a Department of Medicinal Chemistry, Kyoto Pharmaceutical University, Japan. E-mail: firstname.lastname@example.org
Two newly isolated coronaviruses (CoVs) cause the severe pneumonia-like respiratory illnesses, Severe Acute Respiratory Syndrome (SARS) and Middle East Respiratory Syndrome (MERS). Neither therapeutic agents nor vaccines have been developed thus far, and even future pandemics of related infectious diseases are expected through zoonotic virus infections. Since the 3C like (3CL) protease of SARS/MERS CoV, which has structural similarities with the 3C protease of rhinovirus causing common cold in humans, is essential to the proliferation of SARS/MERS CoV, inhibition of the 3CL protease (3CLpro) is thought to be an ideal target for the development of therapeutic agents against SARS and MERS. This article describes the recent achievements in the development of inhibitors of the SARS/MERS 3CLpro mainly based on two different approaches: one by combining a peptide-like structure with a reactive functional group, a so-called “warhead,” and a second one by combining virtual screening and high-throughput screening of a real compound library. A recent approach based on the structure-based rational design of a novel inhibitor scaffold for 3CLpro is also included.
Coronaviruses (CoVs) are enveloped, positive-strand RNA viruses that infect various vertebrates including bats, poultry, and humans. The name “coronavirus” is derived from the crown-like spikes on their surface. There are four main sub-groups of coronaviruses based on phylogenetic analysis of the genome, known as alpha, beta, gamma, and delta. Infectious bronchitis virus (IBV) is the first coronavirus discovered and was isolated from chicken embryos in 1937.1 In the 1960s, two human coronaviruses, human alpha coronavirus 229E (HCoV-229E) and human beta coronavirus OC43 (HCoV-OC43), were discovered.2,3 These human coronaviruses usually cause disorders such as common cold and/or respiratory illnesses of mild to moderate severity. In 2003, a new human beta coronavirus (Severe Acute Respiratory Syndrome or SARS CoV) was identified.4–6 Recently, three additional human coronaviruses were identified: alpha coronavirus NL637–9 and the beta coronavirus HKU110,11 and MERS-CoV (also known as HCoV-EMC, the coronavirus that causes Middle East Respiratory Syndrome, MERS)12,13 (Fig. 1). Among these human CoVs, SARS- and MERS-CoV, in contrast to HCoV-229E and HCoV-OC43, cause a life-threatening atypical pneumonia, termed severe acute respiratory syndrome. The origin of both SARS- and MERS-CoV are strongly suspected to be zoonotic viruses that infect bats or camels.
In 2003, SARS spread worldwide from its likely origin in southern China in a short period of time, showing that the causative CoV, SARS-CoV, is a great threat to human health. The SARS epidemic affected about 8500 patients with more than 800 fatalities, which boosted coronavirus research in all directions. Nevertheless, in 2012, a new respiratory illness similar to SARS was identified in Europe and the Middle East. This respiratory syndrome called Middle East Respiratory Syndrome, MERS, had affected more than 1800 patients with a fatality rate of 36%. Even after these pandemics, no effective therapy exists for infections with these coronaviruses which may cause re-emergence of SARS/MERS or other related severe diseases.
SARS- and MERS-CoVs recognize a specific receptor on the host cell membrane using the spike (S) protein of the virus. Angiotensin-converting enzyme 214 (ACE2) is a functional receptor for the SARS CoV, and dipeptidyl peptidase 415 (DPP4, also known as CD26) is a functional receptor for the MERS CoV. Interaction of the S1 domain of the viral S protein with the host cell receptors is followed by membrane fusion of the virus and host cell to transport the virus RNA into the host cell. Thus, anti-ACE2 or anti-DPP4 antibodies block viral infections of the host cells, strongly suggesting that agents such as a corresponding soluble receptor (ACE2 or DPP4) or its antibody could be promising inhibitors of the virus-cell interactions.14,15 The viral genome is translated and processed to virus-derived structural proteins such as spike (S), envelope (E), membrane (M), and nucleocapsid (N) proteins. A set of nonstructural proteins including virus-derived proteases are also processed by the viral protease. Thus, inhibitors of the processing reaction required to yield proteins necessary for viral replications are other promising agents to suppress the proliferations of the infectious viruses.
In this review article, recent efforts to develop therapeutic agents for SARS/MERS and related respiratory syndromes are discussed focusing on a coronavirus-specific protease, 3CL (chymotrypsin like) protease. The contents cover the following headings: Introduction (this section) Coronavirus 3CL protease 3CL protease of SARS-CoV and MERS-CoV Crystal structure of 3CL protease Maturation of the 3CL protease and enzymatic activity Inhibitors of 3CL protease Peptide-mimetic inhibitors Small-molecule inhibitors Conclusion
In a recent overview regarding SARS-CoV 3CL protease inhibitors,16 a cumulative source of SARS 3CL protease inhibitors is described in detail. Thus, in this article, the author will focus on several typical inhibitors based on the inhibitory mechanism as well as the structural interactions with the 3CL protease. First, the protein chemistry of the corona virus 3CL protease is described as the base of structure analyses of protease-inhibitor interactions.
The 29.7 kb positive-strand RNA genome of SARS-CoV is complexed with the basic nucleocapsid (N) protein to form a helical capsid (Fig. 1). The virus membrane is complexed with three viral proteins: a glycoprotein called spike (S) protein, a membrane-spanning protein called membrane (M) protein, and a hydrophobic envelope (E) protein that covers the entire structure of the CoV. The SARS-CoV genome contains two open reading frames (ORFs 1a and 1b; Fig. 2) encoding two large replicative polyproteins, pp1a (486 kDa) and pp1ab (790 kDa).17,18 Expression of the ORF1b-encoded region of pp1ab is involved in the ORF produced by ribosomal frameshifting into the −1 frame just upstream of the ORF1a translation termination codon.
The pp1a and pp1ab polyproteins are processed by a viral protease to yield the functional components. SARS-CoV encodes two proteases for this processing: papain like cysteine protease (PLpro) and 3C-like cysteine protease (3CLpro, also called main protease, Mpro). The name of 3C-like comes from picornavirus 3C proteases encoded on the non-capsid region 3C since the substrate specificity of both proteases is similar.19 PLpro cleaves at three sites in the N-terminal region of pp1a/pp1ab, and 3CLpro cleaves at eleven sites in the C-terminal region (Fig. 3).20 The SARS-CoV 3CLpro cleaves the precursor protein at three times more sites than PLpro. In addition, 3CLpro is indispensable for viral replication but not found in the host cell.21,22 These features of the SARS 3CLpro make it as an ideal target for antiviral agents.
