Co-crystallization and structure determination: An effective direction for anti-SARS-CoV-2 drug discovery

Graphical abstract

Abstract

Safer and more-effective drugs are urgently needed to counter infections with the highly pathogenic SARS-CoV-2, cause of the COVID-19 pandemic. Identification of efficient inhibitors to treat and prevent SARS-CoV-2 infection is a predominant focus. Encouragingly, using X-ray crystal structures of therapeutically relevant drug targets (PL^pro, M^pro, RdRp, and S glycoprotein) offers a valuable direction for anti–SARS-CoV-2 drug discovery and lead optimization through direct visualization of interactions. Computational analyses based primarily on MMPBSA calculations have also been proposed for assessing the binding stability of biomolecular structures involving the ligand and receptor. In this study, we focused on state-of-the-art X-ray co-crystal structures of the abovementioned targets complexed with newly identified small-molecule inhibitors (natural products, FDA-approved drugs, candidate drugs, and their analogues) with the assistance of computational analyses to support the precision design and screening of anti–SARS-CoV-2 drugs.

1. Introduction

The unprecedented coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spread quickly across continents, with 4,180,161 deaths currently recorded on a global basis [1]. In view of the high virulence and mortality rates, the World Health Organization declared SARS-CoV-2 as the sixth public health emergency of international concern [2]. The pandemic is ongoing, and the effects of this highly contagious virus are widespread. Efforts are continuing to combat the SARS-CoV-2 outbreak with a variety of drugs, including remdesivir [3], dalbavancin [4], clofazimine [5], mycophenolic acid [6], plitidepsin [7], and Chinese herbal medicines such as Lianhuaqingwen capsules [8] and Qingfei Paidu decoction [9]. However, despite considerable effort, no effective countermeasures for controlling or eventually eradicating this dangerous virus have emerged [10], [11]. Rapid discovery and development of novel, efficacious, safe, and stable agents to treat COVID-19 is currently the focus of intense worldwide research.

The ongoing challenge to develop new antiviral agents that are highly specific and effective against SARS-CoV-2 has spurred the medical community to explore multiple techniques and strategies, including co-crystal structure-guided precision drug discovery [12], [13], [14], [15]. Since 2020, this strategy has gained traction in validating bioactive constituents that could serve as COVID-19 agents [16], [17]. The determination of X-ray crystal structures of therapeutically relevant drug targets in pharmaceutical research represents a potentially valuable direction for anti–SARS-CoV-2 drug discovery and lead optimization. SARS-CoV-2 possesses a positive-sense RNA genome, and its genome organization is depicted in Fig. 1. Several promising targets among SARS-CoV-2 enzymes and proteins have been identified and are of significant research interest (Fig. 1), including papain-like protease (PL^pro) [18], [19], main protease (M^pro) [20], [21], RNA-dependent RNA polymerase (RdRp) [22], [23], and the spike (S) glycoprotein, which binds to the angiotensin-converting enzyme 2 (ACE2)-binding domain [24], [25].

Fig. 1.

Organization of the SARS-CoV-2 genome and potential therapeutic targets. (A) Genome organization of SARS‐CoV‐2. (B) The structures of PL^pro (the Protein Data Bank (PDB) ID: 7BRO). (C) The structures of M^pro (PDB ID: 6Y2G). (D) The structures of RdRp (PDB ID: 7BW4). (E) The structures of S glycoprotein (PDB ID: 6YYT).

In particular, PL^pro plays an essential role in the cleavage and maturation of RNA virus polyproteins, dysregulation of the host inflammatory response, and disruption of host immune responses [26], [27]. M^pro, also called 3C-like protease (3CL^pro), plays a crucial role in the polyprotein maturation process and may serve as a prime therapeutic target [28], [29], [30]. RdRp is another essential enzyme that regulates viral replication by catalyzing the RNA template–dependent development of phosphodiester bonds [31]. ACE2 is a host cell surface receptor necessary for SARS-CoV-2 infection [32]. The S protein plays a critical role in mediating membrane fusion and cell entry via interactions with the ACE2 cell surface receptor [33]. The immense therapeutic potential of targeting PL^pro, M^pro, RdRp, and the S protein in the development of effective drugs has been widely explored.

Innovative drug development constitutes an important research direction, and results to date have been outstanding, with several interesting articles already published from the perspectives of potential drugs and drug targets [34], [35], [36], [37], [38], mechanisms of action and inhibition [26], [39], [40], structure-based drug design [41], [42], [43]. Other articles were identified by virtual screening of the ZINC, TCMSP, ChEMBL, Pubchem, NPASS, and Drug bank databases [44], [45], [46], [47], [48]. However, to our knowledge, no systematic reviews of inhibitor-target co-crystal complexes of SARS-CoV-2 have been published. In the current review, we focused on medicinal chemistry efforts based on analyses of X-ray co-crystal structures of the abovementioned targets in complex with newly identified small-molecule inhibitors as a means of supporting the precision design and screening of anti–COVID-19 agents through the characterization of ligand-bound targets. COVID-19 inhibitors of significance from the perspective of medicinal chemistry are prefaced by a brief introduction summarizing the relevant pharmacology, discovery process (drug design), preparation methods, and mechanisms of action. The main objective of this report is to provide a ‘co-crystal structure-guided drug discovery protocol’ to facilitate the development of safe and effective drugs to combat COVID-19.

