Targeting SARS-CoV-2 Proteases and Polymerase for COVID-19 Treatment: State of the Art and Future Opportunities.

The newly emerged coronavirus, called SARS-CoV-2, is the causing pathogen of pandemic COVID-19. The identification of drugs to treat COVID-19 and other coronavirus diseases is an urgent global need, thus different strategies targeting either virus or host cell are still under investigation. Direct-acting agents, targeting protease and polymerase functionalities, represent a milestone in antiviral therapy. The 3C-like (or Main) protease (3CLpro) and the nsp12 RNA-dependent RNA-polymerase (RdRp) are the best characterized SARS-CoV-2 targets and show the highest degree of conservation across coronaviruses fostering the identification of broad-spectrum inhibitors. Coronaviruses also possess a papain-like protease, another essential enzyme, still poorly characterized and not equally conserved, limiting the identification of broad-spectrum agents. Herein, we provide an exhaustive comparative analysis of SARS-CoV-2 proteases and RdRp with respect to other coronavirus homologues. Moreover, we highlight the most promising inhibitors of these proteins reported so far, including the possible strategies for their further development.

Introduction

RNA viruses represent a large and heterogeneous group of human pathogens that periodically jeopardize the public health because of apparent sudden epidemic outbreaks. In the recent years, the World Health Organization (WHO) prioritized the most threatening viral diseases worldwide including Middle East respiratory syndrome (MERS), severe acute respiratory syndrome (SARS), Ebola, Zika, and Dengue[1] while advancing the theory of Disease X, a disease associated with pandemic risk caused by a still unknown pathogen.[2] Lastly, at the end of 2019, an outbreak of severe pneumonia with an uncertain etiopathology in China was indicated as the Disease X by WHO that soon after announced the identification of a new coronavirus (CoV), called 2019-nCoV, as the etiologic agent of COVID-19 disease.[3,4] Genomic sequence of 2019-nCoV has highlighted 79.5% identity to SARS-CoV genome and thereafter the new CoV is known as SARS-CoV-2. The infection has rapidly spread in many countries across Asia, Europe, and Americas, causing the pandemic COVID-19 disease, as announced by WHO on March 11 2020, that currently has caused almost 35 million cases, many hospitalizations and more than one million deaths.[5] The current scenario has highlighted major gaps in the abilities of most countries to face new virulent pathogens. In fact, the most effective procedure to control the spread of the infection is the social distancing: as a consequence, billions of people’s lives have been impacted following such countermeasures.[6−8] Moreover, COVID-19 will likely burden on worldwide health systems and on the global economy due to the lock-down measures, and the damage will be uncountable if the spread of the virus is not effectively controlled.[9] However, together with the social distancing and the lock-down, the availability of effective antiviral drugs would have an enormous beneficial impact on the control of the health emergency and the possibility to come back to normal life.

In this context, drug repurposing offers the opportunity to quickly identify a drug to prevent, control or, hopefully, eradicate the SARS-CoV-2 virus, although a serious clinical validation is required.[10] WHO has launched “Solidarity”, one of the largest international clinical trials aimed at identifying successful treatments against COVID-19.[11] Among the drugs under investigation, Remdesivir was included. It is an intravenous (iv) monophosphoramidate nucleoside prodrug polymerase inhibitor with broad-spectrum activity against RNA viruses, originally under investigation as an Ebola virus disease (EVD) medication. On the basis of the clinical trials results, Remdesivir was granted an emergency use authorization (EUA) for the treatment of patients with severe COVID-19 symptoms being the only pathogen specific drug approved for treatment of the disease at the moment.[12] The authorization was granted in the U.S. first, and then in a growing number of countries, including Japan, India, EU, UK, Australia, and South Korea, based on the promising results from the National Institute of Allergy and Infectious Diseases (NIAID), the Gilead clinical trials, and from the compassionate use programs.[13−15] The FDA subsequently broadened the scope of the existing EUA to include treatment of all hospitalized adult and pediatric patients with suspected or laboratory-confirmed COVID-19, irrespective of their severity of disease. Remdesivir eventually received the approval from the FDA for the treatment of COVID-19 patients requiring hospitalization.

Besides drug repurposing, which can offer the great advantage to translate safe-in-man drugs quickly into clinic with just little experimental evidence of antiviral activity, the rational design of new chemical entities to specifically act against key steps of viral replication, supported by appropriate biological and DMPK screening cascade, represents the main strategy to discover new potent and selective antiviral drugs.[16,17] Immediately after the understanding of the potential pandemic spread of the SARS-CoV-2 infection, the whole scientific community collaborated to characterize the viral genome and viable therapeutic targets in order to rapidly identify provisional treatments and, as a longer-term goal, find a cure.[18] Many efforts have been made in molecular biology, biochemistry, and structural biology of SARS-CoV-2, thus biochemical assays and 3D-structures have been made public for different viral proteins, while it has been shown that different cell lines are permissive toward viral infection thus allowing easy phenotypic screenings to identify antiviral compounds. Moreover, anti-SARS-CoV-2 drug discovery can take advantage from the knowledge and the biology tools acquired on other human pathogenic CoVs such as SARS-CoV and MERS-CoV, to cite a few, that can be utilized as counter screenings to develop broad-spectrum CoV inhibitors to potentially protect also from emerging infection caused by the same virus family.[18]

Researchers all around the world are exploiting different strategies targeting both viral and host factors essential for the pathogen replication to block one or more steps of its life cycles, and their efforts have already produced a great amount of data and studies. Thus, COVID-19 drug discovery is an emerging, rapidly evolving, and intriguing research field for medicinal chemists.[19]

This Perspective will aim to critically describe some of the most promising approaches toward the identification of anti-CoVs agents, especially focusing on the best characterized virus targets and the state-of-the-art on the antiviral agents identified, providing an unbiased point-of-view on the gaps to be filled. The inhibition of viral protease and polymerase functionalities represents a milestone in antiviral drug discovery field and provided most of the effective antiviral drugs for the treatment of HIV and hepatitis C virus (HCV) infections and also allowed for the development of antiviral agents acting against different virus families. Indeed, viral protease and polymerase inhibitors are included in the WHO list of essential drugs. Therefore, an exhaustive comparative analysis of SARS-CoV-2 proteases and RdRp with respect to other human pathogenic CoVs homologues is provided. Moreover, the most interesting compounds targeting the SARS-CoV-2, 3-chymotrypsin-like (3CL), or Main (M) protease (pro), and the nsp12 RNA-dependent RNA polymerase (RdRp), are analyzed in detail, discussing biological activity in vitro/in vivo, DMPK and clinical data, when available. Possible strategies for further development of the inhibitors are also discussed providing an unbiased point-of-view on the gaps to be filled. Additionally, potential strategies targeting the less explored papain-like (PL) pro will be described.

Biochemistry and Replicative Cycle of SARS-CoV-2

Belonging to the subfamily Coronavirinae, family Coronavirdiae, order Nidovirales, CoVs are single-stranded positive-sense (ss-(+)) RNA viruses, further clustered into four classes designated as alpha, beta, gamma, and delta.[20] Bats and rodents are the gene sources of most α-CoVs and β-CoVs, while avian species are those for most γ- and δ-CoVs, indeed, CoVs are able to cross species barriers, sometimes emerging as important human pathogens.[21] Before December 2019, six CoVs were known to infect humans, including α-CoVs, i.e., HCoV-229E and HCoV-NL63, and β-CoVs, such as HCoV-OC43, HCoV-HKU1, SARS-CoV, and MERS-CoV.[21] The β-CoV HCoV-OC43 and HCoV-HKU1 usually cause common cold disease, while SARS-CoV and MERS-CoV and the newly identified SARS-CoV-2, to different extents, may cause severe life-threatening conditions to lower respiratory tract such as pneumonia, but they may also affect the gastrointestinal system, heart, kidney, liver, and central nervous system, leading to multiple organ failure.[4] Moreover, SARS CoV-2, even if associated with a lower mortality rate, is more transmissible than SARS-CoV.[22] Genomic comparative analysis showed that SARS-CoV-2 sequence has identity of 79.5 and 96% to SARS-CoV and a bat CoV, SL-CoV-RaTG13, respectively; thus, its outbreak highlights the ability of viral spillover from animals to humans.[3]

CoVs share many morphological, biochemical, and functional aspects, which help in the understanding of SARS-CoV-2 biology and life cycle.[4,20] CoVs are relatively big (50–200 nm) enveloped viruses containing an encapsidated (ss-(+))-RNA, which is the largest viral genome found so far, with SARS-CoV-2 made-up by 30 kb (Figure ).[23] Similar to other CoVs, the SARS-CoV-2 viral membrane is studded with glycoprotein spikes (S), that give CoVs their crown-like appearance in electron microscope imaging.[23] The virion contains other three structural proteins, known as envelope (E), membrane (M), which form the viral envelope together with S, and nucleocapsid (N), a basic RNA-binding protein that complexes and protects the genome.[24]

Figure 1

Schematic representation of SARS-CoV-2 virion, viral entry, genome translation, and polyprotein processing.

The S protein mediates the virus attachment to the host cell membrane through subunit S1 interaction with the human angiotensin-converting enzyme 2 (ACE2), acting as the receptor for viral entry (Figure ).[25,26] The interactions between the two proteins have been fully elucidated with the support of several 3D structures of the S/ACE2 complex, highlighting the key residues involved in the recognitions process.[27−29]

The host proteins CatB/L, transmembrane protease serine 2 (TMPRSS2) have been shown to be involved in the entry process.[26] Thus, the identification of drugs targeting either the viral or the host factors could represent a valid strategy to inhibit viral entry.[30,31]

Upon the entry, the uncoating by nucleocapsid degradation allows the release into the cytoplasm of the viral RNA, ready for translation. The genome contains a 5′ cap structure with a leader sequence along with a 3′ poly(A) tail that allows it acting as mRNA. The 5′- and the 3′-ends contain highly stem loop structured untranslated regions (UTRs) required for RNA replication and transcription.[20] The 5′- and 3′-UTRs flank the coding region with the two-thirds of the genome from the 5′-end comprising two overlapping open reading frames (ORFs), ORF1a and ORF1b, that encode for polyproteins pp1a (4382 amino acids) and pp1ab (7073 amino acids), respectively. The autoproteolytically processing by 3CLpro and PLpro affords 16 nonstructural proteins (nsp1–16), which form the replicase/transcriptase complex (RTC) (Figure ). The terminal 10 kb toward the 3′-poly(A) tail encodes instead for the structural S, E, M, and N, following this order (Figure ).[24] Among the genes for structural proteins, there are those encoding for accessory proteins that show roles in viral pathogenesis and host immunity suppression, although they are not essential for in vitro replication, as found for other CoVs.[32] SARS-CoV-2 and SARS-CoV display very similar genetic composition, sharing the Orf1ab encoding for the 16 nsps and the four typical CoV structural proteins. They differ in some accessory proteins, with SARS-CoV-2 Orf3b and Orf10 showing low homology to SARS-CoV and an intact Orf8 instead of the two subunits, Orf8a and Orf8b, expressed by the latter virus.[20]

The RTC includes different enzymes and cofactors involved in post-translational polyprotein processing, RNA synthesis, maturation, and virions assembly and egress, which therefore can constitute ideal viral targets for drug discovery, being essential for the virus life cycle and devoid of a close host homologue.[24] The nsps that exert enzymatic activities are the nsp3 PLpro that catalyzes cleavages at the nsp1/2, nsp2/3, and nsp3/4 sites of the polyprotein, and the nsp5 3CLpro (or Mpro), which performs the remaining 11 proteolytic events.[24] PLpro additionally cleaves post-translational modified host proteins involved in innate immune response, thus contributing to immune escape. Predominantly PLpro can recognize and hydrolyze the cellular proteins ubiquitin (Ub) and the Ub-like protein ISG15 from interferon (IFN) responsive factor 3 (IRF3), blocking its nuclear translocation, thus reducing type I IFN response.[33] The nsp3, nsp4, and nsp6 have transmembrane domains and should serve as scaffold to anchor the RTC complex to intracellular membranes, a prerequisite for RNA virus replication. Then, the nsp7 nsp8 primase complex is able to synthesize short oligonucleotides, while the nsp12 RdRp is the elongating polymerase that, together with the nsp13 helicase/triphosphatase, the nsp14 exoribonuclease, the nsp15 endonuclease, and nsp10-N7- and nsp16–2′O-methyltransferases, assembles in the RTC, thus creating the suitable environment for RNA synthesis and maturation.[24] Moreover, nsp1, nsp3 deubiquitinase and nsp16–2′O-methyltransferase mediate suppression of the innate immune response.

