Base excision repair-mediated resistance to cisplatin in KRAS(G12C) mutant NSCLC cells.

KRAS mutations in NSCLC are supposed to indicate a poor prognosis and poor response to anticancer treatments but this feature lacks a mechanistic basis so far. In tumors, KRAS was found to be mutated mostly at codons 12 and 13 and a pool of mutations differing in the base alteration and the amino acid substitution have been described. The different KRAS mutations may differently impact on cancerogenesis and drug sensitivity. On this basis, we hypothesized that a different KRAS mutational status in NSCLC patients determines a different profile in the tumor response to treatments. In this paper, isogenic NSCLC cell clones expressing mutated forms of KRAS were used to determine the response to cisplatin, the main drug used in the clinic against NSCLC. Cells expressing the KRAS(G12C) mutation were found to be less sensitive to treatment both in vitro and in vivo. Systematic analysis of drug uptake, DNA adduct formation and DNA damage responses implicated in cisplatin adducts removal revealed that the KRAS(G12C) mutation might be particular because it stimulates Base Excision Repair to rapidly remove platinum from DNA even before the formation of cross-links. The presented results suggest a different pattern of sensitivity/resistance to cisplatin depending on the KRAS mutational status and these data might provide proof of principle for further investigations on the role of the KRAS status as a predictor of NSCLC response.


INTRODUCTION
Lung cancer figures among the leading causes of mortality worldwide [1] and strongly associates with environmental factors and smoking habits [2]. The 5-year prognosis of NSCLC patients is very poor with a percentage of survivors lower than 15% for all stages and lower than 5% for metastatic patients [3]. Only few NSCLC patients, harboring mutations in the EGFR gene [4] or presenting the ALK-EML4 translocation [5], benefit from targeted therapy with erlotinib/gefitinib [4] or crizotinib, [6] respectively. The remaining patients are currently treated with platinum-based combinations [7].
KRAS is among the most frequently mutated oncogene in NSCLC and its mutations are present in approximately 20% of lung adenocarcinomas and tumors www.impactjournals.com/oncotarget of smokers [8]. KRAS mutations are mainly missense mutations at codon 12 and 13, but rare variants were detected in other codons [9].
KRAS is a member of the RAS gene family which encodes small G-proteins with intrinsic GTPase activity. GTPase activity leads to protein inactivation and control of downstream effectors involved in multiple pathways including proliferation, differentiation and apoptosis [10]. Point mutations occur in tumors resulting in the loss of intrinsic GTPase activity and consequently in the deregulation of cell proliferation signals and increased aggressiveness of tumors [9][10][11].
We have shown, at preclinical level, that the the most frequently altered KRAS codons in NSCLC have a different response in vitro to conventional chemotherapeutic and targeted drugs used in the clinic. In particular the KRAS(G12C) mutation, the most abundant in lung cancer, associates with a weaker response to cisplatin treatment compared to wt and other tested mutations [12]. We have recently shown in a prospective study that NSCLC patients with mutated KRAS tumor had a worse response to first-line platinum-based treatment compared to KRAS(wt) patients [13] and unpublished results.
In the clinic, KRAS mutated patients so far cannot benefit from any targeted therapy and are treated in firstline with platinum based compounds as the KRAS(wt) patients. In this paper we characterized the role of KRAS mutations at position 12, in particular the KRAS(G12C) mutation, in mediating response to cisplatin treatment with the aim to elucidate the mechanisms of cisplatin resistance induced by this mutation and to give the rationale of possible new pilot clinical trials aimed at stratifying patients on the basis of KRAS mutations.