2. Natural products and analogues as SARS-CoV-2 inhibitors

Natural products provide a rich and novel source for drug development and play an integral role as starting points in the development of therapeutic strategies for treating various complex diseases due to their tremendous structural diversity and unique chemical compositions [49]. In the current race to identify safe and efficacious drugs for COVID-19, natural products with a range of valuable bioactivities have attracted significant attention in terms of evaluating and validating their therapeutic effects [45], [46], [47]. As a result of continuing research, several natural products have been highlighted thus far as promising drug leads to combat COVID-19 [50], [51], [52]. As a result of continuing research, several natural products have been highlighted thus far as promising drug leads to combat COVID-19 [53], [54], [55], [56], [57], [58].

Scutellariae radix (Scutellaria baicalensis) is a well-known herbal medicine used to treat lung injury [59] and inflammatory diseases [60]. It is a widely used nutritional supplement with an impressive safety profile. Liu and co-workers recently demonstrated that crude scutellariae radix extract inhibits SARS-CoV-2 replication (half-maximal effective concentration [EC₅₀] of 0.74 μg/mL) with almost no toxicity (selectivity index [SI] > 675.7) against Vero E6 cells [61]. Baicalein is the major component of scutellariae radix and exhibits broad-spectrum antiviral effects (Fig. 2A) [62], [63]. As expected, baicalein was shown to be effective in inhibiting M^pro activity (half-maximal inhibitory concentration [IC₅₀] of 0.39 μM) [61]. Du and co-workers reported that baicalein can be used to treat acute lung injury, as it improves respiratory function, inhibits inflammatory cell infiltration, and downregulates the release of cytokines [64]. A docking model showed that the 6-OH group of baicalein plays a pivotal role in terminating RNA replication via hydrogen bonding interactions with L141 [61]. To elucidate the underlying molecular mechanisms of baicalein, active sites were explored. Su and co-workers generated a crystal structure of baicalein-M^pro at a resolution of 2.2 Å (PDB ID: 6M2N) [65]. Further analysis revealed that the non-peptidomimetic inhibitor baicalein was ensconced in the core of the substrate-binding pocket via multiple interactions. For example, hydroxyl groups of baicalein form hydrogen bond interactions with Leu141/Gly143 and Ser144/His163 of M^pro to stabilize the tetrahedral transition state of the proteolytic reaction. The enone carbonyl group of baicalein forms hydrogen bond interactions with Glu166, and the free aromatic rings of baicalein form π-π interactions with Cys145 and His41 as well as hydrophobic interactions with Gln189, Arg188, Met49, Cys44, and His41 to prevent substrate access to the active site, rather than covalently blocking catalytic Cys145. These interactions could underlie the observed potent activity (IC₅₀ of 0.94 μM, EC₅₀ of 2.94 μM, and SI > 212) of baicalein against SARS-CoV-2 M^pro [65]. Shuanghuanglian oral liquids extracted from three types of traditional Chinese medicines (including Scutellaria baicalensis Georgi) effectively blocked the replication of SARS-CoV-2 in Vero E6 cells (IC₅₀ of 0.064 μM), highlighting the significant contributory role of this active component in the biological activity of Chinese herbal treatments [65]. Importantly, the crystal structure of the M^pro-baicalein complex provided direct data that could facilitate elucidation of the molecular mechanisms underlying the beneficial effects of traditional medicines.

Fig. 2.

Natural products as SARS-CoV-2 inhibitors. (A) Chemical structures of baicalein and X-ray crystal structure of baicalein with SARS-CoV-2 M^pro (PDB ID: 6M2N) [68]. (B) Chemical structures of myricetin and X-ray crystal structure of myricetin with SARS-CoV-2 M^pro (PDB ID: 7B3E) [68]. (C) Chemical structures of shikonin and X-ray crystal structure of shikonin with SARS-CoV-2 M^pro (PDB ID: 7CA8) [74]. (D) Chemical structures of sinefungin and X-ray crystal structure of sinefungin with SARS-CoV-2 Nsp10-Nsp16 (PDB ID: 6YZ1) [82]. (E) Chemical structures of linoleic acid and cryo-electron microscopy structure of linoleic acid with SARS-CoV-2 S protein (EMD ID: 11145) [84].

Myricetin, a flavonol monomer, is a natural product isolated from the “medicine food homology” species Chinese bayberry (Myrica rubra Sieb. et Zucc.) and exhibits a range of pharmacological effects against acute lung injury [66], inflammatory diseases, and infections with pathogenic microbes [67]. Kuzikov and co-workers reported that myricetin inhibits the synthesis of SARS-CoV-2 M^pro in vitro (IC₅₀ of 0.22 μM) [68]. To optimize precision drug design, Su and co-workers generated a crystal structure of the M^pro-myricetin complex at a resolution of 2.1 Å (PDB ID: 7DPP) [68] (Fig. 2B), which unambiguously revealed that myricetin is covalently bound to catalytic Cys145. Su et al. also showed unprecedented inhibitor-target binding patterns, specifically, a covalent bond between the Cys145 sulfur and the 2′ position of myricetin, which was distinct from baicalein and other reported flavonoid structures [68]. In addition, Kuzikov and co-workers reported another crystal structure of the M^pro-myricetin complex with the same covalent binding mode, at 1.77-Å resolution (PDB ID: 7B3E) [69]. Further analysis of the X-ray structure of the myricetin-M^pro complex revealed that the binding pocket is only partially occupied by myricetin, presenting an important signal and ideal lead for structure-based drug design. Based on molecular dynamics (MD) simulations and molecular mechanics Poisson-Boltzmann surface area (MMPBSA) calculations, Sharma et al. [70] revealed that baicalein forms a stable combination with non-structural protein 15 (Nsp15) with low binding energy (−65.663 kJ/mol).

