1. Introduction

KRAS (Kirsten rat sarcoma viral oncogene homolog) gene mutation is widespread in tumors and is one of the most frequently mutated oncogenes in various malignancies, including lung, colon, and pancreatic cancers, etc.; it is found in one out of every seven human cancers ^[1][2][1,2]. Typically, KRAS switches between the GTP-bound “on” state and the GDP-bound “off” state ^[3][4][3,4]. The activating mutations drive the KRAS protein in a GTP-bound active conformation, thereby resulting in a state of sustained activation of KRAS signaling, which in turn causes uncontrolled proliferation of the cells and leads to carcinogenesis ^[4][5][4,5]. Moreover, KRAS-activating mutations can inhibit the apoptosis of tumor cells, modulate cell metabolism, and alter the tumor microenvironment to induce tumor immunosuppression, thus further promoting the metastasis of the tumor ^[6][7][6,7]. Therefore, targeting KRAS mutations is regarded as the “holy grail” of targeted cancer therapies. However, targeting KRAS mutations directly has been a huge challenge since its discovery in 1982, when it was deemed the “undruggable target” ^[3][8][3,8]. Strikingly, due to the improved understanding of the structural biology of KRAS and the development of drug-designing technologies, covalent KRAS inhibitors that block mutant KRAS-G12C have recently been successfully developed. Sotorasib (FDA, 2022) and adagrasib (FDA, 2022) were approved, and exhibited immediate and profound clinical impacts in oncotherapy ^[9][10][9,10]. Furthermore, the inhibitors which target other KRAS-activating mutants including KRAS-G12D (MRTX1133) [11] or KRAS-G12V (JAB-23000) are in preclinical studies [12]. Interestingly, multiple developing drug design strategies have been applied to the KRAS inhibitor discovery, including targeted protein degradation (TPD) technology [13], multivalent binding strategy, and salt bridge strategy ^[14][15][14,15].

As more and more covalent drugs have been successfully used in the clinical setting, covalent binding strategies have attracted great interest in drug development ^[16][17][48,49]. For inhibitors targeting mutant KRAS, the covalent binding strategy represents a major breakthrough in drug development, which greatly improves the binding ability and selectivity of small molecular compounds for mutant KRAS proteins ^[18][50]. Such a strategy focuses on the KRAS-G12C mutation, and this class of compounds can specifically covalently bind to cysteine at previously unreported binding sites, rendering KRAS insensitive to nucleoside transforming factors such as SOS, thus leaving the conformation of KRAS-G12C in the GDP-bound inactivated state. Two compounds based on covalent binding strategies, sotorasib and adagrasib, have been approved by the FDA in 2022 ^[10][19][20][10,51,52]. Currently, a covalent binding strategy has been adopted in not only KRAS-G12C, but also KRAS-G12D and KRAS-G12S [15].

Cysteine is the most nucleophilic amino acid among the 20 common amino acids and is most commonly covalently bound in covalent drug design. The G12C mutation provides the possibility of covalent binding to the 12-position cysteine to enhance the binding affinity, thereby improving the inhibitory selectivity. The first KRAS-G12C-specific inhibitors were developed based on the structure of a fragment of 6H05 in 2013 by Ostrem et al., which specifically bound in the allosteric site with the covalent linkage to C12; therefore, they were disrupting the recruitment of GTP and maintaining the inactive state ^[21][53]. Another type of inhibitor that binds to the allosteric site is the Quinazoline series, with the quinazoline ring as the core structure, which represents the most potent KRAS-G12C allosteric inhibitors, including MRTX849, AMG-510, ARS-1620, and AZD4625 ^{[22][23][24][25]}[54,55,56,57]. Moreover, KRAS-G12C allosteric inhibitors with other core structures have also been in development, including ARS-853, BI-0474, JDQ443, etc. ^[26][27][28][58,59,60].

Recently, due to the success of covalent kinase inhibitors, Lim et al. developed the first series of the covalent substrate (GTP/GDP)-competitive inhibitors targeting the catalytic site by modifying diphosphate compounds with various electrophiles and different linkers. SML-8-73-1 was identified as the lead compound, which could block >95% KRAS-G12C in competition tested in the presence of 1 mM GDP or GTP by irreversible binding and was regarded as a promising inhibitory agent for KRAS-G12C ^[29][61]. However, two negatively charged phosphate groups at SML-8-73-1 cause low membrane penetration, which limits further application in vivo. The optimization of SML-8-73-1 led to SML-10-70-1, which improved the cell permeability and competitively inhibited KRAS-G12C through covalent binding.

