Progress and Challenges in Targeting the SARS-CoV-2 Papain-like Protease.

SARS-CoV-2 is the causative agent of the COVID-19 pandemic. The approval of vaccines and small-molecule antivirals is vital in combating the pandemic. The viral polymerase inhibitors remdesivir and molnupiravir and the viral main protease inhibitor nirmatrelvir/ritonavir have been approved by the U.S. FDA. However, the emergence of variants of concern/interest calls for additional antivirals with novel mechanisms of action. The SARS-CoV-2 papain-like protease (PLpro) mediates the cleavage of viral polyprotein and modulates the host's innate immune response upon viral infection, rendering it a promising antiviral drug target. This Perspective highlights major achievements in structure-based design and high-throughput screening of SARS-CoV-2 PLpro inhibitors since the beginning of the pandemic. Encouraging progress includes the design of non-covalent PLpro inhibitors with favorable pharmacokinetic properties and the first-in-class covalent PLpro inhibitors. In addition, we offer our opinion on the knowledge gaps that need to be filled to advance PLpro inhibitors to the clinic.

Introduction

Coronaviruses (CoVs) are enveloped, positive-sense, and single-stranded RNA (+ssRNA) viruses. CoVs belong to the subfamily Orthocoronavirinae, family Coronaviridae, and order Nidovirales. Seven coronaviruses are known to infect humans: four common human coronaviruses—HCoV-229E, HCoV-NL63, HCoV-OC43, and HCoV-HKU1—that cause mild symptoms[1] and three coronaviruses—SARS-CoV, MERS-CoV, and SARS-CoV-2—that cause severe acute respiratory tract infections.[2,3] Although humans around the world are commonly infected with HCoV-229E, HCoV-NL63, HCoV-OC43, or HCoV-HKU1, the infection generally only causes mild symptoms that do not require medical treatments.[4,5] Accordingly, no major efforts have been devoted to developing vaccines and antiviral drugs against these viruses. Nonetheless, the 21st century witnessed several coronavirus outbreaks that raised the alarm regarding this virus family. In late 2002, SARS-CoV emerged in Guangdong, China, and caused approximately 8000 cases, with a fatality rate of 9.6%.[6] In 2012, MERS-CoV emerged in Saudi Arabia and South Korea, causing approximately 2400 cases in the following 8 years, with a fatality rate of 34%.[7] Notably, in 2019, SARS-CoV-2 emerged in Hubei, China, and quickly ramped up to the coronavirus disease 2019 (COVID-19) pandemic.[8,9] The clinical outcomes of COVID-19 range from non-symptomatic, mild to severe respiratory tract infections, and influenza-like illness, to lung injuries, organ failure, and death.[10] To date, SARS-CoV-2 has spread all over the world and is the most severe pandemic in recent history. As of May 3, 2022, 511 million cases and 6.23 million deaths had been reported worldwide, among which the United States has had 80.5 million cases and 986,298 deaths.[11]

Given the devastating impact of COVID-19 on social life, public health, and the global economy, researchers around the world are working relentlessly to develop countermeasures. This effort has led to the development of vaccines and antiviral drugs in record-breaking times.[12,13] Vaccines mainly target the viral surface spike protein and rely on the production of antibodies to block the viral entry through inhibiting the interaction between the viral spike protein and the host cell angiotensin converting enzyme 2 (ACE2) receptor.[14] Three vaccines received approval from the U.S. Food and Drug Administration (FDA), including two mRNA vaccines, from Pfizer/BioNTech (Comirnaty) and Moderna (Spikevax), and one adenovirus-based vaccine, from Johnson & Johnson/Janssen. In addition, several vaccines from China and Russia have been approved by the World Health Organization (WHO).[15]

For small-molecule antivirals, major progress has been made in targeting the SARS-CoV-2 RNA-dependent RNA polymerase (RdRp), the main protease (Mpro or 3CLpro), and the papain-like protease (PLpro).[16,17] The first RdRp inhibitor, remdesivir (1, Figure A), was identified from a drug repurposing approach and approved for the treatment of severe SARS-CoV-2 infection by intravenous (i.v.) administration.[18] Remdesivir acts as a chain terminator during viral RNA synthesis.[19] Similarly, the second RdRp inhibitor, molnupiravir (2, Figure A) was originally developed as an influenza antiviral and was later shown to have broad-spectrum antiviral activity against several viruses, including SARS-CoV-2.[20,21] Molnupiravir (2) is a mutagen, and when incorporated into the RNA chain, it increases the mutation rate of the virus.[22] Molnupiravir (2) is a prodrug and has the advantage of oral administration.[23] The main protease inhibitor, Paxlovid, developed by Pfizer, is a combination of nirmatrelvir (3, Figure A) and ritonavir.[13] Nirmatrelvir (3) is an Mpro inhibitor, and ritonavir is included as a boosting agent to increase the half-life of nirmatrelvir. A similar approach was explored in the HIV drug combination Kaletra (lopinavir + ritonavir). Ritonavir is an inhibitor of cytochrome P450 3A4 (CYP3A4), and co-administration of ritonavir is required to increase the in vivo concentration of nirmatrelvir (3) to the target therapeutic range.

Figure 1

Chemical structures of FDA-approved COVID-19 antiviral drugs (A) and the schematic representation of the SARS-CoV and SARS-CoV-2 Open Reading Frame (B), the polyprotein replicase (C), and the recognition motifs of PLpro (D). The genome contains two open reading frames, ORF1a and ORF1b, which are directly translated into polyproteins pp1a and pp1ab due to the ribosomal frameshift between the two ORFs. pp1a contains 11 NSPs, and pp1ab contains 16 NSPs. The PLpro is located within the NSP3. The polyproteins are processed into functional NSP units through cleavage by PLpro and Mpro, and the cleavage sites of PLpro are shown in (C). The substrate amino acid sequence alignment of P4–P1′ recognized by PLpro is shown in (D).

The approvals of vaccines and RdRp and Mpro inhibitors are encouraging signs to combat the COVID-19 pandemic and possibly return to the pre-pandemic normalcy.[24] However, the emergence of SARS-CoV-2 variants of concern (VOC) and variants of interests (VOI) poses a pressing need for additional vaccines and antiviral drugs.[25] Multiple studies have shown the reduced efficacy of vaccines against Omicron VOC.[26,27] Drug-resistant mutations have been evolved against remdesivir (1) in cell culture through serial passage experiments[28,29] as well as in an immunocompromised patient.[30] In addition, the therapeutic benefits of remdesivir (1) are still under debate from several clinical trials.[31,32] Molnupiravir (2) has the potential risk of inducing mutations in the host, which is pending validation.[33,34] Molnupiravir (2) was shown to be positive in the Ames test,[35] which is a standard assay to measure mutagenic potential of drug candidates in bacteria. β-d-N4-Hydroxycytidine (NHC), the active metabolite of molnupiravir (2), displayed host mutational activity in mammalian cell culture.[34] Multiple mutations have been identified in Mpro among the SARS-CoV-2 VOC and VOI, including the Omicron Mpro P132H mutant.[36] Although the currently identified Mpro mutants remain sensitive to nirmatrelvir (3),[36−38] the scientific community is on high alert for future mutations, such as H172Y and S144A, that might lead to drug resistance.[39] The genetic barrier to resistance for protease inhibitors is generally moderate to low, as shown by HIV and HCV protease inhibitors.[40] Resistance to Paxlovid is expected to rise with the increasing prescription. In addition, nirmatrelvir (3) is used in combination with ritonavir in clinics to prolong its half-life. Ritonavir is a potent inhibitor of the CYP3A4 isoenzyme and thus poses the risk of drug–drug interactions.[41] As such, additional antivirals with a novel mechanism of action are clearly needed to combat emerging variants and drug-resistant viruses. In this regard, the SARS-CoV-2 PLpro stands out as one of the next-in-line high-profile drug targets.

