key: cord-0289419-5hz39gaz
authors: Chang, Ching-Wen; Parsi, Krishna Mohan; Somasundaran, Mohan; Vanderleeden, Emma; Cruz, John; Cousineau, Alyssa; Liu, Ping; Li, Qi; Wang, Yang; Maehr, Rene; Wang, Jennifer P.; Finberg, Robert W.
title: Cell culture model system utilizing engineered A549 cells to express high levels of ACE2 and TMPRSS2 for investigating SARS-CoV-2 infection and antivirals
date: 2022-01-03
journal: bioRxiv
DOI: 10.1101/2021.12.31.474593
sha: 6bb6c535f2e573c71ebe3bd32611052d966bf71c
doc_id: 289419
cord_uid: 5hz39gaz

Novel pathogenic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) continues to pose an imminent global threat since its initial outbreak in December 2019. A simple in vitro model system using cell lines highly susceptible to SARS-CoV-2 infection are critical to facilitate the study of the virus cycle and to discover effective antivirals against the virus. Human lung alveolar A549 cells are regarded as a useful and valuable model for respiratory virus infection. However, SARS-CoV-2 uses the ACE2 as receptor for viral entry and the TMPRSS2 to prime the Spike protein, both of which are negligibly expressed in A549 cells. Here, we report the generation of a robust human lung epithelial cell-based model by transducing ACE2 and TMPRSS2 into A549 cells and show that the ACE2 enriched A549ACE2/TMPRSS2 cells (ACE2plus) and its single-cell-derived subclone (ACE2plusC3) are highly susceptible to SARS-CoV-2 infection. These engineered ACE2plus showed higher ACE2 and TMPRSS2 mRNA expression levels than currently used Calu3 and commercial A549ACE2/TMPRSS2 cells. ACE2 and TMPRSS2 proteins were also highly and ubiquitously expressed in ACE2plusC3 cells. Additionally, antiviral drugs like Camostat mesylate, EIDD-1931, and Remdesivir strongly inhibited SARS-CoV-2 replication. Notably, multinucleated syncytia, a clinical feature commonly observed in severe COVID-19 patients was induced in ACE2plusC3 cells either by virus infection or by overexpressing the Spike proteins of different variants of SARS-CoV-2. Syncytial process was effectively blocked by the furin protease inhibitor, Decanoyl-RVKR-CMK. Taken together, we have developed a robust human A549 lung epithelial cell-based model that can be applied to probe SARS-CoV-2 replication and to facilitate the discovery of SARS-CoV-2 inhibitors.

Cells were plated on 96-well tissue culture plates (black polystyrene microplates, Corning) and 160 infected with the low passage of SARS-CoV2 or iscSARS-CoV2-mNG at indicated MOI and 161 infection period. Infected cells were fixed with 4% paraformaldehyde for 30 min at room 162 temperature, gently washed 2x in PBS, and permeabilized with 1% Triton X-100 in PBS, and 163 blocked with 5% BSA. Fixed cells were either labeled with a human monoclonal antibody 164 conjugated with Alexa-488 against the spike antigen (10.1371/journal.pmed.0030237) or a mouse 165 monoclonal antibody that recognizes the NP antigen (SinoBiological, #40143-MM08) by 166 incubation for 2 hours at 4 °C. After washing with saline-Tween 20 (0.05%), the cells were 167 labeled with an anti-mouse goat secondary antibody conjugated with Alexa-594 by incubation for 168 1 h at 4 °C. To visualize the cell nuclei, the cells were counterstained with 4′,6-diamidino-2-169 phenylindole (DAPI) (Abcam) for 15 min at 4 °C to visualize the cell nuclei. The images were 170 acquired with the ImageXpress Micro-XL (IXM) system by immunofluorescence with 4x or 10x 171 objectives. The images were processed using MetaXpress Software. 172

For cell fusion assay, cells were plated on regular 24-well tissue culture plates for overnight to 173 reach 90% confluence, 1ug of spike plasmid for each well was transfected into cells using 174 transfection reagent (Mirus, TransIT-LT1) with adding DMSO or antiviral compounds for 24 175 hours. Transfected cells were then fixed and counterstained with DAPI. The entire plate was 176 scanned with the Celigo Image Cytometer (Nexcelom Bioscience) and analyzed by Celigo 177

