Small-molecule inhibition of MAP2K4 is synergistic with RAS inhibitors in KRAS-mutant cancers

Significance

KRAS^G12C inhibitors have shown clinical responses in KRAS^G12C-mutant lung cancer, but much less so in colon cancers having the same mutation. In lung cancer, acquired resistance limits the clinical benefit patients experience. Here, we show that a novel small-molecule inhibitor of a little studied kinase, MAP2K4, while having minor effects in KRAS-mutant cancers when used alone, greatly potentiates the anticancer effects of the clinically used KRAS^G12C inhibitor sotorasib in vitro and in vivo. Combined inhibition of mutant KRAS and MAP2K4 prevents activation of a feedback loop that reactivates KRAS in the presence of its inhibitor, thereby providing more durable suppression of KRAS signaling and deeper therapeutic responses in animal models.

Abstract

The Kirsten rat sarcoma viral oncogene homologue KRAS is among the most commonly mutated oncogenes in human cancers, thus representing an attractive target for precision oncology. The approval for clinical use of the first selective inhibitors of G12C mutant KRAS therefore holds great promise for cancer treatment. However, despite initial encouraging clinical results, the overall survival benefit that patients experience following treatment with these inhibitors has been disappointing to date, pointing toward the need to develop more powerful combination therapies. Here, we show that responsiveness to KRAS^G12C and pan-RAS inhibitors in KRAS-mutant lung and colon cancer cells is limited by feedback activation of the parallel MAP2K4-JNK-JUN pathway. Activation of this pathway leads to elevated expression of receptor tyrosine kinases that reactivate KRAS and its downstream effectors in the presence of drug. We find that the combination of sotorasib, a drug targeting KRAS^G12C, and the MAP2K4 inhibitor HRX-0233 prevents this feedback activation and is highly synergistic in a panel of KRAS^G12C-mutant lung and colon cancer cells. Moreover, combining HRX-0233 and sotorasib is well-tolerated and resulted in durable tumor shrinkage in mouse xenografts of human lung cancer cells, suggesting a therapeutic strategy for KRAS-driven cancers.

The identification of oncogenic driver mutations and the development of drugs that selectively target the oncogenic signals emanating from these aberrant proteins, known as precision medicine, has reshaped cancer treatment strategies and clinical outcomes. Numerous clinical studies have shown that precision medicine is associated with significant prolongation of progression-free survival (PFS) and less toxicity compared with standard-of-care chemotherapies (1–4). Among these oncogenic drivers, KRAS is a frequently altered oncogene, with mutations in approximately 11% of all human cancers, especially in pancreatic, colorectal (CRC), and non–small cell lung cancers (NSCLC) (5). Activating mutations in KRAS promote and maintain tumorigenesis through stimulation of cell proliferation and escape from apoptosis. Given the frequency of aberrant KRAS signaling in cancer, many targeted therapies have focused on directly inhibiting mutant KRAS as well as inhibiting kinases in downstream signaling pathways like the RAF-MEK-ERK (MAP Kinase) pathway and the PI3K-AKT pathway.

Unfortunately, single-agent targeted therapies that act in the MAPK pathway of KRAS-mutant cancers are often relatively ineffective (4, 6–8). Poor single-agent targeted therapy response can often be attributed to adaptive activation of feedback mechanisms that reactivate the pathway upon its inhibition or activation of a compensatory mitogenic pathway. For instance, inhibition of the MEK kinase in KRAS-mutant NSCLC proved ineffective due to a feedback loop that reactivates the MAPK pathway in the presence of the inhibitor (7, 9). Occurrences of such feedback loops are frequent in cancer, and disabling this feedback mechanism by concomitant inhibition can be highly effective to improve clinical responses. Indeed, inhibition of mutant BRAF in colon cancer leads to activation of EGFR, precluding a response to BRAF inhibitor monotherapy, and the combination of BRAF and EGFR inhibitors is now an approved therapeutic strategy for this disease (10, 11).

Recently, small-molecule inhibitors of mutant KRAS with a glycine-to-cysteine substitution at codon 12 (KRAS^G12C) have been developed (4, 12–14). These inhibitors have shown better efficacy in NSCLC than in colon cancers harboring the same mutation, indicating the existence of tissue-specific adaptive responses that involve activation of EGFR in colon cancer cells (15). Indeed, in colon cancer models, the intrinsic resistance to KRAS^G12C inhibitors can be overcome by coinhibition of EGFR (16, 17). For patients with KRAS^G12C-mutant NSCLC, these drugs provide initial progression-free survival (PFS) benefit, but recent data indicate that there is little overall survival (OS) benefit, indicating that acquired resistance is rapid in these patients (4). These data emphasize the need to identify powerful but well-tolerated rational combination therapies to enhance the efficacy of KRAS^G12C inhibitors.

Feedback reactivation of KRAS signaling by HER receptor tyrosine kinases (RTKs) is a recurrent mechanism of intrinsic resistance to several inhibitors targeting the MAPK pathway (9, 18). Consequently, pan-HER tyrosine kinase inhibitors are synergistic with MEK inhibitors to overcome the feedback reactivation in KRAS-driven tumors (9). Based on these encouraging preclinical results, multiple clinical trials were conducted combining MEK and pan-HER inhibitors in cancers harboring KRAS mutations (19–21). Unfortunately, clinical development has been discontinued due to high toxicity of these combinations. On-target toxicities associated with tyrosine kinase inhibitors targeting the MAPK as well as the PI3K-AKT pathway are not uncommon and a major setback for many drug combinations and have led to the failure of several clinical trials (22–26). This highlights the importance of identifying powerful drug combinations with fewer treatment-related adverse events.