The natural naphthoquinone shikonin, derived from Lithospermum erythrorhizon Sieb. et Zucc (Fig. 2C), has numerous pharmacological properties that support its further development as a therapeutic agent [71]. This agent has been shown to exert anti-inflammatory, anti-fungal, and anti-HIV effects [72]. Yang et al. revealed that shikonin inhibits the replication of SARS-CoV-2 M^pro in vitro (EC₅₀ of 15.75 μM) [73]. Li and co-workers [74] initially solved the crystal structure of the shikonin-M^pro complex at 2.45 Å (Fig. 2C). Crystallography data showed that the His41-Cys145 catalytic dyad undergoes large conformational changes upon complex formation, leading to significant differences with other reported structures [75]. Li and co-workers recently revealed that two novel hydrogen-bonding interactions (one with Gln189 and Thr190, another with Met165, His164, and Cys145) and a π-π stacking interaction with His41 play critical roles in the binding of shikonin with M^pro [74]. These findings clearly indicate a distinct binding pattern suggestive of binding site diversity. However, Ma and co-workers recently reported that shikonin may not be a target-specific M^pro inhibitor, as its activity declines markedly in the presence of reducing reagents [76]. Further elucidation of the different binding modes should provide a solid foundation for effective COVID-19 drug design.

The genome of SARS-CoV-2 contains ~ 29 800 bases encoding 16 nonstructural proteins (Nsp1-Nsp16) essential for viral replication [77]. For example, Nsp16 plays a critical role in immune evasion during virus replication [78]. The natural nucleoside antibiotic sinefungin is a structural analogue of S-adenosylmethionine (SAM), which serves as a key methyl group donor to numerous biomolecules [79]. SAM-dependent 2′-O-RNA methyltransferase (MTase) is a potential relevant drug target in COVID-19 chemotherapy [80]. The function of MTase is associated with Nsp16, which requires Nsp10 as a cofactor for activity [80], [81]. Mahalapbutr et al. [80] also reported that sinefungin exhibits very high susceptibility to Nsp16 (especially through electrostatic interactions with the 2′-OH and N3 of the RNA’s adenosine moiety) based on atomistic MD simulations and free-energy calculations. To establish the mechanisms by which sinefungin inhibits the Nsp16 MTase at a molecular level, Krafcikova and co-workers [82] determined the crystal structure of SARS-CoV-2 Nsp10-Nsp16 complexed with sinefungin at a resolution of 2.4 Å (PDB ID: 6YZ1) (Fig. 2D). The co-crystallization structure showed that sinefungin binds the SAM-binding pocket localized in a canyon within Nsp16 and forms several interactions with specific residues (nucleoside-binding pocket: Asp99, Asn101, and Asp114; methionine-binding pocket: Asn43, Asp130, and Lys170). These data provide an important starting point for structure-based inhibitor discovery.

Linoleic acid, an essential free fatty acid, is an important modulator of the inflammatory response [83]. Toelzer and co-workers [84] recently demonstrated that a combination of linoleic acid (50 μM) and the RdRp inhibitor remdesivir (20 to 200 nM) exert a synergistic inhibitory effect in human Caco-2 ACE2 + cells in vitro. To clarify the underlying inhibitory mechanism of action of linoleic acid, the cryo-electron microscopic (cryo-EM) structure of S protein complexed with linoleic acid was determined at 2.85-Å resolution (Electron Microscopy Data [EMD] ID: 11145) (Fig. 2E) [84]. Further analysis of the linoleic acid binding pocket within the S protein revealed that the hydrocarbon tail of linoleic acid binds to hydrophobic amino acids, whereas the acidic head group interacts with a positively charged anchor (Arg408 and Gln409) to irreversibly lock the S protein. The S protein hydrophobic pocket with a tube-like shape fits well with linoleic acid, resulting in reduced ACE2 interactions, thus setting the stage for an intervention strategy based on linoleic acid binding to the S protein. Further studies are warranted to establish whether linoleic acid exerts anti–COVID-19 effects in vivo. Note that dynamic-nonequilibrium MD studies showed that the linoleic acid site forms a stable combination with the functional regions of the S protein, but variation in the S protein could be exploited to tune a targeted allosteric response [85].

Calpain inhibitor II, developed by the Cerro group [78], has shown potential utility against SARS-CoV-2 infection [87]. Initial research on the synthesis of calpain inhibitor II commenced with structural modification of leupeptin (a naturally occurring and relatively non-specific inhibitor of thiol proteases) by replacing the guanidine group with a methyl group to create calpain inhibitor I (Fig. 3), which exhibited more-potent and selective blockade of calpain activity [86]. An even more important contributor to antiviral drug design is the 4′-S substituted calpain inhibitor II, which exhibited more-potent blockade of calpain activity [86]. Calpain inhibitor II, a broad-spectrum antiviral agent, can be used to treat infections with multiple viruses, such as endemic human coronavirus 229E (HCoV-229E) (EC₅₀ of 0.08 μM) and human coronavirus OC43 (EC₅₀ of 1.82 μM) [87]. As for SARS-CoV-2, Wang et al. [88] confirmed that calpain inhibitor II is highly effective against M^pro (IC₅₀ of 0.97 μM).

Fig. 3.