PLpro and the Mpro are the two essential proteases encoded by the SARS-CoV-2 genome. Both PLpro and Mpro cleave the peptide bonds in the viral polyprotein to release functional non-structural proteins (NSPs) for viral transcription and replication. In addition, PLpro is involved in antagonizing the host’s immune response upon viral infection. PLpro has deubiquitinating and deISGylating activities and removes ubiquitin and ISG15 modifications from host proteins, leading to suppression of the innate immune response and promotion of viral replication.[42−44] The deubiquitinating and deISGylating activities of PLpro are indispensable in antagonizing the host’s immune response.[45,46] Recent studies showed that SARS-CoV-2 infection of human macrophages triggers the release of extracellular free ISG15 through the viral PLpro, leading to the subsequent secretion of proinflammatory cytokines and chemokines, which recapitulates the cytokine storm of COVID-19.[47,48] This finding suggests that inhibiting the PLpro activity might alleviate the hyper-inflammation in COVID patients. Thus, targeting PLpro is expected to not only suppress viral replication but also restore antiviral immunity in the host.[45]

There are two types of PLpros: PL1pro and PL2pro.[49,50] The viruses HCoV-229E, HCoV-NL63, HCoV-HKU1, and HCoV-OC43 encode both PL1pro and PL2pro. PL1pro and PL2pro have distinct substrate specificities in different coronaviruses.[51] In contrast, SARS-CoV, MERS-CoV, and SARS-CoV-2 comprise only one functional PL2pro.

PLpro is part of the nsp3, a 215-kDa multidomain viral protein. SARS-CoV-2 PLpro specifically recognizes a consensus cleavage motif, LXGG↓(N/K/X), which is present in between nsp1/2, nsp2/3, and nsp3/4 at the viral polyprotein as well as the C-terminal sequences of ubiquitin and ISG15 with an isopeptide bond (Figure B–D).

The SARS-CoV-2 PLpro contains four domains: the thumb, palm, zinc-finger domain, and an N-terminal ubiquitin-like domain (Figure ). The catalytic triad consists of Cys111, His272, and Asp286, which are located at the interface of the palm and thumb domains. The zinc-finger motif comprises four cysteines coordinating with a zinc ion and is vital for the structural integrity and the protease activity of PLpro. The flexible BL2 loop undergoes conformational changes from open to closed upon substrate binding (Figure A).[52] This site is also the drug-binding site for GRL0617 (4) and its analogues.[16] The X-ray crystal structures for the apo SARS-CoV-2 PLpro, drug-bound form,[52−55] and complex forms with ubiquitin (Figure A) and ISG15 (Figure B) have been solved,[56] paving the way for structure-based drug design and understanding the virology of PLpro.

Figure 2

X-ray crystal structures of SARS-CoV-2 PLpro. (A) X-ray crystal structure of SARS-CoV-2 PLpro C111S mutant with K48-linked Ub2 (PDB: 7RBR). The BL2 loop is colored in magenta. (B) X-ray crystal structure of SARS-CoV-2 PLpro C111S mutant with human ISG15 (PDB: 7RBS).[56]

SARS-CoV-2 PLpro shares a sequence identity of 82.9% with SARS-CoV PLpro and, to a lesser extent, 32.9% identify with MERS-CoV PLpro. Despite the high sequence similarity, SARS-CoV-2 PLpro has enhanced deISGylating activity and reduced deubiquitinating activity compared to SARS-CoV PLpro.[45,46,57] PLpro is a conserved drug target among SARS-CoV-2 variants (Figure ). Although mutations have been identified, top high-frequency mutations are all located distal from the drug-binding site (Figure C). Nonetheless, it remains to be experimentally validated whether these mutations will alter drug sensitivity. In addition, resistance might emerge under drug selection pressure.

Figure 3

Analysis of SARS-CoV-2 PLpro mutations. Based on the data retrieved from GISAID (www.gisaid.org/epiflu-applications/covsurver-mutations-app), 2,487,047 sequences that contains mutations on PLpro have been identified, which fall into 5754 different types of mutations on various positions of PLpro. All numbers shown are accurate as of Jan 25, 2022. (A) Cumulative frequency of SARS-CoV-2 PLpro mutants. (B) The top 30 most common SARS-CoV-2 PLpro mutants. Among these mutants, A145 has the most frequent mutation to D with 1,131,252 occurrences (99.8% on 145); P77L with 372,993 occurrences (95.7% on 77); K232Q with 117,247 occurrences (99.1% on 232); V187A with 87,861 occurrences (97.9% on 187); and K92N with 47,110 occurrences (98.2%). (C) Mapping of the top six SARS-CoV-2 PLpro mutants to the X-ray crystal structure of PLpro in complex with GRL0617 (4) (PDB: 7JRN). The residues are shown as spheres. The BL2 loop in the drug-binding site is colored in marine, and the drug GRL0617 (4) is colored in magenta.

The knowledge accumulated through studying SARS-CoV PLpro provides a foundation for understanding the virology of SARS-CoV-2 PLpro and developing SARS-CoV-2 PLpro inhibitors. For excellent reviews and research articles of the structure, function, and inhibition of SARS-CoV PLpro, please refer to previous publications.[16,58−62] This Perspective covers recent advances in the development of SARS-CoV-2 PLpro inhibitors and their mechanism of action. We also discuss the knowledge gaps that need to be filled to advance PLpro inhibitors to clinic.

It is not the objective of this Perspective to enumerate all SARS-CoV-2 PLpro inhibitors reported in the literature; instead, the focus is on highlighting several well-characterized examples. Non-specific PLpro inhibitors will also be discussed with the intention to alert the scientific community.

SARS-CoV-2 PLpro Assays

Vigorous pharmacological characterization is vital in triaging non-specific inhibitors at the early stage and prioritizing hits with translational potential for further development. For this, we provide a brief introduction of the commonly used assays for the pharmacological characterization of PLpro inhibitors (Figure A).