Our previous work has shown that genetic inactivation of MAP2K4 (also known as MKK4) greatly enhances sensitivity to MEK and ERK inhibitors in KRAS-mutant cancers (27). Mechanistically, our data indicate that MAP2K4 is part of a feedback loop that leads to the RTK-dependent KRAS signaling rebound in the presence of a MAPK pathway inhibitor. MEK inhibition can increase MAP2K4-JNK-JUN pathway signaling through inhibition of DUSP4. JUN activation in turn activates several RTKs, especially ERBB2 and ERBB3, that subsequently reactivate the MAPK pathway in the presence of a MEK inhibitor. Therefore, combined pharmacological inhibition of MAP2K4 and MAP kinases could be a therapeutic strategy for the treatment of KRAS-mutant tumors. Until recently, there were no small-molecule MAP2K4 inhibitors available. However, such compounds have now been developed primarily to enhance liver regeneration (28, 29), which also enables testing the utility of these compounds for anticancer therapy. The experiments presented below provide a strong rationale to combine MAP2K4 and KRAS inhibition as a synergistic drug combination for the treatment of patients with NSCLC and CRC harboring KRAS mutations.

Results

MAP2K4 Loss Disables RTK Feedback Activation by RAS and MEK Inhibitors.

We have shown previously that KRAS-mutant lung and colon cancers have limited sensitivity to single-agent MEK inhibitors due to a MAP2K4-dependent feedback activation of JUN, which in turn activates HER receptor tyrosine kinases (RTKs), especially ERBB2 and ERBB3. This elevated RTK signaling reactivates KRAS and its downstream effectors in the presence of drug, thereby blunting the effect of MEK inhibition and maintaining cancer cell proliferation and survival (9, 27). To investigate whether such a MAP2K4-JNK-JUN-mediated feedback loop also limits the response to KRAS^G12C inhibition, we treated both wild-type and MAP2K4 knockout H358 and SW837 cells (both KRAS^G12C-mutant) with either the MEK inhibitor trametinib or the KRAS^G12C inhibitor sotorasib and studied signaling in the MAPK pathway 48 h after drug exposure using western blotting with phospho-specific antibodies. Both MEK inhibition and KRAS^G12C inhibition resulted in increased activity of JUN, ERBB2, and ERBB3 as evidenced by increases in phospho-JUN (p-JUN), phospho-ERBB2 (p-ERBB2), and phospho-ERBB3 (p-ERBB3) in the lung cancer and colon cancer cells (Fig. 1A). Total levels of ERBB2 and ERBB3 were also increased, indicating elevated ERBB2 and ERBB3 expression upon sotorasib and trametinib treatment. Moreover, monotherapy sotorasib treatment did not result in efficient suppression of ERK phosphorylation after 48 h. This feedback activation of phosphorylated JUN, ERBB2, and ERBB3 upon treatment was attenuated by loss of MAP2K4 and resulted in a more complete inhibition of phosphorylated ERK in MAP2K4 knockout cells (Fig. 1A). These data indicate that sotorasib treatment, like MEK inhibitor treatment, induces a MAP2K4-dependent HER receptor tyrosine kinase feedback activation in KRAS-mutant lung and colon cancer cells that limits suppression of signaling downstream of KRAS.

Fig. 1.

MAP2K4 knockout confers sensitivity to the KRAS^G12C inhibitor sotorasib and MEK inhibitor trametinib (A) Parental (WT) and MAP2K4 knockout (KO) H358 and SW837 cells treated with either trametinib (20 nM) or sotorasib (25 nM) at the indicated concentrations for 48 h. Western blot analysis was performed with the indicated antibodies. Vinculin served as a loading control. (B) Crystal violet long-term viability assay of parental (WT) and MAP2K4 knockout (KO) H358 and SW837 cells treated with increasing concentrations of sotorasib or trametinib. Media and drugs were refreshed every 2 to 3 d, and the cells were cultured for 10 to 14 d before fixing, staining, and imaging.

We next evaluated whether disabling the MAP2K4-mediated feedback activation would increase the antiproliferative effect of KRAS^G12C inhibition in these cells. Indeed, MAP2K4 loss sensitized both H358 and SW837 cells to sotorasib as seen in the long-term cell proliferation assays (Fig. 1B). Consistent with previous results (27), increased sensitivity of MAP2K4 knockout cells was also seen with the MEK inhibitor trametinib (Fig. 1B). While H358 MAP2K4 KO cells exhibited a decrease in proliferation rate compared to the MAP2K4 WT cells (a population doubling time of approximately 61 h vs. 34 h), MAP2K4 loss had no impact on proliferation in SW837 cells (SI Appendix, Fig. S1A).

Besides KRAS^G12C-specific inhibitors, many other KRAS inhibitors are approaching clinical use (30). RMC-6236 is a RAS(ON) multi-inhibitor that targets the active form of both wild-type and mutant RAS proteins (31). Unlike the KRAS^G12C-specific inhibitors, it can target a range of different KRAS mutations, making it an interesting compound for a wider array of KRAS-mutant cancers. Therefore, we were interested to test whether loss of MAP2K4 could also enhance the effect of this RAS(ON) multi-inhibitor. Although H358 and SW837 cells were relatively sensitive to treatment with RMC-6236, MAP2K4 knockout could further increase the sensitivity of the cells to this drug (SI Appendix, Fig. S1B). In summary, our data indicate that loss of MAP2K4 can greatly enhance the sensitivity to multiple KRAS pathway inhibitors, including KRAS^G12C and RAS(ON) multi-inhibitors, as well as the MEK inhibitor trametinib in KRAS-mutant lung and colon cancer cells.

The MAP2K4 Inhibitor HRX-0233 Is Synergistic with RAS Pathway Inhibition.