In vitro cisplatin response as a function of the KRAS status
Using isogenic NCI-H1299 derived clones, expressing comparable KRAS protein levels (Figure 1a), we determined the activity of cisplatin in vitro by using two independent clones for each KRAS variant. As already reported for one set of clones and different chemotherapeutic agents in our previous manuscript [12], the expression of a specific KRAS mutant induced a distinct sensitivity pattern detected by MTS assay. Both clones expressing the KRAS(G12C) showed a weaker response to cisplatin compared to KRAS(wt), KRAS(G12D) or KRAS(G12V) clones (Figure 1b).
Statistical analysis did not detect differences between independent clones harboring the same mutation, but between clones expressing different KRAS variants. Assessment of IC50 values from the mean curves of two clones expressing the same KRAS variant indicated an approximately two-fold difference between KRAS(wt) (IC50 = 16.32 ± 2.78 uM) or KRAS(G12D) (IC50 = 17.53 ± 1.99 uM) and KRAS(G12C) clones (IC50 > 30 uM) and even a three-fold difference between KRAS(G12C) and KRAS(G12V) clones (IC50 = 12.92 ± 3.21 uM). This finding was strengthened by clonogenicity testing, revealing reduced sensitivity of the KRAS(G12C) clone to cisplatin compared to the other clones (Figure 1c). The reduced activity of cisplatin in KRAS(G12C) expressing cells was further confirmed in other isogenic systems expressing the different KRAS mutants and in NSCLC cells with a different KRAS status (Supplementary Figure S5).
We then performed a series of experiments aimed at understanding the reason for the different response to cisplatin of KRAS(G12C) clones. Cisplatin did not induce the central step of apoptosis, namely caspase 3/7 cleavage, in the KRAS(G12C) clone, however, in KRAS(wt) and to a slightly lesser extent also in KRAS(G12D) and KRAS(G12V) clones the drug stimulated the activation of caspase 3/7 ( Figure 1d).
To understand whether differences in response of KRAS(G12C) cells were associated with changes in cell cycle perturbations, we treated the clones with cisplatin and performed flow cytometric DNA content analysis. KRAS(wt), KRAS(G12D) and KRAS(G12V) clones showed an at least two-fold accumulation of cells in G2/M phase of the cell cycle 24 h after treatment, which was not observed in the less cisplatin-sensitive KRAS(G12C) clone. Forty-eight hours after treatment, KRAS(wt), KRAS(G12D) and KRAS(G12V) clones showed a partial re-distribution of G2/M phase cells in the different phases (Figure 1e, Supplementary Table S2). In KRAS(G12C) cells a G2/M phase accumulation was not even observed at this later time-point.
To exclude the possibility that different drug responses in the clones are due to differences in KRAS activation, guanosine-5′-triphosphate-bound KRAS (GTP-KRAS) levels were assessed before and after cisplatin treatment. At basal level, mutated KRAS harboring clones showed elevated levels of GTP-KRAS compared to the KRAS(wt) clone ( Figure 1f). Upon cisplatin treatment, mutant clones did not accumulate activated KRAS whereas the GTP-KRAS in KRAS(wt) clone increased and reached levels similar to KRAS mutated clones.   compatible with animal health status (10% of the body weight). In conclusion, KRAS(G12C) tumors showed a reduced response to cisplatin compared to KRAS(wt) tumors in vivo, confirming the results obtained in vitro.

Evaluation of MAPK and PI3K signaling downstream of KRAS
To further delineate KRAS activation in the clones, activation of PI3K and MAPK pathways was evaluated at different time-points after cisplatin treatment (Supplementary Figure S1a

Intracellular amount of cisplatin
As graphically shown in Figure 2d no differences in the intracellular content of cisplatin were found at least between the mutant expressing KRAS clones at any timepoint of the experiment. Expression of the copper influx transporter CTR-1, that is also a major influx transporter for cisplatin, was correlated to resistance of this drug [14]. Its expression, assessed by real-time PCR, was similar in the clones (Figure 2e) with a modest, not significant, lower expression in the KRAS(G12D) clone.
The GST activity and the intracellular GSH amount were determined in untreated cells but did not reveal significant differences. In agreement, equivalent responses were found in the clones following treatment with the alkylating agent melphalan, whose resistance is dependent on the GST/GSH levels [15] (Supplementary Figure S1b).  Supplementary Table S1. To further investigate whether cisplatin similarly reached its target in the cells, platinum bound to DNA was estimated by DRC-ICP-MS. Maximum levels of platinum bound to DNA were measured 2 h after treatment and were similar in the different clones. On the contrary, 24 h post-treatment, the level of platinum bound to DNA was below the detection limit in the KRAS(G12C) clone, while it was still measurable in all the other clones ( Figure 2f). Altogether these data excluded impaired uptake, faster export or reduced DNA adduct formation of cisplatin despite rapid adduct disappearance in KRAS(G12C) cells.

