The TMPRSS2 Inhibitor Nafamostat Reduces SARS-CoV-2 Pulmonary Infection in Mouse Models of COVID-19

1Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the causative agent of coronavirus disease 2019 (COVID-19), a multiorgan syndrome characterized by severe pneumonia and additional gastrointestinal, cardiovascular, neurological, and systemic manifestations. This novel coronavirus emerged in China in December 2019 and quickly reached pandemic status. As of 27 May 2021, the number of laboratory-confirmed infections has reached 168 million, resulting in nearly 3.5 million deaths worldwide (Johns Hopkins Coronavirus Resource Center; https://coronavirus.jhu.edu/). SARS-CoV-2 is evolutionarily related to SARS coronavirus (SARS-CoV), the etiologic agent of severe acute respiratory syndrome (SARS), which caused an epidemic of severe pneumonia in 2002 and 2003. A second pathogenic member of the coronavirus family, Middle East respiratory syndrome coronavirus (MERS-CoV), emerged in 2012 and has caused over 850 deaths to date according to the World Health Organization.

2Like SARS-CoV, SARS-CoV-2 utilizes the membrane ectopeptidase angiotensin-converting enzyme 2 (ACE2) as its receptor to initiate binding and entry. The virus engages ACE2 using the receptor binding domain of the spike glycoprotein (S). To facilitate fusion with host membranes, the S protein is “primed” by proteolytic cleavage events, first at the S1/S2 cleavage site and then at the S2′ site to liberate the N terminus of the fusion peptide on the S2 subunit. This entry process is cell type dependent. In tissue culture cell lines such as Vero E6, virions bind ACE2 at the cell surface, then enter an endosomal compartment where cathepsins mediate the S2′ cleavage and membrane fusion takes place. In contrast, in respiratory epithelia, the predominant mode of viral entry appears to be membrane fusion at the cell surface, where cell surface proteases such as TMPRSS2 execute the S2′ cleavage.

3Studies in cultured human airway epithelia established that protease inhibitors targeting TMPRSS2 can block SARS-CoV-2 entry (as well as that of SARS-CoV and MERS-CoV), suggesting their therapeutic potential. One candidate is the serine protease inhibitor camostat mesylate, which was initially recognized for its ability to inhibit TMPRSS2-mediated membrane fusion and entry of influenza viruses in airway epithelia. Camostat reduced SARS-CoV-2 pseudovirion entry into the human lung cell line Calu-3, and it has similar activity against SARS-CoV and MERS-CoV. A related serine protease inhibitor, nafamostat mesylate, also potently inhibits SARS-CoV-2 and MERS-CoV in vitro.

4While these data are promising, experiments demonstrating protection against SARS-CoV-2 infection in animal models are lacking. Mice are not naturally permissive to SARS-CoV-2 infection, due to the low affinity of the S protein receptor binding domain for mouse Ace2. Previously, a transgenic mouse expressing the human ACE2 gene under the control of the cytokeratin 18 (K18) promoter (the K18-hACE2 mouse) was generated to study SARS-CoV pathogenesis. Recent studies have demonstrated that these mice also support SARS-CoV-2 infection and develop a lethal respiratory illness with weight loss, inflammation, and associated brain infection. Another model sensitizes mice to SARS-CoV-2 infection by adenoviral vector transduction to deliver a human ACE2 transgene (Ad5-hACE2). Mice infected with SARS-CoV-2 under these conditions recapitulate features of COVID-19 lung disease, including pneumonia and severe lung pathology.

5Here, we tested the hypothesis that protease inhibitors can inhibit SARS-CoV-2 and MERS-CoV infection in primary human airway epithelial cells. Because nafamostat showed relatively greater potency in inhibiting coronavirus infection in vitro, we focused on nafamostat for in vivo efficacy studies. We found that nafamostat reduced lung tissue viral load and disease severity in two complementary mouse models of COVID-19. The inhibitory effects of nafamostat were dependent upon the time of administration and route of delivery, highlighting the importance of these features in clinical studies of COVID-19 prevention or treatment.

