key: cord-0859616-eznsnflj
authors: Fraser, Bryan J.; Beldar, Serap; Seitova, Almagul; Hutchinson, Ashley; Mannar, Dhiraj; Li, Yanjun; Kwon, Daniel; Tan, Ruiyan; Wilson, Ryan P.; Leopold, Karoline; Subramaniam, Sriram; Halabelian, Levon; Arrowsmith, Cheryl H.; Bénard, François
title: Structure, activity and inhibition of human TMPRSS2, a protease implicated in SARS-CoV-2 activation
date: 2021-06-24
journal: bioRxiv
DOI: 10.1101/2021.06.23.449282
sha: f42578b823280a3931165040e8a79b23470df0de
doc_id: 859616
cord_uid: eznsnflj

Transmembrane protease, serine 2 (TMPRSS2) has been identified as key host cell factor for viral entry and pathogenesis of SARS-coronavirus-2 (SARS-CoV-2). Specifically, TMPRSS2 proteolytically processes the SARS-CoV-2 Spike (S) Protein, enabling virus-host membrane fusion and infection of the lungs. We present here an efficient recombinant production strategy for enzymatically active TMPRSS2 ectodomain enabling enzymatic characterization, and the 1.95 Å X-ray crystal structure. To stabilize the enzyme for co-crystallization, we pre-treated TMPRSS2 with the synthetic protease inhibitor nafamosat to form a stable but slowly reversible (15 hour half-life) phenylguanidino acyl-enzyme complex. Our study provides a structural basis for the potent but non-specific inhibition by nafamostat and identifies distinguishing features of the TMPRSS2 substrate binding pocket that will guide future generations of inhibitors to improve selectivity. TMPRSS2 cleaved recombinant SARS-CoV-2 S protein ectodomain at the canonical S1/S2 cleavage site and at least two additional minor sites previously uncharacterized. We established enzymatic activity and inhibition assays that enabled ranking of clinical protease inhibitors with half-maximal inhibitory concentrations ranging from 1.7 nM to 120 μM and determination of inhibitor mechanisms of action. These results provide a body of data and reagents to support future drug development efforts to selectively inhibit TMPRSS2 and other type 2 transmembrane serine proteases involved in viral glycoprotein processing, in order to combat current and future viral threats. SUMMARY PARAGRAPH Viruses hijack the biochemical activity of host proteins for viral invasion and replication. Transmembrane protease, serine-2 (TMPRSS2) is a surface-expressed protease implicated in the activation of influenza A, influenza B, and coronaviruses, including SARS-CoV-2, to drive efficient infection of the lungs1–5. TMPRSS2 is an attractive target for antiviral therapies, as inhibiting its proteolytic activity blocks efficient viral entry5,6. However, a structural and biochemical understanding of the protease has remained elusive and no selective inhibitors are available. We engineered on-demand activatable TMPRSS2 ectodomain and determined the 1.95 Å X-ray crystal structure of the stabilized acyl-enzyme after treatment with nafamostat, a protease inhibitor under investigation as a COVID-19 therapeutic. The structure reveals unique features of the TMPRSS2 substrate recognition pocket and domain architecture, and explains the potent, but nonselective inhibition by nafamostat. TMPRSS2 efficiently cleaved the SARS-CoV-2 S protein at the canonical S1/S2 site as well as two minor sites previously uncharacterized. We further established a robust enzymatic assay system and characterized inhibition by two additional clinical protease inhibitors under study for COVID-19, camostat and bromhexine. Our results provide a body of data and reagents to enable ongoing drug development efforts to selectively inhibit TMPRSS2 and other TTSPs involved in viral glycoprotein processing, in order to combat current and future viral threats.