Software. 178

The SARS-CoV-2 spike pseudotyped virus activity was determined by bright-glo luciferase 180 assay (Promega). The plate reader detected the luminescence two days post virus infection or 181 without virus infection. Cell death was measured by a cytotoxicity detection kit (LDH) from 182

Roche to assess lactate dehydrogenate activity in the culture supernatants of ACE2plus cells. 183

Data are expressed as means ± standard deviations (SD), and the significance of differences 185 between groups was evaluated using ANOVA with Dunnett's multiple comparisons test. All tests 186 were performed using Prism 9 (GraphPad Software). 187 188 3. Results 189

To establish a robust human A549-based cell line for SARS-CoV-2 investigation, we first 191 transduced lentiviral ACE2 and TMPRSS2 genes sequentially into A549 cells ( Figure 1A ). Following puromycin selection, over 50 clones were tested by Clone 43.20 exhibited a higher infectivity rate of approximately 20% and was therefore selected 194 for further ACE2 receptor enrichment by cell sorting. The sorted cell population with a high level 195 of ACE2 expression was then referred to as ACE2plus and exhibited permissibility to CoV-2 virus replication ( Figure 1B ). By contrast, the parental 43.20 clone showed low levels 197 ( Figure 1C ). Next, we conducted further SARS-CoV-2 infections to compare both clones with 198 different MOIs and seeded cell numbers. Infected cells were identified by spike antibody. Images 199 were analyzed to quantify infectivity as described in materials and methods. As Figure 1D shown, 200

ACE2plus cells were more permissive than the parental 43.20 cells and showed higher infectivity 201 even during low MOI virus infections, indicating that the ACE2plus model is highly susceptible 202 to SARS-CoV-2 viral infection. 203

To further determine the ACE2plus model's permissibility, we performed a comparative 204 infection in ACE2plus and Vero E6 cells with SARS-CoV-2 virus (WA1/2020) and recombinant 205 virus (icSARS-CoV-2mNG). Both cell types were infected with WA1/2020 (wild-type), 206 WA1/2020 (mNG), or control growth medium (mock-infected) at low MOIs 0.05 and 0.1. Wild-207 type infected ACE2plus cells exhibited infection rates of approximately 60%, while Vero E6 cells 208

showed infectivity rates around 70% ( Figure 1E ). ACE2plus cells exhibited a range of infectivity 209 from approximately 60-70% across the two MOIs when infected with icSARS-CoV-2mNG, 210

while Vero E6 cells expressed between 70-80% infectivity. To quantify viral nucleoprotein RNA 211 levels, RT-qPCR was conducted. At 48 hours post-infection, supernatants were collected from 212 cells treated with MOI 0.1 wild-type virus and icSARS-CoV-2-mNG, as well as mock-infected, 213 and processed for RNA extraction. Vero E6 cells showed 1.75 fold higher infectivity than 214

ACE2plus when treated with wild-type virus and approximately 1.2 fold higher infectivity than 215

ACE2plus when challenged with icSARS-CoV-2-mNG ( Figure 1F ). Released virus particles 216 from infected ACE2plus cells were also tested using plaque assay to determinate the virus activity 217 ( Figure 1G ). Altogether, those results demonstrate the comparable efficacy of ACE2plus and 218

Vero E6 cells as models for SARS-CoV-2 infections and shows that ACE2plus is a viable human 219 cell model. were similar ( Figure 2A ). However, ACE2plus cells expressed a higher level of TMPRSS2 228 mRNA than IVG-AT and Calu3. Both gene expressions in parental A549 are extremely low. 229

