A Nonconserved Histidine Residue on KRAS Drives Paralog Selectivity of the KRAS^G12D Inhibitor MRTX1133

The discovery (1–3) and clinical development (4, 5) of mutant-selective inhibitors against the small GTPase KRAS is a major breakthrough in targeted therapy. All three family members of the RAS small GTPases, KRAS, HRAS, and NRAS, are oncogenes in human cancer. Oncogenic mutations in RAS protein attenuate its GTPase activity and increase the level of GTP-bound, active form of RAS in the cell. The MAPK pathway, which consists of the RAF/MEK/ERK kinase cascade, is a major downstream effector of RAS; it mediates mitogenic signaling from RAS and is commonly used to assess RAS activity in cells (6). The first KRAS mutant to be successfully targeted was the KRAS^G12C mutant. KRAS^G12C inhibitors rely on their ability to covalently cross-link to the mutant C12 residue on KRAS to achieve their exquisite mutant selectivity (1–3). Recently, MRTX1133 was disclosed as the first noncovalent inhibitor with selectivity toward the KRAS^G12D mutant (7–9). MRTX1133 was evolved extensively from the scaffold of adagrasib by introducing selectivity toward the D12 residue and by optimizing affinity toward the KRAS switch II pocket (7). The discovery of MRTX1133 proves that KRAS mutant without a cross-linkable residue can be successfully targeted with noncovalent inhibitors.

Human cancer cell lines were grown in RPMI1640 medium (Lonza #12-702F) supplemented with 10% FBS (Thermo Fisher Scientific #10438026) and 100 units/mL of penicillin and 100 μg/mL of streptomycin (Thermo Fisher Scientific #15140). Cell lines were authenticated by short tandem repeat DNA profiling (LabCorp) and tested negative for Mycoplasma using the MycoAlert PLUS Detection Kit (Lonza #LT07-710) and positive control (Lonza #LT07-518; Supplementary Table S1). Rasless mouse embryonic fibroblasts (MEF) cells expressing different RAS alleles were generated by and obtained from the NCI RAS Initiative (https://www.cancer.gov/research/key-initiatives/ras/ras-central/blog/2017/rasless-mefs-drug-screens), and were grown in DMEM (Lonza #12-604F) supplement with 10% FBS (Thermo Fisher Scientific #10438026) and 100 units/mL of penicillin and 100 μg/mL of streptomycin (Thermo Fisher Scientific #15140). Puromycin (Thermo Fisher Scientific #A1113803) was added in the culture media to a final concentration of 2.5 μg/mL to maintain the stable expression of wild-type (WT) HRAS in Rasless MEF cells. Blasticidin (Thermo Fisher Scientific #A1113903) was added in the culture media to a final concentration of 4 μg/mL to maintain the stable expression of all the other WT or mutant RAS. All the cells were cultured at 37°C in a humidified 5% CO₂ incubator. MRTX1133 (ChemGood #C-1420) and trametinib (Selleckchem #S2673) were dissolved in DMSO at the stock concentration of 10 mmol/L.

Cell viability assay was performed using CellTiter-Glo One reagent (Promega #G8462) following the manufacturer's instructions. Briefly, human cancer cells (2,000 cells/well) or Rasless MEF cells (3,000 cells/well) were plated and cultured in black-walled clear bottom 96-well tissue culture–treated plates (Corning #3603) for 24 hours. Next day, equal volume of media containing 2X MRTX1133 or trametinib was added into wells such that the final concentrations of the drug reached the desired concentration in culture. Cells were treated for 3 days, and CellTiter-Glo assay was performed at the endpoint. Luminescence signal was detected on a TECAN Infinite M200 plate reader (Tecan Trading AG), and cell viability was calculated by normalizing the luminescence signal to nontreated conditions. Dose–response curve was fitted and plotted using GraphPad Prism 9.0 (GraphPad Software, LLC).