Role of DNA damage signaling in differential cisplatin response
We then wished to investigate whether cisplatininduced DNA damage-signaling differed among KRAS expressing clones by assessing ATM, the best known apical activator in response to DNA DSBs [16]. We analyzed ATM activation by Western blot detection of the phosphorylated form (p-ATM) of the protein. To exclude that KRAS(G12C) expressing cells presented some defects in DNA damage detection, γH2AX was assessed both 24 h after high equitoxic cisplatin doses and at different time-points after IR treatment that directly induces DNA DSBs. Using equitoxic concentrations of platinum resulted in comparable γH2AX levels in the clones (Supplementary Figure S2). Furthermore, all the clones treated with 5 or 7.5 Gy X-ray displayed marked γH2AX signals at both doses and at all the time-points (90 min, 6 h, 24 h) of the experiment (Supplementary Figure S3a, S3b). In agreement, viability after X-ray was similar among clones (Supplementary Figure S3c).

Analysis of NER and DSB repair mechanisms in mutant KRAS expressing cells
DNA crosslink repair following cisplatin treatment was previously reported to involve NER mediated incisions and repair of resulting DSB intermediates by HR [17]. To understand the molecular cause underlying diminished accumulation of the DNA damage marker γH2AX in KRAS(G12C) cells, we dissected DSB repair components by immunofluorescence microscopic analysis. When monitoring nuclear 53BP1 and BRCA1 foci, two antagonistic components involved in pathway choice decisions between NHEJ and HR [18], we found reduced foci numbers of 53BP1 16 h and 24 h and of BRCA1 24 h post-treatment in KRAS(G12C) cells (Figure 4a-4d). Analysis of the HR recombinase RAD51 did not reveal statistically significant differences in foci numbers ( Figure 4e). Given these results, NER and pathway-specific DSB repair activities were evaluated to examine their potential involvement in the peculiar response of KRAS(G12C) clone to cisplatin.
Functionality of the NER system was indirectly evaluated by UV radiation treatment generating DNA lesions which are mainly repaired through NER [19]. Similar responses in the different clones were observed with a higher sensitivity rather than resistance of KRAS(G12C) cells (Figure 5a). We also analyzed relative mRNA expression of the NER genes ERCC1, XPA, XPF and XPG in the clones. Some differences were detected, however, they did not appear to be associated with the differential responses of the clones to treatment (Figure 5b). A trend to reduced XPA and XPF expression was observed in KRAS(G12C) but also in KRAS(G12D) cells. ERCC1 mRNA was low in KRAS(G12C) cells, but did not translate into a statistically significant difference of the ERCC1 protein level (Figure 5c).
Using EGFP-based reporter assays we measured HR, SSA, NHEJ and MMEJ activities. All KRAS mutant cells showed a significant DSB repair decrease in all the pathways compared to the KRAS(wt) cells (Figure 5d). We also investigated endogenous levels of key proteins involved in DSB repair and/or cisplatin responses. Western blot analysis did not reveal significant differences in the levels of 53BP1, BRCA1, KU70, MLH1 or RAD52 in the different clones, even though a trend to reduced KU70, MLH1 and RAD52 was noticed in the KRAS(G12C), for KU70 in the KRAS(G12D) and for MLH1 and RAD52 in the KRAS(G12V) clone. BACH1 levels were significantly lower in KRAS(G12C) and KRAS(G12D) compared with KRAS(wt) cells (p = 0.02) (Figure 5e). Altogether, NER and DSB repair-related features were unlikely to explain cisplatin resistance in KRAS(G12C) cells. www.impactjournals.com/oncotarget