Leupeptin and its analogues as inhibitors of SARS-CoV-2 and the reaction mechanism [89].

Wang and co-workers [89] also reported the crystal structure of calpain inhibitor II–M^pro at 1.65-Å resolution (PDB ID: 6XA4). Further analysis of the complex structure revealed that stabilized thiohemiketal with the (S) configuration occupies the oxyanion hole, which is formed by Gly143, Cys145, and, in part, Ser144 (Fig. 3). Additionally, the structure exhibited several other interactions with active site residues, including multiple hydrogen bonds of the amide group with His164, Met165, and Glu166 (Fig. 3). Interestingly, a weak hydrogen bond between the sulfur group and His163 was detected, which could explain the stronger activity of calpain inhibitor II (IC₅₀ = 0.97 μM) relative to calpain inhibitor I (IC₅₀ = 8.60 μM) [89].

The introduction of X-ray crystallography technology has provided a highly successful strategy for exploration of natural products and their analogues, improving our understanding of drug interactions at the molecular level based on structures of complexes, to facilitate the design of effective COVID-19 therapies. While natural products have demonstrated potential effectiveness, clinical trial evidence of their utility as anti–COVID-19 agents is currently lacking. Future research will therefore focus on obtaining stronger evidence to support the clinical utility of natural products. In this scenario, strategies for the large-scale manufacture of natural products and analogues, such as shikonin, sinefungin, and calpain inhibitor II, are urgently required.

3. FDA-approved drugs as SARS-CoV-2 inhibitors

Remdesivir, an RdRp inhibitor, was the first and only drug against COVID-19 approved by the U.S. Food and Drug Administration (FDA) in 2020 [90], [91], [92], [93]. The highly potent 1′-CN-substituted GS-441524, inspired by the natural product tubercidin, is important in the design of anti–SARS-CoV-2 agents [94]. Triphosphate GS-443902 is established as the active form of GS-441524. However, monophosphate conversion of GS-441524 to GS-441524 monophosphate is extremely difficult [95], highlighting the necessity of developing a prodrug that can overcome this delivery bottleneck. To achieve better cellular uptake in vivo, monophosphorylated remdesivir was developed by Gilead Sciences (Fig. 4A) [96]. The GS-5734 1′–CN group plays a pivotal role in terminating RNA replication by disrupting the exoribonuclease proofreading activity and inhibiting the RdRp polymerization activity [97].

Fig. 4.

Design, mechanism of action, and synthesis of the covalent RdRp inhibitor remdesivir [101].

To increase the efficiency of remdesivir, researchers at Gilead Science initially developed strategies for preparation of the P-stereogenic prodrug using a scalable process through chiral preparative high-performance liquid chromatography (HPLC) or chiral resolution [96], which inevitably led to waste of resources. For further improvement of efficiency, an elegant asymmetric strategy (Fig. 4B) for producing remdesivir via chiral bicyclic imidazole-catalyzed dynamic kinetic asymmetric transformation (DyKAT) with excellent stereoselectivity (S_P:R_P = 22:1) was recently developed by Zhang and colleagues [98]. In the clinical context, development of a large-scale synthesis method for remdesivir is urgently needed. As remdesivir is a phosphoramidate prodrug, conversion into the triphosphate GS-443902 is required. Similar to many prodrugs, the postulated activation pathway of remdesivir is divided into four sequential steps (Fig. 5): cell entry (via increased hydrophobicity with the aid of a phosphorylated group), removal of the masking group (via enzymatic and chemical demasking), phosphorylation (via NMP kinases), and incorporation into the growing SARS-CoV-2 RNA strand. In cells, the triphosphate form, GS-443902, can block SARS-CoV-2 replication by evading the “proofreading” of viral RNA sequences [99].

Fig. 5.

Bioconversion of the covalent RdRp inhibitor remdesivir.

Wakchaure et al. [100] showed that remdesivir has high binding energy (−29.7 kCal/mol) with respect to RdRp, and MD simulations also revealed that remdesivir binds in the catalytic site via the formation of hydrogen bond interactions. Xu et al. [101] reported the cryo-EM structure of RdRp-remdesivir (using its triphosphate metabolite GS-443902) at 2.5-Å resolution (PDB ID: 7BV2). The cryo-EM structure unambiguously demonstrated that remdesivir monophosphate (IV), positioned at the center of the catalytic site of the primer RNA, covalently binds to the 1 + position of the template strand, thus terminating chain elongation (Fig. 4A). Three strong H-bonds with active site residues (ribose –OH groups: Asp623, Ser682, and Asn691; sugar 2′–OH: Asn691) were identified (Fig. 4A). Notably, the implicit, cryptic, and allosteric binding sites provide a versatile platform for designing high-potency inhibitors with kinetic selectivity [102], [103], [104]. Based on thermodynamic profiling, Srivastava et al. found that triphosphate GS-443902 inactivates RdRp by not only interfering with binding in the initial catalytic site but also by significantly blocking the nucleoside 5′-triphosphate entrance site [105]. Based on MMPBSA calculations, Khan et al. [106] proposed that remdesivir inhibits SARS-CoV-2 via a multi-target mechanism, as it binds to M^pro (−7.8 kCal/mol), membrane protein (−7.4 kCal/mol), and RDRP (−7.1 kCal/mol).

