SOS1 inhibitor BI-3406 shows in vivo antitumor activity akin to genetic ablation and synergizes with a KRAS^G12D inhibitor in KRAS LUAD

Significance

Our data demonstrate the absence of overt toxicity and the therapeutic efficacy of the specific SOS1 inhibitor BI-3406 when used as a single therapy agent or in combination with the specific KRAS^G12D inhibitor MRTX1133, not only in allograft assays but, more importantly, in an in vivo immunocompetent mouse model of KRAS^G12D -driven lung adenocarcinoma. These observations confirm SOS1 as an actionable therapy target in various RAS-dependent tumors and identify a therapeutic window for BI-3406-mediated SOS1 pharmacological inhibition, either as monotherapy or as a valuable combination partner along with different specific KRAS^mut drugs, that may yield clinical benefit through both reduction of intrinsic tumor burden and impairment of extrinsic protumorigenic contributions of the surrounding tumor microenvironment.

Abstract

We evaluated the in vivo therapeutic efficacy and tolerability of BI-3406-mediated pharmacological inhibition of SOS1 in comparison to genetic ablation of this universal Ras-GEF in various KRAS-dependent experimental tumor settings. Contrary to the rapid lethality caused by SOS1 genetic ablation in SOS2^KO mice, SOS1 pharmacological inhibition by its specific inhibitor BI-3406 did not significantly affect animal weight/viability nor cause noteworthy systemic toxicity. Allograft assays using different KRAS^mut cell lines showed that treatment with BI-3406 impaired RAS activation and RAS downstream signaling and decreased tumor burden and disease progression as a result of both tumor-intrinsic and -extrinsic therapeutic effects of the drug. Consistent with prior genetic evidence and the KRAS^mut allografts assays in immunocompromised mice, our analyses using an in vivo model of KRAS^G12D-driven lung adenocarcinoma (LUAD) in immunocompetent mice showed that single, systemic BI-3406 treatment impaired tumor growth and downmodulated protumorigenic components of the tumor microenvironment comparably to SOS1 genetic ablation or to treatment with the specific KRAS^G12D inhibitor MRTX1133. Furthermore, markedly stronger, synergistic antitumor effects were observed upon concomitant treatment with BI-3406 and MRTX1133 in the same in vivo LUAD mouse model. Our data confirm SOS1 as an actionable therapy target in RAS-dependent cancers and suggest that BI-3406 treatment may yield clinical benefit both as monotherapy or as a potential combination partner for multiple RAS-targeting strategies.

After decades of research, the first direct inhibitors of KRAS^G12C oncogenic mutants were recently approved, changing the treatment landscape for patients carrying tumors with this KRAS allele (1–4). Unfortunately, rapid appearance of intrinsic and acquired resistance due to multiple potential mechanisms was reported (5–8). Within the MAPK pathway, besides secondary KRAS mutations, resistance was shown to be driven by receptor tyrosine kinase upregulation or activation of the wild-type (WT) RAS isoforms (9–11). To enhance durability of response the quest for new anti-RAS cancer therapies is continuing with a focus on KRAS inhibitor combinations with drugs targeting downstream or upstream components of RAS pathway (12–16).

The ubiquitously expressed members of the SOS Guanine Exchange Factor (GEF) family (SOS1 and SOS2) are the most universal and functionally relevant activators of RAS proteins in a variety of biological contexts (12, 13, 17, 18). Initial analyses of constitutive SOS1 and SOS2 knockout (KO) mouse strains showed that SOS1 is essential for embryonic development (19) but SOS2 is dispensable for reaching adulthood in mice (20). Single SOS1 or SOS2 KO is perfectly viable in adult mice, but concomitant SOS1 and SOS2 ablation leads to precipitous death of the animals (21). While these observations at the organismal level pointed to some partial functional redundancy between SOS1 and SOS2, subsequent detailed analyses at the cellular level demonstrated the functional prevalence of SOS1 over SOS2 for cellular proliferation and viability in most contexts (21–23). Specific functional roles of SOS1 or SOS2 have been observed in different physiological cellular conditions (24–26). In preclinical mouse models of cancer, SOS1, but not SOS2, is critically required for the development of DMBA/TPA-induced skin tumors (27), BCR/ABL-driven chronic myeloid leukemia (28, 29), or KRAS^G12D-driven LUAD (30). Moreover, the therapeutic efficacy of targeting RAS/MAPK signaling has also been shown in various preclinical settings by using genetic inactivation or pharmacological inhibition (12, 31–33) further supporting the consideration of SOS1 as a relevant therapeutic target in RAS-dependent tumors.

In this regard, pharmacological inhibitors have been developed in recent years with proven ability to directly block SOS1::RAS interactions (14, 34–38) or to promote degradation of SOS1 protein (39–41). BI-3406 is an orally bioavailable, potent, and selective SOS1 inhibitor (SOS1i), which has no effect on SOS2 (35). It binds to the catalytic domain of SOS1, thus preventing the interaction with GDP-loaded RAS and exchange to GTP-loaded RAS. In SOS2-proficient cells, treatment with BI-3406 results in approximately 50% MAPK pathway downregulation. In RAS-dependent preclinical models, BI-3406 leads to strongest in vitro and in vivo antitumor activity in combination with other inhibitors of RAS signaling, such as KRAS^G12C or MEK inhibitors (6, 35, 42–44).

Here, we performed an extensive evaluation of the systemic impact of BI-3406 using our SOS1/2^WT, SOS1^KO, SOS2^KO, or SOS1/2^DKO genetic mouse models (21). We show that systemic pharmacological inhibition of SOS1 with BI-3406 is well tolerated in SOS1/2^WT mice. Moreover, unlike SOS1 genetic ablation in vivo, BI-3406 shows no deleterious effects in SOS2-null animals. We also compared the impact of BI-3406 with SOS1^KO using in vitro and in vivo models of KRAS mutant cancer. We demonstrate that SOS1i has an antiproliferative effect in these models that is similar to the effect of SOS1 genetic ablation we have recently described (30). We also find that this is accompanied by downmodulation of RAS downstream signaling in the tumor cells and also results in extensive reprograming of the tumor microenvironment (TME). Previous studies have demonstrated that SOS1 inhibition, in combination with indirect (35, 43, 45) or direct (42) inhibitors of RAS signaling, leads to enhanced antiproliferative effects in KRAS-driven tumors. Here, we also show that a similar effect can be achieved when combining SOS1i with a class of allele-specific KRAS inhibitors - KRAS^G12D targeting drugs. Overall, these observations suggest that SOS1i may be good universal combination partners for multiple RAS-targeting strategies.

Results

BI-3406 Treatment Results in Negligible Systemic Effects in Healthy Naïve or SOS2^KO Mice.

Consistent with our previous characterization of constitutive SOS2^KO mice and tamoxifen (TAM)-inducible SOS1^KO (Cre^ERT2; SOS1^fl/fl) mice (21), single SOS1 or SOS2 ablation did not affect the viability of adult mice, but double KO of SOS1 and SOS2 (SOS1/2^DKO) in adult mice resulted in quick death after only about 2 wk of TAM treatment (Fig. 1A). In contrast to concomitant SOS1/2 genetic ablation, continuous oral administration of BI-3406 (50 mg/kg, bid) in adult mice for 26 d did not affect the survival rate nor produce any noticeable external phenotypic changes in SOS1/2^WT animals or in constitutive-null SOS2^KO mice (Fig. 1B).

Fig. 1.