Since genetic loss of MAP2K4 enhanced KRAS-mutant lung and colon cells to RAS inhibitors, we next sought to evaluate whether the concomitant inhibition of RAS and pharmacological inhibition of MAP2K4 would be synergistic in these cancer models. HRX-0233 is a novel MAP2K4 inhibitor developed to enhance liver regeneration for treatment of acute and chronic liver diseases (28, 29, 32) (SI Appendix, Fig. S2). We first focused on synergy with sotorasib, as this KRAS inhibitor is in routine clinical use. Fig. 2A shows long-term cell proliferation assays of H358 and SW837 cells in the presence of either drug alone or the combination of sotorasib and HRX-0233. The data indicate that HRX-0233 had little single-agent activity in H358 and SW837 cells but greatly enhanced the antiproliferative effects of sotorasib in both cell models. The Bliss synergy scores calculated from three independent replicate experiments (Fig. 2B) show values of nearly 70 and 51, well above the threshold of 10, above which drugs are considered to be synergistic (33) (Materials and Methods).

Fig. 2.

Combination of HRX-0233 and KRAS inhibitors is synergistic and triggers cell death (A) Response to the combination of HRX-0233 and sotorasib in H358 and SW837 cells and to the combination of HRX-0233 and RMC-6236 in DLD1 cells. Media and drugs were refreshed every 2 to 3 d, and the cells were grown for 10 to 14 d before fixing, staining, and imaging. Representative data are shown from three independent experiments. (B) Bliss synergy plots for the combination of HRX-0233 and sotorasib in H358 and SW837 cells and the combination of HRX-0233 and RMC-6236 in DLD1 cells. The average of three independent experiments was used to generate the Bliss synergy plots. The percentage of cell viability was estimated by quantification of three independent crystal violet long-term viability assays as shown in (A) and normalized to untreated controls. Synergyfinder.org web tool was used to calculate the Bliss synergy scores. A Bliss score above 10 indicates drug synergy. (C) Synergy scores for the combination of HRX-0233 and MAPK inhibitors trametinib and SCH772984 or KRAS inhibitors sotorasib and RMC-6236 across three KRAS-mutant lung cancer and three KRAS-mutant colon cancer cell lines. The percentage of cell viability was estimated by quantification of three independent crystal violet long-term viability assays as shown in Supplementary SI Appendix, Fig. S3 and normalized to untreated controls. Synergyfinder.org web tool was used to calculate the Bliss synergy scores. Three independent experiments are represented (D) IncuCyte-based assay for proliferation (Upper) and caspase 3/7 activity (Lower). H358 and SW837 cells were treated with DMSO, HRX-0233, sotorasib, or the combination in the presence of a caspase 3/7 apoptosis assay reagent. DLD1 cells were treated with DMSO, HRX-0233, RMC-6236, or the combination in the presence of a caspase 3/7 apoptosis assay reagent. Confluence (D) and caspase 3/7 activity (E) over time were measured by IncuCyte®. Green fluorescence from the apoptosis assay reagent divided by the total confluence was used to estimate apoptosis for 72 h. Drugs were refreshed every 2 to 3 d for the indicated time period. Data are mean ± SEM of three independent replicates.

Most KRAS alterations in cancer are non-G12C mutations, which limits the clinical use of sotorasib. We therefore also tested several different MAPK and KRAS inhibitors in combination with HRX-0233 in a broader range of KRAS-mutant cell lines. These included three NSCLCs (H358 (KRAS^G12C), H2122 (KRAS^G12C), and A549 (KRAS^G12S)) and three colorectal cancer (SW837 (KRAS^G12C), HCT116 (KRAS^G13D), and DLD1 (KRAS^G13D)) cell lines. We tested trametinib (MEK inhibitor), SCH772984 (ERK inhibitor), RMC-6236 (RAS(ON) multi-inhibitor), and sotorasib (only in KRAS^G12C-mutant cell lines). We performed drug synergy matrices by combining increasing concentrations HRX-0233 and the MAPK inhibitors mentioned above, and calculated the Bliss synergy score from three independent replicate experiments for each drug combination. The results are shown in Fig. 2C (see SI Appendix, Fig. S3 for representative colony formation assays for the drug combinations). The combination of sotorasib and HRX-0233 showed strong synergy in the KRAS^G12C-mutant cell lines H358, H2122, and SW837, as indicated by the Bliss synergy scores above 10 (Fig. 2 A–C and SI Appendix, Fig. S3). Synergy was also observed by combining HRX-0233 and RMC-6236 across the panel of lung and colon cells harboring different KRAS mutations (Fig. 2 A–C and SI Appendix, Fig. S3). Consistently, targeting downstream of mutant KRAS through inhibition of MEK by trametinib or ERK by SCH772984 also showed strong synergy with HRX-0233 in all cell lines tested (Fig. 2C and SI Appendix, Fig. S3). Importantly, these Bliss synergy score plots suggest a strong synergy across a wide concentration range of the drugs for multiple combinations with the MAP2K4 inhibitor HRX-0233.

The efficiency to impair cell proliferation by the combination of HRX-0233 and sotorasib or RMC-6236 was further confirmed by IncuCyte-based cell proliferation assays (Fig. 2 D, Upper). The drugs as monotherapy only had a modest impact but in combination resulted in a more sustained restraint of cell proliferation. Synergistic anticancer effects of drug combinations can be due to suppression of cell proliferation, cell death, or a combination. Induction of cell death is preferred to prevent relapse and the emergence of acquired resistance. We found that H358 and SW837 cells only displayed modest evidence of apoptosis following drug monotherapy, but strong synergistic induction of apoptosis, as measured by caspase 3/7 activity, when HRX-0233 and sotorasib were combined. While we could observe a strong synergy for the combination of HRX-0233 and RMC-6236 in DLD1, only modest increase of caspase 3/7 activity could be observed, suggesting a more cytostatic effect in these cells (Fig. 2 D, Lower). Taken together, these data demonstrate a strong synergistic drug combination effect of the MAP2K4 inhibitor HRX-0233 with several KRAS or MAPK inhibitors in a panel of KRAS-mutant lung and colon cancer models, regardless of their tumor lineage or type of KRAS mutation.