SARS 3CLpro consists of 306 amino acid residues and is a cysteine protease containing the catalytic dyad defined by His41 and Cys145 (Fig. 4). The N-terminal part (1-184, domains I and II) is composed of a two-β-barrel fold similar to that of a chymotrypsin-type protease. The C-terminal part (201–303, domain III) containing five α-helices takes on a globular fold (details of the three-dimensional structures are discussed in the following section). Enzyme activity-concentration relationship studies support the proposal obtained by previous studies on other CoVs that the 33 kDa monomer of the SARS 3CLpro shows remarkably weak enzymatic activity, whereas the dimer of the 3CLpro is the active form.
Most CoV derived 3CL proteases have a conserved (Leu/Ile)-Gln-↓-(Ser, Ala, or Gly) core sequence (canonical sequence, the cleavage site is indicated by -↓-). In addition to the canonical sequences, SARS 3CLpro recognizes three noncanonical cleavage sites with Phe, Val, or Met in the P2 position and one non-canonical site with Asn in the P1′-position.23 Comparison of cleavage efficiencies of synthetic substrates containing the eleven cleavage sites of SARS 3CLpro confirmed that the most suitable substrate is the P1/P2 site, the N-terminal site of the SARS 3CLpro itself. The second one derives from the P2/P3 site, the C-terminal site of the SARS 3CLpro, suggesting that SARS 3CLpro cleaves itself most efficiently. The maturation process of the protease will be discussed later. It was also confirmed that Phe in the P2 position can be a substrate similar to the canonical substrate with Leu in the P2 position, indicating that the corresponding S2 pocket of the SARS 3CLpro is rather large and hydrophobic.
SARS 3CLpro uses its thiol group as a nucleophile for proteolysis (Fig. 5). The initial step is deprotonation of Cys145-thiol by an imidazole group of His41. The resulting nucleophilic sulfur atom attacks the substrate carbonyl carbon. Thus, SARS 3CLpro shows the highest enzymatic activity at around pH 7 and the activity decrease to 10% at pH 5 because of the protonation of His41. Then, the N-terminal substrate fragment is released from the enzyme, while the imidazole group of His41 is restored to the deprotonated form. Next, the resulting thioester consisting of the enzyme and the C-terminus of the substrate is hydrolyzed by nucleophilic attack of a deprotonated water molecule. The following release of a carboxylic acid regenerates the free enzyme. Thus, compounds containing a warhead interacting with the thiol group of Cys145 could be a promising agent against SARS.
MERS 3CLpro is closely related to SARS 3CLpro. It is a cysteine protease consisting of 306 amino acid residues, and like SARS 3CLpro, it has three domains. The catalytic dyad Cys148-His41 and the substrate binding site are located in the cleft between domains I and II in the active dimer form. Substrate specificities are also similar to other β-CoVs. The enzymatic activity however is low and it is 5-fold less efficient compared to SARS 3CLpro. Indeed, MERS 3CLpro is the least efficient protease among the β-CoVs (MERS, HKU5, and HKU4).
In 2003, the X-ray structure of the SARS 3CLpro complexed with a peptide CMK (chloromethylketone, 1; Fig. 6) inhibitor was reported.24 In the same year, the crystal structure of the free 3CLpro of HCoV-229E and porcine transmissible gastroenteritis (corona)virus (TGEV) 3CLpro complexed with 1 were elucidated.21 In addition, compound 2 (AG7088, Fig. 6), known as a human rhinovirus 3Cpro inhibitor, is supposed to interact similarly with CoV-3CLpro as with 3Cpro as suggested by superimposition of several crystal structures of 3CLpro/3Cpro-inhibitor complexes. Because the substrate specificity of picornavirus 3Cpro is similar to that of CoV-3CLpro, compound 1 and 2 had been used as a starting point for the design of peptide-based 3CLpro inhibitors. Following these initial studies25,26 on the crystal structure of 3CLpro, more than hundred structures of SARS 3CLpro with or without inhibitors are registered in the PDB at present.
The SARS 3CLpro monomer is comprised of three domains (Fig. 7a). Domains I and II are six-stranded antiparallel β-barrels forming the chymotrypsin-like architecture as in picornavirus 3Cpro. The substrate-binding site is located in a cleft between these two domains. A long loop connects domain II to domain III, a globular cluster of five helices. The amino acid sequence of SARS 3CLpro displays 40 and 44% sequence identity to 3CLpro of HCoV 229E and TGEV, respectively (Fig. 7b). Especially, the sequences in domains I and II, the catalytic domains, show a higher degree of sequence conservation in coronavirus 3CL proteases than does domain III. The MERS 3CLpro monomer also shows high structural and sequence similarities to the SARS 3CLpro.
SARS 3CLpro forms a tight dimer in the crystal structure (Fig. 8), whereas the protease is expected to exist as a mixture of monomers and dimers in solution and their ratio is dependent on the protease concentration. Several kinetic and biophysical studies have demonstrated that SARS 3CLpro is only active in vitro as a tightly associated dimer. Domain III plays an essential role in dimer formation, since domain III alone is able to form a tight dimer.27 In the mature dimer form, the N-terminal amino acid residues (N-finger) are squeezed in between domains II and III of the partner monomer to hold the specific dimer interactions. Thus, the exact placement of the N-terminus has a structural role to hold the mature 3CLpro, because deletion of residues 1 to 5 led to a decrease in activity. Addition of amino acids via a His-tag at either the N- or C-terminus also drastically reduces the enzymatic activity due to the decreased ability to maintain the dimer structure.28 MERS 3CLpro also shows a similar dimer form in the crystal structure. The protein-protein interactions to stabilize the dimer, however, are rather weak compared to those of SARS 3CLpro, which causes the lower enzymatic ability of MERS 3CLpro compared to SARS 3CLpro. Differences in the maturation process, an auto-processing of the precursor protein, are thought to largely contribute to this discrepancy as discussed in the following section.
Results of analytical gel-filtration and analytical ultracentrifugation (AUC) analyses indicated that a large fusion protein, which has additional protein sequences at the N- and C-terminus of SARS 3CLpro, exists as a monomer in solution. The monomer polyprotein, however, can still retain its enzymatic activity to some extent in vitro,29 although the mature SARS 3CLpro functions in the tight dimer form instead of the monomer. To address this dilemma, formation of a substrate-induced immature “intermediate” dimer is expected to be a key mechanism, which also presents a possible mode of SARS 3CLpro auto-release from the precursor polyprotein in vivo.
The proposed auto-release mode of SARS 3CLpro involves four steps (Fig. 9).30,31 At first, along with the polyprotein synthesis following the viral replication, two “immature” 3CLpro precursor monomers approach one another, and their domain III forms an “intermediate” dimer structure, which triggers the substrate-induced dimerization and insertion of their uncleaved N-termini into the substrate-binding pockets of the opposite monomer (Step 1). Next, with a substrate induced-fit mechanism, the active site at the interface of domains I and II is set to catalyze the cleavage of the N-terminal precursor sequences. This results in the N-terminal fingers slipping away from the active sites to their final positions (Fig. 8) to yield a dimer with “uncleaved” C-termini (Step 2). The product of this cleavage is a dimer which has the mature N-terminal sequence and increased catalytic ability compared to the “intermediate” dimer. Then, the “uncleaved” C-terminus of the resulting dimer can insert into an active site of another dimer to proceed trans-cleavage (Step 3). Once the C-terminus is processed, the final mature dimer with authentic N- and C-termini is formed, which is observed in the crystal structure of highly active wild-type SARS 3CLpro (Step 4).