Assessment of in vivo toxicities resulting from genetic ablation or pharmacological inhibition of SOS1 in mice. (A) Kaplan–Meier survival plot of TAM-induced (TAM-containing chow diet starting at 1 mo of age) SOS1/2^WT, SOS1^fl/fl, SOS2^KO (black, pink and blue lines, respectively) and SOS1/2^DKO mice (orange line). (B) Kaplan–Meier survival plot of SOS1/2^WT or SOS2^KO groups orally treated with vehicle (Natrosol; gavage) or BI-3406 (50 mg/kg bid; gavage) for 26 d. (C) Body weight measurements at various timepoints (days 1 to 15) during TAM treatment (starting at 1 mo of age) in SOS1/2^WT, SOS1^fl/fl, SOS2^KO, and SOS1/2^DKO groups. **P < 0.01 vs SOS1/2^WT; ^#P < 0.05 vs SOS2^KO; ^&&&&P < 0.0001 vs rest of groups. Data expressed as mean ± SD. The multiple t test was used. (D) Body weight measurements during BI-3406 or vehicle administration in SOS1/2^WT or SOS2^KO groups. ^###P < 0.001 SOS1/2^WT (vehicle-treated) vs SOS1/2^WT (BI-3406-treated); ****P < 0.0001 SOS2^KO (vehicle-treated) vs SOS2^KO (BI-3406-treated); ^&&P < 0.01 SOS1/2^WT (BI-3406-treated) vs SOS2^KO (BI-3406-treated). Data expressed as mean ± SD. The multiple t test was used. For TAM treatments: SOS1/2^WT (n = 7); SOS1^KO (n = 9); SOS2^KO (n = 8); SOS1/2^DKO (n = 8). Following vehicle (Natrosol) treatment: SOS1/2^WT (n = 10); SOS2^KO (n = 8). Following BI-3406 treatment: SOS1/2^WT (n = 21); SOS2^KO (n = 30).

We also carefully monitored growth and body weight of 1-mo-old mice of the relevant SOS genotypes during genetic or pharmacological disruption of SOS1 (Fig. 1 C and D). Consistent with our previous observations (21) the pattern of body weight progression for the genetically ablated, single-null SOS2^KO or the TAM-induced SOS1^KO mice was virtually identical to that of SOS1/2^WT mice, whereas SOS1/2^DKO mice exhibited a dramatic loss of body weight in comparison to the remaining experimental groups (Fig. 1C). In an independent experiment, control (vehicle-treated) SOS1/2^WT and control SOS2^KO mice showed comparable body weight gain (Fig. 1D).

We next compared various serum biochemical parameters between the different experimental groups (SI Appendix, Tables S1 and S2). Consistent with our previous reports (21) genetic ablation of SOS1/2 resulted in significantly reduced levels of serum albumin and total protein content, elevated levels of various liver-related enzymatic activities, and altered levels of triglycerides, total cholesterol, LDH, and uric acid (SI Appendix, Table S1). In contrast, BI-3406-mediated SOS1 inhibition showed no significant changes in these parameters in SOS1/2^WT mice (SI Appendix, Table S2) and in SOS2^KO (SOS1^WT/SOS2^KO) mice produced only a slight increase of GOT, CPK, and LDH serum levels (SI Appendix, Table S2). Furthermore, the values of the serum-level parameters in BI-3406-treated SOS2^KO mice were much closer to those of TAM-treated SOS1/2^WT mice than to those of TAM-treated SOS1/2^DKO mice (SI Appendix, Tables S1 and S2), thus underscoring the low impact of BI-3406-mediated administration in SOS2^KO mice at the organismal level. Finally, it should be noted that oral administration of the vehicle in the SOS2^KO group did not have any effect on the analyzed biochemical parameters (SI Appendix, Table S3), indicating that the mild alterations observed in serum parameters of the BI-3406-treated SOS2^KO mice were specifically due to the biological action of BI-3406.

We also analyzed the impact of BI-3406 treatment on hemogram parameters in our SOS1/2^WT and SOS2^KO mouse strains at the beginning (0 d), mid-term (13 d), and the end (26 d) of treatment (SI Appendix, Table S4). Comparison of these parameters between SOS1/2^WT and SOS2^KO animals treated with BI-3406 revealed no significant differences at day 13 but at day 26 several slight changes were observed (SI Appendix, Table S4). A slight, although statistically significant, reduction of lymphocytes was observed, likely resulting from an increased percentage of neutrophils, as well as an increase in the total number of large unstained cells and a slight alteration in erythrocyte-related parameters such as hemoglobin, hematocrit, and red cell distribution width (SI Appendix, Table S4). Importantly, comparison of these parameters in BI-3406-treated SOS2^KO mice with those previously measured in SOS1/2^DKO mice (21) showed that the alterations of hematological parameters measured in BI-3406-treated SOS2^KO mice were quantitatively milder than those measured following combined genetic depletion of SOS1/2.

The almost complete absence of systemic effects of BI-3406-mediated SOS1 inhibition was further confirmed by comparing histological preparations of different tissues obtained from necropsies of SOS1/2^WT or SOS2^KO mice after TAM-treatment or oral administration of BI-3406 (SI Appendix, Figs. S1 and S2). Consistent with our previous reports (21), SOS1/2^DKO mice exhibited obvious histological defects in the thymus, spleen, and liver as compared to SOS1/2^WT controls, whereas other organs such as the kidney, heart, or lung did not display any visible morphological alterations (SI Appendix, Fig. S2, Left panels). These results were in concordance with the increase of transaminases (SI Appendix, Table S1). In contrast, SOS1/2^WT mice or SOS2^KO mice treated with BI-3406 did not show obvious histopathological findings (SI Appendix, Fig. S2, Right panels).

Nevertheless, in-depth examination revealed minor structural alterations in the liver and kidney of SOS2^KO mice treated with BI-3406 (SI Appendix, Fig. S1A). In particular, the liver of BI-3406-treated SOS2^KO mice displayed occasional clusters of morphologically altered hepatocytes characterized by smaller and darker nuclei and retracted cytoplasm (SI Appendix, Fig. S1A). These morphological alterations might be related to the slight increase in transaminase levels detected in BI-3406-treated SOS2^KO mice (SI Appendix, Table S2). These findings were not observed in the SOS1/2^WT BI-3406 treated animals (SI Appendix, Fig. S1A). In addition, we detected structural disorganization of the epithelial cells covering the surface of the proximal and distal tubules of the kidneys of BI-3406-treated SOS2^KO mice (SI Appendix, Fig. S1A), although the nephritic glomeruli did not show any observable morphological alterations (SI Appendix, Fig. S1A). These kidney alterations might correlate with the minor increase in CPK levels detected in the serum of BI-3406-treated SOS2^KO mice (SI Appendix, Table S2). Detailed examination of various other organs from these mice, including the pancreas (SI Appendix, Fig. S1A), did not uncover any other morphological alterations that could be linked to BI-3406 treatment of SOS2^KO mice. In addition, the vehicle control (SI Appendix, Fig. S1B) did not show any gross structural alterations in any of the analyzed organs. A detailed histopathologic scoring of the morphological studies performed in different tissues and organs of our BI-3406-treated or untreated WT animals or animals with genetic ablation of SOS1 and/or SOS2 is shown in (SI Appendix, Table S5).

Overall, all the above observations demonstrate that the effects resulting from in vivo pharmacological SOS1 inhibition in SOS2^KO mice were significantly milder than those observed in the SOS1/2^DKO mice generated by genetic ablation of SOS1 in SOS2^KO mice.

Genetic Ablation and Pharmacological Inhibition of SOS1 Result in Similar Functional and Mechanistic Effects in KRAS^mut Cells.

Following the tolerability experiments, we evaluated the functional impact of BI-3406 treatment in cells driven by different KRAS mutants. For this purpose, we evaluated the effects of BI-3406-mediated inhibition of SOS1 versus genetic SOS1 deletion using immortalized MEFs of the relevant SOS genotypes (SOS1/2^WT, SOS1^fl/fl, SOS2^KO, SOS1/2^DKO) (22) expressing several common mutants of human KRAS (G12C, G12D, and G12V) (SI Appendix, Fig. S3A).