6We studied two coronaviruses that can cause serious disease in humans, SARS-CoV-2 and MERS-CoV. To better understand the protease requirements for entry of these two pathogenic coronaviruses in airway epithelia, we investigated their responses to several inhibitors during the early course of infection. We used primary cultures of well-differentiated human bronchial epithelia, which closely mimic the native surface airway epithelium in vivo. We tested camostat and nafamostat, bafilomycin A1 and chloroquine (both inhibitors of endosomal acidification), and E64d, an inhibitor of lysosomal cathepsins B and L. The inhibitors were added 1 h prior to infection with a multiplicity of infection (MOI) of 0.1 of each respective coronavirus, and 20 h later, we measured viral RNA abundance and the titer of virus released into airway surface liquid. In this setting, only camostat and nafamostat significantly inhibited infection of human airway epithelia (Fig. 1A and B). Thus, both MERS-CoV and SARS-CoV-2 depend on cell surface serine proteases such as TMPRSS2 for entry into primary airway epithelial cells. We found that the SARS-CoV-2 inhibitory activity of camostat and nafamostat were lost when infections were performed in TMPRSS2-negative Vero E6 cells, suggesting that these compounds likely work through inhibition of TMPRSS2 protease activity rather than through direct virucidal action (see Fig. S1 in the supplemental material).

7Previous pseudovirus studies reported that nafamostat mesylate inhibits SARS-CoV-2 S protein mediated entry into Calu-3 cells with ∼15-fold higher efficiency than that of camostat mesylate, and that nafamostat blocks infection by authentic SARS-CoV-2 more effectively than camostat in these cells. We investigated camostat and nafamostat inhibition of authentic MERS-CoV or SARS-CoV-2 infection of Calu-3 2B4 cells (Fig. 1C to E). For both MERS-CoV (Fig. 1C) and SARS-CoV-2 (Fig. 1D), nafamostat pretreatment reduced viral RNA at 20 h postinfection more than camostat pretreatment. The 50% inhibitory concentration (IC50) for nafamostat was 2.2 nM (95% confidence interval [CI], 1.8 to 2.5 nM; R2 = 0.98) and for camostat was 14.8 nM (95% CI, 8.2 to 26.2 nM; R2 = 0.89) (Fig. 1E).

8Given its greater efficacy in vitro, we evaluated whether nafamostat could inhibit SARS-CoV-2 infection in a mouse model of COVID-19. Mice were sensitized to SARS-CoV-2 infection by intranasal transduction (Ad5-hACE2). Nafamostat was delivered via intraperitoneal (i.p.) or intranasal (i.n.) routes, and mice were inoculated intranasally with SARS-CoV-2. Nafamostat pretreatment inhibited SARS-CoV-2 infection more potently when delivered via the i.n. route. Intranasal nafamostat administration resulted in a nearly 2-log reduction in lung tissue viral titers at the highest dose tested (3 mg/kg) compared to a less than 5-fold reduction following i.p. (20 mg/kg) delivery (Fig. 2A and B). Nafamostat reduced titers when administered 2, 4, or 6 h prior to viral inoculation, with moderately better effectiveness if delivered nearer the time of SARS-CoV-2 infection (Fig. 2C).

