Our study proposes a novel combination therapy for KRAS-G12C mutant cancers utilizing two clinically approved drug classes, namely KRAS-G12C inhibitors and farnesyl-transferase inhibitors. These combinations showed synergistic antitumoral effects in all five KRAS-G12C mutant cancer models in 2D in vitro experiments and the findings were recapitulated in lung adenocarcinoma 3D models in vitro and in xenografts in vivo.

The landscape of the treatment of KRAS-G12C mutant solid tumors has changed with the introduction of covalent, allele-specific KRAS-G12C inhibitors, among which sotorasib and adagrasib have already been clinically approved [9]. As there was no targeted therapy for KRAS mutant tumors before, the clinical activity of sotorasib and adagrasib is considered as a breakthrough in these tumor types. However, objective response rate is still relatively small (30–45% in lung and 7–20% in colorectal adenocarcinoma) [6, 7, 10, 11]. Thus, numerous clinical trials are underway investigating potential drug combinations that can improve the antitumoral efficacy of these novel KRAS-G12C inhibitors [18, 32]. These attempts include vertical or horizontal combinational approaches and also combinations of KRAS inhibitors with immunomodulatory therapies [32]. Vertical combinations rely on the observation that blockage of mutant KRAS signaling leads to compensatory reactivation of the pathway [17, 33]. Simultaneous inhibition of upstream elements like blockage of EGFR can diminish this effect and lead to synergistic antitumoral effects [34, 35]. Another promising approach is targeting SHP2 in combination with KRAS G12C inhibitors [32, 35]. Notably, SHP2 has been proven to be an important upstream regulator for proper KRAS signalization and is a promising candidate for combinational therapy, as its blockade in parallel with KRAS inhibitors can inhibit wild-type RAS signaling [35]. Although they act at a different level, SOS1 inhibitors exhibit similar effects when used in combination with KRAS targeting [32].

Horizontal combinations include the targeting of PI3K/AKT/mTOR signaling, which is a well-established escape route upon RAS-MEK-ERK inhibition [32, 36, 37]. Furthermore, preclinical evidence revealed synergistic antitumoral effects of cell cycle inhibitors (e.g. cyclin-dependent kinase inhibitors) combined with KRAS G12C inhibitors [32, 38].

In addition, mutant KRAS signaling has a wide range of immunomodulatory effects [32]. In line with these observations, preclinical evidence shows that the application of KRAS inhibition along with immunomodulatory therapies like immune-checkpoint inhibitors shows enhanced antitumoral effects [32, 39, 40].

However, none of these preclinical investigations or clinical trials has employed farnesyl-transferase inhibitors, likely due to the long history of the failures of FTis as monotherapies in KRAS mutant cancers [3, 32].

Our analysis of publicly available databases showed that lung adenocarcinoma cell lines harboring KRAS-G12C mutation tend to be more sensitive to FTis than KRAS wild-type cells.

Interestingly, comparison of different types of KRAS mutations revealed that KRAS G12C shows the highest sensitivity towards FTis, however, small numbers of LUAD cell lines with other mutations limit the interpretation of this finding. Regarding zigosity, there was a tendency for higher sensitivity towards FTis of heterozygous KRAS mutant cell lines when compared to cells possessing only the mutant allele, however, the number of homozygous cell models is rather small. Of note, upon blockage of farnesylation, KRAS and NRAS (but not HRAS) proteins undergo an alternative prenylation, named geranyl-geranylation and are therefore not affected by FTis [23, 24]. However, FTi sensitivity may also be influenced by the differential rewiring of RAS signalization and vulnerabilities. For instance, we identified a significant inverse correlation between KRAS and HRAS dependency on KRAS G12 mutant LUAD cell lines. Furthermore, it has been shown that intrinsic GTPase activity of KRAS G12C protein is retained in contrast with most of the other mutant KRAS proteins [32, 41], which may lead to distinct outcomes. Dependency on HRAS signaling, RHEB activity in PI3K/AKT signaling and at least 63 farnesylated proteins may modulate sensitivity towards FTis. Our findings warrant a more detailed examination of modulatory factors of sensitivity towards farnesyl-transferase inhibition.

Our observations on FTi sensitivity led us to combine tipifarnib, a clinically approved potent farnesyl-transferase inhibitor with sotorasib, a novel KRAS-G12C inhibitor in lung adenocarcinoma cell lines. Surprisingly, strong synergistic antitumor drug interactions were demonstrated in all cell lines investigated in adherent conditions. This finding was confirmed in 3D spheroid models. We identified similar drug interactions with additional relevant drugs in clinical development, namely adagrasib and lonafarnib. Besides the antiproliferative effect, the combination of sotorasib and tipifarnib also exerted antimigratory activity in vitro in single cell motility of PF139 cells as well as in wound healing assays. In the case of H358, sotorasib and combinational treatment drastically blocked wound closure, likely due to the combined antimigratory and antiproliferative effects. In H358 cells, tipifarnib alone did not reduce wound closure, in line with previous observation that the impact of tipifarnib on cell migration is cell-type dependent [42].

