key: cord-342657-od48cntc
authors: Klemm, Theresa; Ebert, Gregor; Calleja, Dale J.; Allison, Cody C.; Richardson, Lachlan W.; Bernardini, Jonathan P.; Lu, Bernadine G. C.; Kuchel, Nathan W.; Grohmann, Christoph; Shibata, Yuri; Gan, Zhong Yan; Cooney, James P.; Doerflinger, Marcel; Au, Amanda E.; Blackmore, Timothy R.; Geurink, Paul P.; Ovaa, Huib; Newman, Janet; Riboldi-Tunnicliffe, Alan; Czabotar, Peter E.; Mitchell, Jeffrey P.; Feltham, Rebecca; Lechtenberg, Bernhard C.; Lowes, Kym N.; Dewson, Grant; Pellegrini, Marc; Lessene, Guillaume; Komander, David
title: Mechanism and inhibition of SARS-CoV-2 PLpro
date: 2020-06-19
journal: bioRxiv
DOI: 10.1101/2020.06.18.160614
sha: 
doc_id: 342657
cord_uid: od48cntc

Coronaviruses, including SARS-CoV-2, encode multifunctional proteases that are essential for viral replication and evasion of host innate immune mechanisms. The papain-like protease PLpro cleaves the viral polyprotein, and reverses inflammatory ubiquitin and anti-viral ubiquitin-like ISG15 protein modifications1,2. Drugs that target SARS-CoV-2 PLpro (hereafter, SARS2 PLpro) may hence be effective as treatments or prophylaxis for COVID-19, reducing viral load and reinstating innate immune responses3. We here characterise SARS2 PLpro in molecular and biochemical detail. SARS2 PLpro cleaves Lys48-linked polyubiquitin and ISG15 modifications with high activity. Structures of PLpro bound to ubiquitin and ISG15 reveal that the S1 ubiquitin binding site is responsible for high ISG15 activity, while the S2 binding site provides Lys48 chain specificity and cleavage efficiency. We further exploit two strategies to target PLpro. A repurposing approach, screening 3727 unique approved drugs and clinical compounds against SARS2 PLpro, identified no compounds that inhibited PLpro consistently or that could be validated in counterscreens. More promisingly, non-covalent small molecule SARS PLpro inhibitors were able to inhibit SARS2 PLpro with high potency and excellent antiviral activity in SARS-CoV-2 infection models.

centred around ISG15 Trp123 and Pro130/Glu132, docking ISG15 onto the PLpro 132 a7 helix (Extended Data Fig. 5c ). These interactions dislodge the Ubl-fold from the 133 Fingers subdomain (Figure 2a) . While the complex resembles interaction modes 134 observed in SARS PLpro~ISG15 CTD (pdb 5tl7, 25 , RMSD of 0.74 Å for PLpro, see 135 Extended Data Fig. 4) , some interacting residues (especially, Tyr171 on helix a7) 136 are not conserved (Extended Data Fig. 1, 5c) , and seem to improve the contact in 137 SARS2 PLpro. More variability is seen in MERS PLpro, which binds to ubiquitin and 138 ISG15 CTD similarly through its ability to 'close' the Fingers subdomain 26 (see 139 discussion in Extended Data Fig. 4) . Fig. 1, 6a,b) . Consistently, SARS2 PLpro F69S mutation greatly diminished Extended Data Fig. 5f) , are more pronounced. Mutational analysis hence confirms 165 that the S2 site is more important for providing PLpro with Lys48-polyubiquitin 166 activity and specificity. Taken together, our data illuminate in molecular detail how 167 SARS2 PLpro targets ubiquitin and ISG15.

Repurposing known drugs to inhibit PLpro activity 170 We next focussed our attention on the urgent matter of inhibiting PLpro, to confirm its 171 drugability, and to provide new drug candidates with efficacy in treating Ideally, an already clinically approved drug shows a pharmacologically relevant 173 effect on PLpro with sub-µM inhibitory potential, cell penetrance, oral bioavailability, 174 and extensive safety profiles for the required dosage. Such a drug could be 175 expedited for clinical trials.

A 1536-well low-volume high-throughput assay previously used to identify inhibitors Together, our data suggests that a repurposing strategy using 3727 unique known 199 drugs towards SARS2 PLpro is unlikely to yield drug candidates, and highlights the 200 importance of a counterscreen in assessing the validity of hits coming from a screen 201 of known drugs before any conclusions on their therapeutic potential can be drawn. 202 The robust screen and orthogonal assays for PLpro will be instrumental in drug 203 discovery campaigns. Fig. 1, 8a, Fig 8e) and its activity was confirmed by 234 proteolytic cleavage of the GFP tag (Figure 4d, Extended Data Fig 8f) . Nsp3 235 expression depleted Lys48-linked polyubiquitin, which was inhibited by rac5c in a 236 dose dependent manner (Figure 4d, Extended Data Fig 8f) . Since it had previously 237 been shown that these inhibitors are specific for PLpro over human DUBs 1 (also see 238 Figure 3b ) and since treatment with rac5c did not affect Lys48-linked polyubiquitin in 239 the absence of nsp3 expression (Extended Data Fig 8f) , the effect of rac5c on 240 Lys48-polyubiquitin is likely due to inhibition of nsp3/PLpro. Detailed molecular understanding of how PLpro targets ubiquitin and ISG15, and a 286 robust high-throughput screen, pave the way to structure-guided drug discovery.

Indeed, while available clinically tested drugs may not be suitable to target PLpro 288 (Figure 3) 720 Gel-based cleavage assays were performed as previously described 51 with the For gel-based quantitative analysis, Coomassie stained gel images were converted 732 to grayscale and band intensities were quantified using ImageLabä (Bio-Rad).