Evaluation of the monomer-dimer dissociation constant, Kd, for MERS 3CLpro revealed that the capacity of MERS 3CLpro to dimerize is 130-fold weaker than SARS 3CLpro.32 Analytical ultracentrifugation sedimentation velocity (AUC-SV) studies support the weak association of the MERS 3CLpro dimer, which strongly suggests that the enzyme exists mainly as a less active monomer in solution. The AUC-SV studies also revealed that the addition of low concentrations of ligand to the MERS 3CLpro remarkably increases the ratio of the dimer as well as the catalytic ability. Based on these data, the activation mechanism of MERS 3CLpro can be explained by the kinetic model shown in Fig. 10. The mature dimer of MERS 3CLpro, produced from two immature MERS 3CLpro monomers according to the route shown in Fig. 9, dissociates into inactive monomers in the absence of any ligand (substrate, inhibitor, or 3CLpro cleavage sites in the precursor polyprotein). Binding of ligand to the monomer promotes the monomer to dimer switch. The substrate binds to the resulting dimer at the second active site and is cleaved catalytically. A potent inhibitor can competitively bind to the active site, or directly compete with the ligand for binding to free dimer active sites.
Inhibition of the catalytic activity of coronavirus 3CLpro is thought to be the most promising mechanism to prevent virus proliferation, since the 3CLpro is essential for the virus's viability as described above. Thus, after the SARS pandemic in 2003, numerous studies focused on the development of potent inhibitors for SARS 3CLpro. These inhibitors are structurally classified into two types: peptide-mimetic inhibitors and non-peptide small-molecule inhibitors. Considering the mode of interaction, both types of inhibitors can be classified into two types, inhibitors forming a covalent bond with 3CLpro (irreversible inhibitor) and non-covalent type inhibitors (reversible inhibitor). Typical inhibitors of the first generation reported after the first outbreak of SARS are summarized in Fig. 11.33–36 These peptide-mimetic and non-peptide inhibitors provided valuable insights into further modifications based on structure-based design. In the following sections, several typical inhibitory mechanisms as well as the structural modifications of peptide-mimetic and non-peptide inhibitors are described.
Generally, peptide-mimetic inhibitors of proteases can be designed combining a substrate-like sequence with a functional group (a so-called “warhead”) targeting the catalytic center of the target enzyme. The substrate-like sequence of peptide-mimetic inhibitors for the SARS 3CLpro is designed to optimize the specific interactions at the S1′ site and the S1 to S4 non-prime sites of the 3CLpro. Since the SARS 3CLpro is a cysteine protease, functional groups that can interact with the thiol group are selected as the chemical “warhead”. These warhead groups include a Michael acceptor, aldehyde, halomethyl ketone, epoxy, and others.
The general strategy for the design of a peptide-mimetic inhibitor containing a Michael acceptor involves the replacement of a substrate's scissile amide bond with an appropriate Michael acceptor. At the active center of the SARS 3CLpro, the nucleophilicity of the thiol group of the catalytic center Cys145 is increased by a proton-withdrawing effect caused by His41 at the catalytic dyad, which promotes a typical 1,4-addition to the α,β-unsaturated structure of the Michael acceptor (Fig. 12). The resulting protonated His41 gives the proton to an unstable intermediate anion to form a 3CLpro covalently bound to the inhibitor. Thus, the Michael acceptor type compound acts as a suicide substrate to abolish the catalytic activity of the enzyme.
Starting from a prototype compound 2 (Fig. 6) targeting rhinovirus 3CLpro, optimizations of side-chain structures at the P1′ to P4 sites were conducted to develop SARS 3CLpro specific inhibitors (Table 1).37–39 Interactions of the SARS 3CLpro with some of these inhibitors were evaluated based on X-ray crystal structure analyses of the 3CLpro complexed with the inhibitor (for example, PDB codes 2ZU4 and 2ZU5). These structure analyses confirmed that Cys145 attacks the α-carbon of the Michael acceptor (Fig. 12) at the P1′ site to form a covalent C-S bond of 1.99 Å. Glutamine and a five-membered lactam ring were favored as P1 site substitutes with the five-membered lactam ring being much better. For the P2 site, a cyclohexyl or isopropyl substitute was preferred over a phenyl substitute, which suggests that rigid and planar properties were not favorable for binding in the large hydrophobic S2 pocket of SARS 3CLpro. The P3 group is directed toward the bulk solvent and was predicted to have no specificity for binding. A lipophilic tert-butyl group at this site, however, increased the inhibitory activity, probably due to shifting the substituent toward the P4 site, which induced hydrophobic interactions with the 3CLpro at the phenyl ring of N-terminal benzoyl group.
In another study,40 the effect of methylene insertion between the reactive α,β-unsaturated structure and the P1 position was investigated (Table 2). The Michael type analogs elongated toward the prime site showed no inhibitory activities. These results suggest that the recognition of the prime-site structure is strict and no linker structure inserted between the scissile site and neighboring prime site would be tolerated. By comparing the inhibitory activities of both type inhibitors containing the same peptide sequence, it was also suggested that an aldehyde group would be more effective as a warhead than the α,β-unsaturated structure.
An aldehyde group is another effective functional group involved in a nucleophilic addition reaction to yield an alcohol (Fig. 13). Thus, combined with a substrate peptide sequence, an aldehyde group instead of a Michael acceptor can be used as an effective warhead of SARS 3CLpro inhibitors.
Combining the structural analyses (PDB code 3SN8) of the SARS 3CLpro complexed with prototype compound 16 (Fig. 14) and structure-activity relationship (SAR) studies of Michael-type inhibitors (Table 1), a potent peptide aldehyde inhibitor 17 (Ki=53 nM) was developed.41,42 Analyses of the crystal structure of SARS 3CLpro complexed with 17 (PDB code 2GX4) revealed that the thiol group of Cys145 attacks the carbonyl carbon of the aldehyde of 17 to form a covalent C-S bond (1.24 Å). The resulting active site consisting of a Cys145-His45 dyad at the S1′ pocket, an oxyanion hole formed by the aldehyde oxygen and N-H of Cys145 and Gly143 at the S1 pocket, as well as a large hydrophobic cavity at the S2 to S4 pockets were clearly observed. In addition, at the P1 and P4 sites, hydrogen bonds served to hold the interactions between the 3CLpro and inhibitor 17, whereas the P2 site cyclohexyl group formed extensive hydrophobic contacts at the S2 pocket.