Furthermore, though both sotorasib and tipifarnib monotherapy achieved tumor growth inhibition in our xenograft experiments, the combination exerted the most pronounced impact on tumor growth in two distinct xenograft models of human lung adenocarcinoma utilizing the sotorasib-sensitive H358 and sotorasib-resistant SW1573 cells. Regarding tipifarnib dosage, we utilized double the amount of the corresponding human dose (600 or 900 mg b.i.d, equivalent to ~17 or 25 mg/kg daily in a 70 kg human), adjusted to the higher growth rate of xenograft tumors. Notably, this dosage is still half of the dose that could be found in previous studies [43]. In the case of sotorasib dosage, the human dose is 960 mg daily [7], which is equivalent to 14 mg/kg per day in a 70 kg human. When adjusting dosage for our xenograft models, we also took into consideration different sensitivities of the utilized cell lines towards KRAS inhibition. Notably, H358 is a commonly utilized model for KRAS-G12C inhibition, as it exhibits particularly high sensitivity towards KRAS-G12C inhibitors both in vitro and in vivo, which shows growth inhibition even upon 3 mg/kg dosage [39, 44]. In contrast, SW1573 showed high level of resistance towards KRAS inhibition in vitro in 2D conditions and was also found to be refractory in a study utilizing adagrasib [38, 44]. In line with these results, growth of H358 xenografts was limited upon as small as 5 mg/kg i.p. treatment, while 25 mg/kg dose was necessary to achieve a similar effect in SW1573 tumors. Interestingly, we showed that tipifarnib monotherapy is also able to exert a pronounced inhibitory effect on SW1573 xenografts, which is – to our knowledge – the first demonstration of FTi anticancer effects in vivo on KRAS G12C mutant model. Importantly, combined treatments were able to reduce the volume of H358 xenografts and suppress the growth of SW1573 tumors. Furthermore, even though neither tumor volumes measured with caliper nor tumor weights showed significant differences between monotherapies and combinational therapy, histopathological analyses revealed enhanced efficacy of the latter. Specifically, in the case of H358 xenografts, tumors in the combinational treatment group exhibited a significantly higher percentage of necrotic area compared to tumors in both single agent groups. This finding leads to the conclusion that a lower amount of living tumor tissue can be found in combinational therapy group, which could not be differentiated when measuring tumor volume or weight. In addition, the density of mitotic cells was also significantly lower in tumors treated with combinational therapy compared to sotorasib-treated tumors. Histology of SW1573 tumor samples revealed that the combination induced a significantly higher amount of focal necrosis compared to tipifarnib-treated tumors and increased frequency of apoptotic cells significantly higher compared to sotorasib-treated tumors. In this model, no difference was found in mitotic cell density between single agent and combination treatments. In summary, based on histopathological analyses, the combined application of KRAS G12C and farnesyl-transferase inhibition showed significant anti-tumoral effects in both models not only compared to the control but also compared to monotherapies.

However, such combinations may have unique toxicities. In our mice xenograft model, we have not seen major weight loss upon combination therapies (5–10%) and we have not detected histopathological alterations in the liver and kidney tissues, two organs that are affected by FTIs and G12C inhibitors in monotherapy clinical trials [9, 25, 27, 45]. One has to consider that the treatment of animals was less than a month, while the treatments last for several months in the clinical setting. Accordingly, chronic toxicity studies are necessary to test possible side effects in experimental animals, before entering the clinics. Also, one limitation of this present work is the lack of pharmacodynamic and pharmacokinetics data for the treatments, however, they are available for the monotherapies in early clinical studies [9, 46].

To investigate the potential cellular mechanism of this synergism, we performed cell cycle analysis and studied the expression of apoptosis and proliferation markers. We observed that the G2/M phase was slightly elevated in all cell lines upon FTi treatment in line with previous observations [47, 48]. Interestingly, using a modified, M phase preserving protocol, a more pronounced elevation of the G2/M phase was observed in SW1573 cell line suggesting that a significant amount of cells get stuck in the mitotic phase upon farnesyl-transferase inhibition. The ratio of the cells in the SubG1 phase also increased in H358 and SW1573 cell lines upon both FTi and combinational treatments and cleavage of the PARP protein was also detected in these samples. Investigation of the putative proliferation marker PCNA showed that both monotherapies and the combination effectively blocked proliferation in H358 and PF139 cell lines. Thus, treatment with FTis and KRAS-G12C inhibitors can exert antiproliferative, proapoptotic and even anti-migratory effects in vitro.