Next, we used flow cytometry on ACE2plus and IVG-AT cells to determine the cell-surface 230 ACE2 protein expression. As expected, the ACE2plus cell population showed > 95% expression 231 of ACE2 while the IVG-AT cells only exhibited 33% expression ( Figure 2B ). Although the 232 ACE2 positive population can be increased after drugs selection, the IVG-AT cells grow very 233

slowly. Thus, cell growth rates were determined by staining ACE2plus and IVG-AT cultures with 234 DAPI using Celigo imaging and software ( Figure 2C ) to determine cell count in 24-hour time 235

increments over a total period of 96 hours. Both cell lines were seeded in a 96-well plate at 1x10 4 236 cells per well. After 24-hour, the ACE2plus cell grows faster and more consistent than the 237 commercial IVG-AT cell line. 238

Considering cell heterogeneity, we further optimized the ACE2plus model by single-cell 239 sorting. To do this, the ACE2plus cell was incubated with an ACE2-specific antibody for 240

Fluorescence Activated Cell Sorting (FACS). We then successfully expanded 23 clones and 241 challenged them with icSARS-CoV-2-mNG virus to compare the infectivity of each clone ( Figure  242 2D). Of them, clone 3 (ACE2plusC3) was then selected due to its high expression of ACE2 and 243 less variation among infected cells (data not shown). We then used Immunofluorescence staining 244 for ACE2 and TMPRSS2 in ACE2plusC3 cells. A549 was used as a staining control. A549 cells 245 exhibited negligible ACE2 and TMPRSS2, while ACE2plusC3 showed strong and ubiquitous 246 expression for both ( Figure 2E ). Next, we examined if the ACE2 receptor on the ACE2plusC3 247 cell surface can be recognized by SARS-CoV-2 Spike-RBD protein. As Figure 2 F shows, 248 recombinant RBD proteins can specifically bind to the ACE2 receptor and be internalized rapidly 249 within 45 minutes (data not shown). According to recent reports, spike D614G mutation is 250 associated with ACE2 receptor binding and results in an increase in infectivity of the SARS-CoV-251 2 (Cheng et al.) . To test this, the 614D and 614G of Spike-pseudotyped lentiviral particles were 252 prepared and used for infections. As luciferase and ZsGreen genes were designed as reporters in 253 the system, the infectivity can be easily measured by ZsGreen protein expression or luciferase 254 assay (Crawford et al., 2020) . As a result, 614G showed stronger infectivity than 614D in a dose-255 dependent manner ( Figure 2G ). This data supports the use of ACE2plusC3 for SARS-CoV-2 256 lentiviral infection assays. Taken together, those results provide evidence that the ACE2plus cell 257

line is an ideal model for the SARS-CoV-2 study.

Cell-cell fusion allows viruses to infect neighboring cells, and it was recently discovered that 260 the polybasic S1/S2 site of SARS-CoV-2 Spike is required for efficient infection of human lung-261 derived cells and promotes syncytium formation (Cheng et al., 2020; Hoffmann et al., 2020a) . 262

Thus, it might be essential to understand the ability of syncytium formation between SARS-CoV-263 2 spike variants, as the large size of syncytia is reported to constitute a hallmark of COVID-19-264 associated pathology (Bussani et al., 2020) . To address this, we first established a mCherry stable 265 cell line using the ACE2plusC3 model. After sorting, over 98% of cells express mCherry protein, 266 and this cell model can be applied to real-time observe the kinetic of spike-mediated cell-cell 267 fusion without no need to coculture with other cells. Next, we transfected an equal amount of 268 indicated spike plasmids into ACE2plusC3-mCherry cells for 24 hours. WA1/2020 spike used as 269 a reference, and pCDNA empty vector acts as a negative control. To visualize and capture whole 270 well images, an entire 24-well plate was scanned using Celigo Image Cytometer ( Figure 3A ). As 2 (Miller et al., 2021) . Camostat mesylate is regarded as an antiviral agent, as it inhibits many of 287 the serine proteases that SARS-CoV and SARS-CoV-2 use for virus-to-host cell membrane 288 fusion, like TMPRSS2, and TMPRSS11 (Breining et al., 2021) . As a result, treating cells with 289