Genomic ss-(+)-RNA transcription proceeds through (−)-strand intermediates that serve as templates for the production of both genomic and subgenomic RNAs, which are capped and polyadenylated as the full genomic RNA. The subgenomic RNAs are then translated into the four structural and some accessory proteins. Hence, discontinuous transcription may favor CoV genome recombination, as observed in the murine CoV experimental model.[34]

Assembly of new viral particles takes place on intracellular membranes, where N protein binds to genomic RNA and the nucleocapsid associates to M protein, which is thought to cause membrane curvature and drive budding into ER/Golgi membranes. During this process, S and E are also acquired. In particular, E protein by acting as viroporin alters cell secretory pathways and promotes virion egress by exocytosis.[24]

On the basis of the prominent roles in intracellular steps of viral life cycle, the amount of biochemical/structural data and the knowledge acquired on inhibitors of homologues proteins in other CoVs and other RNA viruses, the 3CLpro and the nsp12 RdRp are at moment the most relevant viral targets to identify specific anti-CoVs agents.

The 3CLpro (or Mpro): Structure and Function

The 3CLpro, known also as Mpro, is a 33.8 kDa cysteine protease able to process the polyprotein at no less than 11 conserved sites, starting with the autolytic cleavage from pp1a and pp1ab; thus, it is responsible for the release of most of the nsps CoV functional proteins. This pivotal role in the viral life cycle and the lack of homologous proteins in human cells make 3CLpro a very attractive target for the identification of anti-SARS-CoV-2 agents.

A large number of crystal structures of 3CLpro from SARS-CoV-2 and other CoVs have been solved either in their apo forms or in complex with inhibitors, providing very important information about the folding, the assembly, and the catalytic mechanism of these proteins. In around seven months, between the end of February and the end of September 2020, already 192 3D-structures of the SARS-CoV-2 3CLpro have been released. Comparison of the apo structures of SARS-CoV-2 3CLpro (PDB 6Y2E)[35] and SARS-CoV 3CLpro (PDB 2BX4)[36] reveals analogous tridimensional features, consistent with the 96% sequence identity and the similar catalytic activity of the two enzymes. Indeed, similar to SARS-CoV 3CLpro,[37−39] the active form of SARS-CoV-2 3CLpro is a dimer, with an estimated dissociation constant in the low μM range.[35] Each protomer is formed by three domains:[35,36] six antiparallel β-barrels form domains I and II (residues 10–99 and 100–182, respectively) and host the substrate-binding site, while domain III (residues 198–303) is a globular cluster of five helices that regulates protein dimerization. Arg4 and Glu290 from each protomer establish an ionic bond driving the formation of the dimer, which has contact interface between the perpendicularly oriented domain II of protomer A and domain III of protomer B (Figure ). The N-terminal tail, called the “N-finger”, of molecule B is squeezed in between domains II and III of the parent monomer and domain II of the other one. This peculiar arrangement is stabilized by some key hydrogen bonds, particularly those formed by Ser1 and Glu166 from each unit. In each monomer, the substrate binding site is located in a wide cleft between domains I and II, hosting the Cys145-His41 catalytic dyad. With respect to these residues, a number of subpockets can be identified, namely S4, S3, S2, S1, and S1′, which are occupied, respectively, by the substrate P4, P3, P2, P1, and P1′ amino acids. Notably, dimerization of 3CLpro seems to help the shaping of the substrate-binding site, particularly of the S1 site, explaining why the enzyme is catalytically inactive in the monomeric form, as demonstrated for the SARS-CoV enzyme.[37−39] The substrates of CoVs 3CLpro display almost identical recognition motifs, which is consistent with the high structural similarity of the catalytic domains of these enzymes.[40] In particular, they all share the Leu-Gln-Ser (Ala, Gly) as preferred P2–P1–P1′ sequence, suggesting the possibility to identify very effective broad-spectrum inhibitors of the CoVs enzyme family. The substrate preferences of SARS-CoV and SARS-CoV-2 3CL proteases were recently compared by examining the hydrolytic rate of different peptides/peptidomimetics.[41] In particular, the catalytic performance of each enzyme was evaluated against a combinatorial library of fluorogenic substrates endowed with a glutamine at the P1 position, while displaying different natural and unnatural amino acids as P2, P3, and P4. Remarkably, this analysis showed that both SARS-CoV and SARS-CoV-2 3CLpro have substrate specificity in S2 with Leu as the best P2 residue, although other hydrophobic replacers are tolerated, such as 2-Abz, Phe(4-NO2), 3-Abz, β-Ala, Dht, hLeu, Met, and Ile. Several hydrophobic D and L but also basic amino acids are suitable as P3, but Tle, d-Phe, d-Tyr, Orn, hArg, Dab, Dht, Lys, d-Phg, d-Trp, Arg, and Met(O)2 are mostly preferred.[40,41] Both enzymes have poor substrate specificity at the P4 position, with preference for small hydrophobic amino acids.

Figure 2

Overview of SARS-CoV-2 3CLpro architecture. The X-ray structure of SARS-CoV-2 3CLpro (PDB 6Y2G) is shown as a ribbon model in two different orientations. For clarity, the bound inhibitor has been removed. Protomers A (light-blue) and B (light-orange) associate into a dimer stabilized by a salt bridge between Glu290 and Arg4, while the substrate binding site resides at the interface of domains I and II. The catalytic residues Cys145 and His41 are highlighted.

The SARS-CoVs 3CLpro substrate affinity and specificity can be reasonably explained based on the shape and amino acid composition of the different subpockets of the cleavage site. In both SARS-CoVs 3CLpro, the S1 pocket is formed by the side chains of residues Phe140, Asn142, His163, Glu166, and His172 and the main chains of Phe140 and Leu141. Interestingly, the imidazole of His163 is located at the very bottom of the cleft, suitably positioned to donate a H-bond to the side chain carbonyl of substrate/inhibitor P1 Gln. The S2 subsite of SARS-CoVs 3CLpro is a buried hydrophobic pocket that can host bulky alkyl/aryl substituents as the substrate S1 Leu side chain. This cage is defined by a “lid” comprising the 310 helix residues 46–51, particularly Met49, three walls defined by the main-chain of residues 186–188 and by the side chains of His41, Asp187, and Gln189 and a floor lined by Met165. Notably, the shape and the size of the S2 subsite of SARS-CoVs 3CLpro is highly similar to that of the MERS-CoV homologue. Indeed, only two conservative mutations can be found, specifically the replacement of Met49 and Arg188 in SARS-CoVs with Leu49 and Lys191 in MERS-CoV, respectively. On the other hand, the volume of the S2 subsite in SARS-CoVs 3CLpro (252 Å3) is significantly larger than in other CoVs’ homologues of the α-genus, such as the HCoV-NL63 3CLpro (45 Å3).[42] Actually, this knowledge might be exploited to guide the structure-based drug design (SBDD) of either specific or broad spectrum inhibitors of CoVs 3CLpro. Compared to S1 and S2, the S4, S3, and the S1′ pockets of the SARS-CoVs 3CLpro are more shallow and exposed to the solvent and could accommodate groups of various size and nature. In particular, S3 and S4 are defined by the flexible loops connecting residues 165–168 and 189–192, which can rearrange upon ligand binding, while the S1′ cleft is characterized by a number of threonine residues (Thr24, Thr25, and Thr26), which can potentially form either hydrogen bonds or lipophilic contacts with the substrate/inhibitor P1′ group.

The structure of CoVs 3CLpro resembles that of the main proteases of enteroviruses (EVs), a family of small and naked (+)-ss-RNA viruses.[42,43] In fact, in EVs genome the Mpro is a cysteine protease encoded by the 3C region, hence the name 3C-like of CoVs protease. The EV 3C protein has a chymotrypsin-like fold with two catalytic domains similar in shape to domains I and II of CoVs 3CL. Contrary to CoVs 3CLpro, the EV 3C protease lacks a dimerization domain, and it works as a monomer. Also, the EV 3C protease displays, in addition to the catalytic Cys-His, an acidic (Asp/Glu) residue to form a reactive triad. A very important similarity between CoVs and EVs Mpro is the peculiar almost absolute requirement for glutamine in the substrate P1. This represents an unknown specificity in human proteases that increases the appeal of these viral proteins as a target for safe and selective inhibitors. Indeed, Rupintrivir (Supporting Information, Figure S1), a potent peptidic inhibitor of EV 3Cpro, successfully completed phase II trials for human rhinovirus (HRV) infection in 1999 without showing toxicity in common cold patients.[44] This compound has inspired design of CoV 3CLpro inhibitors, although it shows very weak activity when tested against the CoV protease.[45] A major difference between the 3C and the 3CL proteins is in the S2 subsite, which completely lacks the lid in the 3C protein. For instance, a comparison of SARS-CoV 3CL and EV Coxsackie B 3C highlighted that in EVs the S2 appears as an open hydrophobic channel lined by Arg39, Asn69, and Glu71, forming the back wall, residues 127–132 and His40 as the side walls, and Val162 constituting the floor.[42] Nonetheless, covalent peptidic inhibitors reported for both targets are able to broadly inhibit EVs and CoVs replication in cell cultures.[43,45]

Targeting SARS-CoV-2 3CLpro

The targeting of proteases represents a solid route for antiviral drug discovery as demonstrated by the therapeutic success of HIV and HCV proteases inhibitors. Inhibition of these enzymes can be achieved by different approaches; indeed, the design of reversible, covalent reversible, and irreversible binders has been exploited in different therapeutic areas including antivirals. The identification of potent noncovalent inhibitors generally requires much efforts and it is a time-consuming process, although it is the most favored strategy to obtain safe and efficacious drugs. On the other hand, the identification of irreversible inhibitors is an apparently simpler and attractive approach, but it suffers from a potential serious drawback due to lack of target selectivity that can lead to unpredictable severe toxicity. A similar but less risky approach is represented by the design of covalent reversible inhibitors that requires the knowledge of the catalytic mechanism and the substrate specificity, thus relying on natural substrate modifications. In general, such an approach consists of converting good substrates into good covalent reversible inhibitors. This can be achieved by replacing part of the protease substrate with a moiety, known as a “warhead”, that is an electrophilic reactive group able to form the covalent reversible bond with the catalytic machinery. Indeed, this strategy has afforded the FDA approved α-ketoamide inhibitors of HCV NS3/4A serine protease, Telaprevir[46] and Boceprevir.[47] In fact, the reaction of these compounds with the enzyme affords stable hemiketals mimicking the tetrahedral transition state of the cleavage process. This approach also yielded human cysteine protease cathepsin K inhibitors, such as the clinical candidate Odanacatib,[48] which features a nitrile warhead, for the treatment of osteoporosis and bone metastasis, even if the phase III clinical trial was discontinued in 2016. Similar to irreversible binders, covalent reversible inhibitors might suffer of lack of high selectivity; furthermore, they have generally less favorable pharmacokinetic properties that can lead to a suboptimal dosing regimen. However, because these drugs are designed based on the substrate sequences, they have higher specificity for target proteins, which reduces their potential side toxicity. In addition, considering that the length of COVID-19 patients treatment should be relatively short, this approach appears as a privileged route to discovery drugs targeting SARS-CoV-2 3CLpro with limited side effects compared to the use of covalent binders for chronic treatments, and the suboptimal dosing regimen can be tolerated. Homologous proteins of related CoVs such as SARS- and MERS-CoVs have represented the main targets for the identification of potent and selective inhibitors of these viruses.[45,49,50] Indeed, several compounds targeting the 3CLpro of different CoVs have been reported over the years, thus providing compounds and knowledge useful for the discovery of anti-SARS-CoV-2 agents.

As introduced in the previous section, several crystal structures of ligands bound to 3CLpro have been released in the PDB, including complexes containing peptidic covalent reversible/irreversible inhibitors. These inhibitors act through a two-step mechanism: first, they bind to the active site, similar to a natural substrate, forming a noncovalent complex with the protease so that the C-terminal warhead comes in close proximity to the catalytic Cys145, to form, in the second step, a covalent bond with this residue by a nucleophilic attack. A number of academic groups have reported peptide covalent reversible inhibitors of SARS-CoV-2 3CLpro, biological activities, and their cocrystal structures with the target protein; in these compounds, the explored chemical warheads include Michael acceptors, α-ketoamides, aldehydes, and ketones. To efficiently compete with the natural substrates at the catalytic site, most of the inhibitors span from P1–P4 to establish a considerable number of specific interactions with the binding pockets (namely, the enzyme S1–S4 subsites), to facilitate the covalent reaction between the Cys145 and the warhead (Figure ). This contributes not only to the affinity and potency, but, more importantly to the binding specificity, limiting the cross reactivity with other off targets that could lead to severe side effects. In addition, the reported inhibitors share similar peptide sequence organization in P1–P2 residues, while there are remarkable differences at P3/P4 positions and in the overall compound molecular size (Figure ).