We first assessed the impact of BI-3406 treatment on the viability of SOS/KRAS^mut MEFs by performing drug response assays using a 3-d BI-3406 treatment, which, in the SOS1/2^WT genetic background, resulted in a dose-dependent reduction of viability in KRAS^mut cells (SI Appendix, Fig. S4A). Genetic SOS2 depletion further sensitized KRAS^mut cells to BI-3406 treatment, leading to a stronger antiproliferative effect, likely due to complete loss of SOS GEF activity. As expected, SOS1 genetic deletion abrogated the sensitivity to BI-3406, confirming drug selectivity for SOS1 and suggesting the lack of off-target effects of BI-3406 at the cellular level.

To further investigate the impact of SOS1 disruption on KRAS^mut cells, we performed live cell imaging-based proliferation assays of KRAS^mut MEFs treated with SOS1i and/or subjected to single or double SOS1/2 KO (SI Appendix, Fig. S3 B and C). In agreement with the drug response assays, the genetic disruption of SOS1 caused a mild, but significant decrease in the proliferative ability of the immortalized KRAS^mut cells (SI Appendix, Fig. S3B). Interestingly, pharmacological inhibition of SOS1 exerted a more potent antiproliferative effect than SOS1^KO, with a milder phenotype in KRAS^G12D-driven cells compared to either KRAS^G12C- or KRAS^G12V-driven cells (SI Appendix, Fig. S3C). Of note, BI-3406 treatment had an enhanced antiproliferative effect in KRAS^mut cells in the absence of SOS2 (SOS2^KO/KRAS^mut cells) (SI Appendix, Fig. S3C). As expected, genetic depletion of SOS1 with 4-hydroxytamoxifen (4OHT) treatment, alone or in combination with SOS2^KO, rendered cells insensitive to BI-3406 treatment.

Immunoblotting analysis of KRAS^mut cells revealed a reduction in ERK phosphorylation upon BI-3406 treatment across the different genotypes, with a more pronounced effect in KRAS^mut/SOS2^KO cells, again suggesting functional redundancy or a partially compensatory role of SOS2 upon SOS1 inhibition (http://www.pnas.org/lookup/doi/10.1073/pnas.2422943122#supplementary-materialsSI Appendix, Fig. S4B). Of note, genetic deletion of SOS1 did not show any significant pERK reduction, likely due to prolonged 4OHT exposure (>10 d) allowing for rebound mechanisms to occur before the start of our assays. Our overall comparison of the different KRAS^mut cell lines tested pointed to a stronger, more pronounced impairment of ERK activation in the KRAS^G12C MEFs than in the other two mutant genotypes (SI Appendix, Fig. S4B). In contrast, no consistent changes in Akt or S6 phosphorylation levels were observed in any of the genotypes or experimental conditions tested (SI Appendix, Fig. S4B).

We next evaluated the differential abilities to respond to EGF stimulation displayed by our SOS1/2^WT and SOS-less KRAS^mut MEFs after undergoing genetic or pharmacological ablation of SOS1 (SI Appendix, Fig. S5). Active RAS pull-down assays revealed that individual genetic disruption of SOS1 or SOS2 already resulted in sizeable reduction of RAS-GTP formation as compared to SOS1/2^WT cells, but only the concomitant ablation of SOS1 and SOS2 resulted in almost complete loss of RAS-GTP upon EGF stimulation (SI Appendix, Fig. S5A). Consistent with the cell proliferation assays (SI Appendix, Fig. S3C), these results also point to the dependency on SOS1 or SOS2 genetic depletion of the KRAS^mut cell lines with regard to RAS-GTP formation and activation of RAS downstream signals (pERK, pAkt) in response to external EGF treatment (SI Appendix, Fig. S5A).

Importantly, our evaluation of the impact of BI-3406 on the levels of RAS activation and downstream RAS/MAPK signal transmission in SOS1/2^WT immortalized MEFs harboring different human KRAS^mut forms also demonstrated that BI-3406 treatment resulted in a stronger reduction of RAS-GTP levels as well as pERK and pAkt levels in KRAS^G12C- and KRAS^G12D-expressing MEFs as compared to cells harboring KRAS^G12V mutation (SI Appendix, Fig. S5B). Since only one MEF line was used here for each genotype, further work using different cell lines will be required in future to confirm this observation. In any event, it should be noted that the three oncogenic variants reached comparable levels of SOS1 protein ablation upon 4OHT administration (SI Appendix, Fig. S6A).

In summary, our results demonstrated that BI-3406-mediated pharmacological inhibition of SOS1 exhibited comparable antiproliferative properties to those observed upon genetically mediated SOS1 ablation in oncogenic KRAS mutated MEFs.

Genetic or Pharmacological Inhibition of SOS1 Impairs Intrinsic Tumor Growth and Downmodulates Extrinsic Protumorigenic Signals in KRAS^G12C and KRAS^G12D Allografts.

We next compared the in vivo impact of genetic (4OHT-induced) versus pharmacological (BI-3406) inhibition of SOS1 on the growth of mouse allografts of KRAS^mut MEFs. For this purpose, immunocompromised mice were implanted with either SOS1^fl/fl/KRAS^G12C or SOS1^fl/fl/KRAS^G12D MEFs, and randomized for treatment with vehicle, 4OHT, or BI-3406.

In KRAS^G12C allografts, 4OHT-mediated genetic depletion of SOS1 resulted in significant impairment of tumor growth, as assessed by the markedly reduced area and volume of the tumor allografts in comparison to the vehicle-treated samples (Fig. 2 A and B and SI Appendix, Fig. S7A). In this model, pharmacological inhibition of SOS1 (BI-3406-treated) also resulted in a stronger antitumor effect than genetic ablation with 4OHT in SOS1^KO mice. Of note, the efficiency of SOS1 ablation following in vivo 4OHT administration was lower (~50%) in comparison to 4OHT-treated oncogenic MEFs (~80%) (SI Appendix, Fig. S6 A and B). Characterization of the tumor allograft explants from these assays by means of immunohistochemistry (SI Appendix, Fig. S7) or by immunoblotting (SI Appendix, Fig. S8A) also showed a direct correlation of the observed antitumor effect of genetic or pharmacological ablation of SOS1 with the reduced levels of cell proliferation and ERK phosphorylation detected in the corresponding tumor explants (Fig. 2B and SI Appendix, Fig. S7 B and C).

Fig. 2.

Allograft assays of (genetically or pharmacologically) SOS1-disrupted KRAS^G12C MEFs. (A) Immortalized SOS1^fl/fl MEFs expressing exogenous KRAS^G12C were injected subcutaneously into nude mice. Once tumors reached a size of 200 mm³, animals were treated with vehicle (gavage), 4OHT (80 mg/kg; i.p), or BI-3406 (100 mg/kg, once daily; gavage) for 12 d. The mean fold-change in tumor volume relative to initial tumor volume is shown. The arrow indicates treatment starts. Waterfall plots of individual tumor responses from vehicle, 4OHT, or BI-3406-treated are depicted for day 12 and day 14. n = 5 (vehicle-treated); n = 9 (4OHT-treated); n = 10 (BI-3406-treated). The bar graph represents tumor volume fold change relative to treatment start. Error bars represent mean ± SD. **P < 0.01, ***P < 0.001 and ****P < 0.0001. One-way ANOVA was used. The panel includes representative images of SOS1^fl/fl/KRAS^G12C allografts at day 16 after tumor implantation treated (starting at day 4; arrow) with vehicle, 4OHT, or BI-3406. (B) Quantitation of total tumor area, the percentage of Ki67-positive cells, the percentage of pERK-stained area with respect of total tumor area, the percentage of CD68-stained area, the percentage of collagen-stained area, the percentage of SMA-stained area, the number of CC3-positive cells per 20X field and the percentage of CD31-positive area. n = 5 tumors per group (for H&E studies) and n = 3 tumors per group (for the rest assays). Data shown as mean ± SD. *P < 0.05, **P < 0.01 vs vehicle-treated group; ^#P < 0.05 vs BI-3406-treated group. One-way ANOVA and Tukey’s test. CC3: cleaved-caspase 3; SMA: smooth muscle actin. n.s: not significant.