The analysis of the downstream signaling network of KRAS revealed diverse, cell line-dependent effects of the single agent and combination treatments on the PI3K/AKT/mTOR and RAF/MEK/ERK cascades. In line with previous results, we also observed a marked reduction of ERK phosphorylation upon KRAS-G12C inhibition [35, 38, 39]. However, in contrast with other’s findings, we showed that S6 activation did not decrease upon KRAS-G12C blockade in two out of three of our LUAD cellular models. Only combinational treatment was able to reduce S6 phosphorylation in all cell lines, which is an important hallmark of response to drugs [49]. Moreover, we also demonstrated successful inhibition of RHEB farnesylation, a key regulatory protein in the PI3K/AKT/mTOR pathway. The importance of this finding is underlined by the fact that horizontal combinational inhibition of RAS pathways is also proposed as a potential therapeutic approach [32]. Interestingly, investigation of activation of FAK-SRC signalization revealed increased FAK autophosphorylation upon sotorasib and combination treatments in PF139 and SW1573 cell lines, while all treatments reduced p-FAK in H358. Notably, it was previously shown that oncogenic RAS can lead to dephosphorylation of FAK at the 397 tyrosine through PIN1 and PTP-PEST leading to increased migration and enhanced wound closure [50]. Blockage of oncogenic KRAS can thereby relieve this negative signal, possibly leading to decreased migratory capacity.

Our investigation on the prenylation of the three classical RAS proteins revealed that FTi treatment successfully inhibited prenylation of HRAS but did not affect KRAS and NRAS, in line with previous observations [24]. Specifically investigating the activation of RAS proteins by assessing the level of GTP-bound RAS, we demonstrated the decrease by KRAS-G12C inhibition in parallel with compensatory HRAS activation. GTP-NRAS was changed by none of the treatments. Notably, GTP-HRAS was mainly found in the non-functional unprenylated form upon FTi treatment [23]. Our finding confirms previous demonstration of compensatory activation of wild-type RAS proteins upon KRAS-G12C targeting [35]. In a study by Ryan and colleagues, six cell lines showed elevated GTP-NRAS and GTP-HRAS upon KRAS G12C inhibitor ARS1620 treatment [35]. In our study, we also found elevated GTP-HRAS level but no change was detected in the amount of GTP-NRAS either in SW1573 or in MIAPACA2 cells. These experiments indicated that while in sotorasib-treated cells HRAS and in tipifarnib-treated cells KRAS can contribute to RAS-mediated signaling, the combination blocks the activity of both proteins and in turn leads to the observed synergism. This finding is particularly important as it was shown that inhibition of oncogenic KRAS can result in the reactivation of RAS-controlled pathways through wild-type RAS proteins [33]. However, it should be noted that both models used for the investigation of the role of HRAS in targeting the reactivation of the RAS pathway were homozygous for KRAS G12C mutations. Thus, further studies are needed to explore this phenomenon in a heterozygous setting where wild type KRAS proteins are also present. Furthermore, we also targeted HRAS with siRNA and managed to sensitize cells towards KRAS-G12C inhibition with a concomitant decrease in sensitivity against farnesyl-transferase inhibition. This data further supports the role of HRAS in the observed synergistic drug interactions. In addition, our in silico analysis found a significant inverse correlation between HRAS and KRAS dependency in KRAS-G12C mutant lung adenocarcinoma suggesting that these two proteins can play a compensatory role in tumor growth.

Investigation of the sotorasib-resistant SW1573 cell line with videomicroscopy revealed strong inhibition of cell division upon KRAS-G12C inhibition and FTi treatment. Notably, the combination resulted in the most pronounced, significant decrease in the number of successful cell divisions suggesting that synergistic drug interactions work through inhibition of proliferation. Furthermore, the videomicroscopic analysis revealed that many cells exhibited a dramatic delay in cytokinesis upon FTi and combinational treatment. These cells that struggled to perform cytokinesis likely contributed to the increase in the ratio of G2/M phase cells in cell cycle experiments. Several farnesylated proteins are important in regulating the mechanical aspects of cell division including centromere-associated proteins (CENP-E, CENP-F) and the laminar network of the nucleus [51, 52]. Indeed, we demonstrated that inhibition of farnesylation of lamin A/C proteins by tipifarnib initiated intranuclear accumulation of lamin and aberrant nuclear morphology, similar to the effects of another FTi, lonafarnib [53].