Figure 3

Schematization of the main features of 3CLpro inhibitors and protease subsite specificity. (A) General representation of peptidic covalent reversible inhibitors of SARS-CoV-2 3CLpro summarizing the chemical requirements and SAR of compounds so far reported in literature. (B) Surface representation of the active site pocket of SARS-CoV 3CLpro bound to a peptide aldehyde inhibitor (dark salmon sticks, PDB 3SNE), chosen as a representative substrate-like inhibitor. The S1–S4 and S1′ subsites are indicated with red lines and labeled. The key residues forming the active site pocket are displayed as white sticks; the catalytic residues Cys145 and His41 are labeled.

In the first released X-ray structure, SARS-CoV-2 3CLpro was in complex with the N-term isoxazole capped tetrapeptide 1 (PDB 6LU7, superseded by 7BQY; Figures and 5).[40]

Figure 4

Co-crystallographic pose of compound 1 (violet sticks, PDB 7BQY) covalently bound to the active site of SARS-CoV-2 3CLpro. (A) The key residues forming the active site pocket are displayed as white sticks and labeled. H-bonds are depicted as dashed black lines. (B) Surface representation of the active site pocket with bound N3. The P1–P4 and P1′ moieties are labeled.

Figure 5

Compounds 1–9 with biological activities. aAntiviral activity evaluated by viral plaque assay. bData from ref (42). cAntiviral activity evaluated by viral RNA qRT-PCR quantification. dAntiviral activity evaluated by CPE reduction. eValues in bracket have been obtained in the presence of a P-gp inhibitor. The main structural differences, significant modifications, and warheads are highlighted.

Compound 1 is endowed with a vinyl carboxyl ester acting as the Michael acceptor warhead that traps the catalytic Cys145 through its nucleophilic attack at the vinyl β-carbon. This compound was previously identified by SBDD as inhibitor of 3CL proteases of different CoVs,[51,52] including SARS- and MERS-CoVs, showing in enzymatic assay a Ki = 9 μM[53] and an IC50 = 0.3 μM,[54] respectively, however no data in the same assay were reported against SARS-CoV-2 3CLpro. Nonetheless, the high structure identity between this protein and its SARS-CoV homologue would suggest that 1 may reasonably inhibit also SARS-CoV-2 3CLpro with similar potency. This hypothesis is supported by cell-based assay data showing that 1 inhibits SARS-CoV-2 replication with an EC50 = 16.77 μM. As expected, 1 has modest selective index with respect to other CoVs (EC50 = 4.0, 8.8, 2.7, and 3.4 μM against HCoV-229E, FIPV, MHV-A59, and MHV, respectively), and most likely the variability depends on subtle differences in amino acids sequences in the different CoVs proteases and cellular assay conditions.

Peptidomimetic α-ketoamides have been recently reported as broad-spectrum inhibitors of 3CL and 3C proteases active in cell-based assays against different CoVs and EVs, with derivative 2 showing the most promising activity.[42] Preliminary chemical optimization of 2 by SBDD led to derivatives 3 and 4, the first SARS-CoV-2 3CLpro inhibitors designed ad hoc (Figure ).[35] The X-ray structure solved for the complex 3/SARS-CoV-2 3CLpro (PDB 6Y2F) (Figure ) confirmed the formation of a thiohemiketal obtained by the nucleophilic attack of the Cys145 over the α-carbonyl of the ketoamide warhead.[35] The oxyanion group is stabilized by a H-bond with His41, while the amide oxygen of 3 accepts a H-bond from the main-chain amides of Gly143 and Cys145, which, together with Ser144, form the so-called “oxyanion hole” typical of cysteine proteases. Ketoamide derivative 3 inhibits SARS-CoV-2 3CLpro in biochemical assay with an IC50 = 0.67 μM, showing similar potency against SARS- and MERS-CoVs Mpro. 3 showed EC50 ∼ 4–5 μM against SARS-CoV-2 replication in human lung (Calu3) cells but, unfortunately, no data on cytotoxicity are provided (Figure ).

Figure 6

Co-crystallographic pose of compound 3 (green sticks, PDB 6Y2F) covalently bound to the active site of SARS-CoV-2 3CLpro. (A) The key residues forming the active site pocket are displayed as white sticks. H-bonds are depicted as dashed black lines. (B) Surface representation of the active site pocket with bound compound 3.

In another study, a SBDD strategy led to the identification of dipeptides 5 and 6 having an aldehyde as warhead and exhibiting excellent inhibitory activity against SARS-CoV-2 3CLpro in enzymatic assays, with IC50 = 53 and 40 nM, respectively (Figures and 7).[55] Notably, derivatives 5 and 6 are the smallest and most potent SARS-CoV-2 3CLpro inhibitors reported so far. More interestingly, both compounds potently inhibited SARS-CoV-2 infection in cell culture with sub-μM potencies (EC50 = 0.53 and 0.72 μM for 5 and 6, respectively), coupled with low toxicity (CC50s > 100 μM, SIs > 139) (Figure ). As expected, both compounds bind to the protein with similar poses, as shown in cocrystal structures (PDBs: 6LZE for 5 and 6M0K for 6), where the carbon of the aldehyde and the sulfur of Cys145 form the thiohemiacetal, the oxygen of the resulting tetrahedral adduct is stabilized by interaction with backbone of residue Cys145 and through a water bridge with the Thr26 side chain (Figure ).[55]

Figure 7

Co-crystallographic pose of compound 5 (yellow-orange sticks, PDB 6LZE) covalently bound to the active site of SARS-CoV-2 3CLpro. (A) The key residues forming the binding pocket are displayed as white sticks; water molecules are shown as red spheres. H-bonds are depicted as dashed black lines. (B) Overlay of 5 and 6 (raspberry sticks, PDB 6M0K) co-crystallographic poses.

Additional peptidic reversible covalent inhibitors have been identified through a biochemical screening based on enzymatic assay against SARS-CoV-2 3CLpro using a focused library of 68 known protease inhibitors, including approved and investigational drugs.[56] An investigational veterinary drug (usually administered via a subcutaneous route) for feline infectious peritonitis (FIP),[57,58] compound 7 (GC376), is a α-hydroxy-bisulfite dipeptide able to potently inhibit SARS-CoV-2 3CLpro (IC50 = 30 nM) and viral replication, even though with almost two orders lower potency (EC50 = 3.37 μM in CPE assay) and no cytotoxicity up to 100 μM (Figure ). Thermal shift assay and kinetic enzymatic experiments supported specific binding and reversible covalent inhibition (Ki = 60 nM). This inhibitor is a prodrug, converted into the aldehyde form by the removal of the bisulfite group to alkylate the Cys145 of the 3CLpro. Derivative 7 was originally designed to target 3CL of feline CoV and shares several chemical features with the inhibitors described for SARS-CoV-2 3CLpro.[43,57] It shows activity against multiple 3CL proteases (IC50 FIPV 3CL = 0.72 μM, IC50 TGEV 3CL = 0.82 μM, IC50 SARS-CoV 3CL = 4.35 μM, IC50 MERS-CoV 3CL = 1.56 μM) and against CoVs (TGEV, FIPV, MHV, 229E, and BCV) in cell lines with high nM potency.[43,57,58]7 and two close previously reported analogues, differing only in the electrophilic warhead (i.e., an aldehyde or an α-ketoamide), resulted also in potent inhibition of EV 3C and norovirus 3CL cysteine proteases and of the replication of several of these viruses, including HRVs.[43] Notably, 7 has been crystallized in complex with SARS-CoV-2 3CLpro, revealing an interaction mode similar to the aldehyde derivatives 5 and 6 (PDBs: 7BRR,[59]6WTJ,[60]6WTT,[56]7C8U, 7C6U, and 7CBT) (Figure ). Recently, a small set of new analogues of compound 7, differing for the N-terminal capping moieties, has been reported as inhibitors of 3CLpro from SARS-CoV-2, MERS-, and SARS-CoVs.[61] The new compounds showed biological activities in the same range of parent compound 7.

Figure 8

Co-crystallographic pose of compound 7 (teal sticks, PDB 7BRR) covalently bound to the active site of SARS-CoV-2 3CLpro. (A) The key residues forming the active site pocket are displayed as white sticks; water molecules are shown as red spheres. H-bonds are depicted as dashed black lines. (B) Overlay of 7 bound to SARS-CoV-2 3CLpro structures (PDBs: 7BRR, teal; 6WTJ, wheat; 6WTT, salmon), showing the different conformations of the benzyloxycarbonyl group.

Three further reversible covalent inhibitors of SARS-CoV-2 3CLpro have been identified by applying the same biological screening protocol as for compound 7. The anti-HCV drug Boceprevir[47] shows low μM potency against the isolated 3CLpro and the viral replication with no host cell toxicity up to 100 μM. Note that Boceprevir, originally optimized as HCV NS3 serine protease inhibitor, was 100-fold more potent against this target in comparison to the activity against the 3CLpro. Calpain cysteine protease inhibitors (Calpain inhibitor II: (2S)-2-acetamido-4-methyl-N-[(2S)-4-methyl-1-[[(2S)-4-methylsulfanyl-1-oxobutan-2-yl]amino]-1-oxopentan-2-yl]pentanamide; Calpain inhibitor XII: benzyl N-[1-[[1,2-dioxo-1-(pyridin-2-ylmethylamino)hexan-3-yl]amino]-4-methyl-1-oxopentan-2-yl]carbamate) with aldehyde warheads are slightly more potent in enzymatic and antiviral assays. These three inhibitors are structurally unrelated and deviate from 3CL substrate sequence specificity. Nonetheless, Boceprevir has been recently cocrystallized with SARS-CoV-2 3CLpro (PDBs: 7BRP,[59]6WNP, 7C6S, and 6ZRU), providing hints to improve its enzyme inhibition (Figure ). Noted that none of the HIV protease inhibitors, including Lopinavir, have shown activity in this study, thus providing an explanation to the difficult interpretation of the clinical results for the use of these drugs to treat COVID-19. Indeed, the results published for a clinical trial evaluating Lopinavir–Ritonavir in hospitalized adult patients with severe COVID-19 have shown no benefit.[62]

Figure 9

(A) Structure and biological activity of Boceprevir. aAntiviral activity evaluated by viral plaque assay. bAntiviral activity evaluated by CPE reduction. (B) Co-crystallographic pose of Boceprevir (purple sticks, PDB 6WNP) covalently bound to the active site of SARS-CoV-2 3CLpro. The key residues forming the active site pocket are displayed as white sticks; water molecules are displayed as red spheres. H-bonds are depicted as dashed black lines. (C) Surface representation of the active site pocket with bound Boceprevir.

In spite of their different warheads, compounds 1, 3, and 5–7 share a Gln mimetic γ-lactam, replacing the substrate P1 Gln to specifically fill the S1 subsite of the SARS-CoV-2 3CLpro catalytic pocket. Indeed, the γ-lactam here establishes key H-bonds with the Phe140 main chain and with the side chains of His163 and Glu166 (Figures , 6–8). Notably, Glu166 is involved not only in substrate recognition but also in the substrate-induced dimerization of SARS CoVs 3CL proteases through the interaction with Ser1 from the other monomer;[35,63] therefore, the interaction of the inhibitor γ-lactam with this residue could stabilize the monomeric and inactive form of the enzyme. Literature data on 3CLpro inhibitors of related CoVs clearly indicate that the P1 γ-lactam enhances the inhibitory potency up to 10-fold, probably because the higher rigidity reduces the loss of entropy upon binding if compared to the flexible Gln.[45] In Boceprevir, the γ-lactam is replaced by a cyclobutyl ring that cannot form any hydrogen bond at the S1 subsite, and this is likely one of the reasons for the modest inhibitory 3CLpro activity of this compound. Nevertheless, the presence of a modified proline at the P2 position would suggest the possibility to synthesize proline-based analogues with higher affinity and specificity toward the target enzyme. Indeed, the P2 amino acids of the other reported inhibitors show that this position tolerates a wide variety of residues characterized by similar lipophilicity and size to the substrate Leu and thus able to occupy the S2 subsite. In particular, leucine occupies the P2 position in 1 and 7, while it is replaced by a cyclopropyl alanine in 3 or by cyclohexyl- and m-F-phenyl alanine residues in 5 and 6, respectively. Compared to the other cocrystallized compounds, a difference in the binding mode of 6 is observed. In fact, the m-F-phenyl of this ligand undergoes a downward rotation so that the fluorine atom can form a H-bond with Gln189 (Figure ). It is interesting to report that this residue can alternatively form a hydrogen bond with the backbone nitrogen of the inhibitor P2 residue, as observed in the case of 1.