To investigate potential off-target effects of HRX-0233, we performed cell viability assays with HRX-0233, sotorasib, or their combination in MAP2K4 wild-type and knockout cells. While the wild-type H358 and SW837 cells showed mild sensitivity to HRX-0233 at 6 µM, the MAP2K4 knockout cells showed significantly less effect. Additionally, concomitant inhibition of MAP2K4 by HRX-0233 showed the enhanced effect of sotorasib or trametinib in the MAP2K4 wild-type cells but not in the MAP2K4 knockout cells (SI Appendix, Fig. S1C). These data support the notion that the observed synergy of the combination is largely due to the on-target effect of HRX-0233.

Pharmacological Inhibition of MAP2K4-JNK-JUN Signaling by HRX-0233 Prevents RTK-Mediated Reactivation of KRAS Signaling.

We next investigated the molecular mechanism underlying the strong synergy between MAP2K4 inhibition by HRX-0233 and KRAS or MAPK inhibitors in KRAS-mutant colon and lung cancer cells. MEK inhibition activates JNK-JUN signaling through suppression of DUSP4 and subsequently leads to elevated transcription of several RTKs, most notably ERRB2 and ERBB3 (9, 27, 34). This induction of ERBB2 and ERBB3 expression in turn reactivates KRAS signaling in the presence of MAPK inhibitors, limiting the sensitivity to these single-agent drug treatments (9, 18). The data shown above indicate that MAP2K4 kinase is essential for feedback activation of JUN and RTKs upon KRAS or MEK inhibition (Fig. 1A) (27). Therefore, pharmacological inhibition of MAP2K4 by HRX-0233 should prevent activation of this feedback loop, resulting in a more complete suppression of KRAS signaling. Transcriptional induction of RTKs following MAPK inhibition takes time in a cell line–specific manner. We therefore performed analysis of signaling effects at two time points: at 6 h when RTK activation is not yet evident and after 48 to 72 h (time to full rebound depends somewhat on the cell line) when RTKs are fully activated by MAPK inhibition (Figs. 1A and 3A and SI Appendix, Fig. S4).

Fig. 3.

HRX-0233 efficiently prevents feedback activation of RTKs upon monotherapy KRAS inhibitors and causes a more sustained and complete inhibition of MAPK signaling (A) H358 and SW837 cells were treated with 25 nM sotorasib, 6 µM HRX-0233, or the combination for 6 and 48 h. DLD1 cells were treated with 100 nM RMC-6236, 6 µM HRX-0233, or the combination for 6 and 72 h. Western blot analysis was performed with the indicated antibodies. Vinculin served as a loading control. (B) Schematic representation of the cross talk between the MEK-ERK and MAP2K4-JNK-JUN signaling pathways. Model of MEK-ERK and MAP2K4-JNK-JUN signaling in monotherapy RAS inhibition by either sotorasib or RMC-6236 treatment (Left) or concomitant inhibition of MAP2K4 by HRX-0233 (Right) in KRAS-mutant cancer cells.

Indeed, while monotherapy sotorasib, RMC-6236 or trametinib induced phospho-JUN (p-JUN) upregulation most prominently after 48 to 72 h, cells cotreated with HRX-0233 failed to activate JUN, indicating effective suppression of the MAP2K4-JNK-JUN feedback loop by HRX-0233 upon KRAS or MAPK inhibition (Fig. 3A and SI Appendix, Fig. S4). Consistently, upregulation of total levels as well as the active, phosphorylated form of ERBB2/3 (p-ERBB2/3) was attenuated by HRX-0233 treatment. Additionally, monotherapy HRX-0233 decreased both p-JUN and total JUN levels in H358, SW837, DLD1, and H2122 cells, further supporting effective inhibition of MAP2K4-JNK-JUN signaling (Fig. 3A and SI Appendix, Fig. S4). We next evaluated whether combined KRAS-MAP2K4 inhibition could effectively sustain inhibition of MAPK signaling. Single-agent drug treatment with sotorasib in H358 and SW837 or RMC-6236 in DLD1 efficiently suppressed MAPK signaling after 6 h of treatment as evidenced by decreased ERK phosphorylation levels as well as down-regulated phosphorylation of its direct downstream target RSK. As expected, at later time points of monotherapy drug exposure, we observed a rebound of phospho-ERK and phospho-RSK levels (Fig. 3A). This feedback activation of MAPK signaling was suppressed by concomitant treatment with HRX-0233, as judged by a more sustained inhibition of phosphorylated ERK and phosphorylated RSK after 48 to 72 h compared to the sotorasib and RMC-6236 monotherapies. Similar results were obtained with the MEK inhibitor trametinib in H358, DLD1, and H2122 cells (SI Appendix, Fig. S4). Importantly, HRX-0233 treatment did not affect the MAPK-ERK pathway by itself as drug treatment showed no effect on phospho-ERK or phospho-RSK levels, suggesting that the more complete MAPK inhibition in the combination is due to the prevention of reactivation.