9To assess whether i.n. nafamostat administration altered the course of SARS-CoV-2 infection, mice were transduced with Ad5-hACE2, followed by i.n. infection with SARS-CoV-2 (105 PFU/mouse). Animals received nafamostat (3 mg/kg, i.n.) at 2 h prior to infection, 1 day postinfection, or 3 days postinfection, and were monitored daily for weight loss (Fig. 3A). Nafamostat pretreatment abrogated SARS-CoV-2-induced weight loss (Fig. 3B). Weight loss was also significantly reduced in animals receiving nafamostat at 1 day postinfection (Fig. 3B). Consistent with these findings, lung viral loads were reduced at 1, 2, and 4 days postinfection in mice pretreated with nafamostat, whereas the reductions in lung viral titers were more modest in mice receiving nafamostat after SARS-CoV-2 challenge (Fig. 3C). Histopathological analysis of lung tissue from infected animals at 5 days postinfection suggests that nafamostat treatment reduced lung pathology in the infected mice, primarily in the animals receiving the pretreatment protocol (Fig. 3D and E). These results indicate that in mice expressing hACE2 via Ad5-hACE2 transduction, nafamostat reduces SARS-CoV-2 infection severity, particularly when administered prior to or early in infection.

10We next tested nafamostat in K18-hACE2 mice. K18-hACE2 mice were pretreated with i.n. nafamostat (3 mg/kg) for 2 h, followed by SARS-CoV-2 challenge (2.5 × 103 PFU/mouse) (Fig. 4A). Over a 14-day time course, nafamostat-treated mice lost less weight and exhibited significantly less mortality than vehicle-treated controls (Fig. 4B and C). At 1 day postinfection, virus was detected in the lungs of vehicle-treated mice but was largely undetectable in tissue from nafamostat-treated mice (Fig. 4D). By 7 days postinfection, virus titers could be measured in both lung and brain in 50% of vehicle-treated mice, whereas no virus was detected in lung or brain tissue from mice receiving nafamostat (Fig. 4E).

11These results suggest that nafamostat pretreatment significantly reduced viral loads over the course of SARS-CoV-2 infection. We examined the distribution of SARS-CoV-2-positive cells in tissues from K18-hACE2 mice by immunostaining for viral antigen. In the lungs of vehicle-treated animals, infection was widespread throughout the cells of the small airways and alveoli by 7 days postinfection. In contrast, SARS-CoV-2-positive cells were far less abundant (though not entirely absent) in tissue from nafamostat-treated mice (Fig. 4F and G). Viral infection in the brain was more variable. While there were no virus-positive cells in either treatment group at 1 day postinfection, by 7 days postinfection, at least half of the vehicle-treated animals exhibited profound brain infection (see Fig. S2 in the supplemental material), generally mirroring the virus tissue titers. This finding of later onset of brain infection was previously reported. Infected cells were also found in the sinonasal cavity, with nafamostat-treated mice showing a trend toward fewer SARS-CoV-2-positive cells in the maxillary sinus and olfactory epithelium (Fig. S2). Very few lung lesions were observed in either treatment group (see Fig. S3 in the supplemental material), making it difficult to assess whether nafamostat treatment reduced lung disease severity in K18-hACE2 mice.

12Here, we show that the serine protease inhibitors camostat and nafamostat potently reduce SARS-CoV-2 and MERS-CoV infection in well-differentiated primary cultures of airway epithelia, presumably by inhibiting the activity of cell surface serine proteases (such as TMPRSS2). Both camostat and nafamostat were previously shown to directly inhibit the enzymatic activity of TMPRSS2 and related serine proteases in biochemical assays, strongly suggesting that inhibition of TMPRSS2 catalytic activity is the primary mechanism for this effect. In contrast, we saw little or no effect from bafilomycin A1, chloroquine, or E64d, agents that alter pH and/or protease function in intracellular compartments, including endosomes and lysosomes. Our results agree with those of earlier studies and contribute to the growing consensus that fusion and entry at the plasma membrane is the preferred route of entry into cells of the respiratory tract. Our experiments in Calu-3 2B4 cells indicate that while both camostat and nafamostat are active against SARS-CoV-2 and MERS-CoV, nafamostat is more potent, a trend also observed in other in vitro studies.