. Engineered activation and structural characterization of stabilized TMPRSS2 ectodomain. a Full-length, membrane bound TMPRSS2 zymogen undergoes autocleavage activation at the Arg255-Ile256 peptide bond and the matured enzyme proteolytically processes SARS-CoV-2 Spike protein docked to the ACE2 receptor to drive viral membrane fusion. b Engineered recombinant TMPRSS2 ectodomain containing the low-density lipoprotein receptor type-A (LDLR) domain, a Class A Scavenger Receptor Cysteine-Rich (SRCR) domain and a C-terminal trypsin-like serine peptidase (SP) domain, features an enteropeptidase-cleavable DDDDK 255 substitution to facilitate controlled zymogen activation. The non-catalytic (LDLR+SRCR) and catalytic (SP) chains are tethered by a disulfide bond and the activation status can be interrogated by SDS-PAGE under non-reducing and reducing (5% βmercaptoethanol) conditions. c X-ray crystal structure of activated TMPRSS2 ectodomain pre-treated with nafamostat (yellow sticks). d The interdomain disulfide pair (Cys244-Cys365) maintains covalent attachment of the SRCR and SP domains. e Close-up view of the SP catalytic triad residues (His296, Asp345 and Ser441) and the post-activation Asp440:Ile256 salt bridge showing complete maturation of the protease. Nafamostat treatment results in phenylguanidino acylation of Ser441. Polar contacts are shown as yellow dashed lines.

TMPRSS2 has a unique and accommodating substrate binding cleft 76

The TMPRSS2 SP domain is highly conserved with all TTSPs and conforms to the canonical 77 chymotrypsin/trypsin fold with two six-stranded beta barrels converging to a central active site 78 cleft harboring the catalytic triad (Fig. 1c) 14 . Divergent protein substrate specificity of these closely 79 related proteases is conferred through highly variable, surface-exposed loops, denoted Loop A-E 80 and Loops 1-3 (Extended Data Fig. 2) 14 . Unique subsites formed on the face of the SP domain, S4-81 S3-S2-S1-S1'-S2'-S3'-S4' recognize substrate P4-P3-P2-P1↓P1'-P2'-P3'-P4' amino acid 82 positions spanning the scissile bond (Fig 2a; Extended Data Fig 3a) . To rationally assign these 83

subsites for TMPRSS2, we superposed the peptide-bound hepsin and TMPRSS13 SP domains 84 (40.1% and 41.4% sequence identity of their SP domains, respectively) belonging to the same 85 hepsin/TMPRSS subfamily as TMPRSS2. The S1 position of TMPRSS2 is occupied by the 86 phenylguanidino moiety of nafamostat, forming salt bridges with the highly conserved Asp435, 87

Ser436, and Gly464 residues in the same binding mode as the guanidino of P1 Arg residues 88 observed in hepsin and TMPRSS13 ( Fig. 1e; Fig. 2a This unpaired cysteine is conserved in feline, bovine, mouse, and rat TMPRSS2 orthologs 101

(Extended Data Fig. 2e) . Furthermore, the unpaired Cys379 is bordered by an expansive 360 Å 2 102 patch of exposed hydrophobic surface area in our structure that may serve as an interaction hub 103

for TMPRSS2 binding partners (Extended Data Fig. 2b ).

The SRCR domain confers additional diversity for molecular recognition.

The SRCR domain is found enriched in proteins expressed at the surface of immune cells as well 107

as in secreted proteins, and are thought to participate in protein-protein interactions and substrate 108 recognition 16 . The Class A SRCR domain of TMPRSS2 is located on the backside of the SP 109 domain away from the active site and is structurally similar to that of TMPRSS13 despite sharing 110 only 19% sequence identity (Fig. 2b) . These two SRCR domains adopt a compact, globular fold 111 with similar orientations relative to their SP domains (Fig. 2b,c) . The SRCR of hepsin (7.5% 112 sequence identity) diverges significantly from TMPRSS2/13 with three intra-domain disulfides 113 and a tighter SRCR:SP association dominated by complementary electrostatic patches and buried 114 surface area (Fig. 2c ). These conformational differences may play a role in the orientation of the 115 SP domain relative to the plasma membrane as well as modulate activity through recognition or 116 recruitment of partner proteins. 117 118 119 120 Figure 2 . Divergent surface properties of TMPRSS2 inform putative substrate preferences and relative domain organization. a The subsites of TMPRSS2 (blue) superimposed on the corresponding residues of Hepsin (magenta, PDB: 1Z8G) and TMPRSS13 (orange, PDB: 6KD5). The S1 subsite occupied by the phenylguanidino acyl group has well conserved Ser441, Asp435, and Gly464 residues, whereas discriminatory residues in S2 (Lys342) and S4 (Thr341) are not occupied. b Ribbon representation of the superimposed SRCR domains of TMPRSS2 (blue), Hepsin (magenta, PDB: 1Z8G) and TMPRSS13 (orange, PDB: 6KD5). c Relative orientation of the SRCR (green) and serine protease domains (blue) of TMPRSS2, Hepsin (PDB: 1Z8G), and TMPRSS13 (PDB: 6KD5). The LDLR domain of TMPRSS13 is shown in magenta with bound calcium in orange.