High levels of phosphorylated ribosomal protein S6, a marker of mTOR as well as MAPK activity, have previously been described to cause resistance to apoptosis in BRAF^V600E-mutant melanoma cells treated with RAF and MEK inhibitors (35). This is due to the enhanced activity of JNK-JUN signaling upon MAPK inhibition, which in turn prevents complete inhibition of S6 phosphorylation to protect cells from apoptosis (26). Moreover, several earlier studies have described that compensatory activation of PI3K-AKT-mTOR pathway signaling, which is also downstream of KRAS, can lead to adaptive resistance to MAPK inhibition (36–38). Therefore, we asked how combining MAP2K4 inhibition with KRAS or MAPK inhibitors affects phosphorylated S6 protein in KRAS-mutant lung and colon cancer cells. While monotherapy sotorasib, RMC-6236 or trametinib treatment were able to decrease phospho-ERK and phospho-RSK levels after 6 h, the cells maintained higher levels of phosphorylated S6. The concomitant addition of HRX-0233 to these KRAS or MAPK inhibitors caused a more complete inhibition of S6 phosphorylation, especially at later time points in all four cell lines tested (Fig. 3A and SI Appendix, Fig. S4). Together, these data indicate that combining MAP2K4 inhibition by HRX-0233 with either sotorasib, RMC-6236 or trametinib efficiently prevents feedback activation of RTKs and thereby caused a more sustained and complete inhibition of signaling downstream of RAS (Fig. 3B), as judged by suppression of both phospho-ERK and phospho-S6. This provides a possible mechanistic explanation for the observed drug synergy.

The Combination Therapy of KRAS^G12C and MAP2K4 Inhibition Is Effective In Vivo.

Since we observed that MAP2K4 and RAS inhibitors synergize in KRAS-mutant cancer cells in vitro, we next sought to validate our findings in vivo. We performed an in vivo study in mice xenografted with human H358 lung tumors treated with HRX-0233, sotorasib, or the combination of both drugs for a period of 60 d. The combination of sotorasib and HRX-0233 resulted in tumor shrinkage over prolonged time, whereas single-agent sotorasib treatment only elicited suppression of tumor growth (Fig. 4 A, Left and 4B). HRX-0233 monotherapy showed no effect on tumor growth compared to vehicle. In addition to the reduction of tumor size, immunohistochemistry analysis showed a significant decrease in the proliferation marker Ki67 after combination treatment compared to the other treatment groups (Fig. 4 C and E). Moreover, combination treatment did not cause weight loss, indicating good tolerability (Fig. 4 A, Right).

Fig. 4.

MAP2K4 and KRAS^G12C inhibition causes tumor shrinkage without toxicity in a H358 xenograft model (A) (Left) H358 cells were grown as tumor xenografts in NOD-SCID IL-2Rg(null) (NSG) mice. A total of 5 × 10 exp6 H358 cells were subcutaneously injected into the right flank. After tumor establishment (200 to 250 mm³), mice were randomized and treated with either vehicle, HRX-0233 (250 mg/kg), sotorasib (10 mg/kg), or their combination for the indicated period of time. The tumor size was measured three times per week. Data are shown as mean ± SEM. N = 5 mice per group. (Right) Body weight variation of the H358 xenografts shown in (A) across the experiment. Data are shown as mean ± SEM. (B) Waterfall plot showing the percentage change in tumor volume (relative to initial volume at treatment start) for individual mice following 33 d of continuous treatment with indicated drugs. According to the RECIST criteria, an increase in tumor volume >30% indicates progressive disease and a decrease <30% indicates stable disease, and a tumor volume decrease >30% indicates partial response. (C) Percentage of Ki67-positive cells quantified in tumors of mice treated with vehicle, HRX-0233 (250 mg/kg), sotorasib (10 mg/kg), or with their combination. Results represent mean ± SD. ∗P = 0.0114, ∗∗P = 0.0018, ∗∗∗∗P < 0.0001, as determined by ordinary one-way ANOVA. (D) Quantitative analysis of ERBB2 staining according to the H-score in tumors of mice treated with vehicle, HRX-0233 (250 mg/kg), sotorasib (10 mg/kg), or with their combination. Results represent mean ± SD. ns P > 0.05, as determined by ordinary one-way ANOVA. (E) Representative images of hematoxylin and eosin (H&E), Ki67, and ERBB2 staining in H358 xenografts. (Scale bars represent 100 µM.)

We showed previously that genetic loss of MAP2K4 confers sensitivity to MEK inhibition in vivo (27). As we have also demonstrated that combining the MEK inhibitor trametinib and HRX-0233 synergized in vitro, we next sought to test the combination in vivo. While trametinib alone was already partially effective in suppression of tumor growth in a H2122 xenograft model, the combination resulted in a more sustained tumor suppression over a time period of 60 d (SI Appendix, Fig. S5 A, Left). This is further supported by the proliferation marker Ki67, which did not show a significant difference between trametinib monotherapy and vehicle, whereas the combination did show a significant decrease in Ki67 staining, indicating more profound proliferation arrest when the two drugs were given together (SI Appendix, Fig. S5 B and C). The combination was also well-tolerated as evidenced by body weight measurements across the experiment (SI Appendix, Fig. S5 A, Right).

In order to validate in vivo the proposed mechanism involving feedback upregulation of RTKs following RAS pathway inhibition, and how this is short-circuited by HRX-0233 in the combination regimen, we performed IHC analysis of ERBB2 (Fig. 4 D and E and SI Appendix, Fig. S5 B and D). Consistent with the in vitro results, both sotorasib (in H358) and trametinib (in H2122) monotherapies induced an evident increase in ERBB2 expression, as judged by H-score quantification (Fig. 4D and SI Appendix, Fig. S5E, see Materials and Methods). Interestingly, HRX-0233 alone was already able to induce a decrease in ERBB2 staining as compared to vehicle-treated animals in both models. Moreover, in H2122 xenografts that received the combination, HRX-0233 was able to counteract the increase in ERBB2 induced by trametinib, resulting in a significantly decreased H-score. In H358, a similar trend was observed for the combination of sotorasib and HRX-0233, although statistically not significant. This difference could be due to the stronger proapoptotic effect elicited by the combination with sotorasib in H358 xenografts, as compared to the combination with trametinib in H2122 xenografts, which is reflected both in the smaller size of the residual tumors and by the increased presence of necrotic tissue in the H358 xenografts treated with the combination. We suspect that, in the case of H358 xenografts treated with the combination of sotorasib and HRX-0233 until end point, the vast majority of tumors cells which responded to the therapy had already undergone apoptosis and either be cleared or became part of the necrotic tissue. It is possible that the surviving cells at this point represent a resistant population that maintained a high expression of ERBB2. However, identifying potential resistance mechanisms specific for this combination requires further research.