Figure 4

SARS-CoV-2 PLpro assays. (A) General flowchart for the pharmacological characterization of PLpro inhibitors. (B) Assay principle for the FRET-based enzymatic assay. (C) Assay principle for the cell-based FlipGFP PLpro assay. (D) Assay principle for the Protease-Glo luciferase PLpro assay.

The gold standard assay for protease is the fluorescence resonance energy transfer (FRET)-based enzymatic assay, which is typically used as a primary assay for compound testing. In the FRET assay, a peptide corresponding to the protease substrate is designed with a fluorophore donor and a quencher at the two ends (Figure B). Upon cleavage by the protease, an increase in fluorescence signal is observed. However, the enzymatic assay condition varies among different laboratories in terms of enzyme concentration, FRET substrate sequence, pH, the addition of detergent (to rule out aggregates), bovine serum albumin (to rule out non-specific hydrophobic interactions), and reducing reagent (to prevent non-specific modification of catalytic Cys111). For this reason, the IC50 values from different studies should be interpreted with caution and should not be used for direct comparison. Instead, positive controls such as GRL0617 (4) need to be included as a reference to normalize the results. The assay guidance manual compiled by Eli Lilly & Company and the National Center for Advancing Translational Sciences offers detailed guidance for assay optimization, which might help limit the variations between individual laboratories.[63] In addition, counter screening against unrelated cysteine proteases should be conducted to rule out non-specific inhibitors. Furthermore, compounds that either quench the fluorophore or have overlapping absorbance/emission with the fluorophore will lead to false positive/negative results.

Our studies have shown that reducing reagents such as dithiothreitol (DTT) or glutathione are essential in the FRET enzymatic buffer to rule out promiscuous compounds that have non-specific inhibition against cystine proteases. Our recent studies of validation and invalidation of reported Mpro and PLpro inhibitors demonstrated that the FRET IC50 values obtained in the absence of reducing reagent DTT had poor correlation with the antiviral activity.[64−66] We therefore urge the scientific community to be cautious in interpreting the PLpro assay IC50 values obtained in the absence of reducing reagent.

Several binding assays are also commonly used to determine the binding affinity between inhibitors and the PLpro: the thermal shift assay,[54] the surface plasma resonance (SPR) assay,[67] and isothermal titration calorimetry (ITC).[68] The thermal shift assay measures protein stability, and ligand binding typically leads to an increase of the melting temperature Tm. Nevertheless, a decrease in protein stability is also observed for certain ligand–protein interactions. Compared to the thermal shift assay, SPR is more quantitative, and binding kinetics kon, koff, and KD can be derived from the binding curves. ITC can determine the thermodynamic binding parameters ΔG, ΔH, and ΔS in a single experiment without a need to modify the protein. To gain a molecular-level understanding of the PLpro–inhibitor interactions, a co-crystal structure needs to be solved.

It is expected that the cell-free enzymatic assay or binding assay results can be used to faithfully predict the cellular antiviral activity. However, SARS-CoV-2 is a biological safety level 3 (BSL-3) pathogen, which limits the number of compounds that can be tested in the antiviral assay, given the paucity of the resources. In this regard, there is a need for a cell-based protease assay to help predict the antiviral activity at the BSL-1/2 setting. The cell-based protease assay not only reveals intracellular target engagement but also can rule out compounds that are cell-membrane-impermeable or cytotoxic. The FlipGFP and Protease-Glo luciferase assays are two representative cell-based protease assays that have been applied for the screening and validation of SARS-CoV-2 PLpro inhibitors.[54,69] In the FlipGFP assay, cells are transfected with two plasmids, one expressing the PLpro and another expressing the GFP reporter (Figure C).[70−72] The reporter plasmid encodes three proteins: the GFP β1–9 template, the β10–11 fragment, and the mCherry. The β10–11 fragment is restrained in the parallel orientation through the K5/E5 coiled coil and therefore cannot associate with the β1–9 template. Upon cleavage of the PLpro substrate linker, β10 and β11 become antiparallel and can associate with the β1–9 template, leading to the restoration of the GFP signal. mCherry serves as an internal control to normalize the transfection efficiency. As such, the GFP/mCherry ratio correlates to the enzymatic activity of PLpro. Results from us as well as others have shown that the FlipGFP assay is a valuable assay in characterizing the cellular Mpro and PLpro inhibition without the need of the infectious SARS-CoV-2 virus.[54,64,69,70,72,73] A positive correlation between the FlipGFP IC50 values and the antiviral EC50 values was observed for the PLpro inhibitors,[54] suggesting the FlipGFP assay can be used as a surrogate assay to prioritize lead compounds for antiviral testing.

The Protease-Glo luciferase assay is designed in an analogous way as the FlipGFP assay, in which the luciferase activity depends on cleavage of the substrate linker by the protease.[64] Specifically, the firefly luciferase is engineered with a protease substrate cleavage sequence (Figure D). Before cleavage, firefly luciferase is in the permuted circular inactive conformation. Upon protease cleavage, a conformational change leads to association of the two domains and restoration of the luciferase activity. The Protease-Glo luciferase assay can be performed either in live cells or in cell lysates.[69,74,75] As the readout is luminescence, the Protease-Glo luciferase assay can help rule out compounds that have fluorescence interference properties. Other cell-based assays, including the GFP ER translocation assay, the bioluminescence resonance energy transfer (BRET) assay, and the cell cytotoxicity assay, can be similarly engineered for PLpro.[75−77]

SARS-CoV-2 PLpro Inhibitors

We group SARS-CoV-2 PLpro inhibitors into non-covalent inhibitors and covalent inhibitors. The non-covalent inhibitors are further divided into GRL0617 (4) analogues and non-GRL0617 inhibitors (Table ).

Table 1

SARS-CoV-2 PLpro Inhibitorsa

N.T. = not tested.

Non-covalent SARS-CoV-2 PLpro Inhibitors

GRL0617 Analogues

The naphthalene-containing GRL0617 (4) was a well-characterized SARS-CoV PLpro inhibitor. It was originally developed through lead optimization based on a high-throughput screening (HTS) hit.[62] Several follow-up studies have been conducted with the aim of improving the potency of enzymatic inhibition and antiviral activity as well as pharmacokinetic (PK) properties. However, no significant improvement has been achieved.[60,61] As the SARS-CoV-2 PLpro is 83% identical and 90% similar to SARS-CoV PLpro, GRL0617 (4) became a top candidate as the SARS-CoV-2 PLpro inhibitor. Several groups independently showed the potent inhibition of SARS-CoV-2 PLpro by GRL0617 (4).[45,53,54,68,78] However, the moderate to weak antiviral activity of GRL0617 (4) prevents it from advancing to animal model studies.[54,67] Since the beginning of the COVID-19 pandemic, encouraging progress has been made in redesigning GRL0617 analogues as potent SARS-CoV-2 PLpro inhibitors. The X-ray crystal structure of SARS-CoV-2 PLpro in complex with GRL0617 (4) has also been solved by multiple groups,[53,54,56,57,78] paving the way for structure-based lead optimization.

