key: cord-0887590-7ym9mp16
authors: Bein, Amir; Kim, Seongmin; Goyal, Girija; Cao, Wuji; Fadel, Cicely; Naziripour, Arash; Sharma, Sanjay; Swenor, Ben; LoGrande, Nina; Nurani, Atiq; Miao, Vincent N.; Navia, Andrew W.; Ziegler, Carly G. K.; Ordovas-Montañes, José; Prabhala, Pranav; Kim, Min Sun; Prantil-Baun, Rachelle; Rodas, Melissa; Jiang, Amanda; Tillya, Gladness; Shalek, Alex K.; Ingber, Donald E.
title: Enteric coronavirus infection and treatment modeled with an immunocompetent human intestine-on-a-chip
date: 2021-06-03
journal: bioRxiv
DOI: 10.1101/2021.06.03.446968
sha: ee99c73b8e5424892ef73d854c19ad3aeaf2e599
doc_id: 887590
cord_uid: 7ym9mp16

Many patients infected with coronaviruses, such as SARS-CoV-2 and NL63 that use ACE2 receptors to infect cells, exhibit gastrointestinal symptoms and viral proteins are found in the human gastrointestinal tract, yet little is known about the inflammatory and pathological effects of coronavirus infection on the human intestine. Here, we used a human intestine-on-a-chip (Intestine Chip) microfluidic culture device lined by patient organoid-derived intestinal epithelium interfaced with human vascular endothelium to study host cellular and inflammatory responses to infection with NL63 coronavirus. These organoid-derived intestinal epithelial cells dramatically increased their ACE2 protein levels when cultured under flow in the presence of peristalsis-like mechanical deformations in the Intestine Chips compared to when cultured statically as organoids or in Transwell inserts. Infection of the intestinal epithelium with NL63 on-chip led to inflammation of the endothelium as demonstrated by loss of barrier function, increased cytokine production, and recruitment of circulating peripheral blood mononuclear cells (PMBCs). Treatment of NL63 infected chips with the approved protease inhibitor drug, nafamostat, inhibited viral entry and resulted in a reduction in both viral load and cytokine secretion, whereas remdesivir, one of the few drugs approved for COVID19 patients, was not found to be effective and it also was toxic to the endothelium. This model of intestinal infection was also used to test the effects of other drugs that have been proposed for potential repurposing against SARS-CoV-2. Taken together, these data suggest that the human Intestine Chip might be useful as a human preclinical model for studying coronavirus related pathology as well as for testing of potential anti-viral or anti-inflammatory therapeutics.

To carry out immunofluorescence microscopic imaging, the apical and basal channels of 157 the chips were gently washed with PBS using a micropipettor, fixed with 4% paraformaldehyde 158 Where is the volume of the receiving channel outflow (basal channel), is the concentration 179 of tracer in the receiving channel, is time (sec), is the area of the membrane (cm 2 ), and 180 is the concentration of tracer in the dosing channel outflow (apical channel). 181 Master Mix (Thermo Fisher) or on the CFX96 RT-PCR (Bio Rad) with Sybr green master mix 199 and primers designed against known gene sequences (Supp. Fig. 1) . Expression levels of 200 target genes were normalized to GAPDH or Beta Actin (ACTB). 201

De-identified human patient-derived apheresis collars (a by-product of platelet isolation) 203

were obtained from the Crimson Biomaterials Collection Core Facility under approval obtained 204 from the Institutional Review Board at Harvard University (#22470); informed written consent 205 was not required. PBMCs were isolated using Ficoll density gradient centrifugation and then 206 used immediately or as a frozen stock. Briefly, PBS (2x volume) was added to dilute whole 207 blood in a 50 ml conical tube before 15 ml of the diluted blood was gently added to the top of the 208 density gradient medium, Lymphoprep (Stem cell) and centrifuged at 300 X g for 25 min. 209

Without disturbing the density gradient, the white PBMC layer was collected and suspended in 210 RPMI medium and centrifuged at 120 X g for 10 min. After removing the supernatant, cells were 211 resuspended in fresh RPMI medium and stained with CellTracker Green CMFDA (1:1000 v/v in 212 -PBS per 4 x 10 6 cells, Thermo Fisher) for 10 min at 37°C in a water bath. Stained PBMCs were 213 seeded into the basal channel of the Intestine Chip at 5 x 10 7 /ml and allowed to adhere to the 214 endothelium in an inverted chip for 3 hours before reconnecting the chip back to flow. 215