As described in the previous paragraph, the S3 and S4 pockets of SARS-CoVs 3CLpro are less structured and indeed can rearrange upon the binding of distinct P3/P4 residues. Hence, these amino acids are generally modified to modulate both potency and drug-like properties of the inhibitors. For instance, in peptide 1 the P3 lipophilic Val is solvent exposed, although it can establish van der Waals contacts with P1 γ-lactam to stabilize the inhibitor binding conformation, while the P4 N-capped Ala can form some lipophilic contacts with residues Met165, Leu167, and Gln192 in S4 (Figure ). Remarkably, the P3 backbone of the inhibitor establishes two key hydrogen bonds with the main chain of Glu166. These interactions can also be formed by derivative 3 (Figure ), in which the P2–P3 amide bond is masked by a N-Boc-aminopyridone, likely to prevent cellular proteases cleavage. Indeed, the pyridone CO donates a H-bond to Glu166, which in turn accepts another H-bond from the NH of the ligand carbamate. The analysis of the 3/3CLpro complex shows that the terminal hydrophobic Boc group does not establish significant interactions at the S3/S4 enzyme subsites. However, this group is fundamental to gain cell activity, probably due to the improvement of membrane permeation; in fact, its deletion did not affect biochemical inhibition of the protease but completely abrogated antiviral activity in cell-based assay (compound 4 in Figure ). Nonetheless, in the 3/3CLpro cocrystal structure, a DMSO molecule is found between the Gln189 side chain and the pyridone moiety, suggesting that there is space for larger and more functionalized groups (Figure ). Similar to 3, in aldehydes 5 and 6, the P3 is replaced by a more drug-like moiety, specifically a 2-indolic acid that caps the N-terminal of P2. In the enzyme-bound conformation, the indole ring is exposed to solvent and stabilized by H-bonds with the amide backbone Glu166 (Figure ), as the pyridone of 3. Also, in derivative 7, the P2 Leu is capped by a benzyloxycarbonyl group which can alternatively extend toward the S3/S4 pockets or downward to form intramolecular contacts with the P1 γ-lactam; in both cases, however, this moiety can form only one hydrogen bond with Glu166 with respect to the two formed by inhibitors 1, 3, 5, and 6. To conclude, the crystallographic complexes of SARS-CoV-2 3CLpro with 1 and 3 show that the P1′ benzyl group cannot form specific interactions at the S1′ site. In fact, the aldehydes 5 and 6, which completely lack a P1′ moiety, display the highest inhibitory potency against the enzyme, indicating the role of this substituent needs to be further investigated. It is worth noting that some of the cocrystallized inhibitors, namely derivatives 1, 7, and Boceprevir, fold, bringing the P3 residue or the benzyloxycarbonyl cap close to the P1 γ-lactam, thus suggesting that P1–P3 macrocyclization could be explored as optimization strategy. Indeed, macrocyclization has been widely employed as strategy to preorganize HCV NS3 protease inhibitors in their bioactive conformation, leading to BILN-2061[64,65] and the approved pan-genotypic drug Grazoprevir.[66,67] Moreover, macrocyclization generally show additional advantages potentially leading to greater stability and improved PK properties.

Compounds 3, 5, and 6 have been also evaluated for their in vivo PK properties in mice, and even if dosages, parameters, and administration routes are not fully aligned, it is worth comparing the collected data. Pyridone 3 upon sc administration (3 mg/kg) showed modest half-life (T1/2 < 2 h) and a rather low Cmax (around 126 ng/mL), indicating a suboptimal PK profile and the need of further optimization.[35] More interestingly, indole derivatives 5 and 6 were dosed via iv (5 mg/kg), and ip (5 and 20 mg/kg) administrations (single dose), resulting in a reasonable profile with T1/2 ranging from ∼2 to ∼5 h, depending on the administration route and the compounds, CL = 17 and 21 mL/min/kg, respectively, via iv, high Cmax (>2390 ng/mL) and availability >85%.[55] Overall, by comparing PK profiles of both compounds after iv administration at 5 mg/kg, 5 resulted metabolically more stable than 6, showing a T1/2 = 4.4 h and CL = 17 mL/min/kg with respect to T1/2 = 1.65 h and CL = 21 mL/min/kg, respectively. Compound 5 was further evaluated in rats (10 mg/kg, iv) and dogs (5 mg/kg, iv), showing good half-life values (rat, 7.6 h; dog, 5.5 h), low clearance (rat, 4.01 mL/min/kg; dogs, 5.8 mL/min/kg), and high AUCs (rat, 41 500 h·ng/mL and dog, 14900 h·ng/mL). The PK profile was considered good enough to proceed in acute and seven-day toxicity studies in rats and dogs. In particular, compound 5 was evaluated for acute toxicity in rats by iv administration at 24 mg/kg (1 rat), 40 mg/kg (10 rats), and 60 mg/kg (4 rats); one rat from the last group died. The seven-day studies were carried out in rats by administering via iv at 2, 6, 18 mg/kg of the compound to four rats per study; four dogs were dosed via iv at 10 mg/kg (the first day), 15 mg/kg (the second day), 20 mg/kg (the third day), 25 mg/kg (the fourth day), 25 mg/kg (the fifth to seventh days, randomly two dogs), 40 mg/kg (the fifth to seventh days, other two dogs). All animals were clinically observed during seven days for toxic signs, which include bodyweight, food intake, and hematology, and no anomalies were observed. Therefore, 5 was considered well tolerated after iv administration at different doses in rats and dogs. The plasma concentration and the concentration at trough unfortunately were not reported.

In summary, although there are slight differences in biochemical experimental conditions, in cell-based assays, and in the PK in vivo evaluation, that complicate a rigorous head-to-head comparison between the above mentioned inhibitors, some conclusions can be drawn. Compound 5, having the indole as P2 capping group and aldehyde as the warhead, resulted in the most potent 3CLpro inhibitor, with acceptable PK after iv administration and an acceptable safety profile in preclinical species. Remarkably, compound 5 is reminiscent of SARS-CoV 3CLpro inhibitors disclosed in two patents by Pfizer, following the SARS outbreak in 2002–2003.[68,69] Moreover, the introduction of an indole or other heterocycles in P3 was already reported for peptidomimetic inhibitors of SARS-CoV 3CLpro.[70] Similarly, these compounds having an indole as P2 drug-like cap and either a ketone-based warheads showed nM potency against the isolated target in biochemical assays. The COVID-19 pandemic prompted Pfizer to investigate the efficacy of these compounds and of newly developed analogues against SARS-CoV-2, and the obtained data have been recently reported.[71,72] One of the most potent derivatives is compound 8 (PF-00835231), which exhibits as P3–P1 substituents, respectively, a 6-methoxyindole capping group, a leucine, and the canonical Gln mimetic residue. Interestingly, the latter is functionalized with a hydroxymethylketone, enabling reversible covalent chemistry. Compound 8 exerts very potent inhibitory activity against a wide panel of 3CLpro from α-, β-, and γ-CoVs, with Ki/IC50 ranging from the low-nM to pM range. Particularly, a Ki = 0.27 nM and an IC50 = 6.9 nM against SARS-CoV-2 3CLpro and a Ki = 4 nM against SARS-CoV 3CLpro have been reported. On the other hand, 8 resulted as being inactive on HIV and HCV proteases as well as against a panel of human proteases. Therefore, this molecule can be considered a selective 3CLpro inhibitor with broad-spectrum activity against CoVs. The efficacy of 8 against SARS-CoV and SARS-CoV-2 was evaluated in Vero cells, showing EC50 values of 4.8 and 39.7 μM, respectively. However, 8 has been shown to be substrate of P-gp, which is indeed highly expressed in Vero cells. In fact, in the presence of a P-gp inhibitor, the antiviral activity of 8 was significantly increased, with EC50 values of 0.23 and 0.76 μM against SARS-CoV and SARS-CoV-2, respectively, coupled to low toxicity (CC50 > 100 μM). Interestingly, 8 exhibits additive/synergistic effect in combination with Remdesivir against SARS-CoV-2 in cell-based assays. X-ray structures of 8 bound to the 3CLpro of either SARS-CoVs have been solved (PDBs: 6XHL and 6XHM, respectively). As expected, the crystal poses in both structures are almost identical and are very similar to those reported for 5 and 6. DMPK profiling of 8 highlighted promising results but limited aqueous solubility to enable iv administration.[72] To increase the solubility, a prodrug strategy has been applied to compound 8 and its phosphate ester 9 (PF-07304814) has been reported as a suitable clinical candidate,[72] even though projected effective dose (500 mg/day) is high. On 15 September 2020, Pfizer announced the initiation of a double-blind, placebo-controlled phase Ib clinical trial (NCT04535167) to evaluate the safety, tolerability, and pharmacokinetics of 9.[73]

Besides the X-ray structures of the 3CLpro with the peptidic covalent inhibitors, some other strategies are currently underway to identify reversible inhibitors of the enzyme. Few results have been published and are available in preprint manuscripts, in press release on organization websites, and some crystal structure have been released in the PDB.

Baicalein, one of the main flavones of Scutelarria baicalensis, has been shown to inhibit the proteases of SARS-CoV-2 and SARS-CoV with IC50s of 0.94 and 1.1 μM (Figure A), respectively; its glucuronide analogue Baicalin is still active but less potent (IC50 SARS-CoV-2 3CL = 6.41 μM).[74] These data are partly consistent with those reported in another study, where Baicalein was found to be active against SARS-CoV-2 3CLpro (IC50 SARS-CoV-2 3CL = 0.34 μM), while Baicalin was only a weaker inhibitor (40% inhibition at 50 μM).[75] Perhaps the hydrolysis of the glucuronide moiety in one sample can explain the difference in activity. However, Baicalein was shown to be more potent in both studies. The specific binding of the flavones Baicalein and Baicalin to the protein target has been characterized by ITC showing KDs of 4.03 and 11.5 μM, respectively, and consistently with the compound structures the ITC profiles were typical of reversible binders.[76] In the same study, the X-ray structure of Baicalein noncovalently bound to the active site of the 3CLpro (PDB 6M2N, Figure B) has been solved, revealing that Baicalein mainly occupies the S1 pocket, with the three phenolic hydroxyl groups forming multiple H-bonds with the main chain of Leu141 and Gly143 and the side chains of Ser144 and His163 by water bridges.[76] The catalytic Cys145 and His41 stabilize the flavone binding mode by S−π and π–π interactions, respectively, while the carbonyl accepts a H-bond by the Glu166 backbone amide. Finally, the C-2 phenyl ring extends toward the S2 subsite forming favorable contacts with residues such as His41, Met49, and Tyr54. Therefore, Baicalein fills the core of the substrate binding site including the catalytic dyad and the S1/S2 protein residues sharing several recognition elements with peptide substrates and inhibitors.

Figure 10

(A) Structures and biological activities of flavones Baicalein and Baicalin. aAntiviral activity evaluated by viral RNA measurement by qRT-PCR. (B) Co-crystallographic pose of Baicalein (light-pink sticks, PDB 6M2N) into the active site of SARS-CoV-2 3CLpro. The key residues forming the binding pocket are displayed as white sticks. The buried water molecule is displayed as a red sphere. H-bonds are depicted as dashed black lines.

Interestingly, Baicalein also showed antiviral activity (EC50 = 1.69 μM) against SARS-CoV-2 in infected cells without toxicity (CC50 > 200 μM, SI > 118), while Baicalin was less potent in the same assay (Figure A).[76] It is worth noting that Baicalin tablets have been used as adjuvant therapy for the treatment of acute, chronic, or persistent hepatitis in China and were shown to be well tolerated up to the dose of 2.8 g.[77] Phase IIa clinical trials are currently ongoing to evaluate Baicalin in healthy adults with influenza fever (NCT03830684).

The 3D coordinates of SARS-CoV-2 3CLpro bound to another noncovalent ligand 10 (PDBs: 6W63, 6W79) have been recently released, showing that this compound can extend from S1′ to S3 establishing H-bonds with the backbone of Gly143 and Glu166 as well as with the His163 side chain (Figure ). However, the data on the activity of 10 in enzymatic and antiviral cell-based assays are still unknown. Similar compounds belonging to a Ugi library were previously reported as SARS-CoV3CLpro inhibitors, demonstrating μM inhibitory potency against the isolated target and poor activity in cell lines.[78]

Figure 11

(A) Chemical structure of compound 10. (B) Co-crystallographic pose of 10 (slate sticks, PDB 6W63) into the active site of SARS-CoV-2 3CLpro. The key residues forming the binding pocket are displayed as white sticks; water molecules are displayed as red spheres. H-bonds are depicted as dashed black lines.

