96-well plate (n = 3). The cells were allowed to incubate for 96 hours post-seeding, where the plate was scanned using the image cytometer at 24, 72, and 96 hours. The confluence percentages were measured at each time point to select a seeding density that can generate 50–70% confluence.
High-throughput image-based antiviral testing method
To demonstrate the optimized antiviral testing method, an experiment was performed to measure the antiviral effects of four drugs on SARS-CoV-2-infected Calu3 cells. Remdesivir, Ribavirin, Ambroxol hydrochloride (#A0363700, Sigma-Aldrich, St. Louis, MO), and GENZ-123346 (#28500, Cayman Chemical, Ann Arbor, MI) were selected for testing the image-based method. Remdesivir and Ribavirin were selected for the assay as positive controls since they were already shown to be relatively effective at inhibiting SARS-CoV-2 in vitro (25, 26). Ambroxol HCl and GENZ-123346 are both modulators of glucosylceramide, a lipid metabolite that may be vital for SARS-CoV-2 replication (27). Ambroxol HCl is a molecular chaperone of β-glucocerebrosidase 1, and inhibitor of β-glucocerebrosidase 2, and can modulate glycosphingolipid metabolism in pathophysiological situations by these mechanisms (28). GENZ-123346 is an inhibitor of glucosylceramide synthase. These compounds were chosen as part of an ongoing study in our laboratory investigating the role of sphingolipid metabolism in SARS-CoV-2 infections.
The Calu3 cells were first seeded at a density of 2 x 105 cells/well in 96-well plates and incubated overnight. Approximately 1–3 hours prior to infection, the growth media was removed from cells, and subsequently overlaid with DMEM supplemented with 2% heat-inactivated FBS, 2 mM NEAA, 2 mM L-glutamine, 25 mM HEPES buffer, containing each compound at different concentrations: Remdesivir (10, 5, 1, 0.5 µM), Ribavirin (400, 200, 100, 50 µM), Ambroxol HCl (100, 50, 25, 12.5 µM), and GENZ-123346 (50, 25, 12.5, 6.25 µM). The concentrations selected were based on published IC50 values for Remdesivir and Ribavirin (25) and previous optimization in our studies for Ambroxol HCl and GENZ-123346. The cells were then infected with the mNeonGreen SARS-CoV-2 virus at MOI of 0.1 for 72 hours. Each drug compound and its corresponding vehicle control (DMSO or DMF) were tested in triplicate (n = 3). It is important to note that the 96-well plates were made in duplicate to investigate the antiviral activity as well as assessing cytotoxicity effects of the compounds on host cells in parallel.
Following the infection experiment, the supernatant from infected Calu3 cells was harvested and stored at -80°C. The remaining cells were washed with 1X PBS, and then stained with the ViaStain™ Hoechst 33342 following manufacturer’s instructions. To determine viral replication, the plates were imaged and analyzed using the image cytometer to count the total number of Calu3 cells with Hoechst staining and mNeonGreen SARS-CoV-2 infection. The infected cell count results were normalized to the average total cell count.
In contrast, the cytotoxicity plates were not infected with the mNeonGreen reporter virus, but all other conditions and timeframes were maintained. Following the experiment, the media from uninfected cells was aspirated, cells were washed with 1X PBS, and then were stained with the ViaStain™ Calcein AM/Hoechst/PI Viability Kit following manufacturer’s instructions. The viability kit allowed for determination of the effects of various compounds and their respective concentrations on host cell viability. To determine the cell viability, the plates were imaged and analyzed using the image cytometer to count the total number of cells with Hoechst, the live cells with Calcein AM, and dead cells with PI. The live and dead cell count results were also normalized to the average total cell count.
Statistical Analysis
The statistical analysis method used is noted in each applicable figure. Determination of statistical significance for antiviral effect, cell viability, and live/dead cell counts was performed using a one-way Analysis of Variance (ANOVA) with Dunnett’s multiple comparisons using version 9.3.1 of Prism software (GraphPad Software, La Jolla, California, USA).