Analysis of alternative DNA repair mechanisms in mutant KRAS expressing cells
Having so far failed to identify the mechan ism responsible for KRAS(G12C)-mediated resistance to cisplatin, alternative DNA repair pathways with potential involvement in cisplatin adduct removal were investigated.
We examined the FA repair system with critical involvement in crosslink repair at stalled replication forks [20]. The mRNA expression analysis of FANCA, FANCC, FANCD2 and FANCF genes did not reveal any difference in the KRAS(G12C) clone possibly explaining cisplatin resistance (Figure 5f). Kinetics of FANCD2 foci assembly after cisplatin treatment were similar at least between KRAS(G12C) and KRAS(wt) or KRAS(G12D) clones ( Figure 5g). FANCD2 activation was investigated by Western blot detection of ubiquitylation following cisplatin treatment. No significant differences were found, when determining the ratio between the activated (ubiquitylated) and inactive (not ubiquitylated) forms of the four clones ( Figure 5h). PCNA ubiquitylation is triggered by replication stalling lesions such as DNA crosslinks, whereby monoubiquitylation leads to the polymerase switch between  Supplementary Table S1. replicative and translesion synthesis polymerases, initiating another important repair process contributing to crosslink repair [20,21]. PCNA mono-ubiquitylation as indicated by the appearance of a more slowly migrating band at 40 kDa was assessed after cisplatin treatment detecting monoubiquitylated protein in all clones (Figure 5i).

Protective effect of BER against cisplatininduced cytotoxicity in KRAS(G12C) cells
Finally, a potential role of BER in modulating cisplatin responsiveness was investigated. To this end, clones were treated with MMS that produces DNA   Supplementary Table S1. h. Representative Western blot analysis displaying levels of ubiquitylated, i.e. activated FANCD2 protein (upper band) and unmodified FANCD2 (lower band) in the clones treated or not with cisplatin. Lamin B was used as loading control. Two independent experiments have been performed. i. Representative Western blot analysis showing levels of mono-ubiquitylated, i.e. activated PCNA protein (long exposure, upper band) and the unmodified form of the protein (long exposure, lower band, and short exposure) in the clones treated or not with cisplatin. Tubulin was used as loading control. www.impactjournals.com/oncotarget methylation known to be processed by BER [22]. KRAS(G12C) clone resulted less sensitive to this compound when compared with the other clones suggesting a different activity of BER in clones (Figure 6a). We then treated cells with MA, a compound known to inhibit the initial incision step of BER [23]. Concomitant treatment of cells with MA at non toxic concentration, was able to almost completely restore the sensitivity of KRAS(G12C) clone to cisplatin (Figure 6b, 6c).
The expression of DNA polymerase beta (Polβ), the limiting component of BER activity, was evaluated by western blot and was found 2.7-fold overexpressed in the KRAS(G12C) compared to the KRAS(wt) clone (Figure 6e). Transfection with Polβ-specific siRNA successfully downregulated protein levels only after repeated Polβ siRNA transfection and only within 168 h (transfecting siRNA every 48 h) (Supplementary Figure  S4a, S4b). This caused excessive cytotoxicity excluding the possibility to perform survival assays following siRNA-mediated Polβ knockdown. As an alternative approach, we applied PA, which was described as a potent Polβ inhibitor [24]. Pre-treatment of clones with PA at non-toxic concentration completely restored the sensitivity of KRAS(G12C) clone to the cisplatin treatment (Figure 6b, 6d) further confirming causal involvement of Polβ in altered cisplatin responsiveness.
Trying to elucidate the mechanism(s) by which Polβ was differently expressed in the clones, we analyzed the mRNA and protein stability. We could exclude a different Polβ protein stability among clones considering the experiments presented in Supplementary Figure  S4a, S4b. The mRNA stability analysis of Polβ gene did not reveal as well any significant difference in the clones (Supplementary Figure S4d), whereas Polβ mRNA was more than 2-fold overexpressed in the KRAS(G12C) if compared to the KRAS(wt) (Supplementary Figure S4e), thus suggesting a transcriptional mechanism responsible for the increased expression of Polβ.