Overall, these results establish that the highly synergistic anticancer effect of concomitant inhibition of KRAS^G12C and MAP2K4 results in a strong tumor shrinkage in vivo. Additionally, the combination of trametinib and HRX-0233 also showed a more durable tumor suppression over prolonged time compared to the monotherapies in another NSCLC xenograft model.

Discussion

Recent clinical data have shown that single-agent sotorasib, a KRAS^G12C inhibitor, provided no overall survival (OS) benefit for patients with KRAS^G12C-mutant NSCLC compared to standard-of-care chemotherapy, despite initial promising on target activity and progression-free survival (PFS) benefit (4). Unfortunately, these KRAS inhibitors suffer from the same issue of limited clinical benefit as seen with many other monotherapies that act in the MAPK pathway in KRAS-mutant cancers. In the setting of KRAS^G12C-mutant colon cancer, initial PFS benefit was too modest to warrant clinical use due to feedback reactivation of EGFR. Indeed, adding the EGFR blocking antibody panitumumab to sotorasib appears to deliver some clinical benefit, indicating that combination therapies are required to improve responses to KRAS inhibition (17).

Our previous work has shown that genetic inactivation of MAP2K4, also known as MKK4, sensitizes KRAS-mutant cancers to MEK and ERK inhibitors (27). At that time, we could only use this finding by exploiting inactivating mutations or deletions of MAP2K4 as a potential DNA-based biomarker strategy to identify patients that are most likely to respond to MAPK inhibition, as there were no small-molecule inhibitors of MAP2K4 available. However, with the recent development of the small-molecule MAP2K4 inhibitor HRX-0233, we can now exploit the utility of this finding in a therapeutic setting to target KRAS-driven cancers with novel synergistic drug combinations. We show here that pharmacological inhibition of MAP2K4 by the novel inhibitor HRX-0233 is highly synergistic with several KRAS and MAPK pathway inhibitors across a panel of KRAS-mutant lung and colon cancer cells by disabling the feedback reactivation of KRAS signaling, which limits drug responsiveness. Moreover, our data also indicate that the combination of HRX-0233 and the KRAS^G12C selective inhibitor sotorasib is highly synergistic and resulted in strong tumor shrinkage in H358 KRAS^G12C-mutant NSCLC in vivo, without any apparent toxicity.

MAP2K4 inhibitors were developed to enhance liver regeneration as it was reported that MAP2K4 silencing by small hairpin RNAs (shRNAs) increased the regenerative capacity of hepatocytes due to the increased regenerative capacity of hepatocytes and thus suggested pharmacological MAP2K4 inhibition as a powerful strategy for the treatment of acute and chronic liver diseases (39). Importantly, even though enhanced proliferation and resistance to cell death are characteristics of cancer, no tumors were observed in mice with stable intrahepatic MAP2K4 knockdown after 12 mo. Moreover, our data also demonstrate that pharmacological inhibition of MAP2K4 in mice did not increase tumor growth as monotherapy and resulted in strong synergy in combination with KRAS^G12C inhibition as evidenced by massive tumor shrinkage. Therefore, this suggests that MAP2K4 inhibition monotherapy is unlikely to contribute to tumor initiation or progression in this context. It is worth noting however that some cancers have homozygous deletion of MAP2K4 (40).

It is well-established that rational and powerful combinations can overcome intrinsic drug resistance and the emergence of acquired resistance, thereby improving the often short-lived responses elicited by single-agent therapies and extending the OS benefit of patients (41). Nevertheless, such powerful combinational therapies are frequently accompanied by significant toxicity, causing the failure of many clinical trials despite encouraging preclinical results. In the present study, we found the combination of HRX-0233 and sotorasib to be very well-tolerated in mice, while at the same time resulting in massive tumor shrinkage. However, it has been proven difficult to predict treatment-related toxicities of drug combinations based on animal experiments. For instance, combining pan-HER inhibitors and MEK inhibitors for the treatment of KRAS-driven tumors was highly synergistic and well-tolerated in mice, but the three subsequently conducted clinical trials were discontinued due to frequent treatment-related adverse events (19–21). This is also prevalent when combining MAPK and PI3K-AKT pathway inhibitors, leading to the failure of several clinical trials targeting both pathways (23–25, 42). These observed combination toxicities can often be attributed to overlapping toxicity profiles of the individual drugs, also referred to as supra-additive toxicities (42–44). Recently, another small-molecule inhibitor of MAP2K4, HRX-0215, entered in a phase I clinical trial (EUDRA-CT No. 2021-000193-28) and reported that HRX-0215 was well-tolerated with no treatment-related adverse events being observed at increasing doses in healthy volunteers. These clinical results of pharmacological inhibition of MAP2K4 by HRX-0215 in healthy volunteers could suggest that the structurally related HRX-0233 might have a similar favorable toxicity profile. This makes it a potentially attractive compound for combination drug treatment strategies by minimizing toxicity in the combination while maintaining efficacy. However, we cannot exclude unexpected toxicities specific for this combination and the safety and tolerability must be carefully assessed in a clinical setting. We observed, however, strong synergy at relatively low doses of sotorasib in vitro, suggesting that we could lower the dose in patients to prevent off-target effects of sotorasib and reduce potential combination toxicities while maintaining efficacy (10, 35). Therefore, our data provide an appealing therapeutic approach for the combination of HRX-0233 and sotorasib to target KRAS^G12C-driven lung and colon cancers.