To collect whole-cell extracts for immunoblotting, cells were kept on ice and washed with ice-cold DPBS (Corning #21-031-CV) twice then lysed in RIPA lysis buffer (50 mmol/L Tris-HCl pH7.4, 150 mmol/L NaCl, 1 mmol/L EDTA, 1% NP-40, 0.25% sodium deoxycholate, 2 mmol/L Na₃VO₄, 20 mmol/L sodium pyrophosphate, 20 mmol/L NaF, 20 mmol/L β-glycerophosphate) containing additional protease inhibitors (Sigma-Aldrich #11836170001) and phosphatase inhibitors (Sigma-Aldrich # 4906837001). Total protein concentration was determined by BCA assay (Thermo Fisher Scientific #23225). Cell lysate was mixed with Laemmli Sample Buffer (Bio-Rad #1610747) and denatured at 95°C for 10 minutes. Samples were separated by SDS-PAGE in an 8%–16% gradient Tris/Glycine gel (Bio-Rad #5671105) then transferred onto a nitrocellulose membrane by using the Trans-Blot Turbo RTA transfer kit (Bio-Rad #1704271). To detect protein abundance, membranes were first blocked in 5% blocking milk (Bio-Rad #1706404) in TBS (Quality Biological #351-086-131) containing 0.1% Tween-20 (Sigma #P1379) for 30 minutes, then hybridized with primary antibodies (Supplementary Table S2) following manufacturers’ instructions. After primary antibody hybridization, membranes were washed 5 minutes in TBS containing 0.1% Tween-20 for three times then hybridized with horseradish peroxidase (HRP)-conjugated secondary antibodies (Supplementary Table S2) in TBS containing 0.1% Tween-20 for 1 hour. After secondary antibody hybridization, membranes were washed 10 minutes in TBS containing 0.1% Tween-20 for three times. HRP signal was detected by using the Immobilon Forte Western HRP Substrate solution (Sigma #WBLUF0100) and acquired by using a ChemiDoc Touch Imaging System (Bio-Rad Laboratories, Inc). Chemiluminescence signal was quantified using the Image Lab Software (Bio-Rad Laboratories, Inc).

Plasmids harboring WT and G12D mutant KRAS, HRAS, and NRAS cDNA were obtained from the NCI RAS Initiative. RAS double mutants with point mutations in the 95th and 96th amino acid residues were generated by site-directed mutagenesis (Agilent Technologies #200521) using the primers listed in Supplementary Table S3 per manufacturer's recommended PCR conditions. The coding region of the RAS cDNA and point mutations were verified by Sanger sequencing. To construct N-terminal HA-tagged RAS-expressing plasmids, the coding region of RAS cDNA was first PCR amplified from the plasmids above by using CloneAmp HiFi PCR Premix (Takara #639298) and the primers listed in Supplementary Table S4 using manufacturer's recommended PCR conditions. The PCR products were purified by gel purification (Qiagen #28706) then digested with the restriction enzymes NotI (New England Biolabs #R3189) and BamHI (New England Biolabs #R3136). Digested PCR fragments were gel purified (Qiagen #28104) and ligated into the pLVX-Puro vector (Takara #632164) between the PspOMI and BamHI restriction sites by using T4 DNA ligase (New England Biolabs #M0202). Ligated plasmids were transformed into NEB Stable Competent Escherichia coli (New England Biolabs #C3040) following manufacturer's instructions, and single colonies were isolated and grown for plasmid preparation (Qiagen #28115). The entire coding region of the HA-RAS cDNA was verified by Sanger sequencing.

To produce lentivirus, 293FT cells were cotransfected with 1 μg of the pLVX-Puro vectors carrying the cDNAs expressing HA-tagged RAS variants, 0.5 μg of pMD2.G (Addgene #12259), 0.75 μg of psPAX2 (Addgene #12260) in 6-well plates by using TransIT-293 transfection reagent (Mirus #MIR2705) following manufacturer's instructions. Sixty hours after transfection, virus-containing media was collected, aliquoted, and frozen at −80°C. To generate cell lines stably expressing HA-RAS variants, cells were plated and grown overnight. Culture medium was replaced with lentivirus-containing medium at the ratio of 100 μL virus-containing medium per 25,000 cells seeded in the presence of 6 μg/mL polybrene (Thermo Fisher Scientific #TR1003G), followed by centrifugation at 2,000 rpm in room temperature for 30 minutes. Cells were cultured and incubated with the lentiviral media at 37°C in a humidified 5% CO₂ incubator for 24 hours, and then cultured in normal culture medium for another 24 hours. Cells stably expressing the transduced genes were selected with 5 μg/mL puromycin for 3 days and subsequently maintained in medium containing 2.5 μg/mL puromycin. HA-RAS expression was verified by immunoblotting.