13Importantly, we demonstrate the in vivo efficacy of nafamostat in reducing SARS-CoV-2 infection and pathogenesis. The protective effect of nafamostat was greatest when the drug was administered prior to viral infection. Pretreatment with nafamostat via the i.n. route either completely prevented or significantly reduced infection-induced weight loss, and substantially reduced viral loads throughout the ensuing course of illness, in Ad5-hACE2 transduced mice and K18-hACE2 mice. In Ad5-hACE2 transduced mice, although nafamostat treatment beginning at 1 day postinfection provided some protection, the results were less dramatic. These findings suggest that protease inhibitor treatment may provide its greatest clinical benefit when delivered prophylactically or in the early stages of infection, and that there may be a “treatment window” after which treatment no longer improves outcomes. We note that the course of SARS-CoV-2 infection in hACE2-expressing mice proceeds more rapidly and on a shorter time course than in humans.

14Further highlighting the importance of timing for the in vivo efficacy of nafamostat, we observed that the effectiveness of intranasal nafamostat increased as the time interval between nafamostat delivery and viral inoculation decreased (Fig. 2C). This likely reflects the relatively short half-life of nafamostat; early studies with nafamostat reported a plasma half-life of 8 min in rabbits and 1 min in dogs. Currently, there are no data regarding nafamostat stability in respiratory secretions following i.n. administration. It is possible that the fate of nafamostat is different in airway secretions than that in plasma, potentially contributing to the different outcomes observed via i.n. or i.p. routes in our study. It is also unknown how efficiently nafamostat is transported into airway secretions when delivered systemically, which may influence outcomes following i.n. versus i.p. administration. Pharmacokinetic studies are needed to better understand these aspects of nafamostat activity.

15Based on their encouraging in vitro activity against SARS-CoV-2, both camostat and nafamostat are currently under evaluation as potential therapies for COVID-19 (https://clinicaltrials.gov; NCT04652765, NCT04455815, NCT04353284, NCT04583592, NCT04470544, NCT04435015, NCT04608266, NCT04524663, NCT04750759, NCT04625114, NCT04730206, NCT04321096, NCT04355052, NCT04662073, NCT04681430, NCT04644705, NCT04657497, NCT04374019, NCT04418128, NCT04352400, NCT04390594, NCT04628143, NCT04623021, NCT04473053, NCT04483960). Both compounds are approved treatments for other medical conditions and are therefore attractive candidates for rapid drug repurposing. In Japan, camostat is approved for use in treatment of acute pancreatitis and postoperative reflux esophagitis, and it has a well-characterized safety profile. Camostat treatment was shown to improve survival in a mouse model of SARS-CoV infection. Nafamostat is marketed in Japan and South Korea for treatment of acute pancreatitis and disseminated intravascular coagulation (DIC). In vitro studies indicate that nafamostat does not cause cytotoxicity in cultured human endothelial or airway epithelial cells, and a recent case report describing nafamostat administration in three elderly COVID-19 patients reported no adverse events. The proposed clinical trials generally involve systemic administration of nafamostat. In our studies with the Ad5-hACE2 mice, we observed that nafamostat reduced infection more effectively when delivered via the i.n. route, suggesting that it may be important to consider the route of administration when designing treatment regimens with protease inhibitors in human patients. It is possible that nafamostat’s therapeutic efficacy might be boosted by direct delivery to the airways as a nasal spray or inhaled aerosol. It is of note that nafamostat is a broad-spectrum protease inhibitor with effects on multiple biological processes, which has led to speculation that it may confer benefits beyond blocking SARS-CoV-2 entry. In particular, its anticoagulant properties may reduce or prevent COVID-19-related thrombotic complications. Nafamostat also inhibits proteases involved in inflammatory cascades and the complement system, which may dampen inflammation in severe COVID-19 cases.