A recent elegant structure-based drug design led to the discovery of potent PLpro inhibitors with favorable PK properties.[67] One of the major contributions of this study is the conversion of naphthalene to 2-phenylthiophene, which leads to improved PK properties. In addition, the thiophene substitution extends further into the BL2 groove (Figure A), and when it was coupled with additional substitutions on the aniline amine to engage interaction with Glu167 (Figure B,C), multiple nanomolar PLpro inhibitors have been identified. Among the more than 100 analogues tested, compounds ZN-3-80 (5), XR8-24 (6), and XR8-23 (7) were the most potent ones, with IC50 values of 0.59, 0.56, and 0.39 μM, respectively (Table ). Compounds 6 and 7 also showed a significantly improved antiviral activity against SARS-CoV-2 in A549-hACE2 cells, with EC50 values of 1.2 and 1.4 μM, respectively. In comparison, GRL0617 (4) was not active in the virus yield reduction antiviral assay (EC50 > 20 μM). The complex structure with compound XR8-24 (6) (PDB: 7LBS) revealed several key hydrogen bonds/electrostatic interactions, including the water-mediated hydrogen bonds between the pyrrolidine NH+ and the main-chain carbonyl oxygen of Tyr264 (not shown), the electrostatic interaction between the NH2+ from the azetidine ring and side-chain carboxylate from Glu167 (Figure C), and the hydrogen bond between the amide NH from compound XR8-24 (6) with the Asp164 side-chain carboxylate. When dosed in male C57BL/6 mice at 50 mg/kg by intraperitoneal injection (i.p.), compounds XR8-23 (7) and XR8-24 (6) reached Cmax = 6130 and 6403 ng/mL, respectively, indicating favorable in vivo bioavailability. Further optimization might lead to candidates that are suitable for in vivo antiviral efficacy studies.

Figure 5

GRL0617-based SARS-CoV-2 PLpro inhibitors. (A) X-ray crystal structure of SARS-CoV-2 PLpro with GRL0617 (4) (PDB: 7JRN). (B) Design hypothesis for the 2-phenylthiophene series of PLpro inhibitors based on GRL0617 (4). (C) X-ray crystal structure of SARS-CoV-2 PLpro with compound XR8-24 (6) (PDB: 7LBS). (D) X-ray crystal structure of SARS-CoV-2 PLpro with compound 9 (PDB: 7E35). (E) X-ray crystal structure of SARS-CoV-2 PLpro with Jun9-72-2 (12) (PDB: 7SDR). (F) X-ray crystal structure of SARS-CoV-2 PLpro with Jun9-84-3 (13) (PDB: 7SQE). Panels A and C were adapted with permission from ref (67). Copyright 2022 American Chemical Society.

In another study, Shan et al. reported the structure-based design of SARS-CoV-2 PLpro inhibitors based on the GRL0617 scaffold.[79] The most potent lead compound, 8, inhibited PLpro and SARS-CoV-2 viral replication, with IC50 = 0.44 μM and EC50 = 0.18 μM, respectively (Table ). The Kd was 2.60 μM for compound 8 in the SPR assay, compared to Kd = 10.79 μM for GRL0617 (4). In the counter screening against 10 deubiquitinases (DUBs) or DUB-like proteases, compound 8 was highly selective toward PLpro and did not show significant inhibition toward a panel of host DUBs and DUB-like proteases. The X-ray crystal structure of PLpro with an analogue, 9, showed that compound 9 binds to PLpro in a similar mode as GRL0617 (4) (Figure D). It is noted that compound 9 adapts different binding poses in the two monomers (Figure D).

Our group recently conducted a HTS against SARS-CoV-2 PLpro using the FRET-based enzymatic assay.[54] Two closely related compounds, Jun9-13-7 (10) and Jun9-13-9 (11), were identified as potent hits, with IC50 values of 7.9 and 6.67 μM, respectively (Table ). Subsequent lead optimization led to the discovery of several compounds with IC50 values in the submicromolar range, including Jun9-72-2 (12) (IC50 = 0.67 ± 0.08 μM) and Jun9-84-3 (13) (IC50 = 0.67 ± 0.14 μM). In the cell-based FlipGFP reporter assay, Jun9-72-2 (12) and Jun9-84-3 (13) showed dose-dependent inhibition, with EC50 values of 7.93 and 17.07 μM, respectively, suggesting both compounds are cell-membrane-permeable and can inhibit the intracellular protease activity of PLpro. In agreement, both compounds had potent antiviral activity against SARS-CoV-2 in Vero E6 and Caco2-hACE2 cells (Table ). Significantly, there is a positive correlation between the FlipGFP assay results and the antiviral assay results, validating the FlipGFP as a surrogate assay for the prediction of the antiviral activity of PLpro inhibitors.[54] In the X-ray crystal structure of PLpro with Jun9-72-2 (12) (PDB: 7SDR), the tertiary NH+ in the linker electrostatically interacts with the Asp164 carboxylate group (Figure E). The X-ray crystal structure of PLpro with Jun9-84-3 (13) (PDB: 7SQE) revealed an additional hydrogen bond between the indole NH and the Glu167 side-chain carboxylate (Figure F).

Additional GRL0617 analogues, 14–19, have been reported as SARS-CoV-2 PLpro inhibitors (Table );[52,53,68] however, no significant improvement has been made.

Non-GRL0617 Inhibitors

Three phenolic compounds—methyl 3,4-dihydroxybenzoate (HE9, 20), 4-(2-hydroxyethyl)phenol (YRL, 21), and 4-hydroxybenzaldehyde (HBA, 22)—were identified as allosteric SARS-CoV-2 PLpro inhibitors through a high-throughput X-ray crystallization.[55] The screened library contains 500 compounds from the International Center for Chemical and Biological Sciences (ICCBS) Molecular Bank. Interestingly, HE9 (20), YRL (21), and HBA (22) all bind to the ISG15/Ub-S2 binding site of PLpro (Figure A), an allosteric binding pocket that has not been explored for drug design. The allosteric binding site is located about 30 Å away from the active-site residue Cys111. The superimposition structures of PLpro + inhibitors and PLpro + ISG15 indicate that these compounds might compete with ISG15 for the same binding site. As expected, all three compounds inhibited the deISGylating activity of PLpro, with IC50 values of 3.76 ± 1.13 μM (20), 6.68 ± 1.20 μM (21), and 3.99 ± 1.33 μM (22). However, it remains unknown whether these compounds can inhibit the hydrolysis of viral polyprotein by PLpro. HE9 (20) and YRL (21) inhibited SARS-CoV-2 replication in Vero E6 cells in the qRT-PCR assay, with EC50 values of 0.13 and 1 μM, respectively. However, the antiviral assay results for HBA (22) were not conclusive. In the cytopathic effect (CPE) assay, HE9 (20) had an EC50 of 10.37 μM. In contrast, YRL (21) failed to show inhibition in the CPE assay. The discrepancy of antiviral activity in different assays suggests further validation is needed. Furthermore, these results raise the question of whether inhibiting the deISGylation activity of PLpro alone is sufficient for the inhibition of viral replication.

