A novel class of TMPRSS2 inhibitors potently block SARS-CoV-2 and MERS-CoV viral entry and protect human epithelial lung cells

The host cell serine protease TMPRSS2 is an attractive therapeutic target for COVID-19 drug discovery. This protease activates the Spike protein of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) and of other coronaviruses and is essential for viral spread in the lung. Utilizing rational structure-based drug design (SBDD) coupled to substrate specificity screening of TMPRSS2, we have discovered a novel class of small molecule ketobenzothiazole TMPRSS2 inhibitors with significantly improved activity over existing irreversible inhibitors Camostat and Nafamostat. Lead compound MM3122 (4) has an IC50 of 340 pM against recombinant full-length TMPRSS2 protein, an EC50 of 430 pM in blocking host cell entry into Calu-3 human lung epithelial cells of a newly developed VSV SARS-CoV-2 chimeric virus, and an EC50 of 74 nM in inhibiting cytopathic effects induced by SARS-CoV-2 virus in Calu-3 cells. Further, MM3122 blocks Middle East Respiratory Syndrome Coronavirus (MERS-CoV) cell entry with an EC50 of 870 pM. MM3122 has excellent metabolic stability, safety, and pharmacokinetics in mice with a half-life of 8.6 hours in plasma and 7.5 h in lung tissue, making it suitable for in vivo efficacy evaluation and a promising drug candidate for COVID-19 treatment.


Introduction
Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) is the newly emerged, highly transmissible coronavirus responsible for the ongoing Coronavirus Disease 2019 (COVID- 19) pandemic, which is associated with 136 million cases and almost 3 million deaths worldwide as of April 12, 2021 (https://coronavirus.jhu.edu/map.html). While three vaccines have recently been approved by the FDA, there are still no clinically approved small molecule drugs available for the treatment of this disease except Remdesivir and the effectiveness of the vaccines against immune escape variants might be reduced. Multiple therapeutic strategies have been proposed 2, 3 , including both viral and host proteins but none have yet been fully validated for clinical application. One class of protein targets which have shown promising results are proteolytic enzymes including the viral proteases 2, 4-6 , Papain-Like Protease (PLpro) and the 3C-like or 'Main Protease' (3CL or MPro), and several host proteases involved in viral entry, replication, and effects on the immune system creating the life-threatening symptoms of COVID-19 infection. [5][6][7] .
The latter include various members of the cathepsin family of cysteine proteases including cathepsin L, furin, and the serine proteases factor Xa, plasmin, elastase, tryptase, TMPRSS2 and TMPRSS4.
TMPRSS2 8 is a type II transmembrane serine proteases (TTSP) 9 that has been shown to be crucial for host-cell viral entry and spread of SARS-CoV-2 [10][11][12] , as well as SARS-CoV 13,14 , Middle East Respiratory Syndrome Coronavirus (MERS-CoV) 15 and influenza A viruses [16][17][18][19][20][21][22][23] . Like SARS-CoV and MERS-CoV, SARS-CoV-2 cell entry involves binding of the viral Spike protein to the host cell receptor Angiotensin Converting Enzyme-2 (ACE2). The Spike protein requires proteolytic processing/priming by TMPRSS2 to mediate entry into lung cells, thus small molecule inhibitors of this target offer much promise as new therapeutics for COVID-19 and other coronavirus diseases 10,11 . It has been demonstrated that the TMPRSS2-expressing lung epithelial Calu-3 cells are highly permissive to SARS-CoV-2 infection. The irreversible serine protease inhibitors Camostat 10 and Nafamostat 24 are effective at preventing host cell entry and replication of SARS-CoV-2 in Calu-3 cells through a TMPRSS2-dependent mechanism.
Herein, we report on the discovery of a new class of substrate-based ketobenzothiazole (kbt) inhibitors of TMPRSS2 with potent antiviral activity against SARS-CoV-2 and which are significantly improved over Camostat and Nafamostat. Several compounds were found to be strong inhibitors of viral entry and replication, with EC50 values exceeding the potency of Camostat and Nafamostat and without cytotoxicity. Newly developed compound MM3122 (4) has excellent pharmacokinetics (PK) and safety in mice and is thus a promising new lead candidate drug for COVID-19 treatment.