Remdesivir exhibits broad-spectrum activity against multiple viruses in vitro, such as SARS-CoV, Middle East respiratory syndrome coronavirus (MERS-CoV), respiratory syncytial virus, Ebola virus (EBOV), and HCoV-229E, with EC₅₀ values of 0.069 μM, 0.090 μM, 0.021 μM, 0.012 μM, and 0.024 μM, respectively [107], [108], [109], [110], [111]. Moreover, remdesivir reportedly protects rhesus monkeys from MERS-CoV infection [112] and displays clinical safety and efficacy against EBOV infection [113]. Although remdesivir does not appear to be highly effective in patients with COVID-19 [114], it has the obvious advantage of rapid clinical translation to a rational prodrug strategy against SARS-CoV-2. In addition, multiple lines of evidence suggest that combinations of remdesivir and other agents (such as baricitinib [115] and dexamethasone [116]) exert synergistic therapeutic effects against COVID-19.

Drug repurposing, a time-saving process, is a viable strategy to quickly, safely, and successful discover clinically approved drugs for COVID-19 treatment [117], [118], [119], [120]. Favipiravir, a promising repurposed prodrug, was identified by Toyama Chemical Co., Ltd., and approved in Japan in 2014 for the treatment of influenza virus infections [121]. To date, widespread research attention has focused on the utility of favipiravir as a specific anti–COVID-19 agent [122]. Favipiravir is an orally bioavailable prodrug that exhibits potent activity against COVID-19 (EC₅₀ of 61.88 μM) with moderate selectivity (therapeutic window > 6.46) [123]. Favipiravir was recently shown to protect Syrian hamsters from SARS-CoV-2 infection, with a strong dose effect [124]. Abdelnabi and co-workers reported that a combination of favipiravir and molnupiravir exerted a marked synergistic inhibitory effect in a SARS-CoV2 hamster infection model [125]. Several clinical trials have highlighted the efficacy of favipiravir against COVID-19 [126], [127], [128], [129]. For example, Cai and co-workers reported that favipiravir significantly improved recovery rate (91.43%) and shortened viral clearance time (4 days), with fewer adverse events (11.4%) in patients with mild to moderate COVID-19 (N = 80) [129].

To elucidate the mechanisms underlying SARS-CoV-2 polymerase activity, Naydenova and co-workers examined the structure of SARS-CoV-2 RdRp in complex with favipiravir (using its triphosphate metabolite favipiravir–ribonucleoside triphosphate [RTP]) at 2.5-Å resolution via cryo-EM (PDB ID: 7AAP; EMD-11692) (Fig. 6) [130]. The cryo-EM structure unambiguously showed that favipiravir-RTP pairs with the 1 + nucleotide of the template strand via noncovalent interactions (side-chains and two catalytic Mg²⁺ ions) to terminate chain extension (Fig. 6). Peng and colleagues [131] reported an alternative structural snapshot (PDB ID: 7CTT; EMD ID: 30469) distorted in one direction, suggesting another productive conformation at the catalytic site. These distinct cryo-EM states may account for the slow, weak, and inefficient incorporation of the inhibitor into the RNA primer strand, leading to the high EC₅₀ (61.88 μM).

Fig. 6.

Bioconversion of the non-covalent RdRp inhibitor favipiravir [130].

Similar to the prodrug remdesvir, favipiravir must be converted into the active form, favipiravir-RTP. The activation mechanism of favipiravir involves four sequential steps (Fig. 6): cell entry, phosphoribosylation, phosphorylation, and non-covalent incorporation into the growing SARS-CoV-2 RNA strand [132]. Notably, favipiravir-RTP binds non-covalently to the polymerase in a manner distinct from that of remdesivir-RTP. Furthermore, Celik et al. [133] used MD simulations to highlight the binding site differences of favipiravir and its active triphosphate metabolites (favipiravir-RTP) with RdRp. Their results demonstrated that favipiravir-RTP forms a more-stable complex than favipiravir. This finding indicates an unusual binding mode, which provides an important basis for non-covalent RdRp inhibitor discovery.

Free-energy perturbation (FEP) calculations based on molecular simulations and statistical mechanics represents a promising approach to guide structural modifications in drug discovery research [134], [135], [136]. Based on this method, Jorgensen’s group [137] recently showed that the non-covalent SARS-CoV-2 M^pro inhibitor perampanel analogue 5 effectively treats SARS-CoV-2 infection by inhibiting M^pro activity (IC₅₀ of 0.14 μM) and inducing significant suppression of viral replication in Vero E6 cells (EC₅₀ of 1.5 μM), with moderate cytotoxicity (SI = 14.7). Perampanel, a poorly effective inhibitor of M^pro with an IC₅₀ of 100–250 μM, was an initial hit compound selected for redesign and further optimization based on its relatively simple structure and FEP modeling studies [138]. As shown in Fig. 7A, perampanel analogue 2 was identified as a highly potent target via initial docking analyses using FEP calculations [138], [139].

Fig. 7.

Structure-based design and synthesis of perampanel analogues 2, 4, and 5 [137], [139].