Figure 6

Non-covalent SARS-CoV-2 PLpro inhibitors that do not share structural similarity with GRL0617 (4). (A) X-ray crystal structures of SARS-CoV-2 PLpro in complex with fragments HE9 (20), YRL (21), and HBA (22). (B) X-ray crystal structure of SARS-CoV-2 PLpro in complex with proflavine (25), showing three molecules binding near the BL2 loop (PDB: 7NT4). Two molecules stack on top of each other and fit in the GRL0617 (4) binding pocket, and a third molecule binds at the backside of the BL2 loop. (C) X-ray crystal structures of SARS-CoV-2 PLpro in complex with YM155 (27) (PDB: 7D7L). YM155 (27) binds three different sites located at the zinc-finger domain, thumb domain, and the substrate-binding pocket. Detailed interactions between YM155 and the BL2 loop region residues are shown on the right side.

A drug repurposing screening by Napolitano et al. identified acriflavine (23) as a potent inhibitor of SARS-CoV-2 PLpro with in vivo antiviral efficacy.[82] Acriflavine (23) is a mixture of trypaflavine (24) and proflavine (25).[82] Acriflavine (23) inhibited PLpro with IC50 values of 1.66 and 1.46 μM when RLRGG-AMC and ISG15-AMC were used as substrates, respectively (Table ). Acriflavine (23) also inhibited the deubiquitylating activity of PLpro in gel-based assay, thus ruling out the potential fluorescence interference effect of acriflavine (23). In addition, acriflavine (23) did not inhibit Mpro. The X-ray crystal structure of PLpro with proflavine (25) was determined (PDB: 7NT4), revealing that two molecules of proflavine (25) bind to the S3–S5 pockets of PLpro simultaneously (Figure B). The BL2 loop folds inward toward the substrate-recognition cleft, similar to the binding mode of GRL0617 (4). A third proflavine (25) molecule is located at the surface of the protein on the opposite side of the BL2 loop. Acriflavine (23) inhibited SARS-CoV-2 replication in A549-ACE2 and Vero cells, with EC50 values of 86 and 64 nM, respectively. However, the selectivity index was low (A549-ACE2 SI = 36; Vero SI = 53). The antiviral activity was further confirmed in human airway epithelial (HAE) cells. Acriflavine (23) also showed potent inhibition against MERS-CoV (IC50 = 21 nM, SI = 162) and HCoV-OC43 (IC50 = 105 nM, SI = 27) but not the alphacoronaviruses, including feline infectious peritonitis virus (FIPV) and HCoV-NL63. In the in vivo SARS-CoV-2 infection model with K18-ACE2 mice, acriflavine (23) treatment by either i.p. or intramuscular (i.m.) injection significantly lowered the viral titers in the brain and the lung.

6-Thioguanine (6-TG, 26) was previously reported as an inhibitor for SARS-CoV and MERS-CoV;[92,93] therefore, it was hypothesized that it might also inhibit the SARS-CoV-2 PLpro. Swaim et al. recently demonstrated that 6-TG (26) is a potent inhibitor for SARS-CoV-2 in Vero E6 cells, with an EC50 of 2.13 μM (Table ).[94] Next, to confirm the intracellular inhibition of PLpro by 6-TG (26), a TAP-tagged pp1a protein consisting of nsp1, 2, and 3 was expressed. As expected, TAP-nsp1 was the major product due to the self-cleavage of the pp1a polyprotein by PLpro. Treatment with 6-TG (26) led to dose-dependent inhibition of the cleavage, with an IC50 of approximately 0.5 μM. In addition, 6-TG (26) showed potent inhibition of the deISGylation activity of PLpro in HEK293T cells. No in vitro enzymatic assay was performed. In addition, it was proposed that 6-TG (26) might have a secondary mechanism of action by inhibiting the viral RNA synthesis. Nonetheless, in our recently hit validation study, 6-TG (26) did not show inhibition against SARS-CoV-2 PLpro in the enzymatic assay (IC50 > 50 μM), had no binding to PLpro in the thermal shift assay, and did not inhibit the intracellular PLpro activity in the FlipGFP assay.[69] Therefore, the antiviral activity of 6-TG (26) may not arise from inhibiting the PLpro.

Through screening a library of 6000 compounds using the FRET-based enzymatic assay with the Arg-Leu-Arg-Gly-Gly-AMC substrate, Zhao et al. identified YM155 (27) (IC50 = 2.47 ± 0.46 μM), cryptotanshinone (28) (IC50 = 5.63 ± 1.45 μM), tanshinone I (29) (IC50 = 2.21 ± 0.10 μM), and GRL0617 (4) (IC50 = 1.39 ± 0.26 μM) as SARS-CoV-2 PLpro inhibitors (Table ).[80] All four compounds displayed potent antiviral activity against SARS-CoV-2 in Vero E6 cells, with the most potent compound being YM155 (27) (EC50 = 0.17 ± 0.02 μM, CC50 ≈ 400 μM). The structure of PLpro in complex with YM155 (27) was solved by crystal soaking (PDB: 7D7L). Unexpectedly, YM155 (27) was found in three different binding sites: the orthosteric site, the thumb domain, and the zinc-finger domain (Figure C). The binding at the thumb domain is expected to inhibit the binding between PLpro and ISG15. A conformational change was observed at the zinc-finger domain upon YM155 (27) binding, but the physiological relevance of this binding mode has not been validated.

Similarly, cryptotanshinone (28) (IC50 = 1.34 μM) and two other analogues, dihydrotanshinone I (30) (IC50 = 0.59 μM) and tanshinone IIA (31) (IC50 = 1.57 μM), were shown as potent SARS-CoV-2 PLpro inhibitors through a HTS (Table ).[81] In addition, four additional compounds, PKK1/Akt/Flt dual pathway inhibitor (32) (IC50 = 0.26 μM), Ro 08-2750 (33) (IC50 = 0.53 μM), Cdk4 inhibitor III (34) (IC50 = 0.39 μM), and β-lapachone (35) (IC50 = 0.61 μM), were also identified as potent PLpro inhibitors (Table ). Dihydrotanshinone I (30) inhibited SARS-CoV-2 with an EC50 of 8.15 μM. Unexpectedly, cryptotanshinone (28) and tanshinone IIA (31) had no antiviral activity (EC50 > 200 μM), despite their potent enzymatic inhibition. The antiviral result of cryptotanshinone (28) is also in controversy with the previous study, which showed that cryptotanshinone (28) is a potent antiviral with an EC50 of 0.7 μM.[80] Further validation is needed to test the antiviral activity of cryptotanshinone (28) against SARS-CoV-2 in multiple cell lines.