Scientists from the Diamond Light Source, the UK’s national synchrotron, solved a new apo-structure of the SARS-CoV-2 3CLpro (PDB 6YB7) and then carried out a large X-ray crystallographic fragment screening, which resulted in the identification of 22 noncovalent and 44 covalent active-site binders. The corresponding complexes were deposited in the PDB, providing opportunity for fragments growing and/or merging. Indeed, follow-up medicinal chemistry efforts are currently focusing on a group of fragments functionalized with chloroacetamide warheads, which were selected because they were not identified as hits in previous screens and could thus in principle be selective for 3CLpro.[79]

Additional approaches are based on drug repositioning through performing virtual screenings of approved and investigational drugs libraries, followed by experimental validation against the target protein in biochemical experiments and/or in cell-based SARS-CoV-2 replication assays. A large collection of about 10 000 drugs and clinical candidates has been screened through this approach. Among the compounds identified (see chemical structures in Supporting Information, Figure S2), Ebselen (Figure ),[80] a synthetic organoselenium molecule endowed with anti-inflammatory, antioxidant, and cytoprotective properties, turned out as a covalent inhibitor of the SARS-CoV-2 3CLpro displaying antiviral activity in cell line, but the overall results are not particularly encouraging.[40]

Figure 12

Chemical structures and biological activities of Ebselen and compounds 11–17.

The PLpro as Antiviral Target

The CoV nsp3 multidomain protein is the largest replicase subunit (1922 aa).[81] This macromolecule, having transmembrane domains, is likely to play an essential role during the formation of the replication complex by mediating intracellular membranes rearrangement and by multiple interactions with other nsps.[82] Several domains of nsp3 have been identified and are conserved in all CoVs.[83] Among these, the PLpro domain (residues 1602–1855) is a cysteine protease that cleaves the viral polyprotein in the nsp1/2, nsp2/3 and nsp3/4 cleavage sites. Additionally, PLpro can recognize and hydrolyze the cellular proteins Ub and UbL protein ISG15 from the lysine ε-amino group of ubiquitinated host proteins involved in innate antiviral response.[33,84,85] The PLpro thus acts as a potent suppressor of Ub-dependent host immunity mechanisms, blocking the production of IFNβ and other cytokines as well as IRF3 and NF-κB pathways. By exerting proteolytic activity on part of the immature viral polyprotein and on Ub-modified host cell proteins, inhibitors of PLpro can block CoV replication and potentiate antiviral host cell immunity. In contrast to SARS-CoV-2 3CLpro, which is already well characterized and for which several inhibitors have been described, PLpro has been not investigated in detail so far. However, this protein is quite similar to the SARS-CoV homologue PLpro, sharing 83–86% sequence identity while differing from MERS-CoV PLpro (33% identity).[33] Thus, a description based on the structural and functional data on the SARS-CoV PLpro is possible. SARS-CoV-2 PLpro has the classic catalytic triad of cysteine proteases made by Cys111-His272-Asp286 (corresponding to Cys112-His273-Asp287 in SARS-CoV PLpro) that specifically recognize the P1–P4 sequence GGXL conserved in all the nsp1/2, nsp2/3, and nsp3/4 cleavage sites of the viral polyprotein.[33,85] As anticipated above, PLpro is also able to cleave Ub and the UbL protein ISG15, both showing the GGRL recognition motif at the C-term, from the ε-amino group of host proteins. Indeed, the PLpro overall architecture resembles that of human deubiquitinating enzymes of the Ub-specific protease family. This comprises a three-dimensional right-hand structure that hosts the catalytic triad at the interface of the thumb and the palm subdomains, an Ub binding motif at the N-term, and a fingers region containing an essential structural zinc ion tetrahedrally coordinated by four cysteines.[86] The Trp106 residue (Trp107 in SARS-CoV PL) in the oxyanion hole of the PLpro is required for the enzymatic activity because it stabilizes the hemithioacetal substrate intermediate through a single H-bond established by its indole nitrogen.[33,85] An additional key common feature between SARS-CoV and SARS-CoV-2 PLpro is the presence of the flexible BL2 loop containing a Tyr residue (Tyr268). It is found in an open conformation in the unliganded PLpro, while it is in the closed form upon inhibitor binding.

Like for SARS-CoV-2 3CLpro, the substrate specificity profile of PLpro was evaluated by determining the hydrolytic rate of different substrates from a combinatorial library, in comparison with the SARS-CoV enzyme.[87] This analysis has shown that both proteases are highly specific for Gly in P1–P2, while they can accept a wide panel of P3 residues, tolerating not only basic residues like Arg, Lys Orn Phe(guan), hArg Dap, Dab but also hydrophobic amino acids, such as hTyr, Phe(F5), Cha, Met, Met(O), and Met(O)2. On the other hand, these pockets cannot recognize d-amino acids. The S4 subsite can host only hydrophobic residues, among natural amino acids, with a strong preference for Leu. Notably, SARS-CoV PLpro can bind two P4 unnatural residues, namely hTyr and hTyr(Me), with higher affinity than leucine. On the basis of these specificities, two tetrapeptides, 11 and 12, displaying a GG-Dap-X as P1–P4 sequence and functionalized with a vinyl ester warhead, were designed as covalent inhibitors of the SARS-CoV-2 and SARS-CoV PL PRs. Their IC50s were not reported in the original manuscript but can be derived from the inhibition curve. Thus, IC50 for both compounds can be approximately estimated to be around 10 μM, however their antiviral activity in cell-based assay was not reported (Figure ). As expected, the compounds were inactive against MERS-CoV-PLpro, significantly differing from the SARS homologues. The X-ray structures of compounds 11 or 12 covalently bound to SARS-CoV-2 PLpro (PDBs: 6WUU and 6WX4, respectively) have been released.

Seleno-organic compounds have been described in two preprint papers as irreversible inhibitors of PLpro of both SARS-CoV and SARS-CoV-2, with Ebselen (Figure ) showing IC50s of 8.35 and 2.26 μM, respectively, after incubation (60 min) with each enzyme, whereas no activity was detected without incubation.[88] Later on, benzisoselenazol-3(2H)-one derivatives, differing from Ebselen in the N substituent, and 2,2′-dicarbamoyldiaryl diselenides have been reported as inhibitors of PLpro (Figure ).[89] In particular, o-hydroxy (13) and o-methoxy (14) N-phenyl substituted benzisoselenazol-3(2H)-ones are 10-fold more potent against SARS-CoV-2 PLpro than Ebselen, with IC50s of 0.24 and 0.26 μM, respectively. Similarly, the corresponding 2,2′-dicarbamoyldiaryl diselenide analogues 15 and 16 show inhibitory potency in the same range (IC50 = 0.34 and 0.26 μM). Antiviral activity in cell-based assays was not investigated for Ebselen and its analogues. However, Ebselen has been shown to covalently inhibit also SARS-CoV-2 3CLpro[40] and a plethora of protein targets, while benzisoselenazol-3(2H)-one and 2,2′-dicarbamoyldiaryl diselenide derivatives have been shown to also irreversibly inhibit human methionine aminopeptidase 2,[90] thus hinting at a nonspecific effect on a wide panel of proteins due to the high reactivity of Se atom.

Noncovalent naphthalene-based SARS-CoV PLpro inhibitors have been previously identified through HTS approaches showing high-nM/low-μM potency in biochemical assays and weak antiviral activity in cell lines, with the optimized derivative 17 (Figure ) displaying the best performance (IC50 SARS-CoV PL = 0.6 μM, EC50 SARS-CoV = 14.5 μM, Figure ).[91,92] As demonstrated by enzymatic kinetic experiments, 17 is a reversible competitive inhibitor that in the X-ray complex with SARS-CoV PLpro (PDB 3E9S) is bound at the S3–S4 subsites in a cleft for the access to the active site.[93] Recent results show that 17 is able to inhibit SARS-CoV-2 PLpro (IC50 = 2.4 μM, EC50 SARS-CoV = 27.6 μM Figure ), while it is not able to inhibit MERS-CoV PLpro due to the replacement with a threonine of Tyr268, in the BL2 loop, that is present in both the SARS-CoV and SARS-CoV-2 enzymes where the residue mediates interactions with the inhibitors.[33] However, compound 17 demonstrated weak antiviral activity in cell-based assays (EC50 SARS-CoV-2 = 27.6 μM in plaque assay, in Figure ).[33]

In summary, targeting PLpro may represent a promising approach to tackle SARS-CoV-2 replication and host immune escape. However, no potent and cell-based active covalent inhibitor is yet available, probably because the design of peptidic compounds, resembling the substrate sequence, is hampered by P1–P2 specificity for glycine. On the other hand, known reversible noncovalent inhibitors of SARS-CoV PLpro show only moderate and narrow antiviral activity, being not active against the related MERS-CoV homologue. Therefore, much work is needed to validate PLpro as a viral target for successful drug discovery.

SARS-CoV-2 RdRp: Structure and Function

CoVs rely on a multisubunit machinery for the replication and transcription of the viral genome. This macromolecular assembly is made up of nonstructural proteins (nsp), which are cleavage products of the ORF1a and ORF1ab viral polyproteins.[94] In particular, RdRp activity resides in nsp12 and catalyzes the formation of a phosphodiester bond between nucleoside triphosphates (NTPs) in a primer-dependent manner. The NTP substrates involved in this reaction are coordinated by two metal ions, which are bound by the conserved Asp residues. Consistent with this conserved mechanism, strong amino acid conservation can be observed in regions that are directly involved in nucleotide selection or catalysis.[95]

Upon the formation of a complex with nsp7 and nsp8, nsp12 acquires processivity for the synthesis of large RNAs, a mechanism that in CoVs has evolved to control the viral RNA synthesis during infection.[96] In addition, nsp8 has shown RNA primase activity, that is, the ability to synthesize an RNA primer,[97] enabling de novo RNA synthesis. The structural and functional features of the RdRps among CoV family are highly conserved; for example, the RdRp of SARS-CoV-2 has 96% amino acid identity to that of SARS-CoV.[98,99]

Recently, the structure of SARS-CoV-2 full-length RdRp, in complex with cofactors nsp7 and nsp8, was determined by cryoelectron microscopy (cryo-EM) at 2.90 Å resolution (PDB 6M71)[100] showing the typical right-hand architecture of the viral RdRps (Figure ), constituted by the finger, palm, and thumb subdomains.[101] The nsp12 RdRp domain is connected through an interface to the N-terminal region possessing nucleotidyltransferase activity, named NiRAN, whose exact role in the viral life-cycle remains elusive.[102] SARS-CoV-2 RdRp cryo-EM structure revealed also an additional, unique N-terminal β-hairpin, which was previously not observed in SARS-CoV RdRp.[100]

Figure 13

Architecture of SARS-CoV-2 nsp12. (A) Schematic diagram outlining the domain organization of SARS-CoV-2 nsp12. The domains are colored as: palm, cyan; fingers, yellow-orange; thumb, salmon. The N-terminal NiRAN domain, the interface and the β-hairpin are colored light-pink, wheat, and blue-white, respectively. The polymerase conserved motifs are colored as: motif A, lime green; motif B, violet; motif C, slate; motif D, olive; motif E, sand; motif F, deep-teal; motif G, ruby. (B) Structure of SARS-CoV-2 in two different orientations (PDB 6M71). (left) Ribbon diagram of nsp12 showing the arrangement of palm, fingers, and thumb domains. Domains are colored as in (A). Nsp7 and nsp8 cofactors are shown as pale-green and raspberry ribbon models, respectively. (right) The palm, fingers, and thumb domains are shown as a molecular surface.

The RdRp active site is composed by seven conserved motifs (A–G) lining a central cavity where the template-directed RNA synthesis takes place (Figure ). In particular, the incoming NTP binds within motif F, whereas the RNA template enters the active site through a channel formed by motifs F and G; motif E and the thumb subdomain sustain the primer strand. The product of RNA synthesis leaves the active site through an RNA exit path, situated at the front side of the polymerase.[100]

Figure 14

Overview of SARS-CoV-2 nsp12 active site. For clarity, the RdRp core region is shown as a white ribbon model, whereas the polymerase conserved motifs (A–G) are colored according to the upper schematic diagram. The catalytic residues Asp760 and Asp761 are shown as white sticks. The template entry, NTP entry, product hybrid exits paths are indicated by orange arrows.

Three other groups reported further cryo-EM structures of nsp12 catalytic subunit and nsp7-nsp8 cofactors. The nsp12-nsp7-nsp8 complex has been solved both in the apo form (PDB 7BV1) and bound to the template-primer RNA and Remdesivir monophosphate (PDB 7BV2);[103] this latter is described in more detail below. Subsequently, another apo structure was reported (PDB 7BW4).[104] By using circular dichroism (CD) experiments, the authors pointed out that both the nsp8 and nsp12 subunits of SARS-CoV-2 have shown lower melting temperature values, as compared to the corresponding subunits of SARS-CoV, suggesting the poorer thermostability of SARS-CoV-2 proteins and a possible adaptation of SARS-CoV-2 toward humans with relatively lower body temperatures than the natural bat hosts. Then, the cryo-EM structure of the replicating SARS-CoV-2 RdRp (PDB 6YYT), comprising nsp12, nsp8, and nsp7, and over two turns of RNA template-product duplex was reported, adding another piece of useful structural information (Figure ).[105]

Figure 15

Structure of SARS-CoV-2 replicating RdRp-RNA complex (PDB 6YYT). Nsp12 is shown as a molecular surface (color code as in Figure ), whereas the cofactors nsp7 and nsp8 (protomers 1 and 2) are shown as pale-green and raspberry ribbon models, respectively. RNA turns are shown as an orange ribbon model. The positively charged nsp8 residues, proposed to interact with RNA, are shown as sticks.