Hit identification of TMPRSS2 inhibitors
It is well established that several TTSPs including TMPRSS2 not only play a role in infectious diseases 8 , but also cancer progression and metastasis [25][26][27] which is thought to be mainly through its ability to activate hepatocyte growth factor (HGF), the sole ligand for MET receptor tyrosine kinase. This is accomplished via proteolytic processing of the inactive single-chain precursor pro-HGF to a two-chain active form. TMPRSS2 shares pro-HGF as a protein substrate with the other HGF-activating serine proteases, HGFA (HGF-Activator), hepsin and matriptase. 28,29 Like TMPRSS2, other TTSPs such as matriptase and hepsin have a canonical serine protease domain residing as the C-terminal domain of the protein that is anchored to the cell membrane by an Nterminal type II signal peptide thereby presenting their enzymatic activity outside the cell. We have previously reported on the discovery and anticancer properties of peptidomimetic ketothiazole (kt) and ketobenzothiazole (kbt) inhibitors of HGFA, matriptase and hepsin [30][31][32][33][34] named synthetic HGF Activation Inhibitors or sHAIs. Since TMPRSS2 has an overlapping endogenous substrate specificity profile with HGF-activating proteases, we postulated that our substrate-based sHAIs would also inhibit TMPRSS2.

Antiviral activity of sHAIs (VSV/SARS-CoV-2 Spike Protein Pseudotypes):
Based on our premise that sHAIs would inhibit TMPRSS2, we first selected the two sHAI lead compounds, ZFH7116 (1) 32, 35 and VD2173 (2) 36 (Figure 1) for initial antiviral testing and confirmed that they potently inhibit TMPRSS2-dependent host cell viral entry driven by the spike protein of SARS-CoV-2 (SARS-2 S, Figure 2A) into Calu-3 lung epithelial cells having an EC50 of 307 and 104 nM, respectively. Marked inhibition of SARS-2-S-driven entry was also detected for the irreversible inhibitor Camostat, in line with published data 10,24 . It is noteworthy that neither compound showed activity in TMPRSS2 negative Vero cells or modulated entry into Calu-3 cells by pseudotypes bearing the VSV glycoprotein (VSVpp VSV-G) (Supplementary data) suggesting that the inhibition of SARS-CoV-2 spike-driven entry into Calu-3 cells was due to blockade of TMPRSS2.

Antiviral activity of sHAIs (VSV/SARS-CoV-2 Spike Protein Chimeras):
Using a replicationcompetent, chimeric VSV expressing the SARS-CoV-2 Spike (shown to mimic the Spike-dependent entry of authentic SARS-CoV-2 37 we also demonstrated that 1 and 2 block viral entry into Calu-3  and allows for the establishment of a system for screening antiviral activity of these inhibitors using VSV-SARS-CoV-2.

Inhibition of cell bound TMPRS2 enzymatic activity by sHAIs
We tested the inhibitory activity of 1 and 2 on TMPRSS2 proteolytic activity in a cell-based enzyme-based fluorogenic assay, by overexpression of TMPRSS2 in a human cell line, HEK-293T, commonly used for experimentation because of its high transfectability. As previously demonstrated, expression of TMPRSS2 can be accurately measured 38, 39 using a fluorogenic peptide reporter substrate, Boc-QAR-AMC in cell cultures. 1 inhibited cell based TMPRSS2 enzyme activity in a concentration dependent manner between 10 µM to 10 nM with an IC50 of 314 nM (Supplementary data). 2 was a more potent inhibitor of TMPRSS2 proteolytic activity with IC50 of 57 nM. These data confirm that both 1 and 2 mediate their function by potently inhibiting TMPRSS2.
Selectivity data of TMPRSS2 inhibitors in a panel of proteases. We profiled 1 and 2 for their selectivity against a panel of 43 serine and cysteine proteases (Supplementary material) for comparison to that of Camostat and Nafamostat protease selectivity data which has been published 40 . We found 1 was a potent inhibitor of matriptase-2, plasma kallikrein, proteinase K, trypsin, tryptase b2 and G1 but also was a moderate inhibitor of factor Xa, factor XIIa, kallikrein 5 and 14 as well as cysteine protease cathepsin S while showing some activity against cathepsin L as well. There was overlap of 1 with the selectivity profiles of Camostat except they did not show any activity against the cathepsins. Furthermore, Camostat more potently inhibited factor XIIa, plasma kallikrein, matriptase-2, and plasmin while also having increased activity for trypsin, the tryptases and urokinase. Cyclic peptide 2 also more potently inhibited urokinase and factor XIIa relative to 1 but not cathepsin S or L, and instead was a moderate inhibitor of cathepsin B.
The ramifications of these different selectivity profiles may explain the varied activity in the VSVpseudotype and chimeric viral entry results most notably the urokinase potency for which 1 was significantly weaker.