Effluents from the basal channel of the Intestine Chips were collected, measured for 217 volume, and cytokine protein concentrations were determined using a Luminex kit (R&D 218 System). Nine inflammatory cytokines were selected and the Luminex assay was carried out 219 according to the manufacturer's protocol. The analyte concentration was determined using a 220

Luminex 100/200 TM Flexmap 3D ® instrument with a module, xPONENT software. 221

All drugs tested in this study, nafamostat, (Medchemexpress), remdesivir 223 

We have previously described human Intestine Chips that are created by culturing Intestine Chips was accompanied by much higher expression of ACE2 mRNA compared to the 262 organoid and Transwell cultures (~15-and ~70-fold, respectively) ( Fig. 1B) , and this was 263 confirmed independently by measuring ACE2 protein levels in Western blots (Fig. 1C) . The 264 presence of DM was critical for this induction as ACE2 mRNA levels in intestinal epithelial cells 265 cultured on-chip in EM were 3-fold lower (Supp. Fig. 2) . We also confirmed that the organoids, 266

Transwells, and Intestine Chips all expressed three transmembrane proteases that are involved 267 in coronavirus infection, TMPRSS2, TMPRSS4, and FURIN (Supp. Fig. 3) . 268

To gain insight into the cellular composition and phenotypes in these cultures, single-cell 269 RNA sequencing (scRNA-seq) was carried out on unperturbed Intestine Chips created from 270 duodenal organoids from two different patient donors. Cell identities were defined using key 271 marker genes and published identity scores 18,19 , revealing key mature and stem populations of 272 the small intestine (Fig. 1D, E) . In addition to clearly defined intestinal cell types, we also 273 

Epithelial infection of the Intestine Chip by introducing NL63 virus into the lumen of the 280 upper channel resulted in a significant but transient increase in virus load, as detected by 281

Reverse Transcriptase-quantitative Polymerase Chain Reaction (RT-qPCR) analysis of 282 subgenomic viral RNA transcripts ( Fig. 2A) . While the presence of endothelium was shown to 283 influence influenza virus infection in Lung Airway Chips in past studies 2,20,21 , the presence of 284 endothelium did not significantly alter infection of intestinal epithelium by NL63 (Supp. Fig. 4) . 285

Virus levels were highest 24 hours after infection, decreased by 48 hours, and returned nearly 286 to baseline levels by 72 hours (Fig. 2A) . Consistent with this, we observed a transient increase 287 in tissue barrier permeability by quantitating passage of a fluorescent tracer dye (Cascade Blue) 288 from the apical to the basal channel at 48 hours after infection, which reversed by 72 hours (Fig.  289   2B) . 290

While ACE2 functions as an NL63 receptor, the membrane protease TMPRSS2 also can 291 modulate entry of NL63 virus 22 as well as SARS-CoV-2 9 . We therefore tested an approved 292 protease inhibitor drug, nafamostat, which can inhibit TMPRSS2 in the Intestine Chip NL63 293 infection model. When we perfused nafamostat through the endothelium-lined vascular 294 channel at its reported human plasma maximum concentration (Cmax) to simulate intravenous 295 administration in patients, we found that it significantly reduced viral infection, as measured by 296 quantifying subgenomic viral N protein transcripts using RT-qPCR (Fig. 2C) . In contrast, similar 297 administration of remdesivir, another intravenous drug that has been given emergency use 298 authorization for COVID-19 23 , was not effective and it did not provide an added effect when 299 given in combination with nafamostat (Fig. 2C) . Moreover, this dose of remdesivir also 300 damaged the endothelium, as indicated by detachment of most to the endothelial cell layer 301 Fig. 5) . To further investigate remdesivir induced endothelial cell toxicity, we tested a range of doses on human umbilical vein endothelial cells (HUVEC) in conventional static 303 cultures and found that remdesivir had significant toxicity above a dose of 1uM (Supp. Fig. 6) . 304

While nafamostat appeared to reduce viral load by about 2-fold, neither it nor remdesivir were 305 able to prevent the compromise of intestinal barrier integrity (Fig. 2D) . To confirm the specificity 306 of the nafamostat effects, we carried out similar studies using the laboratory adapted strain of 307 beta coronavirus OC43, which is known to be insensitive to nafamostat, and indeed this drug 308 had no inhibitory activity in this model [24] [25] [26] (Supp. Fig. 7) . 309