Our data also show the combination potential of HRX-0233 beyond just KRAS^G12C inhibitors, as we also observed synergy with MEK and ERK inhibitors as well as with the pan-RAS inhibitor RMC-6236 across a panel of lung and colon cancer cells harboring different KRAS mutations. A recent phase I study demonstrated that the pan-RAS inhibitor RMC-6236 exhibited an acceptable toxicity profile across dose levels (45). Additionally, the study reported an objective response in KRAS-driven tumors, harboring a variety of different KRAS mutations. Nevertheless, this single-agent pan-RAS inhibitor will most likely also suffer from the challenge of resistance and lack of OS benefit as illustrated by the recent report showing this issue for KRAS^G12C monotherapy (4). Therefore, by combining MAP2K4 and pan-RAS inhibition we could enhance drug effectiveness and potentially minimize toxicity by a lower dose than in monotherapy.

In conclusion, our data identify a strongly synergistic drug combination with a potentially more favorable safety profile that consists of a combination of a MAP2K4 inhibitor with one of several different MAPK or KRAS pathway inhibitors. Given the poor clinical responses to single-agent targeted anticancer drugs and the unmet need for treatment options in KRAS-mutant cancers, this strategy shows potential as a combinational drug therapy for the treatment of KRAS-driven lung and colon cancers.

Materials and Methods

Cell Lines and Culture.

The cancer cell lines H358, H2122, A549, DLD1, and HCT116 were obtained from ATCC. SW837 was a kind gift from Alberto Bardelli (Laboratory Molecular Oncology at the Candiolo Cancer Institute IRCCS - Candiolo (Torino)). All cell lines were cultured in RPMI 1640 medium (Gibco) supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin (Gibco), and 2 mM L-glutamine (Gibco) at 37 °C and 5% CO₂. Mycoplasma contamination was regularly tested via a PCR-based method, and all cell lines were authenticated by short-tandem-repeat DNA profiling (by Eurofins Genomics).

Compounds.

Sotorasib (S8830) and SCH772984 (S7101) were purchased from Selleck Chemicals. RMC-6236 (C-1418) was purchased from ChemGood. Trametinib (201458) was purchased from MedKoo Biosciences. HRX-0233 was provided by HepaRegeniX GmbH.

Western Blotting and Antibodies.

After the indicated culture period and drug treatments, cells were washed with cold phosphate-buffered saline (PBS) and lysed with RIPA buffer (25 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, and 0.1% SDS) supplemented with complete protease inhibitor (Roche) and phosphatase inhibitor cocktails II and III (Sigma). Lysates were cleared by centrifugation for 30 min at 15,000 × g at 4 °C and quantified by performing a Bicinchoninic Acid assay (Pierce BCA, Thermo Fisher Scientific), according to the manufacturer’s instructions. Protein samples were denatured with DTT followed by 5 min heating at 100 °C. About 10 to 30 µg of protein samples was loaded and run (SDS-PAGE) for approximately 1 h at 150 V followed by transfer from the gel to a nitrocellulose membrane at 350 mA for 120 min. Then, membranes were placed in 5% nonfat dry milk in TBS with 0.1% Tween-20 (TBS-T) as a blocking solution for 1 h. Subsequently, membranes were incubated overnight at 4 °C with the indicated primary antibodies 5% BSA (Bovine Serum Albumin) in TBS-T. Membranes were then washed three times for 5 min with TBS-T, followed by 1-h incubation at room temperature with anti-rabbit or anti-mouse secondary antibodies (HRP conjugated) in 5% BSA in TBS-T. Membranes were again washed three times for 5 min in TBS-T. Finally, a chemiluminescence substrate (ECL, Bio-Rad) was added to the membranes and the signal was imaged using the ChemiDoc-Touch (Bio-Rad).

Primary antibodies against Phospho-p44/42 MAPK (Erk1/2) Thr202/Tyr204 (#4377), p44/42 MAPK (Erk1/2) (#4695), Phospho-Jun Ser73 (#3270), Jun (#2315), Phospho-RSK Thr359 (#8753), RSK (#8408), Phospho-S6 Ribosomal Protein Ser235/236 (#2211), S6 Ribosomal Protein (#2217), Phospho-HER3/ErbB3 Tyr1197 (#4561), DUSP4/MKP2 (#5149), and MAP2K4 (#9152) were used at a concentration of 1:1,000 and purchased from Cell Signaling Technology. Primary antibodies against Vinculin (#V9131), ErbB-3/HER-3 (#05-390), Phospho-erbB-2/HER-2 Tyr1248 (#06-229), and c-ErbB2/c-Neu (#OP15L) were used at 1:1000 dilution and purchased from Sigma-Aldrich.

Crystal Violet Long-Term Viability Assays and Bliss Synergy Scores.

Cells were cultured and seeded into 24-well plates at a density of 5,000 to 20,000 cells per well, depending on the growth rate. After 24-h incubation, drugs were added in a 5 × 3 matrix of increasing drug concentrations using the HP D300 Digital Dispenser. Culture media and drugs were refreshed every 2 to 3 d for 10 to 14 d. Cells were fixed with 4% formaldehyde (Millipore #104002) in PBS and stained with 0.1% crystal violet (Sigma #HT90132) diluted in water. Plates were then washed in water and left to dry before scanning. Cell confluence in each well was assessed and quantified using ImageJ software. The quantified crystal violet long-term viability assays were then normalized using the positive and negative controls to express the relative viabilities. The normalized values for each concentration of single drugs or drug combinations from three biological replicates were uploaded to the Synergyfinder online tool (https://www.synergyfinderplus.org/) with viability as a readout to calculate the respective Bliss synergy scores. A Bliss score above 10 suggests synergy.