During our studies on the SARS 3CLpro and its inhibitors, we found that the mature SRAS 3CLpro is susceptible to auto-degradation at the Arg188/Gln189 site, which causes a loss of activity.43 Mutation of Arg to Ile remarkably increased the stability while keeping the almost same three-dimensional structure (PDB code 3AW1) as native SARS 3CLpro. Probably due to this stability, the mutant protease (R188I SARS 3CLpro) showed much higher activity than the mature protease. Use of this highly active mutant protease made it possible to quantitatively evaluate peptide-aldehyde inhibitors using a conventional HPLC system combined with a non-modified substrate peptide instead of peptide derivatives with fluorescent substituents.
Initial SAR studies of a substrate-based peptide aldehyde inhibitor revealed that a P1 site imidazole substituent increases the inhibitory activity more than 6-times compared to a simple substrate-based inhibitor, thus yielding inhibitor 20 (IC50=5.7 μM; Table 3) from inhibitor 12 (IC50=37 μM). Structural analyses of the SARS 3CLpro complexed with inhibitor 20 (PDB code 3AW0) demonstrated that the large hydrophobic S2 pocket is not fully occupied and the side chain functional group at the P5 site is not involved in the interactions. Structure-based optimizations provided the potent tetra-peptide aldehyde inhibitor 22 (IC50=98 nM).40 X-ray crystal structure analysis of the SARS 3CLpro complexed with 22 clearly showed the tight hydrophobic interactions of the cyclohexyl group at the P2 site as well as an additional hydrogen-bond interaction at the β-hydroxyl group of the P4 site Thr.
In these X-ray structural analyses of inhibitors complexed with the mutant protease, the carbonyl carbon of the aldehyde was detected at a distance of 2.30 Å in 20 and 2.48 Å in 22 from the thiol of Cys145 at the catalytic center (Fig. 15). The electron density of the aldehyde group could be fitted to an expected sp2 carbon. In addition, no significant difference was detected between the IC50 value obtained after pre-incubation of the inhibitor with the protease prior to the addition of the substrate and that obtained by simultaneous mixing of the inhibitor, protease, and substrate. These results strongly suggested that no stable covalent bonds are formed with the protease. Kinetic data for 22 obtained from Lineweaver–Burk plots also suggested that inhibitor 22 containing an aldehyde functions as a competitive inhibitor without forming a covalent bond.
Based on an initial study on peptide-mimetic inhibitors of 3Cpro from enterovirus 71 (EV71), three peptide-aldehyde inhibitors were identified as MERS 3CLpro inhibitors.44 Compounds 23, 24, and 25 (Table 4) showed IC50 values of 2.4, 4.7, and 1.7 μM against MERS 3CLpro, respectively. These compounds also inhibited SARS 3CLpro at lower IC50 values. In silico molecular docking of 25 against MERS 3CLpro suggested that the γ-sulfur of Cys148 forms a covalent bond with the aldehyde carbon of 25 and the resulting oxyanion is stabilized by His41. In a cytopathic inhibition assay using MERS CoV infected Huh-7 cells, peptide aldehyde inhibitors 23, 24, and 25 suppressed viral replication with EC50 values of 1.4, 1.2, and 0.6 μM, respectively. The lower EC50 than IC50 value was considered to be due to the high concentration of MERS 3CLpro used for in vitro enzymatic assay because of the weak dimerization ability of MERS 3CLpro. These peptide-aldehyde inhibitors were also cytotoxic for other human CoV, 229E and OC43, although they were not as potent as for inhibiting SARS- and MERS-CoV.
Halomethyl ketone is another warhead which can form a covalent bond by an apparent alkylation reaction. The halomethyl group makes the adjacent ketone group more susceptible to a nucleophilic attack, and the initial nucleophilic attack of a thiolate of Cys145 of 3CLpro toward the carbonyl group of the inhibitor (I) leads to reversible formation of a thiohemiketal (E–I*) holding the tetrahedral conformation that resembles the enzyme-substrate intermediate in the catalytic cleavage (Fig. 16). Intramolecular rearrangement leads to the alkylation product (E–I) as the final product. Experimental data on the kinetics indicate that the above mechanism is more conceivable than a direct mechanism in which the thiolate ion attacks the halomethyl carbon adjacent to the warhead carbonyl leading to the alkylated product.
Initial studies of N,N-dimethyl glutaminyl inhibitors with a fluoromethyl ketone group showed that a halogenated methyl group functions as an effective and promising warhead (Table 5).45 These compounds were designed based on their caspase inhibitory activities. Antiviral activity assessed by cytopathic effect (CPE) inhibition in SARS-CoV infected Vero cell cultures revealed that compound 26 can protect the cells against SARS infection with an EC50 value of 2.5 μM and showed low toxicity in mice. P2-Leu of 26 can be replaced by Ile or Val to yield additional compounds 27 and 28 with slightly lower EC50 values. These active compounds were inactive against rhinovirus type-2 in a cell-based assay, which suggests that these compounds are specific against SARS-CoV.
Analyses of the catalytic mechanism based on kinetic evaluation revealed an effect of the P1 site substituent as well as a dependence on a halogen atom in the warhead for the inhibitory mechanism shown in Fig. 16. Kinetic inhibition data of compounds 30–33 containing different P1 site substituents and halogen atoms of the warhead for the SARS 3CLpro are summarized in Table 6.46 Hydrophobic substituents such as aromatic groups (phenyl, naphthyl, or parafluorophenyl) as well as an aliphatic bulky group at P1 site are tolerated, although Gln is traditionally present at this position. This tolerance is consistent with previous results reporting modifications at the P1 site with a lactam ring, keto-glutamine analogs, and an α,β-unsaturated ester at P1 site.
Altering the halogen atom of the warhead had a substantial effect on the kinetic property of the inhibitors. The rate of the irreversible inactivation step (k3 in Fig. 16) is related to the ability of the warhead to accept a nucleophilic attack by the active center thiol leading to the eventual alkylation. The value of k3 of compound 31 (2.8×10−2 s−1) was almost twice that of 30 (1.5×10−2 s−1) and 32 (1.8×10−2 s−1), which suggests that the larger substituent at the P1 site may orient the warhead in a favorable conformation for the interaction with the thiol group of Cys145. In contrast, the k3 value of compound 33 was too small to be measured accurately, indicating that the irreversible step is very slow and that compound 33 behaves as a reversible inhibitor for several hours. Irreversible inhibition of the enzyme activity was only noticed after a 12 h incubation with compound 33 at a high concentration.