Having developed this human preclinical model of intestinal coronavirus infection, we 310 also tested oral drugs, including toremifene, nelfinavir, clofazimine, and fenofibrate, which have 311 been shown to inhibit infection by SARS-CoV-2 and other viruses in vitro 27-30 . Toremifene (10 312 µM) showed similar efficacy to nafamostat in reducing NL63 viral load (Fig. 3A) while again not 313 rescuing barrier compromise (Fig. 3B) . In contrast, nelfinavir (10 µM), clofazimine (10 µM) and 314 fenofibrate (25 µM) were ineffective at the doses tested (Fig. 3C,D) . 315

Viral infection of the GI system induces a coordinated response between multiple cell 317 types including endothelial cells and immune cells. To study immune responses, fluorescently 318 labeled PBMCs were introduced into the endothelium lined vascular channel, and then flow was 319 stopped and the chips were inverted for 3 hours to promote interactions with the endothelium. 320

Quantification of the PBMCs adherent to the endothelium revealed an increase immune cell 321 recruitment in virus infected Intestine Chips 24 hours after infection compared to uninfected 322 (Fig. 4A,B) , and this was accompanied by endothelial damage, as measured by loss of staining 323 for the junctional protein VE-cadherin and increased staining for the apoptosis marker, caspase 324 3 (Fig. 4A) . When we analyzed Intestine Chips that had been pretreated with nafamostat for 24 325 hours prior to infection and the addition of immune cells, we found that while treatment with this 326 drug reduced viral RNA levels (Fig. 2C) , it did not produce statistically significant inhibition of 327 PBMC recruitment to the endothelium (Fig. 4B) . This is consistent with the finding that nafamostat treatment also did not reduce production of multiple inflammatory cytokines (IL-8, I-329 6, MCP-1, MIP-1a, IL33, IFN-g) released into the vascular channel effluent, and only produced 330 modest reduction in IP-10 levels (Fig. 4C) . Interestingly, while NL63 infection induced these 331 cytokines, it moderately suppressed production of the antimicrobial protein, Lipocalin-2, 332 whereas treatment with nafamostat increased its expression (Fig. 4C) . NL63 infection in the Intestine Chip. This finding is likely due to inherent differences between 396 coronavirus subtypes (i.e., NL63 vs. SARS-CoV-2). But surprisingly we found that remdesivir 397 also induced significant endothelial cell toxicity, and we confirmed this in two different types of 398 human endothelium, which should raise some concerns for its use clinically in COVID-19 given 399 the contribution of vascular injury to patient morbidity in this disease. 400

Importantly, when we tested another approved intravenous drug, nafamostat, which is a 401 broad spectrum protease inhibitor that has been reported to inhibit TMPRSS2, we found it 402 significantly inhibited NL63 infection in the Intestine Chip, although it did not prevent 403 compromise of the intestinal barrier. Moreover, similar results were obtained when the oral 404 approved drug, toremifene, which also has been reported to inhibit TMPRSS2, was administered through the lumen of the epithelial channel. Thus, both ACE2 and TMPRSS2 406 appear to be involved in NL63 coronavirus entry into the human intestinal epithelium. 407

Nafamostat also reduced secretion of some cytokines in response to virus infection, but 408 it did not suppress the inflammatory response completely, and recruitment of immune cells was although no adverse effects were observed with these drugs. 419

Taken together, these data suggest that the human Intestine Chip might be useful as a 420 human preclinical model of coronavirus related pathology as well as for testing of potential anti-421 viral or anti-inflammatory therapeutics. While we only studied NL63 coronavirus infection and 422 screened for drugs that inhibit this response in this study, we previously showed that a human 423 lung Airway Chip can be used in a similar manner, and that study led to the discovery of a 424 potential therapeutic for SARS-CoV-2 that is currently in human clinical trials for COVID-19 2 . 425

Thus, the current Intestine Chip model might enable this approach to be used to search for 426 drugs that can target the GI complications associated with both common cold (NL63) and Immunofluorescence micrographs of intestinal endothelium cultured on-chip in the presence or 605 absence of NL63 infection with or without treatment with nafamostat (Naf, 10 M) and stained 606

for VE-cadherin (yellow, top) or nuclei (blue, middle) and caspase-3 (red, middle). Bottom 607 images show the endothelium with adherent fluorescently labeled PBMCs visualized (green,