IncuCyte-Based Cell Proliferation and Apoptosis Assays.

All cell IncuCyte-based proliferation and apoptosis assays were performed in triplicate. Cells were cultured and seeded in 96-well plates at a density of 500 to 1,500 cells per well, depending on growth rate. After 24-h incubation, drugs and the IncuCyte® Caspase-3/7 green apoptosis assays reagent (Essen Bioscience 4440) were added using the HP D300 Digital Dispenser, and the plates were placed in the IncuCyte®. Culture media and drugs were refreshed twice a week for approximately 10 d. Cells were imaged every 4 h in the IncuCyte ZOOM (Essen Bioscience). Phase-contrast images were collected and analyzed to detect cell proliferation based on cell confluence. Apoptosis was measured based on green fluorescent staining of apoptotic cells using the IncuCyte® Caspase-3/7 green apoptosis assay reagent (Essen Bioscience 4440).

CellTiter-Blue Cell Viability Assay.

Cells were cultured and seeded in 96-well plates at a density of 2,000 cells per well. After 24-h incubation, drugs were added at the indicated concentrations using the HP D300 Digital Dispenser. 10 µM phenylarsine oxide (PAO) and DMSO were used as a positive (100% cell viability) and negative control (0% cell viability), respectively. After 2 d of drug exposure, culture media and drugs were refreshed. At the end of the assay (5 d), resazurin (Sigma #R7017) 20× diluted in culture medium was added, and after approximately 3- to 4-h incubation (depending on the cell line), fluorescence (560Ex/590Em) was recorded using the EnVision (Perkin Elmer).

CRISPR-Cas9 Gene Knockout.

H358 and SW837 cells with stable expression of Cas9 were generated by viral transduction of a lentiCas9-blast vector (Addgene #52962), enhanced with 8 µg/mL polybrene. Cells were selected with 10 µg/mL blasticidin until uninfected control cells were dead. We used the lipofection protocol from Synthego (Arrayed CRISPR Multi-guide Library Lipofection using Cas9-Expressing Cells; Synthego, 2020) to make MAP2K4 KO cell lines (Gene Knockout Kit v2—human—MAP2K4—1.5 nmol).

The 3 sgRNAs for transfection were 1:AUGGCGGCUCCGAGCCCGAG, 2:CCG​GCCCCGTAGGGTCCCCG, and 3:GCCCCCTAGCGCGGCAACCC.

After transfection, the KO score was determined with the ICE analysis from Synthego (https://ice.synthego.com/#/) for the polyclonal population the KO scores were 100% for SW837. This represents different KOs which consist of different deletions. MAP2K4 KO in SW837 and H358 was confirmed by western blotting.

Xenografts.

H358 and H2122 cells were resuspended (5 × 106 cells per mouse) in a 1:1 mixture of RPMI and Matrigel (Corning) and injected subcutaneously into the right flanks of 6- to 8-wk-old NOD-SCID IL-2Rg(null) (NSG) (H358) or Balbc/nude (H2122) mice. Tumor volume was monitored three times a week by digital caliper measurement and calculated by the modified ellipsoidal formula [tumor volume = ½(length × width²)]. Mice were randomized when the tumor reached a volume of approximately 200 to 250 mm³ and treated for a maximum period of 60 d. In this experimental setup, HRX-0233 (250 mg/kg) was dissolved in 0.5% HPMC in 0.1 M Citrate buffer, pH 4 with 1% (v/v) Tween 80 and 1% (w/v) Soluplus, Trametinib (0.25 mg/kg) was dissolved in in DMSO: Cremophor EL: water (1:1:8), sotorasib (10 mg/kg) in in 0.5% HPMC, or a drug combination in which each compound was administered at the same dose and schedule as a single-agent. Control groups were treated with the vehicle alone [0.5% HPMC in 0.1 M Citrate buffer, pH 4 with 1% (v/v) Tween 80 and 1% (w/v) Soluplus + DMSO:Cremophor EL:water (1:1:8), 10 μL/g body weight]. All groups were treated daily by oral gavage. Mice were housed under specific pathogen-free conditions in accordance with the European Directive 2012/63/EU. All animal experiments and care were performed in accordance with the guidelines of institutional committees and European regulations and approved by the local authorities: the animal experiment committee of the Netherlands Cancer Institute and the Dutch Central Authority for Scientific Procedures on Animals.

Immunohistochemistry.

For immunohistochemistry analysis, five-micron paraffin sections obtained from xenograft tumors were stained on a Ventana Benchmark Ultra Staining System (Roche), according to the manufacturer’s instructions, with a primary antibody against Ki67 (ab15580, 1:2,000, Abcam) and a primary antibody against ERBB2 (MA5-14509, 1:100, Thermo Fisher Scientific), and stained with hematoxylin and eosin (H&E) according to standard procedures. Quantitative interpretation and image acquisition of nuclear Ki67 and ERBB2 staining were done with the EVOS® XL Core Imaging System. For ERBB2 staining, the percentage of strong positive, weak positive, and negative cells was evaluated in sections from three mice per condition. The H-score was calculated using the formula (0× % negative cells) + (1× % weak positive cells) + (3× % strong positive cells).

Statistical analysis.

GraphPad Prism was used for the statistical analyses. Statistical significance (P-value) was determined as indicated in the figure legends.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.