In addition to targeting KRAS-G12C, the covalent binding strategy has also been applied to KRAS-G12S and KRAS-G12D. Shakat’s group reported that a series of small molecules could suppress the oncogenic signaling of KRAS-G12S through the chemical acylation of a serine residue. The β-lactone group of design compounds such as G12Si-1 and G12Si-2 could acylate the mutant serine of KRAS-G12S to exhibit selectivity, making the compounds show activity in cells expressing KRAS-G12S, but sparing the wild-type KRAS. This technology overcomes the weak nucleophilicity of an acquired serine residue, which may serve as a way to selectively target other inactivated serines in covalent drug development ^[30][62]. Interestingly, Goldsmith et al. developed a covalent inhibitor RMC-9805, for which the structure is not disclosed. RMC-9805 forms a non-covalent complex with KRAS-G12D and cyclophilin A first, and then the cool covalent warhead of RMC-9805 binds irreversibly with the mutant aspartic acid over a matter of minutes to hours [15].

The inhibitors covalently targeting oncogenetic mutant KRAS showed rather good anti-cancer activity. Treatment with AMG-510 caused the regression of KRAS-G12C tumors and enhanced the anti-tumor efficacy of either chemotherapy or targeted drugs. AMG-510 improved the pro-inflammatory tumor microenvironment and generated durable therapeutic effects alone or combined with immune-checkpoint inhibitors. Cured mice may have induced adaptive immunity against shared antigens to reject the growth of isogenic KRAS-G12D tumors ^[19][51]. Moreover, treated with another covalent inhibitor of KRAS-G12C ARS1620, the KRAS-G12C mutant cell would present ARS1620-modified peptides in MHC-I complexes, which could serve as tumor-specific neoantigens to elicit a cytotoxic T cell response against KRAS-G12C cells ^[31][63].

Currently, targeted protein degradation techniques such as proteolysis-targeting chimeras (PROTAC) have attracted great attention in drug development ^[32][64]; they are expected to be applied to undruggable targets. In 2020, Crews’s group developed the first PROTAC LC-2, which covalently bound KRAS-G12C with an MRTX849 warhead, and recruited the E3 ligase VHL to induce the degradation of KRAS-G12C ^[33][34][65,66]. Moreover, Chen’s group developed the pomalidomide-based PROTAC degraders. Of them, compound KP-14 exhibited the highest KRAS-G12C degrading capability in NCI-H358 cells and showed potent antiproliferative activity ^[35][67]. Following this, Lu’s group reported the first reversible covalent KRAS-G12C PROTAC YF135 based on a cyanacrylamide-based reversible covalent bond, which induced the rapid and sustained degradation of KRAS-G12C ^[36][68]. Zhang et al. developed a series of PROTACs based on AMG-510, and compound III-2 was identified to exhibit binding and degradation ability for KRAS-G12C, showing a more potent inhibitory effect on downstream p-ERK signaling ^[37][69]. In addition to KRAS-G12C, Astellas developed a first-in-class degrader, ASP3082 (structure not shown), which efficiently degraded the G12D mutant and has been in clinical trials since June 2022 [15].

Beyond the heterodimeric small molecules such as PROTACs, the monomeric targeted protein degrader for KRAS is also reported, which induces the interaction between KRAS and E3 ligase to lead to the degradation of KRAS. A natural product, Kurarinone, which was reported to have anti-cancer activity against various cancers, was identified to decrease the protein level of KRAS by proteasomal degradation dependent on an E3 ubiquitin ligase WDR76. Knockdown of WDR76 through small interfering RNA siWDR76 restored the level of KRAS as well as the downstream protein p-ERK and c-MYC. Moreover, Kurarinone arrested the cell cycle in the G0/G1 phase in a p53-independent manner. However, the binding data for KRAS of Kurarinone are still not available ^[38][70]. Through the analysis of TOP flash reporter cells and potential toxicity effects on primary neural stem cells, Moon’s group identified CPD0857 from 2000 chemical compounds, which induced ubiquitin-dependent proteasomal degradation of Ras proteins. Moreover, CPD0857 effectively inhibited the proliferation and increased the apoptosis of CRC cell lines, and overcame the resistance of CRC harboring KRAS mutations. Accordingly, CPD0857 also inhibited tumor growth and significantly decreased Ras protein expression in xenograft tumors of mice ^[39][71].