TMPRSS2 displays robust in vitro peptidase activity 123

To evaluate TMPRSS2 inhibitors and provide groundwork for future structure activity relationship 124

(SAR) studies, we established in vitro proteolytic activity and inhibition assays. The generic TTSP 125 fluorogenic peptide substrate Boc-Gln-Ala-Arg-7-aminomethylcoumarine (AMC) was rapidly 126 cleaved by dasTMPRSS2, C-terminal to Arg, thereby releasing AMC product and enabling initial 127 reaction velocities ( ) measurement within 60 seconds of enzyme addition (Fig. 3a) . In Assay 128

Buffer, dasTMPRSS2 had a of (200±80) µM, of (0.7±0.2) nmol min -1 , of (18±4)s -1 , 129 / of (5.4±0.2) µM -1 min -1 and specific activity at (0.22±0.03) µmol min -1 mg -1 enzyme 130 purified to apparent homogeneity (Fig. 3b) . To our knowledge, this level of activity has not been 131 achieved with any previously described recombinant TMPRSS2 enzyme 17-20 , and enzyme activity 132 was unaffected by the presence of Ca 2+ , NaCl concentrations ranging 75-250 mM, EDTA, and 133 tolerant of 2% (v/v) DMSO (Extended Data Fig. 4c ) that is encouraging for use in high throughput 134

inhibitor screening campaigns. 135 136

TMPRSS2 efficiently cleaves the SARS-CoV-2 S protein at the S1/S2 site in vitro 137

Cells expressing TMPRSS2 have been shown to efficiently cleave the SARS-CoV-1 S protein at 138 S1/S2 (SLLR667↓) and multiple peripheral sites to induce the necessary conformational changes 139

leading to virus-host fusion at the plasma membrane 21,22 ( Fig. 1a ; Extended Data Fig. 5a ). This 140 extensive TMPRSS2 processing has also been linked to periplasmic shedding of the S1 fragment 141

to act as an immune decoy in vivo 22 . For SARS-CoV-2, an acquired multibasic RRAR685↓ S1/S2 142 cleavage sequence was hypothesized to confer preferential cleavage by intracellular furin 143 protease 23 , and was corroborated by studies showing that multibasic, peptidomimetic furin 144

inhibitors prevented S1/S2 cleavage and attenuate infection in cellular models 24 . Further studies 145

showed that these inhibitors are promiscuous and disable multiple surface-expressed proteases that 146 process multibasic substrates in addition to furin, and more selective furin inhibitors cannot fully 147 abrogate S activation 25 . Furthermore, furin-deficient cells can still generate S1/S2 cleaved virus, 148

and propagation of SARS-CoV-2 in TMPRSS2-deficient cell lines results in a loss of the 149 multibasic S1/S2 site 26 , attenuating viral infectivity towards TMPRSS2+ cells. We sought to 150 characterize TMPRSS2's proteolytic activity towards S1/S2 by incubating recombinant furin 151

and/or dasTMPRSS2 with stabilized SARS-CoV-2 S protein ectodomain with S1/S2 knocked out 152

(RRAR 685 ->GSAS 685 ; HexaPro construct) or with S1/S2 intact (denoted HexaFurin; Extended 153 Data Fig. 5a; Fig. 3c ). As expected from previous studies using recombinant, S1/S2 intact S 154 protein, HexaFurin sustained partial S1/S2 cleavage during production in HEK293 cells due to 155 endogenously expressed furin 27 (Fig. 3c ). Recombinant furin treatment converted the remaining 156

intact HexaFurin to S1 and S2 band fragments with incubation over 16 hours, but was unable to 157

cleave HexaPro ( Fig. 3c ; Extended Data Fig. 5b ).