Furthermore, the immunohistochemical analysis of these allograft explants also clearly showed that, in comparison with vehicle-treated counterparts, both the inhibition of SOS1 by BI-3406 or the 4OHT-induced genetic ablation of SOS1 caused significant downmodulation of protumorigenic components of the TME (Fig. 2B and SI Appendix, Fig. S7). In particular, CD68-positive tumor-associated macrophages and SMA-immunoreactive cancer-associated fibroblasts (CAFs) were significantly reduced in the stromal TME of 4OHT-treated and BI-3406-treated allograft explants in comparison with vehicle-treated animals (Fig. 2B and SI Appendix, Fig. S7 D–F). Cleaved-caspase 3 (CC3) immunohistochemical assays of the allograft explants also showed that the BI-3406-treated samples displayed higher levels of cell death than the corresponding 4OHT-treated and vehicle-treated samples (Fig. 2B and SI Appendix, Fig. S7G). Finally, as previously described (30), CD31 immunostaining demonstrated that SOS1 ablation did not quantitatively impact the intratumoral vasculature (Fig. 2B and SI Appendix, Fig. S7H).

We next evaluated the in vivo consequences of genetic/pharmacological SOS1 inhibition on mouse allografts harboring the oncogenic KRAS^G12D mutation. As observed in KRAS^G12C allografts, our results demonstrated that blocking SOS1-mediated signaling, both genetically and with BI-3406, resulted in significantly diminished tumor progression (Fig. 3 A and B and SI Appendix, Fig. S9A), which correlated with a reduction in cell proliferation and ERK activation in this model (Fig. 3C and SI Appendix, Figs. S8B and S9 B and C). Moreover, SOS1 inhibition also attenuated the response in the KRAS^G12D-related TME (Fig. 3C and SI Appendix, Fig. S9 D–F).

Fig. 3.

Allograft assays of (genetically or pharmacologically) SOS1-disrupted KRAS^G12D MEFs. (A) Immortalized SOS1^fl/fl MEFs expressing exogenous KRAS^G12D were injected subcutaneously into nude mice. Once tumors reached a size of 200 mm³, animals were treated (arrow) with vehicle (gavage), 4OHT (80 mg/kg; i.p), BI-3406 (100 mg/kg, once daily; gavage), MRTX1133 (30 mg/kg/day; i.p), or combo (BI-3406+MRTX1133; gavage/i.p) for 12 d. Data shown as mean ± SD. One-way ANOVA was used. **P < 0.01 and ****P < 0.0001. n = 5 (vehicle-treated); n = 9 (4OHT-treated); n = 7 (BI-3406-treated); n = 6 (MRTX1133-treated); n = 5 (combo-treated). The mean fold-change in tumor volume relative to initial tumor volume is shown (Left side). The arrow indicates treatment start. The bar graph represents tumor volume fold change relative to treatment start. *P < 0.05 and ****P < 0.0001. One-way ANOVA was used. (B) Waterfall plots represented as Log₂ of the fold change of individual tumor responses from 4OHT, BI-3406, MRTX1133, or combo-treated are depicted for day 12 and day 14. Representative images of SOS1^fl/fl/KRAS^G12D allografts at day 16 after tumor implantation treated (starting at day 4; arrow) with vehicle, 4OHT, BI-3406, MRTX1133, or combo (Right side). *P < 0.05; ****P < 0.0001 vs vehicle-treated group. (C) The bar charts show total tumor area, the percentage of Ki67-positive cells, the percentage of pERK-stained area with respect of total tumor area, the percentage of CD68-stained area, the percentage of collagen-stained area, the percentage of SMA-stained area, the number of CC3-positive cells per 20X field and the percentage of CD31-positive area. n = 5 tumors per group (for H&E studies) and n = 3 tumors per group (for the remaining assays). Data shown as mean ± SD. One-way ANOVA and Tukey’s test. **P < 0.01, ***P < 0.001 and ****P < 0.0001 vs MRTX1133-treated group; ^&P < 0.05 vs 4OHT-treated group; ^#P < 0.05 vs BI-3406-treated group. (D) Growth curves of KRAS^G12D-mutated KPB6, LKR10 or LKR13 cell cultures at 24 h and 48 h of treatment, individually, or combined, with BI-3406 (1 μM) and MRTX1133 (5 nM). DMSO was used as vehicle. n = 5 independent experiments per group. Data shown as mean ± SD. One-way ANOVA and Tukey’s test. ****P < 0.0001 vs drug-treated LUAD cells; ^###P < 0.001 vs combo-treated cells; ^&P < 0.05 vs MRTX1133-treated cells. (E) Bar chart showing the relative levels of RAS-GTP, measured by using RAS G-LISA assay, in extracts of serum-starved (ST), vehicle-pretreated, BI-3406-pretreated, MRTX1133-pretreated, or combo-pretreated, for 2 h, KPB6, LKR10 or LKR13 LUAD cells, upon EGF stimulation (100 ng/mL) for 2 min. n = 3 independent samples per group. Data shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. One-way ANOVA and Tukey’s test. CC3: cleaved-caspase 3; SMA: smooth muscle actin. n.s: not significant.

Combination Therapy with BI-3406 Enhances the Antitumor Response to a Direct KRAS Inhibitor.

We next compared the antitumor effect of SOS1 inhibition with the effect of a KRAS^G12D-selective inhibitor (MRTX1133) (46) and analyzed various parameters characterizing intrinsic and extrinsic antitumor responses in our KRAS^G12D allograft mouse model and in established KRAS^G12D cell lines.

MRTX1133 treatment (30 mg/kg) for 12 d resulted in significantly reduced tumor volume in KRAS^G12D allografts, which was comparable, although slightly lower, to the reduction seen with BI-3406 or 4OHT treatments in (Fig. 3 A and B). Of note, MRTX1133 administration had no effect on mouse body weight (SI Appendix, Fig. S9I). Moreover, we also compared the pharmacokinetics (PK) profile in Hsd:Athymic Nude-Foxn1^nu mice after single treatment with 50 mg/kg BI-3406, single treatment with 10 mg/kg of MRTX1133, and coadministration of both compounds (Dataset S1). BI-3406 showed a slight reduction in area under the curve (<30%) after coadministration with MRTX1133 as compared to BI-3406 alone. Exposure of MRTX1133 was unchanged compared to coadministration with BI-3406. In summary, as there was no significant alteration of the PK curve for MRTX1133 after additional treatment with or without BI-3406 a drug–drug interaction can be ruled out suggesting that the added efficacy is due to a combination effect of both compounds.

The combination of BI-3406 (100 mg/kg) with MRTX1133 (30 mg/kg) led to an even deeper reduction in tumor growth of KRAS^G12D allografts than observed with either treatment alone (Fig. 3 A and B), thus demonstrating a significantly enhanced antitumor effect resulting from the concomitant inhibition of KRAS^G12D and SOS1. As expected, the reduction of tumor growth caused by the single or combined drug treatments correlated with corresponding reduction of cell proliferation (Ki67) and ERK activation detected by immunohistochemistry (Fig. 3C and SI Appendix, Fig. S9 B and C) or immunoblotting (SI Appendix, Fig. S8B) in the treated allograft explants. Although not reaching statistical significance differences, in line with tumor growth data, single MRTX1133 administration showed reduced levels of ERK phosphorylation in comparison with 4OHT- or BI-3406-treated groups (Fig. 3C and SI Appendix, Figs. S8B and S9 B and C). Importantly, a more profound impairment of tumoral RAS/MAPK signaling was always observed upon combined BI-3406 and MRTX1133 administration compared to the single drug treatments (Fig. 3C and SI Appendix, Figs. S8B and S9C). In line with the allograft assays, single or combined BI-3406 and MRTX1133 administration in different murine KRAS^G12D-mutated LUAD cell lines (KPB6, LKR10, and LKR13) resulted in a potent inhibitory effect on growth (Fig. 3D) and EGF-dependent RAS activation (Fig. 3E).