100uM of Camostat mesylate resulted in significant inhibition (Figure 4) . This is consistent with 290 recent reports that Camostat mesylate significantly reduced SARS-CoV-2-driven entry and 291 infection in lung cell line Calu-3 and primary human lung cell (Hoffmann et al., 2020b) . 292

To test potential dose-dependent antiviral activity of those drugs in our cell model, we treated 293

ACE2plusC3 with different concentrations of those drugs and infected the cells with icSARS- and might be potentially developed for high-throughput screening. 302

suppressing spike-mediated cell fusion 304

In the transfection experiments, we found that the spike-mediated cell fusion was reduced by 305 treatment with furin inhibitor Decanoyl-RVKR-CMK ( Figure 3B ). Because several reports have 306 demonstrated that the cleavage of the SARS-CoV-2 spike protein at a putative furin cleavage site 307 (RRARS) at R685/S686 is critical for spike-mediated cell-cell fusion Cheng et al., 308 2020; Hoffmann et al., 2020a) . We also observed extensive syncytial phenotype in SARS-CoV-2-309 infected ACE2plus cells. We next investigated the efficacy of the furin inhibitor Decanoyl-310

ACE2plucC3 cells. To do this, cells were infected with SARS-CoV-2 (WA1/2020) at MOI 0.1 312 and inoculated with a range of concentrations of Decanoyl-RVKR-CMK for 36 hours before 313 staining for DAPI and nucleocapsid protein expression. Mock-treated cells were infected with the 314 virus and received no drug inoculation. In a concentration-dependent manner, cytotoxicity effects 315

were not observed. ( Figure 6D ). As a result, syncytia formation were markedly observed in virus-316 infected cells, and the syncytial phenotype and infected cells were less clearly prominent in the 317 presence of the CMK furin inhibitor ( Figure 6A, 6B) . To quantify the syncytia number, each 318 image captured from 4x object was analyzed based on the size (> 100uM) and the number of 319 nuclei (> 5). As Fig 6C shown , the syncytia number is significantly reduced by CMK inhibitor. 320

To further test inhibition of viral infection in these samples, plaque assays were performed on 321

Vero E6 cells with supernatants collected at 36 hours post-infection. The virus activity was 322 inhibited in CMK-treated samples ( Figure 6E ). Besides, mock-treated samples had a virus titer of 323 approximately 1.75x10 5 plaque-forming units per mL (PFU/mL), and decreased to 0.8x10 5 324 PFU/mL in the 100uM CMK-treated samples ( Figure 6F) . Thus, the furin-dependent process 325 possibly contributes to syncytium formation in ACE2plusC3 cells. 326 327

The current pandemic caused by SARS-CoV2 is completing its second year of global 329 devastation of human lives. While novel vaccine strategies have provided imminent protection 330 and slowed the pace of spreading infection, this approach has been seriously challenged by newly 331 emerging variants. There is critical need for effective antivirals along with the current vaccine 332 strategy to successfully combat this as well as newly emerging pandemics caused by respiratory 333 viruses. A reliable cell model that reproduces the SARS-CoV-2 life cycle is therefore required to 334 help us better understand the virus-host interactions and to discover novel antiviral drugs. Studies 335 have shown that ACE2 receptor is considered essential for SARS-CoV-2 entry, and the serine 336 protease TMPRSS2 for spike protein priming and spike-mediated cell fusion (Hoffmann et al., 337 2020b) . Since the outbreak of the COVID-19 pandemic, many efforts have been focused on to 338 establish various cell models to perform relevant in vitro studies. Several commonly used cell 339 lines such as 293 ACE2 and Vero E6 have been widely utilized for studying SARS-CoV-2 virus 340 entry, replication, and antivirals (Cheng et al., 2020; Hoffmann et al., 2020b; Zeng et al., 2022) . 341