Since current G12C inhibitors are less active in colorectal or pancreatic cancers [19, 54], we also tested the combination of FTis and KRAS-G12C inhibitors on pancreatic and colorectal adenocarcinoma cells. Similar to our observations on lung adenocarcinoma, we found that all drug combinations resulted in strong synergistic antitumoral drug interaction in both types of human cancer cell lines in vitro.

It should be noted that due to pleiotropic nature of FTi action (as showed in the list of farnesylated proteins), a detailed investigation of each FTi target protein (e.g. through siRNA knockdown) in the synergistic drug interaction is not feasible. However, we identified several factors that can contribute to the observed synergism. These particular molecular mechanisms are also targets of other proposed combinational therapies that are curently in clinical development.

One synergistic interaction is the parallel targeting of PI3K/mTOR pathway by inhibiting the farnesylation of RHEB. The blockage of RHEB can induce changes similar to the mTOR inhibitor everolimus, which is currently evaluated in combination with sotorasib for KRAS G12C mutant tumors in clinical trials [32]. Indeed, we observed enhanced inhibition of S6 activation (downstream of RHEB and mTOR) upon the combinational therapy compared to monotherapies in all cell lines investigated, indicating that this combination is successful in reducing the activity of the PI3K/mTOR pathway.

Second, we have shown that FTi can reduce the compensatory reactivation of wild-type RAS proteins, namely HRAS, which is a known target of farnesyl-transferase inhibitors. Targeting the reactivation is also subject to clinical trials, specifically the combination of KRAS G12C inhibitors with SHP2 (e.g TNO155 with both sotorasib and adagrasib or RMC-4360 with sotorasib) and SOS1 inhibitors (e.g. BI1701963 with adagrasib), upstream elements in the RAS pathway [32]. The blockage of oncogenic KRAS G12C proteins relieves negative feedback loops on proteins upstream of RAS proteins leading to increased flux of the signaling through wild-type RAS proteins [17, 33]. Targeting of wild-type RAS activation by specific inhibition of proteins involved in RAS activation like SOS1 and SHP2 was shown to result in robust and profound synergistic combinational drug interactions when applied in parallel with KRAS inhibitors [32]. We also observed compensatory reactivation of HRAS (but not NRAS) using RAS-pulldown assays. We also demonstrated that HRAS farnesylation was blocked in two cell lines with distinct tissue of origin (SW1573 LUAD and MIAPACA2 PDAC cell lines). Furthermore, silencing HRAS with siRNA in the SW1573 cell line led to an increase of the IC50 of tipifarnib in parallel with a reduction of the IC50 of sotorasib. In addition, further supporting the role of HRAS, we found an inverse correlation between HRAS and KRAS dependency in KRAS mutant cell lines based on a publicly available database containing CRISPR sensitivity data (https://depmap.org/portal/interactive/).

Third, we provided evidence that farnesyl-transferase inhibition delayed cell cycle progression likely at least in part due to its impact on nuclear lamina. The combination of cell cycle inhibitors and KRAS G12C inhibition is indeed in clinical development using the cyclin D inhibitor palbociclib [32]. Of note, a couple of centromere-associated proteins (CENPE, CENPF) are also subject to farnesylation and may contribute to the synergistic effect.

In summary, the utilization of FTis in KRAS combinational therapies shares common mechanisms with three combinational approaches that are currently being evaluated in clinical trials.

The limitations of the current study are that our cancer models do not recapitulate intratumoral heterogeneity, like patient-derived organoids or patient-derived xenograft tumors. Intertumoral heterogeneity was addressed by means of utilizing five different stable cell lines from three different tissue of origins, of which two are novel patient-derived cell lines established by our group. Furthermore, the subcutaneous xenograft model – although its evaluation is more straightforward than more sophisticated orthotopic or transgenic models – cannot recapitulate important features as immune response or orthotopic cellular environment. Nevertheless, we performed in vivo experiments using two distinct cellular models – one heterozygous KRAS G12C inhibition-sensitive and one homozygous KRAS-G12C-inhibition resistant model. Another limitation of this manuscript is that we could not provide proof-of-concept for one circumscribed mechanism of action for the observed synergistic drug interactions. Furthermore, we could not precisely decipher the mechanism of action of the proposed combination due to the pleiotropic nature of farnesyl-transferase inhibition. Nevertheless, we demonstrated that several potentially contributing farnesylated proteins are inhibited in the proposed treatment including the FTi-sensitive wild-type HRAS and RHEB proteins or lamins.

Besides its limitations, our work clearly demonstrates that the application of FTis can potentiate antitumoral effects of novel KRAS G12C inhibitors and suggests that combination of farnesyl-transferase and KRAS-G12C inhibitors should be explored in the clinical setting both in KRAS-G12C mutant lung adenocarcinoma and in other KRAS-G12C mutant tumors.