The structure of the wild-type SARS 3CLpro complexed with compound 33 was determined using X-ray crystallography (PDB code 3D62). Although the electron density for only a portion of compound 33 was observed, a thioether bond (1.7 Å) between the carbon that was originally bound to bromine of 33 and the sulfur atom of Cys145 was clearly detected. This structural analysis is consistent with a time-dependent bimodal mode of inhibition for this compound: initial formation of a reversible complex (E–I*) followed by rearrangement to an irreversible complex (E–I) after at least 6 h incubation.
The phthalhydrazide group is another functional group which makes the adjacent ketone carbonyl carbon more susceptible to a nucleophilic attack. Extending the initial studies of a new class of inhibitors against hepatitis A virus (HAV) 3Cpro, several phthalhydrazide type inhibitors for SARS 3CLpro were developed (Fig. 17).47 The phthalyhydrazide warhead functions well against the SARS 3CLpro, especially in combination with a γ-lactam substitute at the P1 site (Fig. 17; 34–37 vs. 38–41).
In the inhibitory reaction by compound 38, no kinetic evidence of covalent inhibition was observed when 10 μM 3CLpro was pre-incubated with 100 μM of 38 for 15 to 60 min. Electron spray ionization-mass spectrometry (ESI-MS) analysis of a SARS 3CLpro crystal complexed with compound 38, however, revealed that the product is a covalent adduct between the 3CLpro and compound 38 without the phthalhydrazide moiety. Detailed analyses of the crystal structures of the SARS 3CLpro-inhibitor complex obtained via the co-crystallization method (PDB code 2Z3E) and via the soaking method (2Z3C) suggested the intermediacy of the episulfide linkage in solution (Fig. 18).48 In these crystal structures, the bond length between the sulfur atom of the Cys145 thiol and the carbon atom of the P1 carbonyl was 1.81 Å in the thioacyl-like structure and 1.83 Å in episulfide structure. The structures obtained from the electron densities also indicated that transformation of the SARS 3CLpro-bound episulfide cation to the thiomethyl ketone structure requires a conformational change which is more attainable in aqueous solution than in crystals.
A trifluoromethyl group is another warhead making a neighboring carbonyl group susceptible to a nucleophilic attack. The substrate-based trifluoromethylketone inhibitor 46 inhibited the SARS 3CLpro (IC50=10 μM) more effectively than a series of low molecule inhibitors (Table 7).49 A Lineweaver-Burk plot showed that the inhibition with 46 was competitive in the initial 4 h reaction. Prolonged incubation of the 3CLpro with 46, however, exhibited a time-dependent decrease in the enzymatic activity as a function of the inhibitor concentration. This time-dependent tightening of inhibition was assumed to be caused by the slow formation of a covalent adduct through the nucleophilic attack of the thiol of Cys145 on the carbonyl carbon (Fig. 19). Although the X-ray crystal analyses were not successful on this compound, computational molecular modeling strongly suggested a covalent bond formation in the final stage of the inhibition.
A substrate-based trifluoromethyl ketone inhibitor containing Glu or Gln at the P1 site (47, Fig. 20) showed only moderate inhibitory activity due to the formation of a cyclic structure which is expected to hardly interact with the active site of SARS 3CLpro.50 In order to block the cyclization, the side chain was modified to increase the bulkiness, yielding the more potent inhibitor 47. Replacement of the trifluoromethylketone group with an electrophilic thiazolyl ketone 49 further increased the inhibitory activity ten times. Following structure optimization at the P4 site combined with a benzothiazole warhead yielded inhibitors 50 and 51, both having low-nano-molar IC50 values.51,52
The natural product E-64 (Fig. 21) from Aspergillus japonicus has been used as a reference inhibitor for many clan CA cysteine proteases including papain, cathepsins, and calpains. The epoxysuccinate structure is a key functional group for the inhibitory activity, but E-64 is not effective in the inhibition of clan CD protease. During the investigations of inhibitors effective against clan CD proteases, aza-peptide epoxides were designed by replacing the nitrogen atom of the scissile amide bond with a reactive epoxy group to receive a nucleophilic attack by an active center Cys.53 Additional conversion of the α-carbon of an amino acid at the P1 site into a nitrogen results in an aza-epoxy peptide (Fig. 21). This replacement induces a trigonal planar geometry to the α-atom of the P1 residue and reduces the electrophilicity of the carbonyl carbon of the P1 residue, which makes the carbonyl group resistant to the nucleophilic attack. Another screening study of peptide derivatives containing electrophilic building blocks indicated that the trans-configured aziridine-glycylglycylate 52 has moderate inhibitory activity against SARS 3CLpro.54
Following studies focusing on inhibitors of the SARS 3CLpro, an aza-epoxy peptide possessing an aza-glutamine at the P1 residue 53 was found that inhibited the 3CLpro with a Ki value of 18 μM. The S, S diastereomer of the epoxide strongly inhibited the cleavage of a substrate peptide, while the R, R diastereomer did not detectably inhibit the 3CLpro. Evaluation of the crystal structure of mature SARS 3CLpro complexed with 53 confirmed the formation of a covalent bond with a distance of 2.01 Å between the sulfur atom of Cys145 and the epoxide C3 atom of 53 (Fig. 22).55,56 Although the length of a normal C-S bond is about 1.8 Å, the difference between the refined (2.01 Å) and expected (∼1.8 Å) distance was not considered significant since the estimated overall coordinate error for the structure analysis was 0.12. Further analyses of the crystal structures revealed an induced-fit binding of an aza-epoxy peptide to the SARS 3CLpro. In the unbound form, the active site and the S1 pocket of the 3CLpro is in a collapsed conformation, whereas they are in an open conformation in the inhibitor-bound form.
A nitrile group, a well-known warhead group of DPP4 inhibitor vildagliptin used as an anti-diabetes agent, was incorporated in the substrate sequence of SARS 3CLpro to develop nitrile-based inhibitors (Fig. 23). Inhibitory activities of four nitrile-based inhibitors with different protecting groups, 5-methylisoxazole-3-carboxyl (Mic), tert-butyloxycarbonyl (Boc), and carboxybenzyl (Cbz) were evaluated.57 The Cbz-tetrapeptide inhibitor Cbz-AVLQ-CN 56 was ten-times more potent (IC50=4.6 μM) than the other inhibitors including the Cbz-hexapeptide inhibitor Cbz-TSAVLQ-CN 57, which suggests that the Cbz group at the P4 position contributed to the suitable interactions at the S4 pocket of SARS 3CLpro.
The crystal structures of the SARS 3CLpro in complex with the nitrile-based inhibitor (PDB codes 3VB7, 3VB4, 3VB5, 3VB6) demonstrated that the inhibitor was covalently bonded to the thiol group of Cys145 via the carbon atom of the nitrile warhead (Fig. 24). In addition, the tetrapeptide inhibitor Cbz-AVLQ-CN 56 can inhibit 3CLpro from human coronavirus strains 229E (IC50=2.3 μM), NL63 (IC50=2.8 μM), OC43 (IC50=1.6 μM), and HKU1 (IC50=1.3 μM), while the same inhibitor had no observable inhibitory effect on caspases. These results suggest that the nitrile-based inhibitor is specific to 3CLpro from coronaviruses and exhibits broad-spectrum inhibition against the 3CLpro.