However, they may not be suitable cell models to investigate the pathological mechanism of the 342 host cell in response to the virus infection, as they were not derived from human lung tissue and 343 lack cytopathic effect (CPE) as well as type I interferon genes expression (Osada et al., 2014) . To 344 address this, we systematically developed a highly permissive human lung-based ACE2plus cell 345 model which originated from human lung epithelial cell line A549. By comparing to other cell 346 lines used for susceptibility to SARS-CoV2 infection and replication, ACE2plus expressed a 347 higher level of ACE2 and TMPRSS2 than Calu3 and IVG-AT (A549 ACE2/TMPRSS2 , InvivoGen) and 348

showed a faster and more consistent growth rate than the commercial IVG-AT cell line, 349

indicating that it is an easily manipulated cell model for large-scale screening applications. 350

As ACE2plusC3 was derived from a single cell colony, with homogenous population, it is 351 ubiquitously expressed ACE2 and TMPRSS2 proteins. In this study, we evaluated the longevity 352 of the ACE2plusC3 model's susceptibility to SARS-CoV-2 infection using passage 16~18 of 353 cells, they exhibited similar infectivity levels of 70~80% as early passage of ACE2plusC3 cells. 354

We also evaluated the sensitivity of ACE2plusC3 model for SARS-CoV-2 antivirals using 355

Camostat mesylate and two FDA-approved drugs, Remdesivir and EIDD-1931(molnupiravir's active metabolite), our results showed potent antiviral effect with IC50=59.98 uM, IC50=0.14uM, 357 and IC50=0.84uM, respectively. We noticed that there is a comparative data of the Remdesivir 358 IC50 in different cell lines. For example, the reference drug Remdesivir has been showed 359 differences in IC50 in between Vero (IC50=10uM) and Calu-3 (IC50=1.3 uM) cells (Jang et al., 360 2021; Ko et al., 2021) . This indicates that our ACE2plusC3 is a more sensitive cell model and 361

better than Vero cells for developing an antiviral screening assay. 362

Furthermore, we clearly observed the syncytium formation in cells infected with wild type 363 WA01/2020 strain. Syncytia were also evident in ACE2plusC3 cells transduced with different 364 variants of Spike proteins. We found that the beta-and delta-Spike both had stronger fusogenic 365 activity than WA1/2020 and others. Notably, extensive cell fusion caused by the delta Spike led 366 to giant syncytium formation and induce cell death. By contrast, cell-cell fusion was barely 367 observed with overexpressed omicron Spike at the same condition. Our results are in line with 368 recent reports that show delta-Spike possess higher fusion activity and syncytium formation 369 ability than the parent WA1/2020, and thus likely induced cell-cell fusion in the respiratory tract 370

to cause severe pathogenicity reported in infected individuals (Mlcochova et al., 2021) . Although 371 very little is understood about the omicron variant, the severity of disease is reportedly much less 372 than the delta variant. Our results strongly suggested that the Omicron spike protein induced 373 relatively poor cell fusion similar to recent reports in 293 ACE2 and VeroE6 TMPRSS2 cells (Meng et 374 al., 2021; Zhao et al., 2021) . 375

Coexpression of ACE2 and TMPRSS2 strongly correlates with the virus susceptibility and 376 cell-to-cell infection. In addition to the Spike protein inducing cell-cell fusion, furin and 377 TMPRSS2 play important roles in this process of spike-mediated cell fusion (Hoffmann et al., 378 2020b) . In this report, using furin convertase inhibitor, we have provided evidence that the 379 relatively robust cell infection efficiency of SARS-CoV-2 in ACE2plus is most likely dependent 380 on its higher cell fusion capability (compared to the DMSO treatment). However, we cannot 381 exclude the possibility that other molecules are involved in viral recognition and entry. Further 382 investigation is warranted to identify and tease out the exact roles played by these factors. 383

Overall, we have established a robust human lung-based cell model for SARS-CoV-2 infection. 384

Our data on SARS-CoV-2 virus production, pseudotyped virus infection, Spike-mediated cell 385 fusion, and antiviral test highlight the importance of our cell model, which might provide as a 386 powerful tool to facilitate the study of the emerging SARS-CoV-2 variants. 387 

CoV-2 and COVID-19-Rationale, dosing and safety

Persistence of viral RNA, 413 pneumocyte syncytia and thrombosis are hallmarks of advanced COVID-19 pathology