As in SOS1-ablated (genetically or pharmacologically) KRAS^G12D allografts, the selective inhibition of KRAS^G12D by MRTX1133 also resulted in downmodulation of protumorigenic TME components such as CAFs, tumor-associated macrophages in comparison to vehicle-treated allografts (Fig. 3C and SI Appendix, Fig. S9 D–F).

Single MRTX1133 treatment led to noticeably increased level of apoptosis (Fig. 3C and SI Appendix, Fig. S9G) in KRAS^G12D allografts at the end of study, whereas tumor samples from animals treated with MRTX1133 and BI-3406 showed similar level of apoptosis to vehicle-treated allografts (Fig. 3C and SI Appendix, Fig. S9G). In contrast, neither SOS1 ablation (upon 4OHT or BI-3406 administration) nor MRTX1133-mediated KRAS^G12D inhibition (single or combined) resulted in a modification in the percentage content of intratumor blood vessels (Fig. 3C and SI Appendix, Fig. S9H).

In summary, BI-3406 potently synergizes with the selective KRAS^G12D inhibitor MRTX1133, significantly enhancing its antiproliferative effects in the tumors.

Therapeutic Efficacy of In Vivo BI-3406 and MRTX1133 Treatment in a Model of KRAS^G12D-Driven LUAD in Immunocompetent Mice.

We have recently demonstrated that genetic SOS1 ablation leads to clear inhibition of tumor development in an immunologically competent KRAS^G12D-driven LUAD mouse model (30). To compare the antitumor effect of genetic versus pharmacologic SOS1 inhibition, we employed the same model here to assess the effect of single BI-3406 or MRTX1133 treatments as well as the potential synergy resulting from combined BI-3406 plus MRTX1133 dual administration (Fig. 4 and SI Appendix, Fig. S10).

Fig. 4.

Impact of BI-3406 and MRTX1133 treatment on KRAS^G12D-driven LUAD progression. (A) Representative images of microCT from 3-mo-old SOS1/2^WT/KRAS^G12D-mutated mice previous drug treatment (pre-treatment) and the very same animal after treatment (posttreatment) with BI-3406 administration (50 mg/kg, bid; gavage, 26 d), MRTX1133-treatment (30 mg/kg, i.p, 17 d), or combo administration (BI-3406, 50 mg/kg, bid; gavage and 30 mg/kg, i.p for 21 d). The graph’s data point corresponding to each individual mouse are identified by a distinctive color in each case. n = 4 mice per genotype (BI-3406-treated); n = 2 mice per genotype (MRTX1133-treated); n = 4 mice per genotype (combo-3406-treated). Data shown as mean ± SD. The paired t test was used. *P < 0.05. (B) Left. Representative images of paraffin-embedded sections from lung lobes of 4-mo-old, SOS1/2^WT/KRAS^G12D-mutated mice treated with vehicle (gavage), BI-3406 (50 mg/kg; gavage; 26 d), MRTX1133 (30 mg/kg; i.p; 17 d), and combo (BI-3406, 50 mg/kg, bid; gavage and 30 mg/kg, i.p for 21 d) stained with H&E. (Scale bars, 1 mm.) Right. Quantitation of percentage of lung tumor area with respect total lung area. One-way ANOVA and Tukey’s test. (C) The bar chart shows the percentage of Ki67-positive cells into the tumor or the percentage of pERK-stained area with respect of total tumor area, the percentage of total tumor area that immunostained for SMA, the percentage of collagen in the tumor, the percentage of CD68-positive cells, the number of CD3-positive cells per 40X field in the tumor, the number of CC3-positive cells per 40X field and the percentage of CD31-stained area in vehicle-treated, BI-3406-treated, MRTX1133-treated, combo-treated samples as detailed in (B) and an additional mouse after 5 d of combo treatment. Data shown as mean ± SD. n = 4/group. *P < 0.05 vs BI-3406-treated group. n = 4/group. (B and C) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Since no tumors were found in the combo-treated group (21 d; n = 4) and because only one animal was measured in the 5 d combo-treated group, both experimental groups have not been considered for the statistical analysis. Data shown as mean ± SD. One-way ANOVA and Tukey’s test.

Similar to the data obtained upon SOS1 genetic depletion (30), the initial analysis using in vivo microCT scans demonstrated a significant shrinkage of lung tumor volume after single BI-3406 or MRTX1133 administration (posttreatment condition) in comparison to the very same animals prior to drug treatment (pretreatment condition). Importantly, this antitumoral effect was even higher when both drugs were concomitantly applied (Fig. 4A). Consistent with this, histological analysis of lungs from KRAS^G12D tumor-bearing animals demonstrated a reduction of lung tumor burden of more than 50% in BI-3406-treated animals as compared to vehicle-treated mice of the same age and genotype (Fig. 4B). Tumor reduction was stronger in MRTX1133-treated mice with only one small tumor found in the lung lobes of a total of three animals (Fig. 4B). Remarkably, complete pathological response was observed in this model following cotreatment of BI-3406 and MRTX1133 as we were not able to detect remaining tumor cells in the explanted tissue (Fig. 4B). Consistent with our prior observations in this animal model upon SOS1 KO (30), the lung tumor shrinkage upon BI-3406 treatment was associated with a noticeable reduction of cell proliferation (Fig. 4C and SI Appendix, Fig. S10A) and ERK phosphorylation (Fig. 4C and SI Appendix, Fig. S10B). Notice that, since mice showed complete pathological response following treatment for 26 d following coadministration with BI-3406 plus MRTX1133, we evaluated biomarker modulation in the combination following only 5 d of treatment (Fig. 4C).

We also explored the potential effects of BI-3406 treatment on modulation of the TME of the KRAS^G12D lung tumors. Consistent with our previous data (30), immunochemical characterization of lung tumor sections derived from the BI-3406-treated mice further confirmed the positive therapeutic impact of SOS1 pharmacological inhibition with BI-3406 by identifying a clear in vivo downmodulation of various protumorigenic components of the TME (Fig. 4C and SI Appendix, Fig. S10 C–E). We observed a modification of the morphology of the SMA-immunopositive CAFs in the tumors as a consequence of BI-3406 treatment, suggesting a BI-3406-dependent reduction of CAF activation (Fig. 4C and SI Appendix, Fig. S10C). Furthermore, the levels of CAF-dependent collagen deposition (Fig. 4C and SI Appendix, Fig. S10D) and CD68-reactive macrophages (Fig. 4C and SI Appendix, Fig. S10E) present in the TME of the tumors were also strongly reduced in the lungs of BI-3406-treated as compared to vehicle-treated mice. No differences in the number of CD3-positive cells were observed upon BI-3406 single administration (Fig. 4C and SI Appendix, Fig. S10F). MRTX1133 administered alone did not alter the number of infiltrating T cells either (Fig. 4C and SI Appendix, Fig. S10F), differing from other recent data carried out an oncogenic KRAS^G12D pancreatic cancer model showing an increase of T cell tumor infiltration following MRTX1133 treatment (47). Only slightly higher (not statistically significant) levels of CC3-positive cells were detected in the BI-3406-treated tumors as compared to vehicle-treated tumors (Fig. 4C and SI Appendix, Fig. S10G), suggesting that the antitumoral effect of BI-3406 against KRAS^G12D-driven LUAD does not involve a significant priming of apoptosis in tumor cells.