To determine the inhibitory effect of MRTX1133 on the level of active RAS proteins in Rasless MEFs expressing a single RAS allele, a colorimetric-based RAS G-LISA activation assay was used (Cytoskeleton #BK131). Briefly, cells were treated with MRTX1133 or DMSO control for 3 days, and cell lysates were collected on ice, aliquoted, and frozen in liquid nitrogen. Total protein concentration in the lysates was determined by using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific #23225). Briefly, 25 μg of totally lysates were subjected to G-LISA assay in duplicated wells following the manufacturer's instructions, and another 15 μg of total lysates were subjected to immunoblotting to determine Ras expression level as the input. Colorimetric signal in each well was quantified using a TECAN plate reader, and a background signal from wells containing no lysate was subtracted to yield a specific signal. If the signal of a sample well was below that of the back background signal, the specific signal of that well was treated as being zero (i.e., below detection limit). Relative level of active RAS protein was calculated by normalizing against DMSO-treated samples and adjusted by the input level from the corresponding immunoblot. To determine the inhibitory effect of MRTX1133 on HA-tagged RAS with double mutations, a RAS-binding domain (RBD) pulldown assay was used (Cytoskeleton #BK008). Briefly, KRAS^G12D Rasless MEFs stably expressing HA-RAS double mutants were treated with MRTX1133 or DMSO for 24 hours. Cell lysate was collected on ice, aliquoted, and frozen in liquid nitrogen. Total protein concentration in the lysate was determined with BCA assay, and 300 μg of totally lysate was subjected to RBD pulldown assay following the manufacturer's instruction. The level of active RAS protein in the pulldown was analyzed by immunoblotting using RAS antibody (Sigma-Aldrich #OP40). Total was lysate was also subjected to immunoblotting to determine Ras expression level as the input.

The data generated by the listed authors in this study are available upon request from the corresponding authors.

We first tested the activity of MRTX1133 in a panel of human cancer cell lines harboring different KRAS mutations. Consistent with the initial reports (7, 8), two KRAS^G12D mutant cancer cell lines, Panc10.05 and AsPC-1, exhibited exquisite sensitivity to MRTX1133 (Fig. 1A). MRTX1133 at concentration up to 1 μmol/L showed no toxicity in a BRAF^V600E mutant cell line, although nonspecific toxicity was apparent at 10 μmol/L (Fig. 1B). Cancer cell lines with other KRAS mutant alleles, including H358 (KRAS^G12C) and SW620 (KRAS^G12V), also showed partial sensitivity to MRTX1133 (Fig. 1C). In sensitive cell lines, MRTX1133 caused a corresponding decrease in the level of phosphorylated ERK (pERK; Fig. 1D). Thus, MRTX1133 has inhibitory activities, albeit less potent, toward non-G12D KRAS mutants. We next tested an NRAS^G12D mutant multiple myeloma cell line, INA-6, that was previous shown to be dependent on the NRAS oncogene by CRISPR knockout (Supplementary Fig. S1; ref. 10). We were surprised that it was insensitive to MRTX1133 despite being highly sensitive to the MEK inhibitor trametinib (Fig. 1E and F). This indicates that MRTX1133 may have strong paralog selectivity toward KRAS.