D614G Substitution of 418 SARS-CoV-2 Spike Protein Increases Syncytium Formation and Virus Titer via Enhanced Furin-419 Mediated Spike Cleavage

Furin Inhibitors Block SARS-CoV-2 Spike Protein Cleavage to Suppress Virus 424 Production and Cytopathic Effects

Protocol and Reagents for Pseudotyping Lentiviral Particles with SARS-CoV-2 Spike Protein for 429 Neutralization Assays

A Multibasic Cleavage Site in the Spike 432 Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

437 SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically 438 Proven Protease Inhibitor

SARS-CoV-2 Infection of Pluripotent Stem Cell-444

Derived Human Lung Alveolar Type 2 Cells Elicits a Rapid Epithelial-Intrinsic Inflammatory 445 Response

Drugs repurposed for COVID-19 by virtual 448 screening of 6,218 drugs and cell-based assay

Comparative analysis of antiviral efficacy of FDA-450 approved drugs against SARS-CoV-2 in human lung cells

SARS-CoV-2 Initiates Programmed Cell Death in 454 Platelets

The emergence, genomic diversity and global spread of 457 SARS-CoV-2

The Genotype 461 to Phenotype Japan

SARS-CoV-2 Omicron spike mediated immune escape, infectivity 464 and cell-cell fusion. bioRxiv

Interact with Human Equilibrative Nucleoside Transporters 1 and 2: 468 Implications for Reaching SARS-CoV-2 Viral Sanctuary Sites

SARS-CoV-2 B.1.617.2 Delta variant replication 472 and immune evasion

The genome landscape of the african green monkey kidney-derived vero cell line

An Infectious cDNA Clone of SARS-CoV-2

COVID-19 and Multiorgan 484 Response

SARS-CoV-2 spreads through cell-to-cell transmission

SARS-CoV-2 Omicron 492 variant shows less efficient replication and fusion activity when compared with delta variant in 493 TMPRSS2-expressed cells

Dose-dependent inhibition in SARS-CoV-2 infection

Representative images from two independent experiments. Cells were plated in a 96-well plate 613 overnight as described in materials and methods. The next day, cells were infected with icSARS

CoV-2-mNG at MOI 0.1 and treated with a range of concentrations of antiviral drugs for 48

After that, cells were fixed and stained with anti-NP antibody, followed by secondary 616 antibody incubation (conjugated with Alexa-594). DAPI was used for nuclear counterstain

Images were scanned by ImageXpress using 10x magnification, 100uM scale bar

Antiviral efficacy of Camostat, Remdesivir and EIDD1931 in ACE2plusC3. Cells were 634 infected with icSARS-CoV-2-mNG at MOI 0.1 and treated with a range of concentrations of 635 antiviral drugs for 48 hours. Supernatants were collected for LDH cytotoxicity assay. Fixed cells 636 were stained with anti-NP antibody, followed by secondary antibody incubation

DAPI was used for counterstain. The images were acquired with the ImageXpress 638 system by immunofluorescence with 4x and processed by MetaXpress Software to calculate the 639

GraphPad was used to draw the dose-response curve and determine the IC50. The data 640 represent the mean (±SD) from two independent experiments

Decanoyl-RVKR-CMK inhibits SARS-CoV2 infectivity by suppressing Spike-651 mediated cell fusion in ACE2plusC3. (A) Microscope images showing viral 652 nucleocapsid protein expression (red) in infected cells, DAPI (blue) was used for nuclear 653

Images were scanned by ImageXpress using 10x magnification. Cells were infected 654 with WA1/2020 strain at an MOI of 0.1 in the presence of furin inhibitor drug dilutions for 36

Syncytia 656 were determined based on the size (> 100uM) and the number of nuclei (> 5). (D) Cytotoxicity 657 was measured by LDH assay. (E) Plaque assay. Supernatants were collected at the 36-hour 658 timepoint post-infection. (F) Virus titer was quantified from plaque numbers. Data are 659 representative of the mean and SEM of two independent experiments

The authors declare that they have no known competing financial interests or personal 390 relationships that could have appeared to influence the work reported in this paper. 391 392