Lead identification of TMPRSS2 inhibitors
We published 33 that Camostat and Nafamostat ( Figure 3B) are inhibitors of matriptase and hepsin, and analog Nafamostat is the most active published inhibitor of SARS-CoV-2 cell entry 11 .
While we previously pursued this chemical series as HGFA, matriptase and hepsin inhibitors 33 , the best series we have developed are small peptide-based molecules like 1, 2 and 3 (Ac-SKLRkbt; Figure 3B) which exhibit low nM to picomolar IC50s. These compounds contain a serine trapping ketobenzothiazole (kbt) warhead [30][31][32]34 which reacts covalently with the protease, but importantly in a reversible manner, unlike Camostat or Nafamostat in which the inhibition is irreversible. Therefore, we pursued the kbt class of inhibitors for lead identification studies towards more potent and selective TMPRSS2 inhibitors.

Rational
design of novel TMPRSS2-selective

ketobenzothiazole (kbt) inhibitors
We have employed X-ray co-crystal structure data in the rational design of optimized HGFA, matriptase and hepsin inhibitors with increased potency and selectivity [30][31][32][33] . Figure 3, here we used a homology model of human TMPRSS2 41 to computationally model our existing inhibitors (using Glide in Schrödinger) where we docked compound 3 (yellow; Figure 3A) and Nafamostat (grey).

Shown in
For Nafamostat, the naphthyl portion extends deep into the S2 pocket with the benzguanidine in the S1 pocket binding to the conserved S1 Asp189. The P4 Ser of 3 makes two H-bonds to Asp162 and Asn163 outside the S4 pocket, and the Ser backbone carbonyl is also within H-bonding distance to the P2 Lys98. The P3 Lys is near the S4 Glu134 making an electrostatic interaction while the P2 Leu resides in the S2 pocket and the benzothiazole fills the S1ʹ area. The reactive ester of Nafamostat and ketone of the kbt are adjacent to the Ser195-His57-Asp102 catalytic triad. Our model of 3 and Nafamostat bound to TMPRSS2 ( Figure 3A) shows a nice fit to the S1-S4 and S1ʹ pockets of TMPRSS2. For 3, it appears that the P4 Ser is making a dual H-bond to the Asp162 and Asn163 residues in TMPRSS2. This area is occupied by a Gln in both hepsin and matriptase but an Asp in HGFA and interestingly the additional Asn163 residue is unique to TMPRSS2. This suggests that a free amine in the P4 could provide selectivity over matriptase and hepsin. The structure also shows that Lys87 resides in the S2 pocket, suggesting a P2 Glu or Asp might be ideal for an inhibitor of TMPRSS2 and lead to potential selectivity over HGFA (P2 Ser), matriptase (P2 Phe) and hepsin (P2 Asn). This Lys87 is also near the P4 Ser sidechain indicating an Asp or Glu in this position could potentially make a third electrostatic interaction to Lys87. Further, the S2 pocket is large and appears that it could accommodate larger sidechains such as Phe, Tyr, or Trp.

PS-SCL Protease Substrate Specificity Profiling of TMPRSS2
To augment our rational design of more potent and selective substrates. This data reveals TMPRSS2 is tolerant of many different P2 sidechains but prefers Phe and Ala/Thr like matriptase (Figure 4) which can also tolerate both large and small groups (prefers Ser/Ala) but hepsin and HGFA both prefer Leu. For P3 TMPRSS2 prefers Gln/Glu and Met whereas it is Lys/Gln for hepsin and matriptase and Lys/Arg for HGFA (Figure 4). The clearest distinction is in the P4 position where HGFA, matriptase and hepsin all prefer basic residues like Lys/Arg while TMPRSS2 prefers Ile/Gly and Pro, which is shared attribute with hepsin. Using protease substrate terminology, the cleavage site for protease protein substrates is between the P1 and P1ʹ positions of the Peptide substrate, while S1, S1ʹ refers to the Subsite 1 of the protease where the P1, P1' amino acid sidechain binds.