Xu et al. recently reported the discovery of tanshinone IIA sulfonate (36) and chloroxine (37) as SARS-CoV-2 PLpro inhibitors from a drug-repurposing screening.[83] Tanshinone IIA sulfonate (36) was identified in the fluorogenic assay using the ALKGG-AMC substrate, with an IC50 of 1.65 μM (Table ). Chloroxine (37) was discovered in the fluorescence polarization-based assay using the fluorescein 5-isothiocyanate (FITC)-labeled ISG15, with an IC50 of 7.24 μM. Tanshinone IIA sulfonate (36) and chloroxine (37) also showed binding to PLpro in the biolayer interferometry and thermal shift assays. The antiviral activity against SARS-CoV-2 was not reported.

We performed hit validations for YM155 (27), cryptotanshinone (28), tanshinone I (29), dihydrotanshinone I (30), and tanshinone IIA (31).[69] Our study found that YM155 (27) (IC50 = 20.13 μM), cryptotanshinone (28) (IC50 = 52.24 μM), tanshinone I (29) (IC50 = 18.58 μM), dihydrotanshinone I (30) (IC50 = 33.01 μM), and tanshinone IIA (31) (IC50 = 15.30 μM) had much higher IC50 values against SARS-CoV-2 PLpro in the FRET assay compared to the previous reports. The intracellular PLpro inhibition by YM155 (27) and cryptotanshinone (28) in the FlipGFP assay was not conclusive due to cell cytotoxicity, while tanshinone I (29), dihydrotanshinone I (30), and tanshinone IIA (31) had no intracellular PLpro inhibition at non-toxic concentrations. Collectively, our results suggest that YM155 (27), cryptotanshinone (28), tanshinone I (29), dihydrotanshinone I (30), and tanshinone IIA (31) are weak PLpro inhibitors and tanshinone I (29), dihydrotanshinone I (30), and tanshinone IIA (31) lack intracellular target engagement.

In agreement with our results, Brewitz et al. applied a mass spectrometry assay to monitor PLpro-mediated cleavage of the nsp2/3 substrate.[95] Among the list of compounds tested, YM155 (27), tanshinone I (29), and tanshinone IIA sulfonate sodium (36) were not active (IC50 > 50 μM), while cryptotanshinone (28) showed moderate activity, with an IC50 of 19.4 μM.

Through virtual screening of a library of naphthoquinoidal compounds followed by enzymatic assay validation, Santos et al. identified three compounds, 38, 39, and 40, as potent SARS-CoV-2 PLpro inhibitors, with IC50 values of 1.7, 2.2, and 3.1 μM, respectively (Table ).[84] Among the three hits, compound 40 had moderate inhibition against Mpro, with an IC50 of 66 μM; therefore, it was considered as a dual inhibitor for further optimization. Molecular dynamics (MD) simulations predicted that compound 39 binds non-covalently to the S3 and S4 subsites in PLpro. However, the detailed mechanism of action remains to be characterized. When tested in the antiviral assay against SARS-CoV-2 in two different cell lines, Vero E6 and HeLa-ACE2, none of the identified Mpro and PLpro inhibitors had antiviral activity, suggesting these naphthoquinoidal compounds might have off-target effects. It is noted that no reducing agent such as DTT was added in the Mpro enzymatic assay; however, 0.1 mM DTT was included in the PLpro assay. Therefore, the observed PLpro inhibition might not be due to non-specific modification of the PLpro C111 residue. Further validation studies are warranted to confirm their enzymatic inhibition.

Cho et al. reported SJB2-043 (41) as a SARS-CoV-2 PLpro inhibitor, with an apparent IC50 of 0.56 μM.[85] However, no complete inhibition was achieved at high drug concentration. Therefore, it remains to be validated whether SJB2-043 (41) is a specific PLpro inhibitor.

Commercial mouth rinses are known to inactivate SARS-CoV-2,[96,97] but the detailed mechanism remains elusive. Lewis et al. tested the active ingredients of mouth rinses against the SARS-CoV-2 Mpro and PLpro.[86] Although none of the compounds were active against Mpro, two compounds, aloin A (42) and aloin B (43), inhibited PLpro, with IC50 values of 13.16 and 16.08 μM, respectively, in the enzymatic assay. Aloin A (42) and B (43) also inhibited the deubiquitinating activity of PLpro, with IC50 values of 15.68 and 17.51 μM. MD simulations suggest that aloin A (42) and B (43) bind to the GRL0617 (4) binding site and mainly interact with Glu167, Tyr268, and Glu269.

Specific Covalent PLpro Inhibitors

The cleavage of PLpro substrate occurs after the second glycine in the Leu-X-Gly-Gly sequence.[57] As a result, the binding pockets for the S2 and S1 subsites are absent, which leaves the S4 and S3 subsites for inhibitor binding. Accordingly, to develop a covalent inhibitor to react with the catalytic C111, a linker is needed to conjugate the S4/S3 pocket binder with a reactive warhead.[53,57]

A positional scanning was conducted to identify the optimal substrate of SARS-CoV-2 PLpro.[57] A total of 19 natural and 109 non-proteinogenic amino acids were screened at each position. It was found that the P2 and P4 positions have high preference for glycine and hydrophobic residues, respectively, while the P3 position can tolerate both charged residues, including Phe(guan), Dap, Dab, Arg, Lys, Orn, and hArg, and hydrophobic residues, including hTyr, Phe(F5), Cha, Met, Met(O), Met(O)2, and d-hPhe. Leveraging this information, two covalent inhibitors, VIR250 (44) (Ac-Abu(Bth)-Dap-Gly-Gly-VME) and VIR251 (45) (Ac-hTyr-Dap-Gly-Gly-VME), were designed by incorporating the optimal P3 and P4 substitutions with the vinyl methyl ester (VME)-reactive warhead (Table ). VIR250 (44) and VIR251 (45) showed dose-dependent inhibition against both SARS-CoV-2 and SARS-CoV PLpros; however, the IC50 values were not quantified. The X-ray crystal structures of SARS-CoV-2 PLpro in complex with VIR250 (44) (PDB: 6WUU) and VIR251 (45) (PDB: 6WX4) were solved (Figure ), revealing covalent thioether linkage of the C111 thiol and the β carbon of the vinyl group. Although no antiviral assay results were reported, this is an elegant rational design that led to the first covalent SARS-CoV-2 PLpro inhibitors.

Figure 7

X-ray crystal structures of SARS-CoV-2 PLpro in complex with peptidomimetic covalent inhibitors VIR250 (44) (PDB: 6WUU) (A) and VIR251 (45) (PDB: 6WX4) (B).