The sensitivity of human cancer cell lines toward MRTX1133 could be confounded by other coexisting mutations in the cell and by the variable expression levels of WT KRAS, HRAS, and NRAS alleles that could transmit RAS signaling (11). We therefore examined the activity of MRTX1133 in a panel of isogenic “Rasless” MEFs that express only a single RAS allele. Rasless MEFs were previously generated by deleting all three Ras genes, and this led to growth arrest (12). The NCI RAS Initiative stably reexpressed WT and mutant RAS cDNAs in Rasless MEFs, and this rescued their proliferation (13). Thus, in the reconstituted Rasless MEFs, proliferation is strictly dependent on the sole RAS allele being expressed. This isogenic system is uniquely useful for evaluating the selectivity of a RAS inhibitor.

Consistent with human cancer cell lines, KRAS^G12D Rasless MEFs were highly sensitive to MRTX1133 with an IC₅₀ of 70 nmol/L (Fig. 2A). As a control for its specificity, we found that MRTX1133 had no impact on the viability of BRAF^V600E Rasless MEFs at concentration up to 2 μmol/L, and nonspecific toxicity was only evident at 10 μmol/L (Fig. 2A). KRAS^G12C, KRAS^G12V, and KRAS^G13D Rasless MEFs exhibited partial sensitivity to MRTX1133, whereas KRAS^G12R, KRAS^Q61L, and KRAS^Q61R mutants were insensitive (Fig. 2B). Concordantly, MRTX1133 significantly decreased pERK level in sensitive cell lines but had minimal impact on pERK level in insensitive cell lines (Fig. 2C and D).

Next, we tested the sensitivity of KRAS^WT, HRAS^WT, or NRAS^WT Rasless MEFs. Unlike KRAS^WT human cancer cell lines that showed insensitivity to MRTX1133 (8), KRAS^WT Rasless MEFs were sensitive to MRTX1133, with an IC₅₀ of 296 nmol/L that was approximately 4-fold higher than that of KRAS^G12D Rasless MEFs (Fig. 2E). Thus, the mutant D12 residue is important, but not essential, for MRTX1133 binding. In contrast, HRAS^WT and NRAS^WT Rasless MEFs were completely insensitive to MRTX1133 (Fig. 2E). Concordantly, MRTX1133 caused a dose-dependent decrease in pERK level in KRAS^WT Rasless MEFs but not in HRAS^WT and NRAS^WT Rasless MEFs (Fig. 2F and G). To directly evaluate the impact of MRTX1133 on RAS activity, we measured the level of active RAS in cells with a RAS activation assay, G-LISA. We found that MRTX1133 was most effective at reducing the level of active KRAS^G12D and KRAS^WT protein in the cells, it also reduced the level of active KRAS^G13D protein to a lesser extent. In contrast, MRTX1133 had little impact on the level of active KRAS^Q61R and that of HRAS^WT and NRAS^WT protein (Fig. 2H). These results support the notion that the activity of MRTX1133 is not confined to KRAS^G12D, and it can inhibit KRAS^WT as well as other KRAS mutants including G13D and G12C, albeit at lower potency.

The striking difference in the selectivity of MRTX1133 for KRAS over HRAS and NRAS prompted us to investigate the underlying mechanism. Prior structural studies on MRTX1133 showed that the compound makes multiple contacts throughout the G-domain of KRAS (8). All amino acid residues in KRAS^G12D that made direct interaction with MRTX1133 are conserved among the three Ras paralogs except H95 (Supplementary Fig. S2A). The imidazole side chain of H95 on KRAS makes at least three important contributions to drug binding: it forms a direct hydrogen bond with the MRTX1133 pyrimidine ring; it forms a tripartite hydrogen bonding interaction with both the MRTX1133 pyrimidine ring and the Y64 phenol hydroxyl group; and it forms a T-shaped interaction with the Y96 phenol ring to stabilize the hydrogen bonding between Y96 and MRTX1133 (Supplementary Fig. S2B). In HRAS and NRAS, the corresponding residues are Q95 and L95, respectively (Supplementary Fig. S2A). Thus, we hypothesized that the nonconserved H95 on KRAS provides a critical selectivity handle for MRTX1133, and substitution of H95 with Q and L, as seen in HRAS and NRAS, respectively, could disrupt MRTX1133 binding.