As expected, the anti–SARS-CoV-2 activity of perampanel analogue 2 was markedly improved, with an IC₅₀ of 9.99 μM, which was further enhanced by a factor of ~ 2 upon addition of a second chlorine atom to perampanel analogue 2 to generate perampanel analogue 4. Jorgensen et al. [137] determined the crystal structure of a complex of SARS-CoV-2 M^pro and perampanel analogue 4 at a resolution of 1.6 Å (PDB ID: 7L10), which unambiguously revealed that the chlorophenyl edge packs well against the imidazole ring of His41 and that the other meta-chlorine near Gln189 should be replaced with an alkyl or alkoxy group. These data provided valuable guidance for the synthesis of propoxy perampanel analogue 5, leading to a notable improvement in the IC₅₀ value from 4.02 μM to 0.14 μM for perampanel analogue 4. Examination of the crystal structure of M^pro in complex with perampanel analogue 5 (1.8 Å resolution, PDB ID: 7L11) revealed that the terminal methyl group extended into the hydrophobic region at the juncture of Met165, Leu167, and Pro168. In addition, a combination of perampanel analogue 5 and remdesivir exhibited a significant therapeutic effect against COVID-19, with a synergy volume/antagonism volume ratio of 30.8/0 μM² %. Jorgensen et al. [137] additionally explored several structurally novel and non-peptidic inhibitors with IC₅₀ values in the 20 nM range identified from FEP calculations and crystal structures.

To increase the efficiency of target perampanel analogues 2, 4, and 5, Jorgensen et al. [137] further developed a general route for synthesis, as shown in Fig. 7C. Briefly, Suzuki cross-coupling of 5-bromo-2-methoxypyridine with (2-cyanophenyl) boronic acid yielded S1, followed by deprotection to produce intermediate S2, which was subjected to Chan-Lam coupling to yield intermediate S3. Bromination of S3 yielded key intermediate S4, which was subjected to Suzuki cross-coupling with aryl boronic acids to generate target perampanel analogues 2, 4, and 5. Notably, upscaling of the total synthesis strategy remains an urgent requirement.

4. Candidate drugs and analogues as SARS-CoV-2 inhibitors

A wide range of drug candidates may be effective against COVID-19 via broad-spectrum effects, both in vitro and in vivo [140], [141]. These candidate drugs provide a basis to further explore and develop SARS-CoV-2 inhibitors, which could reduce the period of treatment and minimize costs. The peptidomimetic α-ketoamide inhibitor 11r, an antiviral agent, can be used to inhibit multiple viruses, such as SARS-CoV (EC₅₀ of 1.4 μM), enterovirus A71 (EC₅₀ of 0.8–0.9 μM), coxsackievirus B3 (EC₅₀ of 0.45 μM), and MERS-CoV (EC₅₀ of 0.0004 μM) [142]. Hilgenfeld et al. showed that 11r, a specific M^pro inhibitor, is effective against SARS-CoV-2 infection (EC₅₀ of 0.18 μM), thus presenting a lead for further development [20]. To improve its plasma half-life, lead 11r was modified by hiding the P3-P2 amide bond within a pyridone ring. As expected, the plasma half-life in mice increased from 18 min to 1 h by preventing the cellular protease–mediated cleavage of the amide bond. In addition, to enhance solubility and antiviral activity, compound 13b (EC₅₀ of 0.67 μM) was produced by replacing the hydrophobic cinnamyl and cyclohexyl moieties with a low-hydrophobic Boc group and smaller cyclopropyl group, respectively. The researchers reported total synthesis of compound 13r on the ~ 50-mg scale [20]. The key components of the strategy were nucleophilic substitution (SN₂) and nucleophilic addition of isocyanides to aldehydes (Fig. 8).

Fig. 8.

Structure-based design and synthesis of the peptidomimetic inhibitor 13b [20].

Notably, compound 13b exhibited significantly improved plasma protein binding (reduced from the 99% of 11r to 90%) and plasma half-life in mice (increased from the 18 min of 11r to 1.8 h). Compound 13b exhibited favorable pharmacokinetic properties in mice, including good tropism to the lungs after subcutaneous administration, suitability for direct administration via inhalation, long-term efficacy, and good tolerability without adverse reactions. The hydrophobic Boc group appeared to play an essential role in crossing of the cell membrane and binding to viral M^pro, without which the compound was almost inactive [20]. The above studies offer a platform for research into more-effective pyridone-containing anti-SARS-CoV-2 drugs to suppress disease progression and support the importance of structure-based approaches in the design and optimization of SARS-CoV-2 inhibitors. Rungrotmongkol et al. [143] reported that compound 13b was a suitable template for structural modifications at the P1′ and P4 sites to enhance interactions.

The Hilgenfeld group [20] determined the crystal structure of 13b-M^pro at 1.95-Å resolution (PDB ID: 6Y2F), which unambiguously revealed the presence of a stabilized thiohemiketal generated via nucleophilic attack of Cys145 over the α-carbonyl of the ketoamide warhead. Moreover, several interactions with active site residues (such as the amide oxygen of 13b forming hydrogen bond interactions with Gly143, Cys145, and partly Ser144 to form the “oxyanion hole”; the α-keto group of 13b forming a covalent bond with catalytic Gly145 to yield the thiohemiketal group) were identified. Apparently, there is a large space (distance > 3.6 Å) between Thr190, Gln189, and the pyridone ring of compound 13b, suggesting that a group larger than the pyridone ring could be designed based on the P3 moieties. In addition, a small model of the thiol moiety (active center of M^pro) with inhibitor 13b was generated (Fig. 9) [21]. In this model, the C-S bond can be formed directly (without deprotonation of the thiol group), and water molecules play an important role in the kinetics, as the energy barrier decreases by 9.0 kCal/mol for the intermediate when assisted with an explicit water molecule, which was facilitated via a 6-membered transition state instead of a constrained 4-membered state. Furthermore, Kumari et al. [144] used MD simulations to highlight the binding site differences of 13b with monomeric and dimeric M^pro. Their results showed that dimeric M^pro exhibits higher affinity for 13b via residue interactions between S1 and F140, E166, and H172.