Our previous results indicate that single MRTX1133 treatment increases the number of apoptotic cells, whereas the combo treatment resulted in a lower number of CC3+ cells (Fig. 3C). Due to the enhanced efficacy observed in combination, we hypothesize that the combined MRTX1133 plus BI-3406 treatment induces apoptosis early on and therefore, at the time of evaluation at the end of the study, many apoptotic cells could have already been cleared. In this regard, our measurements of intratumor cell death rates in the experiment involving a short-term (5 d) treatment showed an increase in the number of CC3-positive cells at the initial steps of the combination treatment when compared with single MRTX1133-treated mice (Fig. 4C and SI Appendix, Fig. S10G), thus providing support of the above interpretation of the data in Fig. 3. No differences in content of intratumor blood vessels were observed among experimental groups (Fig. 4C and SI Appendix, Fig. S10H). Consistent with the data previously obtained in the PDX models, these results clearly demonstrate a potent antitumoral in vivo effect of BI-3406, exerted through both reduction of intrinsic tumor burden and impairment of extrinsic TME in an immunocompetent KRAS^G12D-mutated mouse model of LUAD.

BI-3406 and MRTX1133 Exhibit Synergistic Therapeutic Activity in KRAS^G12D LUAD Human Cell Lines.

Having determined the antiproliferative effect, both in vitro and in vivo, of BI-3406 in KRAS mutant MEFs and murine cell lines and its strong synergistic effect with the mutant selective KRAS^G12D inhibitor MRTX1133, we next aimed at extending these observations in KRAS^G12D mutant human LUAD cell lines. Single BI-3406 or MRTX113 treatment did not strongly affect the proliferation and viability of A427 and SK-LU-1 human LUAD cell lines (Fig. 5A and SI Appendix, Fig. S11A). However, both compounds exhibited synergistic effects on the viability of both tumor cell lines when concomitantly administered SI Appendix, Fig. S11B). Combined BI-3406 and MRTX1133 resulted in a potent inhibitory effect on growth of those cell cultures (Fig. 5A). Consistently, immunoblotting analysis of human LUAD cells revealed a moderate reduction of GTP-loaded RAS as well as of ERK and Akt phosphorylation upon single BI-3406 or MRTX1133 treatment, with a more pronounced effect when both inhibitors were concomitantly administered (Fig. 5B).

Fig. 5.

Determination of BI-3406 and MRTX113 sensitivity in KRAS^G12D mutant human LUAD cell lines. (A) Growth curves of KRAS^G12D-mutated A427 or SK-LU-1 cells at 12, 24, 36, and 48 h of the treatment, individually, or combined, with BI-3406 and MRTX1133 at the indicated concentrations. DMSO was used as vehicle. n = 5 independent experiments per group. Data shown as mean ± SEM. For A427 cells: **P < 0.01 vs BI-3406-treated (10 µM) cells; ***P < 0.001 vs BI-3406/MRTX1133-treated cells (1 µM/10 nM); ****P < 0.0001 vs BI-3406/MRTX1133-treated (10 µM/100 nM) cells. ^#P < 0.05 MRTX1133 (100 nM) vs BI-3406/MRTX1133-treated cells (1 µM/10 nM) or BI-3406-treated (10 µM) vs BI-3406/MRTX1133-treated (10 µM/100 nM); ^####P < 0.0001 MRTX1133 (100 nM) vs BI-3406/MRTX1133-treated cells (10 µM/100 nM). For SK-LU-1: **P < 0.01 vs BI-3406/MRTX1133-treated (0.01 µM/0.1 nM) cells; ****P < 0.0001 vs BI-3406/MRTX1133-treated (10 µM/100 nM) cells; ^#P < 0.05 MRTX1133 (100 nM) vs BI-3406/MRTX1133-treated cells (0.01 µM/0.1 nM) or BI-3406-treated (10 µM) vs BI-3406/MRTX1133-treated (10 µM/100 nM); ^##P < 0.01 MRTX1133 (100 nM) vs BI-3406/MRTX1133-treated cells (10 µM/100 nM). One-way ANOVA and Tukey’s test. (B) (Left) Representative western blots of active RAS pull-down assays and the corresponding signaling of pERK and pAkt, as well as the expression of SOS1 and SOS2 in each case. Tubulin was used as a loading control. (Right) Bar charts showing the relative levels of RAS-GTP, pERK, and pAkt in extracts of ST, vehicle-pretreated (EGF), BI-3406-pretreated, MRTX1133-pretreated, or combo-pretreated (for 2 h) A427 or SK-LU-1 cells, upon EGF stimulation (100 ng/mL) for 2 min. n = 3 independent samples per group. Data shown as mean ± SD. *P < 0.05, **P < 0.01, ****P < 0.0001; ^&&P < 0.01 vs MRTX1133-treated cells; ^#P < 0.05, ^###P < 0.001 vs BI-3406-treated cells. One-way ANOVA and Tukey’s test.

Discussion

Previous reports have documented the functional importance of SOS1 and SOS2 in normal cell development (21–26) and in the development of various tumor types (27–31, 48) supporting the notion of SOS inhibition as a potentially useful therapeutic approach for RAS-dependent malignancies. However, the rapid lethality of genetic double-null SOS1^KO/SOS2^KO mice and the observation that SOS2 can only provide limited functional compensation for the lack of SOS1 under physiological conditions, warranted further investigation of the potential impact of pharmacological SOS1 inhibition in the presence or absence of SOS2. Here, we evaluated the tolerability and therapeutic efficacy of single or combined SOS1/2 inhibition by comparing genetic or pharmacological inhibition of SOS1 in SOS1/2^WT and SOS2^KO tumor cells and in genetically engineered mouse tumor models. We also compared the antitumor effect of genetic or pharmacological inhibition of SOS1 with a KRAS^G12D-specific inhibitor in monotherapy and in combination with SOS1i.

In stark contrast to the rapid lethality suffered by genetically engineered SOS1/2 double KO mice (21), the inhibition of SOS1 with BI-3406 did not affect animal survival and only exhibited minimal physiological effects in SOS2^KO animals that were not observed in BI-3406-treated SOS1/2^WT mice. Furthermore, our data indicate that pharmacological SOS1 inhibition results in significant intrinsic and extrinsic therapeutic benefits by blocking the growth of the tumor cells and impairing the protumorigenic effect of the TME.

An obvious mechanistic question is the reason for the observed differences between genetic and pharmacological SOS1 inhibition. BI-3406 inhibits the interaction of SOS1 with M, H, N, and KRAS resulting in sensitivity of the tumor cell lines dependent on these GTPases (35, 42). In contrast, SOS1 genetic depletion or SOS1 degradation removing the entire protein, not only disrupts the SOS1::RAS interaction with both oncogenic and WT RAS isoforms (12, 22, 27, 31, 39–41, 48), but likely also affects the interaction between SOS1 and RHO GTPases, such as RAC (28, 49, 50). In this context, genetic SOS1 disruption in SOS1/2^DKO cells resulted in a significant decrease in RAC-GTP levels (22), while an effect on RAC-mediated signaling after pharmacological SOS1 modulation has not yet been studied. Furthermore, BI-3406 is quickly metabolized in the organism (35), whereas SOS1 depletion is sustained over time.

Taken together, while genetic SOS1 depletion has stronger physiological impact in comparison to BI-3406-dependent pharmacologic SOS1 inhibition, it can lead to deleterious biological outcomes, especially in the absence of SOS2, an effect that was not detected in vivo upon BI-3406 administration in healthy, nontumor bearing mice. This could be especially relevant in the context of targeted SOS1 therapies that do not just inhibit but degrade the protein, such as SOS1 PROTACs, which have recently been described (40, 41, 51). In this regard, it could be argued that systemic SOS1 inhibition may exert toxicity in humans. As previously demonstrated, SOS1 and SOS2 show some functional redundancy (21) and therefore the presence of SOS2 in SOS1i-treated patients could attenuate potential harmful consequence of SOS1 inhibition. In addition, our PK studies demonstrated that BI-3406 is rapidly cleared from the plasma, and therefore BI-3406-dependent SOS1 inhibition is not sustained over time, thus reducing the potential deleterious effects of a sustained SOS1 depletion.