The active site cleft of nsp12 binds the first turn of RNA, whereas two subunits of nsp8 bind to opposite sides of the cleft, flanking the exiting RNA duplex with long α-helical extensions, called “sliding poles”. These nsp8 extensions are rich in positively charged residues and form multiple RNA backbone interactions. Notably, this structure revealed large additional portions in nsp8 that become ordered upon RNA binding, while nsp8 nsp7 complexes are usually more flexible.[105] The mutation of one of these positively charged residues, namely Arg58, with Ala, was previously reported as lethal in SARS-CoV because of the strong decrease of polymerase activity, leading to a nonviable phenotype (as evaluated by immunofluorescence microscopy and plaque assays).[96]

Targeting SARS-CoV-2 RdRp

The SARS-CoV-2 RdRp represents an ideal drug target due to its critical role in virus replication and the absence of an enzymatic counterpart in the host cell. In fact, viral polymerases inhibitors represent the cornerstone of antiviral therapeutics; indeed, most of the approved drugs for the treatment of viral infections, including HIV and HCV, belong to this class.

Inhibitors of viral polymerases fall into two main categories, according to their mode of action and structure, with nucleoside inhibitors (NIs) acting at the substrate site and the non-nucleoside inhibitors (NNIs) interacting with allosteric binding sites.

Therefore, building on the experience and achievements obtained in the past, these approaches might lead to strategies that effectively control SARS-CoV-2 infection. NIs represent an attractive avenue toward disrupting viral RNA replication because of the high degree of conservation of active sites of RdRp and relatively low mutation rate in these regions, thus allowing for a broad antiviral activity and a high barrier to resistance. To compete with the natural substrates for incorporation into the viral RNA, NIs requires activation to the pharmacologically active triphosphate form (NTP) by a multistep process, carried out by host kinases (Figure , orange box). The first phosphorylation step is often the rate-limiting step in the activation.[106] To overcome this drawback, the application of the ProTide strategy, based on the synthesis of ester phosphoramidate prodrug of monophosphate nucleotides, has been extensively explored in NIs to bypass the initial phosphorylation step. After cellular uptake, ProTide ester hydrolysis is mediated by host esterases (Figure ), followed by spontaneous cyclization, phenol release, and phosphoroamidase cleavage to deliver the nucleotide monophosphate, which is then converted into the active NTP form.[106]

Figure 16

Putative mechanism of ProTides in vivo metabolism. Upon diffusion into the cell, the amino acid ester of the ProTide is cleaved by intracellular esterases, then a cyclization occurs onto the phosphorus, with the release of the phenoxide moiety. The unstable cyclic intermediate is then hydrolyzed by water to the alanine metabolite, whose P–N bond is hydrolyzed by phosphoramidase-type enzymes to unmask the NI monophosphate form. The NI monophosphate is routed to further phosphorylation steps, yielding the active triphosphate form (NTP) and thus circumventing the endogenous phosphorylation pathway (orange box). X, aromatic substituents; Y, O, or CH2; R, ester substituents.

The viral RdRp binds the drug in the form of NTP; then, the hydrolysis and release of a pyrophosphate group provides the energy source for nucleotide monophosphate incorporation into the nascent RNA chain, with consequent inhibition of strand elongation and viral replication. Antiviral NIs fall into three categories: obligate chain terminators, nonobligate chain terminators, and mutagenic.[107] Obligate chain terminators do not possess the 3′-hydroxyl group at the riboside moiety of the molecule. Nonobligate chain terminators possess instead a natural base and a 3′-hydroxyl on the sugar, but they display an additional substituent at the C-1′ or the C-2′ positions of the ribose ring, blocking the formation of the phosphodiester linkage with the incoming NTP.[107] The mechanism of lethal mutagenesis involves the inability to recognize the nucleoside analogues as regular nucleobases, thus inducing a mismatch in base pairing and an increase in mutations, ultimately leading to nonviable genomes.

Despite the large amount of structural data produced for the SARS-CoV-2 RdRp, this protein may appear as a still underexplored antiviral target in comparison with the 3CLpro of CoVs. Indeed, no focused medicinal chemistry programs, aimed at identifying specifically designed inhibitors of CoV RdRp, have been reported so far. On the other hand, the few inhibitors identified derive from a large repurposing campaign on known broad spectrum, well characterized NIs which can be readily used in the clinic. In fact, Remdesivir (GS-5734, developed by Gilead Sciences) was recently granted EUA[12] for the treatment of SARS-CoV-2, following the encouraging results from the National Institute of Allergy and Infectious Diseases (NIAID) and Gilead clinical trials and from the compassionate use programs.[13−15] Other attractive candidates include Favipiravir (T-705, Avigan), which was granted emergency use approval in Russia and India, Galidesivir (Galidisvir, BCX4430), and the compound β-d-N4-hydroxycytidine-5′-isobutyrate ester 18 (EIDD-2801). These compounds can be grouped in adenine (Remdesivir and Galidesivir), guanine (Favipiravir), and cytosine (18) analogues (Figure ). Remdesivir and Galidesivir are C nucleosides obtained by introducing a C-glycosidic bond to link the sugar to the nucleobase, showing higher resistance to phosphorolysis mediated by intracellular phosphorylases.

Figure 17

Chemical structures of antiviral nucleoside analogues in clinical development. The parent bases are highlighted in light-blue (adenine), green (guanine), and orange (cytosine).

Remdesivir is the Sp isomer ester monophosphoramidate prodrug of 1′-cyano-substituted adenine C-nucleoside ribose analogue 19 (GS-441524, Figure ).[108] It was designed to deliver the nucleoside monophosphate 21 into the cell, circumventing the rate-limiting first phosphorylation step and allowing for efficient formation of the active triphosphate form (Figure ).[109,110] This advantage is not present in Favipiravir, 18, and Galidesivir. Structurally, the 1-cyano group of Remdesivir provides potency and selectivity toward viral RNA polymerases. The Sp isomer was selected because of the high potency across multiple cell lines and the crystalline nature of an Sp prodrug reagent that allowed rapid synthesis scale-up, a key element for the current request of active pharmaceutical ingredients (API).[110]

Figure 18

Schematization of Remdesivir metabolic conversion to the active triphosphate form. Remdesivir is first transformed into the intermediate alanine metabolite 20, then to the monophosphate form 21 and finally to the triphosphate form 22. The phosphoramidate prodrug moiety is shown in magenta. 19 (GS-441524) is the parent nucleoside of Remdesivir.

The cryo-EM structure of Remdesivir in complex with RdRp from SARS-CoV-2 and a 50-base template-primer RNA (PDB 7BV2) has been recently released.[103] This structure, generated using the drug’s triphosphate metabolite 22, shows the nucleotide analogue in its monophosphate form, covalently bound to the primer strand at the 3′-end (Figure ) to terminate the chain elongation.

Figure 19

Binding mode of Remdesivir into the SARS-CoV-2 nsp12 active site (PDB 7BV2). (left) nsp12 is shown as a molecular surface, colored according to the schematic diagram in Figure . For clarity, nsp7 and nsp8 cofactor have been removed. The template and primer RNA are shown as ribbon models and labeled. (right) Zoom-in of the nsp12 active site. The covalently bound monophosphate form of Remdesivir (slate) and the pyrophosphate group are shown as sticks. Magnesium ions are shown as green spheres. The RNA bases interacting with Remdesivir are shown as orange thin sticks, while protein residues are shown as white thick sticks. Hydrogen bonds are shown as black dashed lines.

Next to the covalently bound Remdesivir, there is the leaving pyrophosphate group; the two catalytic magnesium ions, coordinated by two aspartic acid residues Asp760 and Asp761, are also visible. Strikingly, nsp7 and nsp8 do not form interactions with RNA, although they are required for RNA binding by RdRp. Remdesivir is positioned at the center of the catalytic site, with the adenosine analogue forming stacking interactions with the upstream bases U-1 and A-1 of the primer and template strands. In addition, Remdesivir engages three strong H-bonds with the uridine bases U–1 and U+1, whereas the sugar 2′-OH group forms a further H-bond with Asn691 (Figure A). The Asp623, Ser682, and Asn691 residues are involved in the recognition of the ribose hydroxyl groups of the incoming NTP and are responsible for the proper positioning of the ribose moiety. Such residues represent a conserved feature among viral RdRps.[111] Among them, Ser682, located on motif B, is considered a “fidelity checkpoint” because it is sensitive to the conformational changes of motif B upon binding of the NTP, which, in turn, triggers the active site closure.[112] This structure shows for the first time a clear picture of how the template-primer RNA is recognized by the enzyme and how the covalently bound Remdesivir interacts with the active site residues.

Figure 20

A close view of the covalently bound Remdesivir within the RdRp active site. (A) Remdesivir is incorporated into the primer strand and terminates chain elongation (PDB 7BV2). (B) Remdesivir incorporation induces a mechanism of delayed-chain termination (PDB 7C2K). RNA is shown as orange ribbon model. For clarity, only key residues (thick white sticks) and bases (orange thin sticks) that interact with Remdesivir are shown. Hydrogen bonds are shown as dashed lines.

A “delayed” chain termination has been proposed for Remdesivir, according to which the RNA synthesis is terminated after the addition of three more nucleotides (at position i+3, where i corresponds to the position of the first incorporated Remdesivir monophosphate);[113,114] this mechanism has been reported for SARS-CoV, MERS-CoV, and also SARS-CoV-2.[111,115] This hypothesis is consistent with a recently released cryo-EM structure of the SARS-CoV-2 RdRp in complex with RNA, before and after RNA translocation (PDBs: 7C2K and 7BZF, respectively).[116] In the pretranslocated catalytic complex structure, the incorporated Remdesivir has been translocated to the −1 position, whereas the 3′-guanosine occupies the +1 position (Figure B). Remdesivir engages four H-bonds, with the upstream G-2 base, the uridine base U-1, and with Ser759. The primer strand with the incorporated Remdesivir may translocate without obstruction to positions i+1, i+2, or i+3, allowing the incorporation of three subsequent nucleotides. However, at position i+4, a putative steric clash was postulated between the 1′-CN substituent of Remdesivir and Ser861 along the RNA exit tunnel.[111] Ser861 is highly conserved among CoV RdRps, and its important role in Remdesivir induced RdRp inhibition was ultimately supported by mutagenesis studies, revealing that the Ser861Ala RdRp mutant yields a smaller fraction of i+3 termination compared with the wild-type RdRp, possibly supporting the steric clash hypothesis.[116]

The analysis of the pretranslocated complex cryo-EM maps revealed the two bound nsp8 protomers in two distinct states, named conformations I and II;[116] this latter resembles the previously observed “sliding poles” (PDB 6YYT), whereas in conformation I, the nsp8 protomer, which lays on the fingers and interface domains, show an orientation change of about 45°. Taken together, the structural information reported so far suggest that, besides targeting nsp12, an alternative approach for RdRp inhibition would be the disruption of the polymerase complex integrity, for example, targeting the interaction between nsp8 and nsp12. Indeed, it would be interesting to explore the presence of “hot spots”, which are small areas of the protein–protein interface providing most of the binding energy. In fact, protein–protein interactions (PPIs) are today a validated target for drug discovery, and in this perspective, the structure of the replicating RdRp (Figure ) may offer hints for the development of peptidomimetics or small molecules capable of impairing the PPIs between nsp7-nsp8-nsp12. For instance, the recognition between the upper part of nsp8–1 and the fingers domain of nsp12 is mediated by a β sheet, suggesting the possibility to design small peptidomimetics capable of resembling a β-strand motif.

Remdesivir was originally developed within antiviral research program against HCV and respiratory syncytial virus (RSV) but has shown a broad antiviral activity against other different viruses, including Dengue, parainfluenza type 3 virus (hPIV3), Ebola virus, and SARS-CoV with EC50 in the sub-μM range.[108,117,118] Remdesivir showed efficacy in preclinical rhesus monkey models of Ebola virus infection;[109] it was evaluated in humans on a compassionate basis in some isolated cases of EVD[119,120] and in a large-scale clinical trial for EVD conducted in Congo.[121] Although the efficacy of Remdesivir was lower than the antibodies treatments tested, the survival rate was significantly higher than in the EVD outbreak.