Fig. 9.

Transition states of direct C-S bond formation between the inhibitor 13b and the thiol moiety.

Chirality is one of the most crucial attributes for a vast majority of pharmaceutical compounds in the natural world and chiral bioactive substances play a key role in disease control [145]. Chiral agents have attracted significant research attention due to enantiomers often possessing differential binding affinities for SARS-CoV-2 enzymes and proteins.

Kitamura and co-workers [146] recently demonstrated that the non-covalent SARS-CoV-2 M^pro inhibitor 23R can be used to treat the virus infection. The compound inhibited M^pro (IC₅₀ of 0.31 μM) and exerted significant suppressive effects in Vero E6 cells (EC₅₀ of 1.27 μM, SI > 78.7). Initial research on 23R began with structural modification of the SARS-CoV M^pro agent ML188 (R), which was separated using chiral stationary-phase supercritical fluid chromatography from ML188 enantiomers (S/R). In 2013, Jacobs and co-workers [147] demonstrated that the lead compound ML188 (S/R) and its single stereoisomer ML188 (R) effectively inhibited replication of SARS-CoV M^pro with IC₅₀ values of 4.8 μM and 1.5 μM, respectively. However, the single stereoisomer ML188 (S) was inactive and had no inhibitory effect on SARS-CoV replication. The X-ray crystal structure of the M^pro-ML188 (R) complex (PDB ID: 3V3M) revealed that an optimal shape complementary to SARS-CoV was formed through several key interactions (Fig. 10A) [147]. Among these, the 3-pyridyl group occupied the S₁-subpocket via a critical hydrogen bond with the His163 side chain. Knowledge of the structure of this non-covalent SARS-CoV M^pro inhibitor could further guide effective inhibitor design.

Fig. 10.

Structure-based design and synthesis of the non-covalent SARS-CoV-2 M^pro inhibitor 23R [146], [147].

Based on the high-affinity binding (between the hydrogen bond and His-163 side chain) strategy, 39 compounds (a mixture of enantiomers or diastereomers) were prepared using the Ugi four-component reaction (Ugi-4CR), which provides an expeditious strategy for lead optimization [146]. Among these compounds, 23 (S/R) significantly inhibited viral replication, with an IC₅₀ of 0.66 μM and low toxicity (SI > 303). This elegant process of 23 (S/R) synthesis is depicted in Fig. 10B. After separation of the single stereoisomers, 23R and 23S, via reverse-phase HPLC, IC₅₀ values were determined as 0.31 μM for the diastereomer 23R and 5.61 μM for the less-active 23S diastereomer [146]. To further elucidate the binding forces, Kitamura and co-workers [146] generated a crystal structure of 23R complexed with M^pro at 2.6 Å resolution (PDB ID: 7KX5). The P1 pyridinyl ring occupied the S1 pocket, forming a close (2.9 Å) hydrogen bond with the His163 side chain. Another point of particular interest is the location of the terminal α-methylbenzyl group between the S2 and S4 sites, which represents a previously unreported mechanism of binding pocket formation. The S4 pocket remained mainly unoccupied by 23R, indicating the presence of sufficient space that could be explored for further drug development. The COVID-19 pandemic has highlighted the urgent need for rapid structure-based drug discovery and development, and this strategy achieved expedited design of a non-covalent M^pro inhibitor through coupling of co-crystal structures, structure-based design, and Ugi-4CR methodologies.

5. Other active compounds as SARS-CoV-2 inhibitors

Numerous small-molecule inhibitors have shown promising results as efficient therapeutic agents for SARS-CoV-2. Earlier, Hoffman and co-workers [148] at Pfizer Worldwide Research and Development highlighted the efficacy of the novel M^pro inhibitor, PF-00835231, in treatment of infections. Potent activity of this compound against hCoV 229E (EC₅₀ of 0.090 μM), SARS-CoV (EC₅₀ of 4.8 μM) and SARS-CoV-2 (EC₅₀ of 0.13 μM) with no observable cytotoxicity was reported. Examination of the X-ray crystal structure of the M^pro-PF-00835231 complex (PDB ID: 6XHM) revealed that the stabilized tetrahedral carbinol complex is generated via nucleophilic attack of Cys145 over the carbonyl carbon of the hydroxymethylketone warhead [148]. Furthermore, compound PF-00835231 exhibited synergistic activity against COVID-19 in combination with remdesivir in HeLa-ACE2 cells. Preclinical experiments confirmed that the absorption, distribution, metabolism, and excretion profile as well as safety and activity profiles of PF-00835231 were suitable to warrant further development as a potent M^pro inhibitor. Clinical trials of PF-00835231 have been registered (NCT04627532 and NCT04535167) for in-depth studies of the compound’s anti–SARS-CoV-2 activity. Additionally, the Dittmann group [149] reported that statistically, PF-00835231 is a more-potent inhibitor than remdesivir. In addition to the abovementioned active compounds, several other small-molecule inhibitors with known crystal structures have been shown to exhibit significant anti–SARS-CoV-2 activity (Table 1). Further research is required to establish the clinical safety and efficacy of these inhibitor candidates, however.

Table 1.

Other small-molecule SARS-CoV-2 inhibitors with known crystal structures.