Both in the allograft assays with KRAS^mut transformed MEFs and in the in vivo KRAS^G12D LUAD model in immunocompetent mice, the genetic and pharmacologic inhibition of SOS1 resulted in a significant reduction of intrinsic primary tumor growth and cell proliferation that directly correlated with decreased levels of activation of RAS/MAPK signaling pathway, in line with previous observations (12, 30, 38, 42, 52). Here, we found that the strong antitumor effect of BI-3406-mediated SOS1 inhibition was not restricted to the tumor tissue, but also impacted the surrounding stroma. In particular, we observed a significant reduction in CAF activation and in the number and collagen deposition of macrophages recruited to the tumor. Previously, BI-3406 has been shown to modulate macrophages in the TME of NF1-driven neurofibroma (43). These observations are also consistent with our prior data showing that genetic SOS1 depletion in LUAD mouse models reduced the activation of CAFs, tumor-associated macrophages, and T-lymphocytes in the TME (30). Importantly, immune checkpoint blockade is part of the current standard of care for LUAD. Therefore, the potential impact of SOS1i on immune-related tumor progression should be additionally considered to their already demonstrated antiproliferative properties on cancer cells. It should be noted however that these results in LUAD differ from previous studies in models of pancreatic cancer reporting that depletion of CAFs lead to aggressive tumors and poor survival (53, 54). Nevertheless, it has been also reported that oncogenic KRAS promotes CAFs transformations promoting tumor progression and resistance (55–57) and therefore, inhibiting the oncogenic signal might attenuate these protumorigenic effects of CAF population. This discrepancy could reflect tissue-specific or cancer type–specific differences in responses of CAFs following SOS1/KRAS^G12D inhibition. Our results in LUAD also correlate with the previously characterized role of SOS1 in the physiological maintenance/homeostasis of various cell types that may be eventually recruited to the TME, including fibroblasts, macrophages, neutrophils, or lymphocytes (21–24, 26, 27, 58). It could be argued that the decrease of infiltrating CAFs is due to an intrinsic effect of SOS1 inhibition, with this reduction not having a deleterious effect on tumor progression or aggressiveness in our model. Taken together, BI-3406 may elicit a dual antitumor effect through its antiproliferative role in the tumor cells and its potential modulation of the protumorigenic TME, as observed with the selective KRAS^G12D inhibitor (59).

Interestingly, our results showed differential susceptibility of different KRAS mutants to BI-3406 treatment. In particular, BI-3406 exhibited enhanced antiproliferative effects in KRAS^G12C cells, likely because this KRAS allele has a higher intrinsic GTPase activity than other KRAS mutants (60), and therefore is more dependent on SOS1 GEF activity to maintain the GTP-state. Therefore, impairing nucleotide exchange in this context could provide additional benefit together with direct KRAS inhibition and more durable clinical response (42). Interestingly, G12C cells showed incomplete SOS1 depletion even after long exposure to TAM in an SOS2^KO context, indicating positive selection of cells with inefficient Cre-dependent recombination. Consistent with the in vitro studies, SOS1 inhibition (upon TAM or BI-3406 administration) also limited tumor growth in vivo in allograft models expressing G12C or G12D human oncogenic variants of KRAS. In this context, BI-3406 treatment exerted stronger therapeutic effect than TAM exposure, likely due to positive selection of KRAS^mut cells retaining SOS1 as a result of incomplete recombination, similarly to the in vitro setting. Moreover, BI-3406 significantly reduced lung tumor size in a murine model of KRAS^G12D-driven LUAD, confirming previous observations with genetically mediated SOS1 inhibition in the same model (30). Similarly, recent works reported that BI-3406 also exhibited antitumor properties in different allograft models expressing KRAS^G12C, demonstrating a wider therapeutic impact of targeting SOS1 in KRAS-dependent cancers (42).

Despite promising preclinical and clinical activity of selective KRAS^G12C and KRAS^G12D targeting agents (61), the emergence of drug resistance (5–8, 62) is an important concern for most of these new small molecules (15). Combined administration of inhibitors that cotarget upstream or downstream components of the RAS signaling pathway appears to be a promising strategy to enhance the efficacy of RAS inhibitors or to overcome resistance to these compounds (14, 42, 63–65). In fact, to address this challenge, KRAS^G12C inhibitors are currently tested in clinical trials in combination with drugs targeting fundamental nodes in the RAS signaling pathways, including SOS1 (NCT05578092). Genetically, we and others have demonstrated a therapeutic effect of SOS1 depletion in KRAS^G12D-driven tumors, but the effect of combined pharmacological SOS1 and KRAS^G12D inhibition had not been previously evaluated. Here, we demonstrated a potent antitumor effect of the combination of BI-3406 and the selective KRAS^G12D inhibitor MRTX1133, which has recently entered clinical trials in advanced solid tumors (NCT05737706). Overall, our observations strongly support the consideration of SOS1 as an actionable target for pharmacological intervention in the context of RAS-dependent cancers, either as monotherapy or in combination with other drugs.

Finally, the results presented here and previously suggest that combined SOS1/SOS2 (pharmacologic/genetic) inhibition leads to antitumor effects in RAS-driven diseases (12, 27, 30, 32, 48). These observations support further evaluation of SOS2 as a potential therapeutic target for oncogenic processes in vivo, particularly to enhance the therapeutic effect of SOS1 inhibition. In contrast to our genetic model of combined SOS1/2 ablation, pharmacological inhibition of SOS1 in SOS2^KO mice was tolerated, suggesting that a therapeutic window may exist. Hence, the development of new drugs capable of modulating SOS1 and/or SOS2 activity in vivo could be a promising therapeutic avenue for the near future.

Materials and Methods

Animal Models.

Our conditional, TAM-inducible SOS1/2^KO system (21) involving the use of a mouse strain harboring a floxed SOS1 exon flanked by LoxP sites (SOS1^fl/fl) controlled by Cre^ERT2 (Cre recombinase fused to a triple mutant form of the human estrogen receptor, from the endogenous Polr2a locus), as well as constitutive SOS2^KO mice (20), were used to generate mice of the relevant SOS genotypes (SOS1/2^WT, SOS1^KO, SOS2^KO, or SOS1/2^DKO) for these studies. All animals were kept on the same C57BL/6 J background, and maintained under identical experimental conditions. Genotypes were monitored by PCR (21). For SOS1 disruption in living mice, the Cre recombinase was activated by feeding the animals with TAM-containing chow diet (Harlan; Teklad CRD TAM400/CreER). For SOS1 disruption in cell cultures, the active metabolite 4OHT was used.

BI-3406 administration was performed as follows. For one set of animals (SOS1/2^WT, SOS1^fl/fl, SOS2^KO, and SOS1^fl/fl/SOS2^KO (hereafter named SOS1/2^DKO), a TAM-containing chow diet (Harlan; Teklad CRD TAM400/CreER) was administered to 5-wk-old mice (no gender selected) to achieve SOS1 depletion in SOS1^KO and SOS1/2^DKO groups. All experimental groups were TAM-fed to avoid undesirable off-target effects. Another set of 5-wk-old mice with SOS1/2^WT and SOS2^KO genotypes were orally treated (gavage) on a 5 d on/2 days off bid (6 h apart) schedule with vehicle (0.5% Natrosol) or BI-3406 at 50 mg/kg for 26 d as described (35).

For additional in vivo tests, SOS1/2^WT mice were cross-mated with KRAS^LA2 mice spontaneously developing KRAS^G12D-driven LUAD (30, 66). All mice were kept on the same C57BL/6 J background (RRID:IMSR_JAX:000664) and maintained under identical experimental conditions. 3-mo-old mice (no gender selected) were orally treated (gavage) on a 5 d on/2 d off bid (6 h apart) schedule with vehicle (0.5% Natrosol) and BI-3406 at 50 mg/kg for 26 d or with MRTX1133 (30 mg/kg, i.p.) for 17 d. Combo (BI-3406 plus MRTX1133) drug treatment was also performed for 21 d using the same treatment regimen as described for single administration.