16In conclusion, we provide evidence that camostat and nafamostat potently inhibit SARS-CoV-2 and MERS-CoV infection in cultured human airway epithelia; nafamostat exhibited greater potency than camostat in reducing SARS-CoV-2 infection, suggesting that it may be a more attractive candidate for COVID-19 lung disease prevention or treatment. Nafamostat inhibited SARS-CoV-2 infection and improved disease outcomes in two COVID-19 mouse models. Our experiments in these animal models highlight the importance of route and timing of administration in the design of effective treatment regimens. These preclinical data support further investigation of protease inhibitors as antiviral prophylactic or therapeutic strategies for COVID-19.

17Camostat mesylate, nafamostat mesylate, bafilomycin A, chloroquine, and E64d were purchased from the Sigma-Aldrich Corporation (St. Louis, MO). The EMC/2012 strain of MERS-CoV was provided by Bart Haagmans and Ron Founchier (Erasmus Medical Centre, Rotterdam Netherlands). The USA-WA1/2020 strain of SARS-CoV-2 was obtained from the BEI Resources Repository (https://www.niaid.nih.gov/research/bei-resources-repository; catalog no. NR-52281).

18Primary human airway epithelia were prepared from bronchi as previously described. Briefly, epithelial cells were dissociated and seeded onto collagen-coated, semipermeable membranes with a 0.4-μm pore size (Costar Transwell, surface area, 0.33 cm2; Corning) in 24-well plates maintained in Ultroser G (USG) medium at 37°C and 5% CO2. At 24 hours after seeding, the mucosal medium was removed, and cells were grown at the air-liquid interface. Only well-differentiated cultures (>3 weeks old; resistance, >1,000 Ω · cm2) were used in this study. Calu-3 2B4 cells were maintained in minimal essential medium (MEM) supplemented with 20% fetal bovine serum (FBS), 0.1 mM nonessential amino acids (NEAA), 1 mM sodium pyruvate, 2 mM l-glutamine, 1% penicillin and streptomycin, and 0.15% NaHCO3 at 37°C with 5% CO2. Vero E6 cells were grown in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS, 0.1 mM NEAA, and 1% penicillin and streptomycin at 37°C in 5% CO2.

19To investigate proteases required for MERS-CoV or SARS-CoV-2 infection in primary human airway epithelia, cells were incubated in medium (50 μl in the apical compartment and 500 μl basolaterally) containing bafilomycin A1 (50 nM), chloroquine (20 μM), E64d (25 μM), camostat (25 μM), or nafamostat (25 μM) at 37°C for 1 h. After 1 h of pretreatment, the apical medium was removed and replaced by medium containing MERS-CoV or SARS-CoV-2 (MOI of 0.1) in the presence of the indicated inhibitor/chemical for another 1 h incubation. The apical medium (containing unbound virus) was then removed, and cells were washed 2 times with phosphate-buffered saline (PBS) at the apical surface. At 20 h postinfection, the apical surface of infected cultures was rinsed with PBS to collect airway surface liquid (ASL), and titer was determined to verify the release of progeny virions into the ASL. Total cellular RNA was harvested in TRIzol reagent (Invitrogen, Waltham, MA).

20To compare the efficacy of camostat and nafamostat against MERS-CoV or SARS-CoV-2 infection, Calu-3 2B4 cells were cultured in 96-well plates and pretreated with the indicated concentrations of inhibitors for 1 h. Cells were then infected with MERS-CoV or SARS-CoV-2 (MOI = 0.1) in the presence of the inhibitors for 1 h, followed by overnight incubation with the inhibitor. Total cellular RNA was harvested in TRIzol reagent at 20 h postinfection.