In contrast, using both the HexaFurin and HexaPro constructs, we observed that dasTMPRSS2 159 could cleave the S protein at 3 distinct sites with variable efficiency (Fig. 3d-e) . HexaFurin was 160 cleaved to only the S1 and S2 fragments within 5 minutes of dasTMPRSS2 addition (Fig. 3d) , 161

demonstrating the S1/S2 site was best recognized by TMPRSS2 across a minimal incubation. 162

HexaPro, lacking S1/S2, was cleaved across 30 min to generate a larger 150 kDa band, denoted 163 fragment X, and 80 kDa fragment Y when analyzed under non-reducing conditions (Fig. 3e ). 164

Reducing conditions revealed an additional cleavage site hidden within fragment X that is spanned 165 by two cysteine residues participating in a disulfide bond, splitting fragment X into 120 kDa 166 fragment X'a and 35 kDa fragment X'b. Exhaustive HexaPro treatment (120 min) completely 167

converted fragment X into X'a and X'b (Extended Data Fig. 6d ). 168 Figure 3 . dasTMPRSS2 displays robust proteolytic activity towards peptide and SARS-CoV-2 S protein substrates. a The generic Boc-Gln-Ala-Arg-AMC fluorogenic peptide substrate is efficiently cleaved by dasTMPRSS2. Each progress curve was performed in quadruplet. b Michaelis-Menten plot of initial reaction velocities for kinetic parameter estimation after curve fitting in GraphPad. c S1/S2 intact S protein ectodomain (HexaFurin; S 0 ) is partially cut at the S1/S2 site to produce S1 and S2 fragments and can be quantitatively converted by recombinant furin treatment over 16 hours. d The addition of 30 nM TMPRSS2 to furin treated HexaFurin produces no additional bands, but furin untreated HexaFurin is exhaustively cleaved at S1/S2 within 5 minutes. e dasTMPRSS2 cleaves HexaPro at two additional sites peripheral to S1/S2, with the first cleavage producing X and Y band fragments under non-reducing SDS-PAGE conditions. Reducing conditions reveal band fragments X'a and X'b derived from fragment X. f dasTMPRSS2 peptidase activity is blocked with varying potencies by clinical protease inhibitors, with no inhibition seen for bromhexine. All data are shown as mean ± s.e.m., n = 3 biological replicates g Apparent melting temperatures (as determined by differential scanning fluorimetry) are increased for benzamidine, camostat, and nafamostat at 1 µM concentration but are not increased by SFTI-1 or bromhexine. Samples were in triplicate.

To visualize all of these cleavage sites simultaneously, we treated HexaFurin 30 min with 171 dasTMPRSS2 and compared SDS-PAGE banding to a western blot using an antibody directed 172 towards the S protein receptor binding domain (RBD; Extended Data Fig. 5c ). At least 7 bands 173

were observed on reducing SDS-PAGE and the western shows that both the S1 fragment as well 174

as an S1-derived 50 kDa fragment contain the RBD. The banding patterns observed (S1/S2, X/Y, 175 and X'a/X'b cleavages) are consistent with western blot studies monitoring SARS-CoV-1 S 176 protein processing by TMPRSS2 that enables shedding of the S1 fragment 22 to act as an immune 177 decoy.

Nafamostat rapidly acylates TMPRSS2 and slowly hydrolyzes 180

Nafamostat by Differential Scanning Fluorimetry (DSF) 28 (Fig. 3g) and was a key stabilizing feature to enable 187 protein crystallization (Methods). Nafamostat demonstrated enhanced potency over camostat with 188 IC50 values of (1.7±0.2) and (17±4) nM, respectively, with 5 min assay pre-incubation (Fig. 3f) . 189

However, IC50 values were time-dependent and required further kinetic interrogation to assess 190 their divergent potencies (Extended Data Fig. 6b-c) . Nafamostat was 40-fold more potent than 191 camostat with respective inact / i values of (0.024±0.006) and (0.00059±0.00003) s -1 nM -1 .