In general, a low molecular weight inhibitor for infectious viruses and bacteria is considered a promising agent against infectious diseases. Usually, small-molecule inhibitors have various chemical structures even when the same protein/protease is targeted. Thus, many of these small-molecule inhibitors were discovered from natural products and/or related derivatives or by high-throughput screening of synthetic compound libraries. In this section, several examples of small-molecule inhibitors of SARS/MERS 3CLpro found by these procedures are included.
Isatin (2, 3-dioxindole) is an indole derivative found in many plants, such as Isatis tinctoria and Calanthe discolor. It is known that certain isatin compounds are potent inhibitors of rhinovirus 3Cpro. Because the active site architecture of the 3Cpro is similar to that of SARS 3CLpro, isatin derivatives were expected to be good candidates for SARS 3CLpro inhibitors. Using initial studies on isatin derivatives which have N-1 and C-5 substituents, potent inhibitors for SARS 3CLpro 58 (IC50=0.37 μM) and 59 (IC50=0.95 μM) were developed (Fig. 25).58,59 Computer modeling analyses suggested that the isatin scaffold is docked in the S1 site, and the N-1 substituent is located in the S2 site of the 3CLpro. The sulfur atom of Cys145 in the active center of the 3CLpro is estimated to be located in hydrogen-bond distance from the isatin oxygen at C-3.
Related high-throughput screening revealed that 5-bromoisatin was a potent inhibitor of the 3CLpro and could be soaked into a crystal of the 3CLpro. Based on these results, a replacement of the carboxamide group using a series of sulfonamide groups was achieved, and isatin 5-sulfonylamide derivatives 60 (IC50=1.18 μM) and 61 (IC50=1.04 μM) were identified as promising inhibitors for SARS 3CLpro (Fig. 26).60 Docking studies, however, suggested a mode of docking different from that expected for the above derivatives 58 and 59. The 2,3-dioxindole scaffold of the isatin 5-sulfonylamide derivatives is docked at the S1′ site of SARS 3CLpro instead of the S1 site for the above derivatives. The sulfonamide substituent and N-1 substituent were located at the S2 and S1 sites, respectively. These docking model could support the differences of inhibitory activities in a series of isatin 5-sulfonamide derivatives.
During the search for positional anti-SARS CoV agents from medicinal plants and foodstuffs, a series of chalcones isolated from an ethanol extract of Angelica keiskei were found to be moderate inhibitors of SARS 3CLpro. Among the 13 isolated and structure-determined chalcones, compounds 62 and 63 (Fig. 27)61 showed relatively high inhibitory potencies for the SARS 3CLpro (IC50=27 μM for 62 and 11 μM for 63). Comparing the inhibitory activities of both compounds, the perhydroxyl group was expected to be more effective in the interactions with the hydrogen bonding site of the 3CLpro. Lineweaver-Burk and Dixon plots suggest that both chalcones 62 and 63 are competitive inhibitors with Ki values of 35 μM and 16 μM, respectively. In an in silico docking simulation, both chalcones fitted nicely into the substrate-binding pocket of the 3CLpro. The perhydroxyl group of compound 63 formed strong hydrogen-bonds with Cys145 (3.45 and 2.72 Å) as one of the key motifs of this inhibitor. In a separate study on natural products isolated from Broussonetia papyrifera, a few polyphenols were found to be moderate inhibitors for SARS and MERS 3CLpro with IC50 values of 28∼65 μM (Fig. 28).62
A series of active heterocyclic ester analogs was identified as a novel class of mechanism-based irreversible inhibitors with activities in the nanomolar range.63 The possible irreversible acylation of Cys145 by the typical ester derivatives 67 and 68 was verified by ESI-MS (Fig. 29).64,65 The acylation by the ester ligand was also confirmed by an X-ray crystal structure of the SARS 3CLpro complexed with the benzotriazole ester.66 Further studies on 3-chloropyridyl ester-based inhibitors revealed that the position of the carboxylic acid ester is critical for potent inhibitory activity (Fig. 30).67 Carboxylate substitution on indole rings at the 4-position gave the most potent inhibitor 69 with an IC50 of 30 nM. This inhibitor also showed a SARS CoV antiviral activity with an EC50 value of 6.9μM. Covalent modification by this inhibitor was confirmed by Matrix-Assisted Laser Desorption Ionization (MALDI) time of flight (TOF) MS analysis of the 3CLpro obtained after incubation with the inhibitor.
Although the above mentioned active-ester analogs act as potent covalent inhibitors of the 3CLpro, they are supposed to be susceptible to hydrolysis catalyzed by various enzymes such as esterases, lipases, and other enzymes in mammalian cells. These ester analogs initially bind competitively to the active site and are then hydrolyzed by the 3CLpro as a suicide substrate. Thus, the corresponding ketone derivative might form a hemithioacetal with the thiol of Cys145 to act as a reversible inhibitor. Based on this idea, a series of methylene ketones and fluoro-methylene ketones were synthesized and those inhibitory activities were evaluated (Fig. 31).68 The methylene ketone 73 and its fluorinated methylene ketone analogs 74 and 75 were good inhibitors of the 3CLpro with IC50 values of 13–57 μM. ESI-MS analysis of the reaction mixture strongly suggested that these compounds utilize a non-covalent reversible mechanism of inhibition. Docking model studies also suggest that these compounds interact with the 3CLpro in a S4–S1 binding mode using the three-aromatic ring structure to block the entry of substrates into the active site.
Metal-conjugated compounds have been used as pharmaceutical excipients and antimicrobial preservatives in parenteral and topical pharmaceutical formulations. From the screening of a compound library consisting of 960 commercially available drugs and biologically active substances, some metal ions (Cu2+, Hg+, Zn2+) and their metal-conjugated compounds [phenylmercuric acetate (PMA), toluene-3,4-dithiolato zinc (TDT), and N-ethyl-N-phenyldithiocarbamic acid zinc (EPDTC)] were identified to show inhibitory activity against SARS 3CLpro with Ki values of around micromolar and sub-micromolar levels (Fig. 32).69 Crystal structure analyses of these compounds as well as two additional zinc-based inhibitors complexed with SARS 3CLpro revealed the binding mode with the 3CLpro (PDB codes 2Z9J, 2Z9K, 2Z9L, 2Z9G, and 2Z94).39,70 Hg+-PMA was coordinated with Cys44, Met49, and Tyr54 with a square planar geometry at the S3 pocket, whereas each Zn2+ of the four zinc-inhibitors was tetrahedrally coordinated with the His41–Cys145 catalytic dyad. Although the electron density of the bulky substituent of the metal ion was not detected, computer modeling study showed the entire EPDTC could be accommodated in the active site pocket of the 3CLpro.