The efficacy of Remdesivir against CoVs was only recently reported.[122,123] In particular, it showed dose-dependent inhibition of SARS-CoV and MERS-CoV replication in primary human airway epithelial cell (HAE) cultures, with average EC50 = 0.069 μM (SARS-CoV) and 0.074 μM (MERS-CoV).[122] While typical chain terminators are excised by the viral exonuclease nsp14, Remdesivir seems to be able to escape the exonuclease activity. Two amino acid substitutions were found in the nsp12 polymerase providing low-level resistance to Remdesivir, corresponding to the SARS-CoV residues Phe480Leu and Val557Leu. Such residues are identical across CoVs and have been found to cause an impairing of fitness and virulence.[124]

The possible structural and function implications of such resistance mutations have been recently analyzed.[112] The two mutations do not have a direct impact on the catalytic site or substrate-binding pocket but rather cause minor structural alterations, which likely impact an NTP fidelity checkpoint performed by the polymerase before catalysis. Phe480 is located at the interface between the fingers and palm domains, facing a patch of hydrophobic residues that indirectly impact motif B. The mutation Phe480Leu likely reduces the interactions within the hydrophobic core, thereby reducing its structural rigidity. Therefore, one possible interpretation is that the phenylalanine to leucine mutation would have an impact on the positioning of the Ser682 of motif B, altering the fidelity checkpoint. On the other hand, Val557 is located at the end of motif F and forms a hydrophobic wall upon which the template base is stacked. The consequence of a valine to leucine mutation is a more extended hydrophobic side chain, possibly generating a steric hindrance with the template RNA. As a result, the RNA would be expected to be deviated from the groove, away from the serine of motif B. Remdesivir also showed potential to treat SARS-CoV-2 infections, with different studies reporting slightly different antiviral activities for this drug (Table ).[125−128]

Table 1

Remdesivir Antiviral Activities against SARS-CoV-2 Measured in Different Cell Lines and Assays

	EC50 (μM)
cell line	plaque assay	genome copy number	log10 TCID50 (mL)	cytopathic effect (CPE)
Vero E6		0.77a
		26.90b	23.15b
				0.651c
	1.65d	1.49d
Calu3 2B4	0.28d	0.60d
HAE	0.010d

From ref (125),

From ref (126),

From ref (127),

From ref (128),

The potency of Remdesivir is highly dependent on the intracellular concentration of the pharmacologically active triphosphate metabolite, which can vary in different cell systems. Indeed, such concentration is markedly higher in primary HAE cultures compared to human lung cells (Calu3 195 2B4) and monkey kidney cells (Vero E6). Therefore, the differences in EC50 may be due to intrinsic differences of SARS-CoV-2 virus isolates, methods of quantification, and assay conditions. Because SARS-CoV-2 does not readily infect wild-type mice because of incompatibilities between virus spike and the murine orthologue of ACE2, which is needed for SARS-CoV-2 entry, a recombinant SARS-CoV-1 chimeric virus, encoding the SARS-CoV-2 RdRp, was engineered to rapidly assess the in vivo efficacy of Remdesivir in mice.[128] Remdesivir inhibited chimeric virus replication in a dose-dependent manner in human epatoma (Huh7) cells, with mean EC50 = 3 nM. Moreover, Remdesivir treatment significantly ameliorated loss of pulmonary function in female C57Bl/6 Ces1c–/– (carboxyl esterase 1c deficient) mice administered 25 mg/kg Remdesivir subcutaneously 1 day postinfection (dpi) and BID thereafter. The transgenic mice model better recapitulates the DMPK profile of Remdesivir in humans. Indeed, carboxyl esterase 1c is a serum esterase which dramatically reduces half-life of Remdesivir and is absent in humans.[122] For instance, the plasma stability of Remdesivir was significantly improved in Ces1c–/– mice (T1/2 ∼ 25 min) with respect to wild-type mice (T1/2 < 5 min). In vivo experiments in mice infected with MERS-CoV proved the higher efficacy of Remdesivir than a combination of Lopinavir/Ritonavir and IFN-β in improving lung function.[129] Remdesivir in vivo efficacy against SARS-CoV-2 was gauged in rhesus macaques: therapeutic treatment started 12 h after inoculation with SARS-CoV-2 and continued once daily through 6 days post inoculation (dpi).[130] One group of rhesus macaques was treated with a loading dose of 10 mg/kg Remdesivir, followed by a daily maintenance dose of 5 mg/kg, while the other group, serving as control, received a vehicle solution. Remdesivir was able to reduce virus titers in bronchoalveolar lavages already 12 h after the first treatment, also reducing clinical disease and damage to the lungs. The treatment resulted in reduced virus replication in the lower respiratory tract but not in the upper tract, suggesting the need to explore drug delivery strategies to improve Remdesivir distribution. Furthermore, the study suggested that treatment of patients with Remdesivir should be initiated early during infection.

Remdesivir received the approval from the FDA for the treatment of hospitalized patients diagnosed with COVID-19.[12]

The optimal duration of treatment is still under evaluation in clinical trials. Both 5-day and 10-day treatment durations are suggested, based on the severity of disease, with 200 mg iv on the first day, followed by 100 mg each day thereafter. The approval was granted on the basis of the available data from the Gilead’s SIMPLE trials (NCT04292899), the Adaptive COVID-19 Treatment Trial (ACTT), sponsored by NIAID (NCT04280705), in patients with severe manifestations of COVID-19, and from the compassionate use program.

The two SIMPLE trials evaluated the safety and efficacy of 5-day and 10-day dosing duration of Remdesivir in hospitalized patients with severe (first study) and moderate manifestations (second study) of COVID-19. The results from the first study showed that patients receiving a 10-day treatment course had a distribution in clinical status similar to those taking a 5-day treatment course.[13] Thus, priority should be given to a 5-day regimen for patients at the early stages of severe disease (patients receiving supplemental oxygen but not yet been intubated).[14] During the trial, the side effects were mild and, however, the absence of a placebo control group and the open-label design of this study did not permit an overall assessment of the benefit of Remdesivir. Another limitation is the lack of SARS-CoV-2 viral-load results during and after treatment due to the variability in local access to testing and practices across the global sites.

The recently published results of the second SIMPLE trial study showed that among patients with moderate COVID-19, those randomized to a 5-day course of Remdesivir plus standard of care had higher odds of having an improvement in clinical status compared to those randomized to standard of care alone. Patients randomized to a 10-day course of Remdesivir did not show a statistically significant difference in clinical status compared to standard care at 11 days after initiation of treatment.[131]

The ACTT trial is a randomized, double-blind, placebo-controlled multicenter trial conducted in around 100 sites globally, evaluating the time to recovery of hospitalized adults diagnosed with COVID-19 up to day 29. The recently published results indicate that Remdesivir was superior to placebo in shortening the time to recovery; mortality was numerically lower in the Remdesivir group than in the placebo group (11.4% with Remdesivir and 15.2% with placebo by day 29).[15] On the other hand, according to the published results of a trial carried out at 10 hospitals in Hubei, China, Remdesivir was not associated with statistically significant clinical benefits beyond those of standard of care treatment.[132] However, the trial was underpowered because the study terminated earlier, before attaining the prespecified sample size, as there were no further patients meeting eligibility criteria.[132,133]

It has to be noted how challenging it is provide robust findings when conducting trials during a pandemic: in this light, it is highly valuable to share results across multiple, even smaller, but high-quality studies. Some of the variation in results from the RCTs of Remdesivir could be due to differences in study design.[134] The overall results suggest that Remdesivir improves the recovery of patients hospitalized with COVID-19 and may prevent the progression to more severe respiratory disease.[15]

Moreover, there is room for further studies involving potential combination therapies with other antivirals and anti-inflammatory agents in appropriate regimen. In this regard, the ACTT3 trial (NCT04492475) will evaluate the combination of INF-β1a and Remdesivir compared to Remdesivir alone.

Significant efforts are devoted to find an alternative route of administration to the currently iv used. Because the upper respiratory tract is the most prevalent site of SARS-CoV-2 infection in early disease, an inhaled, nebulized formulation of Remdesivir is being seen as a valid noninvasive alternative. This should afford a more targeted administration with a lower systemic exposure, increasing the therapeutic index of the drug. Inhaled Remdesivir would be particularly suited for early stage COVID-19 patients who do not need to be hospitalized. Additional clinical trials (GS-US-553-9018 and NCT04539262) have thus been initiated to evaluate this inhaled formulation.

Favipiravir contains a carboxamide moiety, reminiscent of guanine; it is mainly incorporated in the salvage pathways for purine nucleotides through the purine phosphoribosyltransferases, then it needs to be converted to its ribofuranosyl monophosphate 23 and to be further processed to the active triphosphate form 24 (Figure ).[135]

Figure 21

Schematization of Favipiravir metabolic conversion. Favipiravir is first converted to its ribofuranosyl monophosphate derivative 23 and subsequently to the triphosphate active form 24.

This drug possesses a broad-spectrum activity toward RNA viruses.[136] High concentrations (EC50 of 61.88 μM) were required to reduce viral infection in Vero E6 cells infected with SARS-CoV-2.[125] Conversely, no apparent antiviral effect against SARS-CoV-2 was observed in vitro at concentrations under 100 μM. Favipiravir is thought to act as chain terminator;[135] however, a putative mechanism of lethal mutagenesis has been proposed in the activity of this drug against influenza A (H1N1) viruses in vitro.[136,137] Thus, antiviral nucleoside mechanisms of action may well vary from one virus to another. Favipiravir received approval in Japan for restricted use in uncomplicated influenza virus infections. The established dose for influenza in Japan is 1600 mg twice a day on day 1 followed by 600 mg twice a day for four days.[135] An important feature of this drug is the apparent lack of resistant mutants in cell cultures, as well as in clinical trials. Some studies, however, found mutants of influenza A (H1N1)[138] and chikungunya virus[139] with reduced susceptibility to Favipiravir in cell cultures, suggesting the need to implement diagnostic approaches to control the emergence of Favipiravir resistance in clinical settings.

Regarding the use of Favipiravir against COVID-19, to date, the only published clinical results appeared in a preprint, describing an open-label randomized trial carried out in China, comparing the efficacy and safety of Favipiravir (1600 mg BID first day, followed by 600 mg BID) and Arbidol (200 mg TID) to treat COVID-19 patients.[140] The clinical recovery rate on day 7 did not significantly differ between the Favipiravir group and the Arbidol group; patients administered Favipiravir experienced faster resolution of fever and cough but similar rates of respiratory failure compared to the control group receiving Arbidol. On 1 June 2020, Russia’s Ministry of Health temporarily approved a Favipiravir-based medication for use in patients with COVID-19. Favipiravir also received accelerated approval in India for the treatment of mild to moderate COVID-19 patients. Despite the quite fast approval, more robust data, coming from well-designed studies and possibly involving larger population, are needed to provide strong evidence about the clinical benefit of this drug.

Galidesivir’s (Galidisvir, BCX4430) effect on the inhibition of RNA transcriptional activity was first assessed in a cell-free isolated HCV RdRp assay.[141] Galidesivir acts as a nonobligate RNA chain terminator; the switch from furanose to an azasugar is thought to alter the electrostatic interaction of the ring so that the viral RdRp is unable to add more than one or two further nucleotides.[141,142] Galidesivir was shown to possess weak antiviral activities against many RNA viruses, also including SARS-CoV (EC50 = 57.7 μM, measured by NR uptake assay) and MERS-CoV (EC50 = 68.4 μM assessed by high-content image analysis) and to protect nonhuman primates from lethal filovirus challenge.[141] On the other hand, Galidesivir does not inhibit SARS-CoV-2 viral replication at concentrations below 100 μM.[126] Notably, Galidesivir can be administered by both parenteral (im and ip) and oral routes.[142] Galidesivir is efficiently taken up into cells and converted to the active triphosphate. In mouse, rat (both treated with 2 mg/kg), guinea pig (50 mg/kg), and cynomolgus macaque (20 mg/kg), Galidesivir DMPK is characterized by rapid clearance from the plasma with a T1/2 < 5 min. Conversely, the T1/2 of the active triphosphate form in the liver in rats (administered 30 mg/kg) is substantially longer at 6.2 h.[141] Despite the apparently weak anti-CoV activity, on 10 April 2020, the company BioCryst Pharmaceuticals, Inc. announced that a clinical trial will be conducted in Brazil to evaluate the antiviral efficacy of Galidesivir (administered (iv) in patients with SARS-CoV-2 infection (NCT03891420).

The isobutyrate ester derivative 18 is the orally available prodrug of the ribonucleoside analogue 25 (β-d-N4-hydroxycytidine, NHC, EIDD-1931) (Figure ).

Figure 22

Chemical structures of the prodrug 18 and its parent compound 25. The isobutyrate ester prodrug moiety is shown in blue.