These results emphasize that single timepoint IC50 values are insufficient for evaluating 193 mechanism-based, covalent inhibitors of this highly active protease in SAR studies. As previously Non-covalent trypsin-like serine protease inhibitors benzamidine and sunflower trypsin inhibitor-204 1 (SFTI-I) were less potent with respective IC50 values of (120±20) µM and (0.4±0.2) µM ( Fig.  205 3f), and values of (80±10) µM and (0.4±0.2) µM (Extended Data Fig. 7a-b) . 6-amidino-2-206 napthol also disabled dasTMPRSS2 activity with an IC50 of (1.6±0.5) µM and of (1.1±0.3) µM 207

(Extended Data Fig. 7a ). Bromhexine hydrochloride, another agent under investigation for anti-208 TMPRSS2 COVID-19 therapy 30,31 , showed no inhibition in either the peptidase or HexaPro 209 cleavage assay formats (Extended Data Fig. 7c-d) , corroborating reports of its ineffectiveness in 210

blocking SARS-CoV-2 pseudovirus entry 32 and further underscores the need for novel, selective 211 TMPRSS2 inhibitors.

Future prospects 214 We have produced and characterized a source of TMPRSS2 enzyme that will enable rapid inhibitor 215 development as antivirals and thorough molecular interrogation of coronavirus and influenza virus 216 activation. Although nafamostat potently neutralizes TMPRSS2 activity, it is non-selective and 217 disables trypsin-like serine proteases involved in coagulation such as plasmin, FXa, and FXIIa, as 218 well as other TTSPs through its generic arginine-like engagement with the S1 subsite 19,33,34 . 219

Furthermore, nafamostat requires continuous intravenous infusion to approach therapeutic 220 concentrations for COVID-19 owing to its short biological half-life of 8 minutes (NCT04418128; 221 NCT04473053). These features, although undesirable as a selective therapeutic, make nafamostat 222 an extremely useful and sensitive reagent for in vitro kinetic characterization of trypsin-like 223

proteases, and sufficiently stabilized our protease for crystallization and structural determination. 224

Nevertheless, selective and biologically stable drugs for TMPRSS2 must be explored, and may be 225 achieved through inhibitors engaging the more TMPRSS2-specific S2, S3, and S4 subsites 226 identified in our crystal structure.

We observed no electron density for the LDLR-A domain of TMPRSS2, despite a similar construct 229 design to that which afforded the TMPRSS13 crystal structure (PDB: 6KD5). The LDLR-A 230 domain of TTSPs is responsible for tethering the protease to the plasma membrane and most 231

TTSPs have a conserved ability to bind calcium. Interestingly, a key Asp residue in TMPRSS13 232

involved in calcium chelation is absent in human and other mammalian TMPRSS2 proteins, 233 substituted instead with His or Gln residues (Extended Data Fig. 8) . These data suggest that 234 TMPRSS2 may have lost the ability to bind calcium at this site.

Our demonstration that TMPRSS2 can cleave the multibasic S1/S2 site of the S protein suggests 237 that instead of conferring furin dependence, the virulent properties of this site may derive from 238 promiscuous recognition and cleavage by airway-expressed TTSPs, which is supported by the 239 demonstrated roles that TMPRSS4 35,36 , TMPRSS11d 20,37,38 , and TMPRSS13 20,37 , which colocalize 240 with ACE2 36 , play in enabling SARS-CoV-2 infection across various tissues.

Our characterization of dasTMPRSS2 did not reveal an obvious mechanism by which the native, 243

membrane-bound enzyme could be autoproteolytically processed peripheral to the activation motif 244

and thereby shed as a soluble enzyme into the extracellular space. However, studies using 245 TMPRSS2-specific antibodies have reported detection of a secreted enzyme product in prostate 246 sera that is expected to play a functional role in pericellular activation 39 . Due to the disulfide-linked 247 nature of activated TMPRSS2, former studies may have mischaracterized the catalytic subunit as 248 a shed SP domain when it would instead resolve to the intact species under non-reducing 249 conditions (Fig. 1b) Competing interests 314

The authors declare no competing interests.