Screening of a compound library gives an opportunity to identify a novel scaffold or unexpected activity of old compounds, while experimental screening of real compounds generally provides clear-cut data on the potential activity and experimental base for rational structure optimization. A few examples of the SARS 3CLpro inhibitors found by virtual or high-throughput screening are listed in Fig. 33 and the screening processes are summarized in the following sections.
Cinanserin (Fig. 11), a well-characterized serotonin antagonist, was identified as an inhibitor of the SARS 3CLpro by virtual screening of a database.35 The substrate-binding pocket formed by residues within a radius of 6 Å around the catalytic dyad (His41 and Cys145) was used as the target site for virtual screening using a docking program. The comprehensive Medicinal Chemistry Database of Molecular Design Limited (MDL-CMC) containing the structure information of more than 8000 compounds was searched, and finally cinanserin was selected as a potential candidate compound. After the selection, it was confirmed that cinanserin can indeed inhibit the activity of SARS 3CLpro with an IC50 value of 5 μM. The binding affinity (KD=49.4 μM) of cinanserin was determined by kinetic analyses using surface plasmon resonance (SPR). The antiviral activity of cinanserin was further evaluated in tissue culture assays. The level of virus RNA and infectious particles was reduced by up to 4 log units, with IC50 values ranging from 19 to 34 μM.
In a study on the screening of small molecular inhibitors for SARS 3CLpro conducted after the above work, it was reported that cinanserin showed no inhibition of the SARS 3CLpro.71 This discrepancy was assumed to stem from the protease preparation used in each experiment: the N-terminal extended 3CLpro was used in the above initial screening work, whereas mature 3CLpro without the incorporation of additional sequences at the N-terminus was used in the latter small-molecule inhibitor screening experiment. Thus, cinanserin may be effective at inhibiting a non-dimeric form of the SARS 3CLpro.
In another study, structure-based virtual screening of 308 307 chemical compounds was performed using the computation tool Autodock 3.0.5 on a WISDOM (Wide In Silico Docking On Malaria) production environment.72 From the top 1,468 ranked compounds selected through the hydrogen bond interaction at the active site of the 3CLpro, 53 compounds were tested in an in vitro assay. Two potent 3CLpro inhibitors 81 and 82 (Fig. 33) were finally identified as competitive inhibitors of 3CLpro with Ki values of 9.11 and 9.93 μM, respectively.
Another validated docking protocol based on the Gold Docking Program was used for a virtual screening of 120 000 compounds to select 108 candidates to be tested in vitro.73 Two compounds, 83 and 84 (Fig. 33), were finally selected as promising inhibitors for SARS 3CLpro with IC50 values of 18 and 17 μM, respectively. Using the docking simulation carried out in the screening procedure, these two inhibitors were expected to show occupancy of the S1′, S1, and S2 pockets of the SARS 3CLpro, but not the S4 pocket.
A combination of virtual screening (VS) and high-throughput screening (HTS) techniques identified a novel, non-peptide small molecule inhibitor 85 (Fig. 33) of SARS 3CLpro.74 A structure-based VS approach integrating docking and pharmacophore based methods was used to screen 621 000 compounds from the ZINC library, a free database of commercially-available compounds for virtual screening. The screening protocol was validated using known 3CLpro inhibitors and was further optimized. Subsequently, a fluorescence-based enzymatic HTS assay was used to screen approximately 41 000 compounds chosen based on the VS results. After eliminating the false positives from the initial HTS hits by a secondary orthogonal analysis using SPR, the final candidate compound 85 was identified as a reversible small molecule inhibitor exhibiting mixed-type inhibition with an IC50 value of 13.9 μM and a Ki value of 11.1 μM.
In a study of Jacobs and co-workers71 on a search of noncovalent small-molecule inhibitors of the SARS 3CLpro, an initial high-throughput screening of the NIH molecular libraries sample collection (∼293 000 compounds) at the Scripps Research Institute Molecular Screening Center (SRIMSC) produced 406 hits. The second screen of these initial hits produced 136 active compounds, and the following dose-response testing yielded 44 active compounds with IC50 values below 10 μM. Among the several scaffold clusters obtained by structural clustering analysis of these active compounds, dipeptide compound 86 was selected for further structure-activity relationship studies (Fig. 34). Initial structure optimization, however, only produced one compound 87 with slightly weaker inhibitory activity than 86. X-ray crystal structure of the SARS 3CLpro complexed with compound 87, however, revealed that the binding orientation of 87 is overall similar to a known covalent peptide-mimetic inhibitor. Inhibitor 87 preferentially occupied the S3-S1′ pockets of the 3CLpro as the R enantiomer (Fig. 34). The catalytic Cys145 was positioned beneath both the amide carbonyl carbon and the furan oxygen at a distance of 3.5 Å. Based on these structural analyses, a second chemical library focusing on P1′ and P1 replacement was constructed and the inhibitory activities were evaluated. Following enantiomer separation of the selected active compound finally provided 87-R as a potent SARS 3CLpro inhibitor with an IC50 value of 1.5 μM (Fig. 34). The mechanism of inhibition was determined to be a competitive mode with a Ki value of 1.6 μM. Compound 87-R can effectively inhibit SARS CoV replication in cell culture with an EC50 value of 12.9 μM.
Further screening of the above NIH molecular library gave the related diamide compound 88 having an IC50 value of 6.2 μM (Fig. 35).75 X-ray crystal structure analysis of the mature SARS 3CLpro complexed with 88 revealed a unique induced-fit reorganization of the S2–S4 binding pockets. This induced fit accommodated the syn N-methyl pyrrole and anilido acetamide moieties of the inhibitor within the pockets that can be characterized as S2–S4 and S2–S1′ pockets, respectively. Additional SAR studies afforded compound 89 with a P3 truncated structure as a minimum pharmacophore as a noncovalent nanomolar inhibitor.
Few examples of structure-based rational design of inhibitors for SARS 3CLpro have been reported, and the inhibitory activities are still moderate. Nevertheless, the rational approach would largely contribute to designing a novel scaffold of a non-peptide low molecular inhibitor of the SARS/MERS 3CLpro which has a higher potential as a therapeutic agent than peptide-based inhibitors. In this section, two approaches for the design of non-peptide inhibitors starting from a potent peptide aldehyde inhibitor are included.
For the structure-based rational design of a SARS 3CLpro inhibitor, Konno et al.76 selected a serine-based scaffold as a suitable candidate for small-molecule inhibitors. A highly potent peptide-based inhibitor was selected as a starting derivative for the design. Side chain structures at the P1, P2, and P4 sites of tetrapeptide inhibitor 22 were used for the design of the serine derivative, since serine, a commercially available proteinogenic amino acid, has three variant reaction sites: an alcohol, an amino, and a carboxylic acid moiety, which can be orthogonally connected to various functional groups (Fig. 36).