Severe acute respiratory syndrome 458 coronavirus-2 (SARS-CoV-2): a global pandemic and treatment strategies

A human-airway-on-a-chip for the rapid identification of candidate antiviral 461 therapeutics and prophylactics

Prevalence of Gastrointestinal Symptoms and Fecal Viral Shedding in 464

Patients With Coronavirus Disease 2019: A Systematic Review and Meta-analysis

The Gastrointestinal Tract Is an Alternative Route for SARS

Clinical Characteristics of COVID-19 Patients With Digestive Symptoms in 470

Digestive symptoms of COVID-19 and expression of ACE2 in digestive tract 473 organs

SARS-CoV-2 productively infects human gut enterocytes

TMPRSS2 and TMPRSS4 promote SARS-CoV-2 infection of human 479 small intestinal enterocytes

Infection of bat and human intestinal organoids by SARS-CoV-2

Development of a primary human Small Intestine-on-a-Chip using 483 biopsy-derived organoids

Seq-Well: portable, low-cost RNA sequencing of single cells at high 485 throughput

Second-Strand Synthesis-Based Massively Parallel scRNA-Seq 487

Reveals Cellular States and Molecular Features of Human Inflammatory Skin 488

Cumulus provides cloud-based data analysis for large-scale single-cell and 490 single-nucleus RNA-seq

The Architecture of SARS-CoV-2 Transcriptome

Comprehensive Integration of Single-Cell Data

Integrated analysis of multimodal single-cell data. bioRxiv

Self-organization and symmetry breaking in intestinal organoid 499 development

A single-cell survey of the small intestinal epithelium

Human lung-on-chips: Advanced systems for respiratory virus models and assessment 504 of immune response

Endothelial cells are central orchestrators of cytokine amplification 506 during influenza virus infection

Entry of Human Coronavirus NL63 into the Cell

Remdesivir: A Review of Its Discovery and Development Leading 510 to Emergency Use Authorization for Treatment of COVID-19

Cleavage of group 1 coronavirus spike proteins: how furin cleavage 513 is traded off against heparan sulfate binding upon cell culture adaptation

Wild-type human coronaviruses prefer cell-516 surface TMPRSS2 to endosomal cathepsins for cell entry

SARS-CoV-2 variants with mutations at the S1/S2 cleavage site are 519 generated in vitro during propagation in TMPRSS2-deficient cells

Repurposing of FDA-Approved Toremifene to Treat COVID-522 19 by Blocking the Spike Glycoprotein and NSP14 of SARS-CoV-2

HIV protease inhibitor nelfinavir inhibits replication of SARS-525 associated coronavirus

Clofazimine is a broad-spectrum coronavirus inhibitor that antagonizes 528

SARS-CoV-2 replication in primary human cell culture and hamsters

SARS-CoV-2 induces transcriptional signatures in human lung epithelial 531 cells that promote lung fibrosis

Gastrointestinal symptoms associated with COVID-19: impact on the gut 534 microbiome

Gastrointestinal symptoms of 95 cases with SARS-CoV-2 infection

COVID-19 symptoms predictive of healthcare workers' SARS-CoV-2 538 PCR results

Coronaviruses and gastrointestinal diseases

Gastrointestinal involvement attenuates COVID-19 severity and 542 mortality. medRxiv

Multisystem inflammatory syndrome in children is driven by zonulin-544 dependent loss of gut mucosal barrier

Systematic review: the use of proton pump inhibitors and 546 increased susceptibility to enteric infection

Association Between Acute Gastroenteritis and Continuous Use of 549

Proton Pump Inhibitors During Winter Periods of Highest Circulation of Enteric Viruses

Proton pump inhibitors as a risk factor for norovirus 552 infection

The management of adult patients with severe chronic small 554 intestinal dysmotility

Viral infections of the gastrointestinal tract

Single-cell analyses reveal SARS-CoV-2 interference with intrinsic 558 immune response in the human gut

A cross-talk between epithelium and endothelium mediates human 561 alveolar-capillary injury during SARS-CoV-2 infection

Harnessing Colon Chip Technology to Identify Commensal 564

Bacteria That Promote Host Tolerance to Infection

Enabling out-of-body experiences for living organs

Bar, 100 µm; similar results were obtained in two independent experiments

Quantification of the PBMC recruitment results shown in A. Each data point represents a field of 610 view; gray and black dots show different experiments