The dimerization of oncogenic KRAS plays an important role in the activation of MAPK signaling to promote the proliferation of cancer cells. The salt bridge between D154 of one KRAS molecule and R161 of a partner is key for the formation of KRAS dimerization ^[40][72]. Marshall’s group identified that the α-β dimerization of KRAS-G12D is induced through the second phosphatidylserine-dependent interface, which is sensitive to small molecule inhibitors. Therefore, targeting the dimerization interface to inhibit KRAS–KRAS interaction has become a promising strategy ^[41][73]. It was reported that targeting the α4-α5 dimerization interface using RAS-specific monobody would inhibit oncogenic KRAS and inhibit tumor formation in vivo ^[42][74]. Specifically, the KRAS inhibitor BI-2852 was identified bound to a pocket between switch I and II on RAS with the nanomolar affinity, which is distinct from that of covalent KRAS-G12C inhibitors such as AMG-510. BI-2852 dismisses GEF, GAP, and effector interactions with KRAS, thereby causing the inhibition of downstream signaling and the proliferation in KRAS-mutant cells ^[43][75]. Interestingly, BI-2852 was identified to induce the β-β dimerization of KRAS-G12D, which may be responsible for the inhibitory activity of BI-2852 ^[41][73].

In addition to self-dimerization, the interaction of KRAS with other proteins plays an important role in the activity of KRAS. Therefore, blocking the interaction of KRAS and its activators would downregulate the downstream signaling of KRAS. Using a combination of computational and biochemical approaches, Stockwell’s group identified a site adjacent to proline 110 (P110 site). A compound KAL-21404358 was suggested to target the P110 site of KRAS-G12D, which disrupted the interaction of KRAS-G12D with B-Raf to inhibit the RAF-MEK-ERK and PI3K-AKT signaling ^[44][76]. Interestingly, Shokat et al. developed a bifunctional small molecular ligand (Compound-12), which was able to serve as a molecular glue to promote the association between cyclophilin A and GTP-bound Ras, which disrupted the interaction with B-Raf. Such a strategy sheds light on the development of novel KRAS inhibitors ^[45][77]. Revolution Medicine developed a non-covalent pan-RAS inhibitor RMC-6236 that sticks cyclophilin A to all forms of KRAS. RMC-6236 exhibits a cancer-killing feature. However, the pan-inhibitory aspects, including affecting wild-type KRAS, induces additional safety risks [15].

Unlike the KRAS-G12C mutant, which could form a cysteine-based covalent bond, there was a challenge to develop selective inhibitors targeting KRAS-G12D. Considering the α-carboxylic acid moiety of Asp12 is deprotonated under physiological conditions, targeting the Asp12 residue of KRAS-G12D may produce the selectivity for KRAS-G12D. Zhang’s group replaced the acryloyl moiety group of the G12C inhibitor MRTX22 with a piperazine moiety, which was near to Asp12 to form a salt bridge interaction. As a result, both ITC and enzymatic assays indicated the selectivity of the designed inhibitor TH-Z816 for KRAS-G12D over KRAS-WT, and a salt bridge was identified through the crystal structure (PDB ID: 7EW9) ^[46][78]. Further optimization of such inhibitors led to more potent compounds including TH-Z827, TH-Z835, and TH-Z837. This strategy has also been applied to the development of other KRAS-G12D inhibitors, including the MRTX1133 ^[11][47][11,79], which bears a piperazine moiety to improve the selectivity for KRAS-G12D. Strikingly, the protonated piperazinyl group produces 10-fold selectivity over KRAS-WT via forming a salt bridge with Asp12. The piperazine adopts the twist-boat conformation, and its salt bridge to Asp12 is stabilized by an additional hydrogen bond interaction with Gly60 (Crystal structure, PDB ID:7RPZ) [11]. MRTX1133 efficiently inhibited the phosphorylation of KRAS downstream factor ERK1/2 and exhibited more than 1000-fold selectivity cell viability in KRAS-G12D-mutant cell lines compared to KRAS-WT cell lines. Meanwhile, MRTX1133 indicated dose-dependent inhibition of KRAS-mediated signal transduction and induced significant regression of the tumors (≥30%) in the patient-derived KRAS-G12D-mutant xenograft models ^[48][80].

Multivalency for ligands such as glycans may enhance both affinity and binding specificity, which has become a useful strategy in drug development ^[49][81]. For the inhibitor of KRAS, through molecular docking, multiple fragments were predicted to bind with the D38 site, A59 Site, or Y32 Site, respectively. A series of multivalent compounds were designed by conjugating the fragments and were further filtered through drug-like properties prediction. The candidate compounds were synthesized and tested. The compound 3144 was identified as the most promising inhibitor. Compound 3144 was identified to bind to RAS proteins as a result of microscale thermophoresis, nuclear magnetic resonance spectroscopy, and isothermal titration calorimetry, demonstrating cytotoxicity in RAS-dependent cells. It showed metabolic stability during liver microsomes assay and exhibited anti-tumor activity in vivo ^[50][82]. Therefore, the multivalent strategy is supposed to be an effective method for targets including KRAS ^[50][82].