Various molecular mechanics calculations with SPARTAN from Wavefunction and docking simulations of protein interactions by GOLD were carried out to determine whether a series of serine derivatives could adopt an energetically favorable conformation mimicking parent inhibitor 22. The initial trial however gave a result contrary to expectations: the cyclohexyl group of the serine derivative occupied the S1 pocket instead of the expected S2 pocket of the 3CLpro. In contrast, a cinnamoyl derivative reported by Bai et al. was located deep inside of the S1′, S1, and S2 pockets with appropriate cinnamoyl functionalities, which was identified via simulation using Autodock 3.0. Combining the contrary results of these simulations, a hybrid scaffold with Bai's derivative and the serine derivative was designed for the following SAR studies (Fig. 37). Detailed SAR studies at the P1′ and P4 positions provided two optimized compounds, 94 and 95, as novel scaffolds of small molecular inhibitors for the SARS 3CLpro with IC50 values of 30 and 65 μM, respectively. Docking simulations of these compounds confirmed the expected interactions at the S1′, S1, and S4 pockets (Fig. 38).
In a separate study,77 another approach starting from the same peptide inhibitor 22 was examined. Previous analyses of the crystal structure of the SARS 3CLpro complexed with inhibitor 22 (PDB code 3ATW) revealed that a cyclohexyl substituent at the P2 site was well packed in the corresponding S2 pocket of the 3CLpro, an interaction critical for making 22 a potent competitive inhibitor. Detailed analyses of this hydrophobic interaction at the S2 pockets revealed that the P2 site cyclohexyl structure was situated rather close to the peptide backbone in the active site cleft. The distance of the α-amide nitrogen of the P2 cyclohexylalanine (Cha) to the position 2 carbon (C2) of the cyclohexyl ring of Cha was estimated to be 3.48 Å in the crystal structure. Thus, connecting the C2 carbon of the cyclohexyl ring to an α-nitrogen of the P2 site Cha via a methylene linker is considered to yield a novel fused ring structure acting as a core hydrophobic substituent at the P2 position (Fig. 39). The resulting decahydroisoquinoline scaffold is expected to maintain the hydrophobic interactions at the cyclohexyl ring of 22 in the S2 pocket. It is also assumed that this fused ring scaffold will be able to arrange the P1 site imidazole and active site functional aldehyde at each required position. The acyl substituent on the nitrogen in the decahydroisoquinoline scaffold may add an extra position for additional interactions with the SARS 3CLpro.
Compounds containing different enantiomers of the fused ring structure were separately prepared to examine the effect of configurations at the fused ring scaffold. For these preparations, a combination of the enantiomer resolution by salt formation and Pd-catalyzed stereoselective cyclization reaction was used. As summarized in Table 8, the synthesized decahydroisoquinoline derivatives all showed moderate but clear inhibitory activities toward SARS 3CLpro, which strongly suggests that the decahydroisoquinoline can function as a hydrophobic core structure at the P2 site. Clear differences of the inhibitory activities due to the difference of stereostructure at the fused ring moiety were also observed. In addition, a limited range of differences of the inhibitory activity caused by the substituent on the nitrogen atom of the decahydroisoquinoline scaffold were also observed.
These differences of the inhibitory activities were rationalized by X-ray crystal structure analyses of the SARS 3CLpro complexed with decahydroisoquinoline derivatives, compounds 96, 101, and 97 (PDB code 4TWY, 4TWW, and 4WY3). The decahydroisoquinoline inhibitor 101 was at the active site cleft of the 3CLpro as observed in the parent peptide aldehyde inhibitor 22. The carbonyl carbon of the aldehyde group of 101 was detected at a distance of 2.31 Å from the thiol of Cys145, suggesting that the decahydroisoquinoline inhibitor would function as a competitive inhibitor to the parent peptide aldehyde inhibitor 22 (Fig. 40). It was also clearly confirmed that the decahydroisoquinoline scaffold of 101 took a trans-fused configuration and was inserted into the large S2 pocket of the 3CLpro. Most of the S2 pocket was occupied by the fused-ring structure. The nitrogen atom of the P1 site imidazole of 101 formed hydrogen bonds with nearly the same mode as the parent inhibitor 22, resulting in close fitting at the S1 pocket. Thus, the hydrophobic interaction of the decahydroisoquinoline scaffold in the S2 pocket functions to hold the P1 site imidazole and terminal aldehyde group inside the active site cleft.
Crystal structures of the 3CLpro complexed with 96 and 97 were compared to evaluate the effects of the decahydroisoquinoline scaffold configuration (Fig. 41). Compared to the interaction of the active compound 96, the decahydroisoquinoline scaffold having less active configuration (4aS, 8aR) in 97 was located in the S2 pocket in a more twisted position. This conformational change of the fused ring in the S2 pocket was transferred in the direction of the N-substituent of the fused ring scaffold, and the N-substituent of the less active compound 97 was directed toward the outside of the surface of the 3CLpro. In contrast, the substituent of the active compound 96 was located on the surface of the 3CLpro, where an additional interaction with the protease might be possible. These conformational differences at the N-substituent derived from the configurational change at the decahydroisoquinoline scaffold would explain the discrepancy in the inhibitory activity between 96 and 97.
Two procedures have been used in the development of potent inhibitors of SARS 3CLpro: (i) a combination of substrate-based peptide structures and effective warheads and (ii) a combination of in silico virtual screening and high-throughput screening of actual compound libraries. The data obtained through these approaches summarized in this article are the basis for the development of therapeutic agents for the infectious SARS coronavirus. Inhibitory potencies against the proliferation of the CoV in cells as well as the inhibitory effects towards the 3CLpro were confirmed for several candidate compounds including peptide-based inhibitors and small-molecule inhibitors. The low toxicity for cells was also examined for a few compounds; however, further in vivo studies should be conducted. These inhibitors are potential starting points for the design of inhibitors as they have high inhibitory potency against the CoV as well as good physicochemical and pharmacodynamics properties necessary for in vivo use. Recent studies on the rational design of a novel scaffold starting from a peptide-based inhibitor revealed a possibility to yield inhibitors based on another category. Compounds with good viral inhibitory activity can also be discovered by screening a library composed of approved drugs or therapeutics still in clinical development.
MERS 3CLpro has proteochemical properties different from SARS 3CLpro, especially regarding the stability of the dimer structure. This might interfere with the exact evaluation and design of specific inhibitors for MERS 3CLpro. Instead, broad-spectrum inhibitors might be promising since the sequence similarity of the 3CLpro of SARS and MERS is estimated to be more than 60%. Development of novel anti-SARS/MERS CoV inhibitors with drug-like properties can be attained based on the achievements summarized in this article.
© The Royal Society of Chemistry 2018 (2017)