No.	Name	Structure	EC50 or IC50 (μM)	Target	PDB ID	Refs
1	Boceprevir		4.13	Mpro	6XQU	[150], [151], [152]
2	Calpain inhibitor XII		0.45	Mpro	6XFN	[89], [153], [154]
3	Calpeptin		1.56	Mpro	7AKU	[155]
4	Carmofur		1.82	Mpro	7BUY	[156], [157]
5	Ebselen		4.67	Mpro	7BAK	[158], [159]
6	GC-373		0.40	Mpro	6WTK	[160]
7	GC-376		0.030	Mpro	6WTJ	[89], [160], [161]
8	GRL0617		2.20	PLpro	7CMD	[19], [162], [163], [164]
9	GS-441524		0.70	RdRp	7BF6	[165], [90]
10	Isofloxythepin		4.80	Mpro	7AY7	[155]
11	Masitinib		2.50	PLpro	7JU7	[166], [167]
12	MI-23		0.0076	Mpro	7D3I	[168]
13	ML188		2.5	Mpro	7L0D	[146], [169]
14	MPI1		0.100	Mpro	7JPZ	[170]
15	MPI3		0.0085	Mpro	7JQ0	[170]
16	MPI4		0.015	Mpro	7JQ1	[170]
17	MPI5		0.033	Mpro	7JQ2	[170]
18	MPI6		0.060	Mpro	7JQ3	[170]
19	MPI7		0.047	Mpro	7JQ4	[170]
20	MPI8		0.105	Mpro	7JQ5	[170]
21	Narlaprevir		5.10	Mpro	6XQT	[150], [152]
22	N3		16.77	Mpro	6LU7	[73], [171]
23	Pelitinib		1.25	Mpro	7AXM	[155]
24	PF-00835231		0.13	Mpro	6XHM	[172]
25	Suramin		0.26	RdRp	7D4F	[173], [174]
26	Telaprevir		18.0	Mpro	6XQS	[150]
27	UAWJ246		0.045	Mpro	6XBG	[89]
28	UAWJ247		0.042	Mpro	6XA4	[89]
29	UAWJ248		0.012	Mpro	6XBI	[89]
30	VIR250		not reported	PLpro	6WUU	[27]
31	VIR251		not reported	PLpro	6WX4	[27]
32	5 h		4.2	Mpro	7JKV	[175]
33	11a		0.053	Mpro	6LZE	[75], [176]
34	11b		0.040	Mpro	6M0K	[75]
35	15 l		0.019	Mpro	7MBI	[177]
36	2		5.10	PLpro	7JIT	[19]
37	3		6.40	PLpro	7JIV	[19]
38	14		0.128	Mpro	7L12	[137]
39	21		0.018	Mpro	7L13	[137]
40	26		0.17	Mpro	7L14	[137]

6. Conclusion and future perspectives

COVID-19 is an ongoing global health crisis, with a current estimated 4,180,161 fatalities worldwide. Efficacious anti–SARS-CoV-2 drugs are not yet available, despite remdesivir having been granted full authorization in October 2020 by the U.S. FDA [114]. Over the past year, the scientific community has made remarkable inroads into developing promising anti–COVID-19 agents by employing multiple techniques and strategies. Among these approaches, the generation of X-ray crystal structures of relevant drug targets has provided effective guidance for anti–SARS-CoV-2 drug discovery and lead optimization.

Theoretically, components associated with each stage of the SARS-CoV-2 replication cycle could be comprehensively explored as promising therapeutic targets. Current anti–SARS-CoV-2 drug discovery and development strategies primarily focus on preventing viral replication via the targeting of M^pro, PL^pro, or RdRp and blocking binding of the S protein to ACE2 receptors. To improve the precision design and screening of anti–SARS-CoV-2 drugs, we focused on all available X-ray co-crystal structures of the abovementioned targets in complex with newly identified small-molecule inhibitors, including natural products, FDA-approved drugs, and candidate drugs. The determined co-crystal structures offer a direct molecular-level perspective that could help elucidate the underlying mechanisms of action and provide guidance for future structure-based drug design. In the case of compound 13b, for example, a larger group at the P3 moiety could be designed to further improve the anti–SARS-CoV-2 activity in view of the space between Thr190, Gln189, and the pyridone ring of 13b. On the other hand, examination of the structures revealed no strong or durable effects for some non-covalent inhibitors (such as favipiravir, 23R, and perampanel analogue 5), supporting the need for further research focusing on covalent inhibitors to reduce “off-target” risks. In addition, combinations of several small molecules (such as linoleic acid and PF-00835231) with remdesivir have shown significant additive/synergistic anti–SARS-CoV-2 activities in vitro, presenting candidate drug combinations that may be effective in treating COVID-19 patients. Furthermore, the high mutation rate of SARS-CoV-2, resulting in the generation of novel variants, has led to increased pathogenicity [178], [179]. This may be one of the reasons for the failure of inhibitors against SARS-CoV-2. In this case, we suggest that collaborating groups should pay close attention to conserved binding sites available near the binding pocket during structure-based drug development, as this highly conserved region has promising druggable value but often goes unnoticed [180], [181], [182]. Despite substantiation of in vitro effectiveness and extensive clarification of associated mechanisms and interactions, direct clinical evidence of therapeutic efficacy for many of these agents is lacking at present. Nevertheless, we suggest that small-molecule inhibitors (including drug combinations) represent useful agents to prevent and control the COVID-19 pandemic.

CRediT authorship contribution statement

Zhonglei Wang: Conceptualization, Writing – original draft, Writing - review & editing, Visualization. Liyan Yang: Conceptualization, Writing - review & editing. Xian-En Zhao: Conceptualization, Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.