MSP-MS profiling of TMPRSS2 for extended substrate sequence specificity
To strengthen our compound design, we obtained further information on the extended substrate specificity of TMPRSS2 using multiplex substrate profiling by mass spectrometry (MSP-MS). 44 In the MSP-MS assay, a physiochemically diverse library of 228 tetra decapeptides was incubated for several hours with human recombinant TMPRSS2 in activity buffer. At different time points (15,60 and 240 minutes) aliquots of the reaction mixture were extracted, quenched with 8M guanidium HCl and analyzed by tandem mass spectrometry to allow monitoring of proteasegenerated cleaved products. As a TTSP with a trypsin fold, TMPRSS2 is known for its high affinity towards substrates containing an arginine residue in the P1 position. 25 This prominent P1 specificity was confirmed by our MSP-MS analysis, which showed a nearly exclusive occupancy of the P1 position by either Arg or Lys on TMPRSS2-generated cleaved products ( Figure 5A).

Peptide sequencing by LC-MS/MS enabled the identification of the 25 most preferred substrates
for TMPRSS2 in our peptide library ( Figure 5B). An IceLogo frequency plot ( Figure 5C) displays the extended substrate specificity profile of TMPRSS2 at pH 7.5, which reveals the preference for hydrophobic amino acids in both P2 and P1' positions, which flank the cleavage site. The analysis indicates a preferred cleavage of N-terminal tetrapeptides to be PLFR with other P4 amino acids H and M and P3 amino acids G, Y, V, and Q but only F at P2 which is striking. Largely this data recapitulates what was elucidated in the PS-SCL study (Figure 4). Based on careful analysis of the computational modeling work with these and the PS-SCL results, we selected additional existing compounds in our library of HGFA/matriptase hepsin inhibitors for testing and also synthesized four new analogs specifically designed for TMPRSS2: Ac-GQFR-kbt (4), Ac-PQFR-kbt (5), Ac-QFRkbt (6) and Ac-IQFR-kbt (7).
The majority of these compounds contain a Phe (F) residue in the P2 position as predicted from the PS-SCL study while others contain a Gln (Q) at the P3 position. When tested in the VSV pseudotype assay all displayed improved activity relative to 1 ( Table 1) where the best compound was 13. 13 is a tripeptide containing an unnatural amino acid D-Trp in the P3 position and Phe in the P2 position, predicted from PS-SCL and MSP-MS analysis to be preferred. 17 which contains a P3 Gln also predicted from PS-SCL shows the next best activity with an EC50 of 78 nM. The remaining compounds were tested in the VSV SARS CoV-2 chimeric assay. Excitingly, the most promising results ( Table 1)

Inhibition of VSV-MERS viral entry
To demonstrate broad coronavirus activity, we tested Camostat, 2 and our best new compounds, 4-7 (Figure 6) based on the VSV chimera results for their activity in blocking host cell entry of a MERS-CoV chimeric virus. Shown in Table 2, all compounds inhibited MERS-CoV viral entry with nearly equivalent potency as compared to SARS-CoV-2 ( Table 1). Though the TMPRSS2dependence of MERS-CoV has been demonstrated previously, this provides strong evidence that our compounds have broad-spectrum potential against other coronaviruses in the clinic. To confirm that this activity is indeed TMPRSS2-related, we next set out to produce recombinant TMPRSS2 protein for testing our compounds for their enzyme inhibitory activity.