Data Availability 317

The coordinates and structure of the phenylguanidino TMPRSS2 

Microplate reading at 320: 490 nm excitation:emission were used to 590 calculate the number of dasTMPRSS2 active site residues and calibrate peptidase activity and 591 inhibition assays

Nafamostat inhibition half-life 594 The half-life of the phenylguanidino acyl-enzyme complex after nafamostat treatment was 595 measured for dasTMPRSS2 using methods established for camostat with enteropeptidase 10

After incubation, unbound 598 nafamostat was removed by passage and 3x washes in a 3 kDa MWCO Amicon filter centrifuged 599 at maximum speed. Acylated and untreated dasTMPRSS2 samples were then transferred in 600 quadruplet to a microplate containing either 125 µM or 250 µM substrate (final concentration of 601 3.2 nM enzyme). Fluorescent reads were carried out immediately, analogous to IC50 assays, but 602 across a period of 8 hours

GraphPad to derive the half-life for activity recovery, normalized to the uninhibited initial 604 reaction velocity

Differential Scanning Fluorimetry 607 Apparent melting temperature ( , ) shifts were measured for various dasTMPRSS2-inhibitor 608 coincubations using SYPRO Orange dye (Life Technologies; cat. S-6650) and monitoring 609 fluorescence at 470:510 nm excitation: emission using the Light Cycler 480 II

Samples were prepared in triplicate in 384 well plates (Axygen; Cat# PCR-384-C; Cat# 611 UC500) at a final volume of 20 µL containing 0.05 mg/mL dasTMPRSS2, 1 µM compound or 612 vehicle control, and 5X SYPRO Orange. Thermal melt curves were generated between 25 °C to 613 95 °C at a gradient of 1 °C /min and plots prepared with the DSFworld application 11 for , 614 determination

5 mg/mL in Assay Buffer and incubated with the indicated concentrations of furin protease 619 (NEB) or dasTMPRSS2. Digestions took place over 16 hours for furin and from 5-120 minutes 620 for dasTMPRSS2. Furin digestions were terminated by the addition of 4 mM EDTA whereas 621 dasTMPRSS2 digestions were terminated with 5 µM nafamostat, then SDS-PAGE samples were 622 immediately prepared with the addition of 4X SDS-PAGE loading buffer and boiled for 5 min at 623 95 °C

Coomassie blue staining. For anti-RBD western blotting, 2 µg S protein were loaded per well

For cleavage inhibition assays, dasTMPRSS2 diluted to 320 nM in Assay Buffer was pre-626 incubated 15 minutes with inhibitor (final 1% DMSO (v/v)) or DMSO control, then assays were 627 started by transfer of enzyme:inhibitor mixes to S protein. S protein:protease mixtures were 628 incubated at room temperature for 2 hours with nafamostat and 30 minutes for bromhexine

Multiple Sequence Alignments 631 Multiple sequence alignments were prepared to compare the human TTSP family members and 632 TMPRSS2 mammalian orthologs. Human TTSP FASTA sequences (isoform 1) were accessed 633 from UniProt and TMPRSS2 orthologs identified with UniProt BLAST. Sequences were aligned 634 with

Protein Visualization and Property Calculation 637 The structure of dasTMPRSS2 was inspected and compared to other TTSPs using PyMol 638 (Schrodinger) and the Molecular Operating Environment (MOE; Chemical Computing Group) 639 software suite. The exposed hydrophobic patches of TMPRSS2 were calculated using the MOE 640 Protein Patch Analyzer tool 14

Structure-based design of prefusion-stabilized SARS-CoV-2 spikes. 1505

Phaser crystallographic software

Features and development of Coot

MolProbity : More and better reference data for improved all-atom 652 structure validation

Use of a fluorescence plate reader for measuring kinetic parameters with 654 inner filter effect correction

A two-point IC 50 method for evaluating the biochemical potency of 656 irreversible enzyme inhibitors. bioRxiv (2020) 657 doi

Divergent inhibitor 659 susceptibility among airway lumen-accessible tryptic proteases

Targeting Enteropeptidase with Reversible Covalent Inhibitors To Achieve 662 Metabolic Benefits

Three Essential Resources to Improve Differential Scanning Fluorimetry ( 664 DSF ) Experiments

The EMBL-EBI search and sequence analysis tools APIs in 2019. 666

Deciphering key features in protein structures with the new 668 ENDscript server

Homology modeling and structure-based design improve hydrophobic 670 interaction chromatography behavior of integrin binding antibodies