Compound 25 is a broad-spectrum antiviral agent with in vitro activity against multiple unrelated viruses, including influenza, RSV, chikungunya virus, Ebola, Venezuelan equine encephalitis virus, and Eastern equine encephalitis virus.[98,143−146] In a recent study, 25 displayed potent antiviral activity in two cell lines infected with SARS-CoV-2, namely Vero cells, with EC50 = 0.3 μM and low toxicity (CC50 > 10 μM), as well as Calu3 cells, with EC50 = 0.08 μM (plaque assay). Moreover, 25 turned out to be active against SARS-CoV (average EC50 = 0.14 μM), MERS-CoV (average EC50 = 0.024 μM), and SARS-CoV-2 in HAE cultures[98] and shows remarkable potency against a model CoV mouse hepatitis virus (MHV) bearing resistance mutations to Remdesivir.[124] Thus, the lack of cross-resistance suggests that 25 and Remdesivir may select for exclusive and mutually sensitizing resistance pathways.[98]25 was reported as a potent broad-spectrum inhibitor of respiratory viruses during a HTS campaign for influenza A virus (IAV) and RSV inhibitors. Derivative 25 was found to be active at concentrations in the nM to low-μM range and selectivity indices (SIs) of ≥89 against a broad panel of RSV, IAV, and IBV laboratory strains and isolates.[147] DMPK profiling for 25 demonstrated dose-dependent oral bioavailability of 56, 43, and 36% as well as T1/2 of 5.2, 3.2, and 2.7 h following oral administration of 50, 150, and 500 mg/kg, respectively. The dose dependency of prodrug exposure was consistently observed in lung tissue, however, the peak of 25 triphosphate concentrations saturated above the oral dose levels of approximately 150 mg/kg, suggesting an anabolism bottleneck of 25 in mouse respiratory tissue at higher prodrug levels. Lung tissue distribution assessment revealed sustained levels of the active triphosphate anabolite of approximately 10 nmol/g for over 8 h after administration of 150 and 500 mg/kg. Compound 25 was also well tolerated after extended dosing at 800 mg/kg of body weight/day. While PK properties were good in rodents, oral bioavailability of 25 in cynomolgus macaques was limited. This liability led to the development of the prodrug 18, a 5′-isobutyrate ester of 25, which demonstrated good oral bioavailability in nonhuman primates. Upon intestinal absorption, 18 can be efficiently hydrolyzed to free 25.[148]

Compound 18 has shown a significant reduction of the virus titer and ameliorated the pulmonary function in animal models of SARS-CoV and MERS-CoV infection.[98] In prophylactic efficacy studies, mice were administered 18 at doses of 50, 150, or 500 mg/kg by oral gavage 2 h before infection with SARS-CoV and MERS-CoV, and then every 12 h. At the dose of 500 mg/kg, the most complete protection from disease was observed, therefore the same dose was used for therapeutic efficacy studies for both viruses, with treatment starting 2 h before infection or 12, 24, or 48 h post infection (hpi). The overall results showed that early treatment with 18 provides higher clinical benefit, highlighting a window of 24 to 48 h within which to treat emerging CoV infection in mice before peak replication. The decrease of MERS-CoV yields both in vitro and in vivo were correlated with an increase of the frequency of mutation in the viral RNA but not in the host RNA, highlighting a putative mechanism of lethal mutagenesis. The interesting preclinical results, suggesting that 25/18 might be a promising clinical candidate, prompted the FDA and the UK Medicines and Healthcare Products Regulatory Agency to quickly give approval to initiate human safety testing. Inhibitor 18 has been evaluated in a phase I trial, recently completed (NCT04392219), whereas in late May, two phase II randomized, double-blind, placebo-controlled trials were activated to evaluate the antiviral efficacy of 18 in adults diagnosed with COVID-19 (NCT04405570, NCT04405739). To this aim, the Drug Innovation Ventures at Emory (DRIVE) LLC, which developed the drug, partnered with the biotech company Ridgeback Biotherapeutics LP and with Merck to advance 18 through clinical development and to optimize the drug’s availability during the pandemic.

The structural information on CoV viral proteins available to date, evolving at unprecedented speed, has prompted a number of in silico studies attempting to provide new lead compounds or to repurpose known drugs as possible antiviral agents against SARS-CoV-2.[149,150] However, experimental validation of the findings from such studies is still missing.

So far, NIs reached the most advanced stages of development; even if the activity of the currently evaluated candidates could be improved by medicinal chemistry efforts, taking into account the wealth of structural information that become available. Conversely, NNIs have not yet been reported for SARS-CoV-2 and for other human pathogenic CoVs. NNIs generally show lower resistance barrier with respect to NI/NTP and are characterized by drug-like heterocyclic scaffold; thus, the identification of such agents may offer the advantage to handle molecules with higher optimization potential. However, no allosteric sites have been mapped on the RdRp protein surface for CoVs. Because viral RdRp shows the typical 3D right-hand shape, organized in palm, thumb, and fingers subdomains, the knowledge acquired on other RNA virus polymerase may allow similar considerations also for CoVs RdRp. In this regard, the best characterized viral RdRp is HCV NS5B that, despite working in absence of a primer in cell environment, shows four distinct binding sites for different classes of NNIs that are able to allosterically modulate the enzyme.[151] A very large number of inhibitors, specifically binding one of these sites, have been reported, with some of them reaching late stages of clinical trials and also approval (for instance, Dasabuvir).

The integrated computational and experimental analysis of the nsp12 surface may point toward the discovery of potential allosteric sites. The functional replicase complex requires the tight association of nsp12 with cofactors nsp7 and nsp8, as shown by cryo-EM, therefore these interfaces could represent druggable sites to be explored.

Concluding Remarks

Concurrent to the pandemic outbreak of the SARS-CoV-2 infection, the scientific community in collaboration with private and public funding organizations, started an unprecedented task force to find effective antiviral agents to stop the COVID-19 crisis. Impressive efforts in molecular biology, biochemistry, and structural biology of SARS-CoV-2 have indeed provided in a very short time a huge amount of data. Both biochemical assays and 3D structures have been made public for several viral and host proteins, whereas newly developed phenotypic screenings might allow rapid identification of alternative molecular targets and antiviral compounds.

The knowledge acquired in the past years on other human pathogenic CoVs, such as SARS-CoV and MERS-CoV, to cite a few, can accelerate COVID-19 drug discovery. Considering the information available to date, 3CLpro and nsp12 RdRp are among the most characterized SARS-CoV-2 targets, from both a structural and biochemical perspective, prompting the discovery of specific inhibitors as promising anti-CoV agents. Moreover, 3CLpro and nsp12 RdRp show the highest degree of conservation across different CoVs, with 96% sequence identity between the SARS-CoV-2 and the SARS-CoV proteins, thus fostering the identification of broad-spectrum inhibitors.

At a glance, NIs targeting the nsp12 RdRp appear to be the most advanced candidates in the therapeutic arsenal for COVID-19 treatment. Indeed, the broad-spectrum viral RdRp inhibitor Remdesivir has received EUA for the treatment of patients diagnosed with COVID-19, being approved as the first specific antiviral drug. Remdesivir is iv administered, and this can be considered a main drawback limiting a fast large-scale production and forcing its use in hospital setting. To overcome this hurdle, an inhalation formulation is currently under evaluation in phase I clinical trials. The second most advanced NI is the orally available derivative 18, namely the 5-isobutyrate ester of 25, acting as a broad RdRp inhibitor that could be administered outside of a clinical setting. Both 18 and 25 showed interesting preclinical results against CoVs in in vitro/in vivo models; thus, a phase I trial started in mid-April and two phase II randomized, double-blind, placebo-controlled trials launched in late May are currently ongoing. The findings collected from both clinical and preclinical studies highlight that these drugs should be administered as early as possible after infection, as they might not be as effective in the later stages of the disease. Notably, these are the most advanced compounds coming from repurposing of molecules already profiled for other viruses, indicating that there is still room for the identification and development of further CoV RdRp inhibitors. In fact, on the basis of the experience with other viral polymerases, an alternative route to NIs would be represented by the discovery of NNIs. Such an approach would require the identification of allosteric pockets, which, however, have not been validated so far, although many efforts are ongoing in this direction. The NNIs have several advantages comparable to the NIs, particularly the higher chemical diversity and likely po administration. A major drawback of these compounds could be instead the low genetic barrier for resistance and narrow-spectrum activity, as previously found for NNIs targeting other viruses such as HIV-1 and HCV.

The second most studied essential viral target is the 3CLpro. This protease shows a unique substrate sequence specificity for Gln-Leu as P1–P2 and maintains an overall high degree of conservation across the CoV genus. Remarkably, the considerable amount of data generated on other human pathogenic CoVs together with the available structural data on SARS-CoV-2 3CLpro enable SBDD of novel inhibitors of this enzyme. In the initial effort to identify 3CLpro inhibitors, several research groups are mainly focusing on the development of peptidic covalent and covalent reversible inhibitors. However, the inhibitors reported so far do not significantly deviate from the sequence specificity, showing only small differences in P3/P4. The most relevant examples are the indole dipeptide derivatives with an aldehyde and ketone-based warheads, with compounds 5 and 8 resulting in the leading SARS-CoV-2 3CLpro inhibitors. Interestingly, these molecules are endowed with nM potency against the isolated target and in cell-based assays that are coupled to promising in vivo DMPK and safety properties. However, the efficacy of derivatives 5 and 8 in animal models are still unknown. It is worth noting that the prodrug of compound 8, i.e., its phosphate ester 9, has been reported as a suitable clinical candidate for iv administration, advanced into a phase Ib clinical trial. The additive/synergistic antiviral effect observed in vitro for the combination of 8 plus Remdesivir represents a promising result to plan possible multidrug regimen.

Nonetheless, so far, there is no report describing peptidic inhibitors of CoVs 3CLpro that are orally available, an important feature required for the domestic management of a widespread infection such as COVID-19. Through additional approaches, such as fragment screenings and HTS of libraries of approved and investigational drugs, Bacalein, a flavone derivative widely employed in traditional Chinese medicine, has been described in a preprint as noncovalent active site inhibitor of 3CLpro, as shown by X-ray crystallography, able to selectively inhibit SARS-CoV-2 replication in cells. Due to the stage of the development of such molecules, it is, however, clear that current 3CLpro inhibitors cannot be considered a near-term strategy to tackle COVID-19, although they represent a suitable starting point for lead optimization programs.

Finally, another putative emerging target is the PLpro, which is nevertheless a more challenging and underexplored protein for the identification of inhibitors. Furthermore, this protease is less conserved across the CoVs family, and for this reason it may be less attractive in view of possible future CoV outbreaks.

A summary table reporting all inhibitors for each of the three targets, with references and the current stage of development, is provided (Table ).

Table 2

List of All the Compounds, Including Repurposed Drugs, Discussed in the Text

compd	target	phase	clinical trial ID	refs
1	3CLpro	preclinical		(40,51,52)
2	3CLpro	preclinical		(42)
3	3CLpro	preclinical		(35)
4	3CLpro	preclinical		(35)
5	3CLpro	preclinical		(55)
6	3CLpro	preclinical		(55)
7 (GC376)	3CLpro	preclinical		(43,56,57,59,60)
8	3CLpro	preclinical		(71)
9	3CLpro	Ib	NCT04535167	(72)
Boceprevir	3CLpro	preclinicala		(47,59)
Baicalein	3CLpro	preclinicalb		(76)
Baicalin	3CLpro	preclinicalc		(76)
10	3CLpro	preclinical
11	PLpro	preclinical		(87)
12	PLpro	preclinical		(87)
Ebselen	3CLpro PLpro	preclinicald		(40,88)
13	PLpro	preclinical		(89)
14	PLpro	preclinical		(89)
15	PLpro	preclinical		(89)
16	PLpro	preclinical		(89)
17	PLpro	preclinical		(91,92)
Remdesivir	RdRp	FDA approved[12]	NCT04292899, NCT04280705	(109,110)
Favipiravir	RdRp	IIIe	ChiCTR2000030254	(135)
Galidesivir	RdRp	I	NCT03891420	(141,142)
18 (EIDD-2801)	RdRp	II	NCT04405570, NCT04405739	(98,143−146)

Marketed for HCV.

Phase II for influenza fever (NCT03830684).

Phase I as an ingredient of Moist Exposed Burn Ointment (MEBO) for the treatment of pain and re-epithelization (NCT03728244, NCT02737943).

Phase II for brain disorders.

Marketed for influenza virus infections.

The current scenario has emphasized that CoV infection coming from the spillover such as MERS, SARS-CoV, and SARS-CoV-2 could benefit from broad-spectrum antivirals, which today is still an unmet necessity. Therefore, stimulating research efforts toward this direction is of pivotal importance in order to face potential future pandemics.

Looking at the current panorama, the drug repurposing approach identified one approved drug, Remdesivir, and a covalent inhibitor with the hydroxyl methylketone prodrug, which entered in phase I. Other compounds are under investigation but lagging behind in their development, and there is no certainty that they will conclude the process. Instead, the structural information acquired on the targets will be of great inspiration for the medicinal chemists that will design ad hoc molecules, especially for the proteases where this approach has been successful for other viral infections. The antiviral approach put in place for this emergency will also be very important for future pandemics.