21Ad5-hACE2 was generated by the University of Iowa Viral Vector Core Facility. Six- to eight-week-old BALB/c mice were lightly anesthetized with ketamine-xylazine and transduced via the i.n. route with 2.5 × 108 PFU of Ad5-hACE2 in 75 μl DMEM. At 5 days postransduction, mice were infected i.n. with SARS-CoV-2 (3 × 103 or 1 × 105 PFU, as indicated). To make a stock of nafamostat for in vivo studies, the compound was dissolved in H2O at a concentration of 10 mg/ml. This nafamostat stock (or H2O for vehicle control animals) was diluted in PBS prior to i.p. injection; for i.n. delivery, the nafamostat stock was diluted in DMEM and delivered as a liquid bolus in a total volume of 50 μl. After SARS-CoV-2 infection, mice were monitored and weighed daily. All work with SARS-CoV-2 was conducted in the biosafety level 3 (BSL3) Laboratory of the University of Iowa. All protocols were approved by the Institutional Animal Care and Use Committees of the University of Iowa.

22Transgenic mice expressing human ACE2 under the control of the cytokeratin 18 promoter were previously reported. The 6- to 8-week-old mice used in these studies were obtained from the Jackson Laboratory [034860-B6.Cg-Tg(K18-ACE2)2Prlman/J] and are congenic on the C57BL/6 background.

23Viral preps and lung or brain homogenate supernatants were serially diluted in DMEM. Vero E6 cells in 12-well plates were inoculated at 37°C in 5% CO2 for 1 h with gentle rocking every 15 min. After removing the inoculum, wells were overlaid with 1.2% agarose containing 4% FBS. After further incubation for 3 days, overlays were removed, and plaques were visualized using 0.1% crystal violet stain. Viral titers were calculated as PFU per lung or brain, as indicated.

24Total cellular RNA was isolated using the Direct-zol RNA miniprep kit (Zymo Research, Irvine, CA) following the manufacturer’s protocol and including a DNase treatment step. Total RNA (200 ng) was used as the template for first-strand cDNA synthesis. The resulting cDNA was used to quantify the MERS-CoV or SARS-CoV-2 RNA levels by real-time quantitative PCR using Power SYBR green PCR master mix (Applied Biosystems, Waltham, MA). Average values from duplicates of each sample were used to calculate the viral RNA level relative to the HPRT gene and presented as 2−ΔCT or 2−ΔΔCT, as indicated (where CT is the threshold cycle). The primers used were as follows: MERS-CoV-F, 5′-CCACTACTCCCATTTCGTCAG-3′, and MERS-CoV-R, 5′-CAGTATGTGTAGTGCGCATATAAGCA-3′; 2019-nCoV-F, 5′-GACCCCAAAATCAGCGAAAT-3′, and 2019-nCoV-R, 5′-TCTGGTTACTGCCAGTTGAATCTG-3′; and hHPRT-F, 5′-AGGATTTGGAAAGGGTGTTTATTC-3′, and hHPRT-R, 5′-CAGAGGGCTACAATGTGATGG-3′.

25Mice were anesthetized and perfused transcardially with PBS. Tissues (lungs, brain, and nasal cavity) were harvested and fixed in 10% neutral buffered formalin (for 7 days), nasal cavities were decalcified in EDTA, and then all tissues were dehydrated through a series of alcohol and xylene baths, paraffin embedded, sectioned at ∼4 μm, and stained with hematoxylin and eosin (HE) stain. Serial sections were immunostained using a rabbit monoclonal antibody (catalog no. 40143-R019; Sino Biological, Beijing, China) against SARS-CoV-2 N protein, as previously described. Tissues were examined by a board-certified veterinary pathologist using a postexamination method of masking and following principles for reproducible tissue scores. Immunostaining of SARS-CoV-2 infection in the lung was scored using distribution-based ordinal scores: 0, absent; 1, <25%; 2, 26 to 50%; 3, 51 to 75%; and 4, >75% of lung fields.

26Results are reported as mean ± standard error (SE). Data were tested for significant differences using Student’s t test, the Mann-Whitney test, and one-way analysis of variance (ANOVA) followed by Tukey’s or Dunnett’s tests of multiple comparisons, or by 2-way ANOVA followed by Dunnett’s or Sidak’s posttests, as indicated. All statistical tests were performed using GraphPad Prism 7. P values of <0.05 were considered statistically significant (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).