Protein Patch Analyzer and 2D Maps

Protein crystallization and structural determination 512After size-exclusion purification of activated dasTMPRSS2, samples were pooled and 513 concentrated to 2 mg/mL. Protein was treated with 3:1 nafamostat:dasTMPRSS2 for 10 minutes 514at room temperature and exchanged into Assay Buffer supplemented with 3:1 nafamostat using 4 515 spin cycles in 30 kDa Amicon MWCO filters (14,000 rpm, 15 min, 4 °C) to remove low Mw 516 autolytic fragments from the 42 kDa enzyme (Extended Data Fig. 1b) . Acylated enzyme was then 517 concentrated to 8 mg/mL and centrifuged (14,000 rpm, 10 min, 4 °C) prior to automated screening 518 at 18 °C in 96-well Intelliplates (Art Robin) using the Phoenix protein crystallization dispenser 519 (Art Robbins). Protein was dispensed as 0.3 µL sitting drops and mixed 1:1 with precipitant. The 520RedWing and SGC precipitant screens were tested and amorphous, non-diffracting crystals were 521 consistently produced when grown over 30% Jeffamine ED-2001 (Hampton Research) with 100 522 mM HEPES pH 7.0. To acquire a diffraction quality crystal, acylated dasTMPRSS2 was treated 523with 50 U PNGase F (NEB; 37 °C for 45 min) to trim N-glycan branches, then centrifuged (14,000 524 rpm, 4 °C, 10 min) prior to setting 2 µL hanging drops with 1:1 protein: precipitant and grown for 525 10 days. Crystals were then cryo-protected using reservoir solution supplemented with ~5% (v/v) 526 ethylene glycol, and cryo-cooled in liquid nitrogen. X-ray diffraction data were collected on the 527 beamline 24-ID-E at the Advanced Photon Source (APS). Data were processed with XDS 2 . Initial 528 phases were obtained by molecular replacement in Phaser MR 3 , using (PDB: 1Z8G) as a starting 529 model. Model building was performed in COOT 4 and refined with Buster 5 . Structure validation 530 was performed in Molprobity 6 . Data collection and refinement statistics are summarized in 531Extended Data Table 2 . were mixed with 4x Laemelli buffer (BioRad) and subjected to differential reducing (± 5 mM β-537 mercaptoethanol; Gibco), then boiling at 95 °C for 5 min in order to probe the covalent nature of 538 protein complexes and subunits. The Precision Plus Protein marker (BioRad) was used as a 539standard. 540For SARS-CoV-2 RBD western blotting, SDS-PAGE was carried out as described, followed by 541 wet transfer in Transfer Buffer (25 mM Tris pH 8.3, 192 mM glycine, 20% MeOH (v/v)) to PVDF 542 membrane (80 V, 53 min, 4 °C). Membranes were incubated in Blocking Buffer (5% skim milk 543in TBST) for 1 hr at room temperature, washed 5x with TBST, then probed overnight with 1/3000 544 mouse anti-RBD primary mAb (Abcam ab277628) solution at 4 °C. Membranes were then washed 545 5x with TBST and probed with 1/5000 FITC-labelled goat anti-mouse IgG secondary pAb (Abcam 546 ab6785) and imaged for fluorescence on the Typhoon FLA7000 biomolecular imager (GE 547 healthcare). to a product AMC concentration using standard curves at each substrate concentration to correct 555for the inner-filter effect 7 (Extended Data Fig. 4d ). All assays contained 2% (v/v) DMSO and 556initial reaction velocities were tabulated over the linear portion of the first 60 seconds of progress 557curves. 558To determine Michaelis Menten kinetic parameters, 50 µL 4x enzyme stock was added through 559 automated addition to microplates containing 150 µL substrate (0.5-1000 µM) in triplicate and 560 initial reaction velocities were plotted against substrate concentration and curve fit using GraphPad 561Prism. Time-dependent IC 50 measurement and / determination 574Camostat IC50 curves were generated using 7 concentrations of inhibitor ranging 0.1-1000 nM 575inhibitor and nafamostat between 0.01-100 nM with a DMSO control as described. The time 576dependence of inhibitor potencies was measured by using Flexstation Flex kinetic reads that 577automatically transferred dasTMPRSS2-inhibitor mixes to substrate wells at the indicated pre-578 incubation timepoints (Extended Data Fig. 6c ). For the 10s timepoint, a kinetic read was 579performed after manual addition of enzyme, followed by substrate, using a multichannel pipette. 580Kinetic parameters Kiapp and kinact were determined with the simplified Equations 2 and 3,