Sanders et al. recently reported the rational design of the first-in-class drug-like covalent SARS-CoV-2 PLpro inhibitors.[87] An N,N′-diacetylhydrazine linker was designed as a mimetic of the Gly-Gly to conjugate the GRL0617 methyl group with different reactive warheads. A series of commonly used cysteine-reactive warheads, including fumarate methyl ester, chloroacetamide, propiolamide, cyanoacetamide, and α-cyanoacrylamide, have been exploited. Among the designed covalent PLpro inhibitors, compounds 46 and 47 with the fumarate methyl ester, and compound 48 with the propiolamide, showed significantly improved potency, with IC50 values of 0.094, 0.230, and 0.098 μM, respectively (Table ). Compound 49, with the chloroacetamide, and compound 50, with the cyanoacetamide, were less active, with IC50 values of 5.4 and 8.0 μM, respectively. In contrast, compound 51, with the α-cyanoacrylamide, was not active (IC50 > 200 μM). As expected, covalent protein adducts with inhibitors were observed for compounds 46–50 in electrospray ionization (ESI) mass spectrometry. The X-ray crystal structure of PLpro with compound 46 was solved at 3.10 Å resolution (PDB: not released), showing a covalent adduct between the C111 thiol and the C1 of compound 46. The N,N′-diacetylhydrazine linker forms four hydrogen bonds with Gly163 and Gly271, highlighting the importance of this rationally designed linker. In SARS-CoV-2-infected Vero E6 cells, compound 46 had an EC50 of 1.1 μM, which is comparable to the potency of remdesivir (EC50 = 0.74 μM). Surprisingly, compound 47 had insignificant cytoprotective effects, despite its potent enzymatic inhibition. Compound 48 was cytotoxic; therefore, its antiviral activity was not conclusive. Similar to GRL0617 (4), compound 46 also inhibited the deubiquitinating and the deISGylating activities, with IC50 values of 76 and 39 nM, respectively. Selectivity screening against a panel of DUBs showed that compound 46 is highly selective, and no inhibition was observed up to 30 μM. In vitro PK profiling showed that compound 46 is stable in human liver S9 and microsomes, with T1/2 = 60 and 50 min, respectively. This study represents the first rational design of drug-like covalent PLpro inhibitors with potent antiviral activity, and the X-ray crystal structures are invaluable in guiding the lead optimization.

Liu et al. reported the design of peptide–drug conjugates (PDCs) as covalent inhibitors of SARS-CoV-2 PLpro.[88] The PDCs consist of GRL0617 and cyclic sulfonium-containing peptides derived from PLpro substrate Leu-Arg-Gly-Gly (Table ). The sulfonium serves as a warhead and is designed to react with the C111. Among the examined PDCs, EM-C (52) and EC-M (53) were the most potent against SARS-CoV-2 PLpro, with IC50 values of 7.40 ± 0.37 and 8.63 ± 0.55 μM, respectively (Table ). Both conjugates were cell-membrane-permeable and inhibited the deISGylating activity of PLpro. In-gel digestion of the PLpro + PDC mixture followed by MS/MS analysis confirmed that C111 is the enriched conjugation site. No antiviral assay results were presented. Although the results presented convincingly demonstrated the covalent labeling of PLpro C111, their binding mode remains unknown. The EC-M (52) and EM-C (53) PDCs contain the GRL0617 and the Leu-Arg dipeptide sequence, both of which are S3 and S4 subsite binders. It is not clear why the design contains duplicate binding elements. The X-ray crystal structure might solve the puzzle.

A tryptophan-containing dipeptide, compound 54, was recently reported as a dual inhibitor of SARS-CoV-2 Mpro and PLpro.[89] Compound 54 inhibited Mpro and PLpro with IC50 values of 1.72 and 0.67 μM, respectively, while it had no binding to the viral spike protein (KD > 25 μM). In the antiviral assay, compound 54 inhibited two SARS-CoV-2 clinical isolates, UC-1074 and UC-1075, with EC50 values of 0.32 and 1.37 μM, respectively. Given the lack of structural similarities between Mpro and PLpro, coupled with the high reactivity of the α-chloroacetamide warhead in 54, it remains to be investigated whether the inhibition of Mpro and PLpro by compound 54 is specific. Nevertheless, the potent antiviral activity of compound 54 is encouraging, which warrants further optimization.

Non-specific Covalent PLpro Inhibitors

Ebselen Analogues

Given the broad-spectrum antiviral activity of ebselen against several viruses, Weglarz-Tomczak et al. explored ebselen and its analogues as SARS-CoV-2 PLpro inhibitors.[90] Ebselen (55) inhibited PLpro with an IC50 of 2.02 ± 1.02 μM, and dialysis experiment showed that no enzymatic activity was recovered, suggesting irreversible inhibition. Subsequently, a library of analogues was designed, among which two ebselen derivatives, 56 (IC50 = 236 ± 107 nM) and 57 (IC50 = 256 ± 35 nM), and two diselenide orthologs, 58 (IC50 = 339 ± 109 nM) and 59 (IC50 = 263 ± 121 nM), had improved enzymatic inhibition against SARS-CoV-2 PLpro compared to ebselen (55) (IC50 = 2.02 ± 1.02 μM) (Table ). In this study, 2 mM DTT was added in the enzymatic assay buffer. However, our previous studies showed that ebselen (55) only inhibited SARS-CoV-2 PLpro in the absence of DTT but not with DTT.[65] This discrepancy needs to be further validated.

In another study, a similar strategy has been exploited for the development of dual inhibitors targeting both SARS-CoV-2 Mpro and PLpro based on the ebselen scaffold.[91] Among the 23 ebselen analogs, seven showed dual inhibition with the Mpro IC50 values in the nanomolar range and the PLpro IC50 values in the single digit to submicromolar range (60–66, Table ). No reducing reagent was added in either the Mpro or the PLpro enzymatic assay. The antiviral activity of the potent hits was not reported. Nonetheless, ebselen (55) was previously reported to inhibit SARS-CoV-2 replication, with an EC50 value of 4.67 μM in the plaque assay, albeit the proposed mechanism of action is through Mpro inhibition.[98]

The inconsistent PLpro enzymatic inhibitory activity of ebselen (55) and its analogues from several groups, coupled with their antiviral activity against SARS-CoV-2, suggest that further characterizations are needed to confirm their cellular PLpro target engagement and additional targets that might contribute to the antiviral activity.