For allograft experiments, Crl:NU-Foxn1^nu mice (females, 8-wk-old) were purchased from Charles River. KRAS^G12C;SOS1^fl/fl;SOS2^+/+;Cre⁺ or KRAS^G12D;SOS1^fl/fl;SOS2^+/+;Cre⁺ cells (2 × 10⁶) were injected subcutaneously in the flanks of recipient mice. Once tumors were detectable, mice were randomly assigned to either BI-3406, 4OHT, or vehicle treatment [For KRAS^G12C: n = 5 (vehicle-treated); n = 9 (4OHT-treated); n = 10 (BI-3406-treated) and For KRAS^G12D: n = 5 (vehicle-treated); n = 9 (4OHT-treated); n = 10 (BI-3406-treated); n = 6 (MRTX1133-treated); n = 5 (combo-treated)] and measurements were taken every 2 d using calipers. Body weight was assessed every other day in control mice treated with MRTX1133. BI-3406 (Boehringer-Ingelheim) was resuspended in 0.5% Natrosol (provided by Boehringer-Ingelheim), 1 equimolar HCl (1 M) and administered once daily by oral gavage at a dose of 100 mg/kg. TAM was dissolved in corn oil and administered as intraperitoneal (i.p) injection (100 µL) 4 consecutive days/week at a dose of 80 mg/kg. MRTX1133 (provided by Boehringer-Ingelheim) was dissolved in 10% Captisol in 50 mM citrate buffer pH5.0 and administered as intraperitoneal (i.p) injection every day at a dose of 30 mg/kg. Mice were killed and tumors were resected. Half of the tumors were snap frozen and the second part was fixed in formalin and embedded in paraffin for further analysis.

Mice were kept, managed, and killed in the Nucleus animal facility of the University of Salamanca or at the MBC Animal Facility of the University of Torino, according to current European (2007/526/CE) and Spanish (RD 1201/2005 and RD53/2013) legislation in accordance with the guideline for Ethical Conduct in the Care and Use of Animals as stated in The International Guiding Principles for Biomedical Research Involving Animal. The mice were housed in cages with adequate space, bedding material for comfort and maintained under specific pathogen-free conditions, while maintaining 12-h dark/light cycle. Ambient temperature was kept within 20 to 24 °C, and humidity levels ranged from 45 to 65%. All experiments were approved by the Bioethics Committee of the Cancer Research Center (#596) and by the Italian Health Minister (authorization n° 1227/2020-PR).

Cell Lines.

Human LUAD cell lines SK-LU-1 (ATCC, Cat#HTB-57) and A427 (ATCC, Cat#HTB-53) holding KRAS^G12D mutation were grown in minimum essential medium containing 10% FBS, 1% penicillin/streptomycin, 0.5 mg/mL Fungizone (Amphotericin B), and 50 μg/mL of Plasmocin® (Invivogen) at 37 °C and 5% CO₂. Murine cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) (4.5 g/L glucose + 10% FBS + 1% penicillin/streptomycin) at 37 °C and 5% CO_2.. The murine KPB6 cells (harboring KRAS^G12D and p53 mutations) were generated in Sergio Quezada´s Laboratory (University College London), used by Julian Downward’s lab (67) and provided by Dr. Esther Castellano. The LKR10 and LKR13 are mouse KRAS^G12D-mutant lung cancer cells derived by serial passage of minced LUAD tissues from two tumors isolated from separate lobes of the Kras^LA1 mouse model (66) and provided by Dr. Esther Castellano. To evaluate cell proliferation, AlamarBlue™ Cell Viability Reagent (Invitrogen, Cat#DAL1100) was added (10 µL/well) and incubated in dark for 1 h at 37 °C and 5% CO₂. Fluorescence intensity was evaluated at 590 nm using TECAN Infinite® 200 PRO microplate reader taking measurements at 0, 24, and 48 h. In particular, 1 × 10⁶ of murine KRAS^G12D-mutant LUAD cells (LKR10, LKR13, or KPB6) were treated with vehicle (DMSO), BI-3406 (1 μM), and/or MRTX1133 (5 nM) and cell proliferation quantitated as described above. In addition, the effect of BI-3406 on RAS activation was measured using the RAS G-LISA™ assay (Cytoskeleton Inc, Cat#BK131) in ST KPB6, LKR10, and LKR13 (KRAS^G12D-mutated) LUAD cells treated with vehicle (DMSO), BI-3406 (1 μM), and/or MRTX1133 (5 nM) for 2 h and stimulated with EGF (100 ng/mL) for 2 min.

Mouse Embryonic Fibroblast Immortalization.

This methodology is detailed in SI Appendix.

Generation of SOS-Less MEFs Expressing Human WT or Mutant KRAS.

Immortalized SOS-less MEFs including control SOS1/2^WT (SOS1^+/+;SOS2^+/+;Cre⁺), single SOS1^KO (SOS1^fl/fl;SOS2^+/+;Cre⁺), single SOS2^KO (SOS1^+/+;SOS2^–/–;Cre⁺), and SOS1/2^DKO (SOS1^fl/fl;SOS2^–/–;Cre⁺) were established as described (21).

KRAS^G12C, KRAS^G12D, and KRAS^G12V retroviral plasmids were created by point mutagenesis into pBABE-hygro-HA-tagged KRAS^WT plasmid (provided by David Santamaria) using the QuikChange XL Site-Directed Mutagenesis Kit (Agilent, Cat#200516). Retroviruses were generated by cotransfecting pBABE plasmids together with pAmpho plasmid into 293 T cells using Effectene Transfection Reagent (Qiagen, Cat#301425). The retroviruses were transduced into SOS-less MEFs followed by 2 wk of hygromycin selection (200 µg/mL) in complete DMEM. To induce SOS1 ablation, cells were then cultured for at least 11 d in the presence of 4OHT (Sigma, 600 nmol/L, Cat#H6278). Cells were grown at 37 °C and 5% CO₂ in a humidified incubator in DMEM supplemented with 10% FBS, 100 mg/mL penicillin, and 100 units/mL streptomycin.

Pharmacokinetic Study.

Pharmacokinetic assays were performed as detailed in SI Appendix.

Histology and Immunostaining.

Processing and staining of the sections (SI Appendix), as well as the histopathological evaluation of the samples (68) were performed by the Comparative Molecular Pathology-Biobank Network of Oncological Diseases of Castilla y León.

Hematological and Biochemical Parameters.

The analysis of hematological and biochemical parameters was performed as previously described (21), and detailed in SI Appendix.

Micro-CT Scanning.

In vivo micro-CT scanning assays were performed as described (30) and detailed SI Appendix.

IncuCyte Growth Assays.

Growth rate of KRAS^G12C, KRAS^G12D, and KRAS^G12V SOS-less immortalized MEFs cells was assessed as previously described (10). A more detailed description is shown SI Appendix.

Drug-Response Assay.

This methodology is described in Dataset S1.

RAS Activation Pull-Down Assay and Western Blot.

All antibodies and conditions used are described SI Appendix.

Quantification and Statistical Analysis.

GraphPad Prism 8.0.1 software was used. Statistical significance was determined by one-way ANOVA using the Tukey’s method to correct for multiple comparisons. For comparisons established only between two groups, we used Student’s two-tailed, unpaired t test. To compare tumor growth inhibition in the very same animals, paired t test was performed. Survival analysis was performed by the Kaplan–Meier method and between-group differences in survival were tested using the Log-rank (Mantel-Cox) test. IncuCyte experiments and western blot assays were performed in triplicate. n values mentioned in the figure legends indicate the number of animals used per experimental group. Results are expressed as mean ± SD. Significant differences are considered at P value < 0.05. The investigators were blinded during evaluation of tumor size variations following treatments. For dataset analysis, the natural logarithm of the half-maximal inhibitory concentration, Ln(IC50), represents drug sensitivity. SynergyFinder web application (v 3.0) (69) was used to calculate and represent the potential synergy of BI-3406 and MRTX1133. Synergy scores were calculated based on the Zero Interaction Potency model (70) and represented as a 2D synergy map.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.