Background intensities were automatically subtracted using a base line relative to 794 We assessed the activity of 5,577 compounds contained in commercially available CoV-2  TI V  TVD  N  T  TY  Q G  GAD  P  G  V  H  VV  Q  Y  VT  KT  E R  K FT  NI L  Q  DMSM  G  F PT LD  KIK HNSHE  F  SARS-CoV  TI V  TVD  N  T  TY  Q G  GAD  P  G  V  H  LV  Q  Y  VT  KT  E K  K FT  NT L  Q  DMSM  G  F PT LD  KIK HVNHE  F  MERS-CoV  TI V  TVD  N  T  TY  Q G  GAD  P  G  L  R  VL  S  F  IS 

A L S V C CG L G A Y G V M L H K V M S E K RET SY FQ N D. C R LNVV KT QQQTT K E VM TL Y QF SARS-CoV D A L S V C CG L G A Y G V M L H K V M S D K RET TH LQ N E. A R LNVV KH QKTTT T E VM TL Y NL MERS-CoV D A L S V C CG L G A Y G A L V K R L V T E R SRL

the Thumb domain, to adopt a similar orientation and interaction as seen for 1012 ISG15 bound to MERS. MERS PLpro ubiquitin complexes have been determined 1013 with 'open' and more 'closed' Fingers 26

Extended Data Figure 5. Separation of function mutations in SARS2 PLpro 1017 a, Ubiquitin and ISG15 binding site analysis based on PISA analysis, indicating 1018 interface residues on SARS2 PLpro

PA (salmon) as bound to SARS2 PLpro highlights the different binding 1020 modes with a ~40º rotation between the two proteins. c, Details of the binding of 1021

Interacting residues shown as 1022 sticks. d, Mutations in S1 and S2 sites were introduced in PLpro, and the enzyme 1023 variants were expressed in bacteria, purified, and tested for integrity by assessing 1024 the inflection temperature, indicating the transition of folded to unfolded protein. With 1025 exception of mutating the catalytic Cys to Ala, which was severely destabilised and 1026 precipitated during purification

Inflection temperature values were determined in technical duplicate from 1028 experiments performed twice. e, f, Triubiquitin cleavage to mono-and diubiquitin 1029 (left), and proISG15 cleavage to mature ISG15 (right), were compared side-by-side 1030 over a time course, resolved on SDS-PAGE gels, and visualised by Coomassie 1031 staining. Experiments were performed in duplicate with 250 nM enzyme and 2 µM 1032 substrate

reproduced from Extended Data Fig 2b, 2c, for comparison. d, S1 site 1034 mutants as indicated

Extended Data Figure 6. The S2 site in SARS2 PLpro 1037 a, A previous structure of SARS PLpro bound to a non-hydrolysable, Lys48-linked 1038 diubiquitin probe (pdb 5e6j, 20 ) explained the noted preference of PLpro for longer 1039

While the proximal ubiquitin unit occupies the S1 site in a highly 1040 similar fashion in SARS~Ub and SARS2~Ub structures (see b, Figure 2a

Data Fig. 4), the second, distal, ubiquitin unit binds to the a2 helix of PLpro, through 1042 a common binding mode involving the ubiquitin Ile44 patch

Phe70 in SARS PLpro residue is flanked by residues involved in polar contacts

Structure of the SARS2 PLpro~Ub complex. The S2 site consisting of a2 helix 1045 with Phe69 residue, is fully conserved in SARS2 PLpro. Mutation of Phe69 to Ser 1046 severely impacts triubiquitin and proISG15 hydrolysis (see Extended Data Fig

Fluorescence polarisation assay on ISG15-TAMRA for PLpro wild-type 1048 (reproduced from Extended Data Fig. 2a) and PLpro F69S. A ~3-fold lower 1049 efficiency for F69S is similar to cleavage of ISG15 CTD -TAMRA

suggesting that the S2 site contributes the difference in binding for the N-1051 terminal Ubl-fold. Experiments for F69S were performed in technical triplicate and 1052 biological duplicate. d-f

triubiquitin (e) and proISG15 (f) using wild-type PLpro (top, gels reproduced from 1054

Extended Data Fig. 2b-d to enable direct comparison) or PLpro F69S (bottom)

Mutation of the S2 site has no marked effect on hydrolysis of proISG15 CTD (d) and 1056 reduces proISG15 cleavage to the same levels as proISG15 CTD

Experiments were 1058 performed in duplicate, see Supplementary Figure 1

Extended Data Figure 8. SARS PLpro compounds inhibit SARS2 PLpro 1084 a, Structure of SARS2 PLpro bound to the ubiquitin C-terminal tail in the active site, 1085 compare with Figure 4a. b, Superposition of ubiquitin tail in SARS2 PLpro, and 1086 compound 3j in SARS PLpro (pdb 4ovz, 29 ) shows an identical binding for 1087 compounds in SARS2 PLpro and highlights the change in Tyr268

Compounds rac3j and rac3k, racemic 1089 versions of 3j and 3k from 29 , and their in vitro biochemical IC50 values determined by 1090 the HTS assay technical triplicate and in three independent experiments (as for 1091 rac5c in Figure 4c). e, Immunoblot characterisation of the PLpro antibody on HEK 1092 293T cells overexpressing PLpro from MERS, SARS or SARS2. Cell lysates were 1093 immunoblotted 48 h post transfection. f, Immunoblot analysis showing the effect of 1094 rac5c (10 µM for 24 h) on Lys48-polyubiquitin chain disassembly by nsp3, 48 h post 1095 transfection in HEK 293T cells

Values in 1130 parentheses are for highest-resolution shell. b, c (Å) 64.99, 144.41, 49.60 124