Recombinant TMPRSS2 (Full-length) Enzyme Assay
Using active recombinant full-length TMPRSS2 as described above, along with Boc-QAR-AMC as a fluorogenic substrate 40 , we found the Km was 85.6 µM using an enzyme concentration of 3 nM (Supplementary material). We used this substrate to test all inhibitors for inhibition of TMPRSS2 proteolytic enzyme activity in a standard kinetic assay using Nafamostat and Camostat as controls, where we determined the compound IC50s over the period of one hour, following 30 min compound incubation with enzyme. The IC50 values for Camostat and Nafamostat are 1.5 nM and 0.14 nM which is similar but slightly improved compared to values previously reported of 6.2 nM and 0.27 nM, respectively. 40 We found the IC50 data we generated for TMPRSS2 inhibition closely correlated with the VSV pseudotype and chimera assay data with some exceptions.
Multiple compounds were significantly more potent than Camostat and several were equipotent to Nafamostat ( Table 1). The new rationally designed lead compounds 4-7 (Figure 6) all displayed exquisitely potent sub nanomolar IC50 values of which 7 was the best with an IC50 of 250 picomolar similar to that of Nafamostat. It should be noted however that Nafamostat is an irreversible inhibitor while the kbt class of inhibitors are reversible, so it is difficult to directly compare IC50 values between the two series having different mechanism of inhibition. While the initial hit compound 1 still shows good potency against TMPRSS2 (IC50 74 nM), it was significantly weaker when compared to the rationally designed TMPRSS2 inhibitors which was expected. However, the other initial hit compound 2 is a much more potent inhibitor of TMPRSS2 activity than 1 with an IC50 of 2.6 nM, only about 10-fold less active than 4-7. It is noteworthy that the P3-P1 tripeptide 6 without the P4 sidechain was almost as active as either of the tetrapeptides 4, 5 or 7 which have a P4 residue. However, this compound 6 and the other tripeptides including 8 and 13 suffer from high inhibition of the plasma protease Factor Xa which is undesirable for further drug discovery due to potential bleeding side effects in patients.
The least active compound was the aryl ether cyclic peptide 21 with an IC50 of 197 nM which is however still respectable. It is presumed the larger, more constrained aryl ether ring system is not an ideal fit for binding to the TMPRSS2 active site through bridging the S2-S4 pockets. It is important to note that while TMPRSS2 activity is potent in all inhibitors, most compounds tested are all still relatively active against the other proteases HGFA, matriptase and hepsin.
The most selective TMPRSS2 analogs are 16 with an IC50 of 6.3 nM, being 40-fold more active than HGFA, 2-fold over matriptase and 12 which is 60-fold more selective over HGFA and 2-fold against matriptase. As with other compounds in this series of peptidyl kbt inhibitors, in our experience we have found that deriving selectivity over hepsin is challenging but one of our best examples turns out to be initial hit 2, which is 4-fold more active against TMPRSS2 relative to hepsin and almost 10-fold more active against matriptase. The consequences of inhibiting these other serine proteases, especially the other HGF-activating proteases, in the COVID-19 scenario is unclear at this time but may be beneficial for treatment rather than detrimental.
There are clear structure-activity relationships (SAR) from the 21 compounds tested. For example, it appears that TMPRSS2 prefers large groups extending beyond the kbt S1' C-terminal portion of the inhibitor since 1 have increased activity relative to unsubstituted kbt analog 3.
Strengthening this hypothesis, the compounds with the smaller ketothiazole (kt) warhead in the P1' position like seen in 9-11 also lose potency relative to analogs with the larger kbt. Also, the P3 position is important for activity where it seems basic groups like Lys (K) are undesired (9, 3, 1) which seems also true for the P4 position as seen with compound 17. Finally, as predicted from modeling and the PS-SCL data, Phe is clearly preferred in the P2 position. Therefore, it is likely that diverse new analogs can be pursued with other aromatic sidechains like Trp and Tyr which will also produce inhibitors with exquisite TMPRSS2 activity and potentially more selectivity.

Antiviral activity of lead compounds (Wild-type SARS-CoV-2)
After confirming and quantitating TMPRSS2 enzyme inhibition of all compounds against recombinant protein and their selectivity over HGFA, matriptase, hepsin, thrombin, and Factor

Figure 8. A) Inhibition by lead TMPRSS2 inhibitors of Calu-3 viral toxicity of wild-type SARS-CoV-2 using the CellTiter-Glo (Promega) assay. B) Viability of Calu-3 cells with compounds in the absence of virus.
Xa, we assessed the activity of our most promising lead compounds 2, 4-7 towards wild-type SARS-CoV-2 by employing a CellTiter-Glo (Promega) cellular viability assay using lung epithelial Calu-3 cells. 45 We used Remdesivir and Nafamostat (n=1) as a positive control. While all compounds showed excellent activity in this assay (Figure 8 and Table 1), the most potent compounds were the rationally designed TMPRSS2 inhibitors 4 and 5 (Figure 6) with EC50s of 74 and 52 nM, respectively which is >20-fold improved over Remdesivir and also significantly more active than Nafamostat (note: based on n=1). Importantly, all five kbt inhibitors showed no signs of toxicity to Calu-3 cells (up to 50 µM, which is nearly 1000 times the IC50 for 5, while Remdesivir (and Nafamostat) displayed toxicity at the highest concentration tested (50 µM).

Pharmacokinetics and metabolic stability of VD2173 (2) and MM3122 (4)
VD2173 (2) and lead compounds 4-7 were tested for their in vitro stability in mouse and human blood plasma ( Table 2) was selected as a lead candidate and was tested for its in vivo pharmacokinetics (PK) in mice. The mouse PK of VD2173 in plasma has been published elsewhere. 36 Shown in Figure 9 in both the TMPRSS2 enzyme assay and viral assays is magnitudes higher (500-fold) making MM3122 an ideal lead candidate drug suitable for in vivo efficacy studies of COVID-19 and further optimization to be reported in a future communication.