Zinc Ejector

PLpro contains a zinc-binding domain (ZBD) in which the zinc ion is coordinated by four conserved cysteine residues: Cys189, Cys192, Cys224, and Cys226. The ZBD is essential for the structural integrity and hence the enzymatic activity of PLpro. As such, the cysteine-rich ZBD was also proposed as a putative drug target.[99]

Disulfiram (67) and ebselen (55), together with 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB, 68), 2,2′-dithiodipyridine (69), and 2,2′-dithiobis(benzothiazole) (70), were found to eject zinc from PLpro, as shown by the increase in fluorescence emission signal from the zinc-specific fluorophore, FluoZin-3.[100] The matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrum further confirmed formation of the covalent adduct between disulfiram and ebselen with PLpro and nsp10. The LC-MS/MS experiment mapped the ebselen and disulfiram conjugation sites to C189 and C192, both of which are involved in zinc chelation in the ZBD of PLpro. In the FRET-based enzymatic assay, disulfiram (67) and ebselen (55) inhibited PLpro, with IC50 values of 7.52 and 2.36 μM, respectively. It is noted that the enzymatic inhibition might be a combined effect of targeting both the catalytic C111 and the cysteines in the ZBD. A combination experiment showed that ebselen and disulfiram had synergistic antiviral effects when combined with hydroxychloroquine. This study suggested that clinically safe zinc ejectors could potentially target the conserved ZBD in multiple viral proteins and could potentially be exploited as broad-spectrum antiviral drug candidates. Following studies from the same group further showed that disulfiram (67) and ebselen (55) are zinc-ejectors of the SARS-CoV-2 nsp13 and nsp14 and consequently inhibit nsp13 ATPase and nsp14 exoribonuclease activities.[101] The antiviral activity of ebselen (55) and disulfiram (67) against SARS-CoV-2 was synergistic with remdesivir.

As discussed above, ebselen analogs have also been extensively exploited as Mpro and PLpro inhibitors by targeting the active-site cysteine.[102,103] Combined with the zinc-ejecting property, the antiviral activity of ebselen (55) and its derivatives might be due to its polypharmacology in targeting the ZBD, PLpro, Mpro, and others.

Perspectives on Targeting the SARS-CoV-2 PLpro

The COVID-19 pandemic is a timely call for the immediate need for antivirals. As the SARS-CoV and MERS-CoV epidemics subsided, the interest in developing coronavirus inhibitors unfortunately waned, and no significant efforts were devoted to optimizing the hits identified from early high-throughput screening campaigns. Nevertheless, the COVID-19 pandemic re-ignited the interest in PLpro drug discovery, and the past 2 years have seen encouraging progress in the field. Although drug repurposing largely failed to identify potent and selective PLpro inhibitors, rational design based on the X-ray crystal structures led to major breakthroughs, including the design of 2-phenylthiophene PLpro inhibitors with favorable PK properties and the first-in-class covalent PLpro inhibitors since the pandemic. In light of this encouraging progress, we hereby share our opinions on the further development of SARS-CoV-2 PLpro inhibitors, and we hope to clarify some of the confusions in the field based on our experience.

First, there is a need to broaden the antiviral spectrum of PLpro inhibitors to target MERS-CoV. The BL2 loop located at the drug-binding site is poorly conserved among SARS-CoV and MERS-CoV,[104] explaining the lack of activity of the GRL0617 (4) series of compounds against MERS-CoV PLpro. No potent and specific MERS-CoV PLpro inhibitors have been reported until now. In the search for PLpro inhibitors with a broader spectrum of antiviral activity, it is worthwhile to include MERS-CoV PLpro in the secondary assays. It might be possible to identify allosteric inhibitors with dual inhibitions against both SARS-CoV-2 PLpro and MERS-CoV PLpro. Alternatively, PLpro inhibitors can be developed specifically for SARS-CoV-2 and SARS-CoV, and MERS-CoV PLpro inhibitors can be pursued separately.

Second, structurally disparate PLpro inhibitors are needed to advance PLpro inhibitors to the clinic. Compared to PLpro, Mpro is a more amenable drug target, and structurally disparate inhibitors have been identified from HTS as potent Mpro inhibitors. In contrast, several recent HTS failed to identify additional potent and selective SARS-CoV-2 PLpro inhibitors other than GRL0617 analogues.[67,85] GRL0617 (4) contains the naphthalene ring, which is a known metabolic labile group and a possible toxicophore.[105] Therefore, it might present a challenge in PK optimization. To increase the chances of success, additional structurally disparate PLpro inhibitors are needed as backups. The recent elegantly designed 2-phenylthiophene and the covalent PLpro inhibitors are prominent examples in this direction.[67,87]

Third, target selectivity needs to be addressed at an early stage of development. Although there is a lack of sequence or structural similarity between PLpro and human DUBs, both PLpro and human DUBs bind ubiquitin at the extended C-terminus with the consensus sequence Leu-X-Gly-Gly, raising the potential concern about off-target effects of PLpro inhibitors against human DUBs.[106] Consequently, it is important to conduct counter screening of PLpro inhibitors against a panel of related human DUBs to avoid potential toxicity. Along this line, counter screening should also be conducted with other cysteine proteases like the Mpro, cathepsin L, calpains, etc. to rule out promiscuous inhibitors that non-specifically inhibit unrelated proteases.

Fourth, be aware of promiscuous inhibitors and compounds with polypharmacology. Promiscuous compounds are defined as compounds that lack a defined mechanism of action or compounds that showed inconsistent results in different assays. PLpro is a cysteine protease that is prone to non-specific inhibition by redox cycling compounds (quinone, arylsulfonamide, tolyl-hydrazide, etc.),[107,108] alkylating reagents, and other pan-assay interference compounds (PAINS).[109−111] In addition, compounds such as acriflavine and YM155 are cationic amphiphilic drugs (CADs), which could cause phospholipidosis and disturb endosome/lysosome functions. This effect may explain the improved antiviral potency over biochemical potency. In this regard, the antiviral activity of acriflavine and YM155 might be a combined effect of PLpro inhibition and endosome/lysosome disruption. Furthermore, it is better to perform the antiviral assays in different cell lines, especially in physiologically relevant cell lines such as Calu3 or normal human airway epithelial cells. This eliminates the cell-type-dependent antiviral activity of certain compounds.

Fifth, for translational drug discovery, we need to differentiate chemical probes from drug candidates. Compounds such as ebselen and disulfiram, having non-specific inhibition against PLpro and Mpro as well as other unrelated cysteine proteases, should not be classified as PLpro inhibitors. Nevertheless, this does not indicate that these promiscuous compounds should not be further pursued as SARS-CoV-2 antivirals. Instead, they should be defined as chemical probes for mechanistic studies. The aforementioned cell-based protease assays, such as the FlipGFP and Protease-Glo luciferase assays, are valuable tools to help rule out promiscuous compounds like ebselen and disulfiram and delineate the cellular target engagement of the specific PLpro inhibitors.

In summary, despite the encouraging progress in the past 2 years, there is still a long journey to advance PLpro inhibitors to the clinic. No rationally designed drug-like PLpro inhibitors have been shown to have in vivo antiviral efficacy against SARS-CoV-2 infection in animal models yet. In addition to the RdRp and Mpro inhibitors, PLpro inhibitors are expected to enrich our armamentarium in flighting the current COVID-19 pandemic and future unforeseeable coronavirus outbreaks. Combination experiments need to be planned to characterize the combination therapy potential of PLpro inhibitors with RdRp or Mpro inhibitors. Furthermore, the knowledge accumulated in developing SARS-CoV-2 PLpro inhibitors can be similarly applied to MERS-CoV PLpro.