To test this hypothesis, we generated reciprocal mutations in G12D mutant of KRAS, HRAS, and NRAS where residue 95 was “swapped” among these paralogs. We stably expressed HA-tagged cDNAs of these double mutants in the parental KRAS^G12D Rasless MEFs so we could directly compare their sensitivity against KRAS^G12D single mutant in the same cells (Supplementary Fig. S2C). Expression of the HA-KRAS^G12D/H95Q and the HA-KRAS^G12D/H95L double mutants rendered cells fully insensitive to MRTX1133 (Fig. 3A). Using RBD binding assay to directly measure the level of active RAS protein, we found that MRTX1133 was ineffective at reducing the level of active KRAS double mutants (Fig. 3B). Consequently, expression of these double mutants significantly attenuated the downregulation of pERK by MRTX1133 (Fig. 3C). Consistent with the notion that MRTX1133 is unable to inhibit HRAS and NRAS, expression of HA-HRAS^G12D and HA-NRAS^G12D single mutant rendered cells insensitive to MRTX1133. In contrast, cells expressing the HA-HRAS^G12D/Q95H and the HA-NRAS^G12D/L95H double mutants remain fully sensitive to MRTX1133 (Fig. 3D). Concordantly, the level of active HA-HRAS^G12D and HA-NRAS^G12D single mutants remained unchanged with MRTX1133 treatment, whereas the level of active HA-HRAS^G12D/Q95H and the HA-NRAS^G12D/L95H double mutant was effectively reduced by MRTX1133 (Fig. 3E). This differential sensitivity was also reflected by corresponding changes in pERK level in the cell (Fig. 3F).

In patients treated with the KRAS^G12C inhibitor adagrasib, acquired resistance can arise via secondary mutation at H95 and Y96 because these residues form hydrogen bonds with the covalently bound ligand (Supplementary Fig. S2D; refs. 14, 15, 16). We therefore tested whether secondary mutation at Y96 could confer resistance to MRTX1133. Expression of a HA-KRAS^G12D/Y96D double mutant rendered KRAS^G12D Rasless MEFs fully resistant to MRTX1133 (Fig. 4A). MRTX1133 was unable to reduce the level of active HA-KRAS^G12D/Y96D mutant (Fig. 4B), and the level of pERK in cells expressing the double mutant was resistant to MRTX1133 (Fig. 4C). To further corroborate these findings, we tested the effect of HA-tagged KRAS^G12D/H95Q, KRAS^G12D/H95L, and KRAS^G12D/Y96D double mutants in three human pancreatic cancer cell lines Panc10.05, KP4, and SW1990. These cell lines harbor endogenous KRAS^G12D mutation and are sensitive to MRTX1133. Expression of these KRAS double mutants conferred functional resistance to MRTX1133 and preserved pERK level to various degrees in the presence of MRTX1133 (Fig. 4D and E; Supplementary Fig. S3). In conclusion, our results indicate that the non-conserved H95 residue in KRAS presents an essential selectivity handle for MRTX1133 to effectively inhibit KRAS while sparing HRAS and NRAS protein.

In conclusion, our results indicate that the nonconserved H95 residue in KRAS presents an essential selectivity handle for MRTX1133 to effectively inhibit KRAS while sparing HRAS and NRAS protein. Amino acid diversity at residue 95 can be exploited for the development of pan-KRAS inhibitors as well as HRAS and NRAS paralog-selective inhibitors. Additional implications of our finding are discussed in the Supplementary Discussion section.

No disclosures were reported.

M.A. Keats: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft. J.J.W. Han: Data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft. Y.-H. Lee: Data curation, formal analysis, validation, investigation, visualization, methodology. C.-S. Lee: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. J. Luo: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

This research was supported by the Intramural Research Program (ZIABC011732) of the NCI, NIH, to J. Luo.

The authors would like to thank the NCI RAS Initiative for generously providing the panel of Rasless MEF cell lines and human RAS cDNA plasmids for cloning, the CCR Genomics Core facility for sequencing support, and Dr. John S Schneekloth at the NCI for helpful discussion.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.