Acute 7-day toxicity study of MM3122 (4) in mice
We further tested lead compound MM3122 for its safety in mice. MM3122 was administered daily to mice at three different single dose levels of 20, 50, and 100 mg/kg via intraperitoneal (IP) injection over a period of 7 days. No adverse effects were observed in any of the treatment groups and no weight loss or changes to harvested organs (liver, spleen, and kidneys) were noted compared to the control group (Supplementary material).

Conclusions
TMPRSS2 has been shown to be essential for host cell viral entry and replication of SARS-CoV-2.
Based on molecular docking studies using a published homology model of TMPRSS2 41 and substrate specificity data from PS-SCL, we hypothesized that a set of our existing peptidyl ketobenzothiazole inhibitors of HGFA, matriptase and hepsin would also inhibit TMPRSS2.
Indeed, we demonstrated that these compounds not only potently inhibit TMPRSS2 enzyme activity of recombinant and cell-surface protein, but also host cell entry and toxicity driven by the but also potentially represent broad-spectrum antivirals.

TMPRSS2-Protease domain expression and purification
Periplasm secreted bacterial expression of serine proteases has been widely reported to yield correctly folded high-quality enzyme due to its reducing environment allowing for disulfide formation and fewer proteases being present compared to the cytoplasm. 53, 54 We therefore

*Expression of Human Recombinant TMPRSS2 in Pichia pastoris
Human Recombinant TMPRSS2 was expressed and purified from yeast following protocols previously described 25

Antiviral activity (VSV SARS-CoV-2 and MERS chimeras):
Viruses used: VSV-eGFP, a recombinant VSV expressing a GFP reporter (depends on the VSV glycoprotein G for entry), has been previously described. 58 VSV-SARS-CoV-2, a replication competent infectious VSV chimera which employs the SARS-CoV-2 Spike (S) protein for viral entry in place of VSV G and expresses eGFP, has also been previously described. 37

Antiviral activity (Wild-type SARS-CoV-2):
We seeded Calu-3 cells onto a white, flat bottom 96-well plate at 10,000 cells per well in Dulbecco's Modified Eagle Medium supplemented with 10% heat-inactivated fetal bovine serum, 100 U/mL Penicillin-Streptomycin (Life Technologies, Inc., 15140-163), and buffered with 10 mM HEPES and incubated in a humidified 5% CO2 incubator at 37°C. 16 hours later, the media was removed and replaced with the 50 µl/well of the same media but containing 2% heat-inactivated fetal bovine serum (D-2) instead of 10%. In D-2, compounds were serially diluted 3-fold in a 9point series and added in a 25 ul volume to the 96-well plate to make final concentrations ranging from 50 µM to 7.6 nM on the assay plate for one hour. We then infected the cells (with compounds still on them) with 25 ul/well of SARS-CoV-2 at an MOI of 0.1 PFU/cell, with a final total volume of 100 µl/well. We incubated the plates for 72 hours as described above and then added 25 μL/well of CellTiter-Glo® Reagent (prepared as directed by manufacturer, Promega, G7573), followed by 5 min shaking and 20 min room temperature incubation before detecting luminescence with a Biotek Synergy H1 plate reader. Compound cytotoxicity was determined using the same assay, but instead of adding virus, we added 25 µl of D-2. These experiments were performed three times, and we performed nonlinear regression on log(inhibitor) vs response with variable slope and four parameters on compiled results from the three experiments.
Likewise, IC50s were calculated based on all three biological replicates.

Animal studies
All animal procedures were performed following the guidelines and approval of the Washington University's Institutional Animal Care and Use Committee. Animals were maintained in a controlled temperature (22°C± 0.5°C) and lighting (12L:12D) environment. Standard laboratory chow and water were supplied ad libitum.

Mouse toxicity studies
NSG male and female mice (3 per group) were injected with 0, 20, 50, or 100 mg/kg MM3122 (in 1-5% DMSO/95% saline; IP) daily for 7 days. Animal health was assessed daily for drug toxicity (ruffled fur, discharge from the eyes, weight loss, dehydration, lethargy, hypothermia, abnormal tissue growth) and body weights were measured every 2-3 days. At the end of the experiment animals were sacrificed and organs inspected. Liver, spleen, and both kidneys were collected and weighed.

Mouse pharmacokinetic (PK) studies
MM3122 was tested for its PK in mice (single IP dosing, actual dose 16.7 mg/kg). Eight animal groups contained three animals each (male, CD-1) with plasma drawn and lung tissue harvested