Either BER and Polβ inhibitor treatments are able to rescue the phenotype of KRAS(G12C) cells
In order to strengthen the role of BER and in particular of Polβ as responsible of all the findings described above, we applied either MA or PA in combination to cisplatin to rescue the phenotypes of KRAS(G12C) clone. Caspase 3/7, as previously reported, were not activated by cisplatin in the KRAS(G12C) clone. By applying either the PA or MA, although the latter was less effective, the caspase 3/7 were cleaved also in KRAS(G12C) clone (Figure 7a). Pre-treatment with PA or MA was also able to restore the ability of KRAS(G12C) cells to block the cell cycle and accumulate in G2/M upon cisplatin treatment as shown for other clones (Figure 7b). When we analyzed ATM activation, the less cisplatin sensitive KRAS(G12C) clone showed only a two-fold increase in the p-ATM compared to a 4-5-fold activation of the KRAS(wt), KRAS(G12D) and KRAS(G12V) clones 16-24 h after cisplatin treatment. By adding either MA or PA to cisplatin, the combinations were able to restore to activate ATM following cisplatin (Figure 7c, Supplementary Figure S4f). Similarly, the concomitant treatment with cisplatin and PA or MA restored the ability of KRAS(G12C) cells to activate H2AX (Figure 7d). Finally we set an experiment with a limited number of animals as a proof of principle for the activity of the combination cisplatin plus PA in xenograft model. Although the experiment was underpowered to have a statistical significant result, we observed an higher activity for the combination compared to cisplatin alone in the KRAS(G12C) clone injected mice. For the combination schedule we obtained a best T/C of 49% at day 14 compared to 70% of the cisplatin alone (data not shown).

BER or Polβ inhibitor treatment is able to sensitize the KRAS(G12C) cells to cisplatin in other systems
To corroborate the potential role of KRAS(G12C) in modulating cisplatin responsiveness we extended our study to other systems.
The second independently generated NCI-H1299 KRAS(G12C) clone (cl.4), presented in Figure 1a, 1b and showing a lower response to cisplatin comparable to the KRAS(G12C) cl.2, was further investigated to confirm previous results. The KRAS(G12C) cl.4 was treated with MMS and, as reported for KRAS(G12C) cl.2, resulted less sensitive to this compound when compared with the wt clone (Supplementary Figure S5a). We then applied PA to restore the sensitivity to cisplatin. Pre-treatment with PA at non-toxic dose completely restored the sensitivity of KRAS(G12C) cl.4 to cisplatin (Supplementary Figure S5b) Figure S5i).
Finally we compared two NSCLC cell lines harboring a different KRAS status, KRAS(wt) (NCI-H1299) or KRAS(G12C) (NCI-H358). The Polβ expression was higher, both at mRNA and protein levels, in the NCI-H358 cell line (Supplementary Figure S5j, S5k). The response to cisplatin was also weaker in the NCI-H358 cell line expressing the KRAS(G12C) mutant compared to KRAS(wt) expressing NCI-H1299 cells. When we applied PA we were able to induce, in the NCI-H358 cell line, a strong response to cisplatin comparable to the KRAS(wt) cell line NCI-H1299 (Supplementary Figure S5l) whereas PA treatment did not change the activity of cisplatin in the already sensitive cell line NCI-H1299.

DISCUSSION
Platinum-based therapy remains so far the best treatment option for the majority of patients with NSCLC [25,26]. Targeted therapy is available for a limited number of patients presenting specific mutations or translocations and these alterations are mutually exclusive with KRAS mutations [27][28][29]. This implies that patients with NSCLC presenting KRAS mutations are almost invariably treated in first-line with platinum-containing drugs. Given that KRAS mutations associate in several types of tumors with a more aggressive phenotype and/or resistance to treatment, [13,30] it is mandatory to study the consequences of KRAS mutations on response to treatments. Mutations in the KRAS gene have been found mostly at codon 12 and 13 resulting in a pool of mutations differing in the replaced base and the substituted amino acid [31]. This spectrum of changes might be one of the reasons why mutated KRAS has so far not been an helpful marker to further classify patients in different cancer subgroups [32].
Here we provide evidence that cisplatin responsiveness indeed depends on the type of KRAS mutation. We found KRAS(G12C) mutant expressing cells, i.e. carrying the KRAS mutation most frequently found in NSCLC, to be the least responsive ones when compared with cells expressing other KRAS mutants or the KRAS(wt). These observations support our previous data indicating that cell lines expressing different KRAS mutations differently respond to drugs with different mechanisms of action [12]. The isogenic system we have used is based on a similar expression of exogenous KRAS (wt or mutated). The robustness of our systems is increased by the evidence that the introduction of exogenous KRAS (either wt or mutated) does not change the expression of endogenous KRAS (see Supplementary Figure S5m) which is similarly expressed in all the clones utilized avoiding the interference linked to the oncosuppressive role of KRAS(wt) as previously reported [33].
It is worth noting that the degree of resistance in our experimental settings is moderate, which, however, most likely reflects the clinical situation. Importantly, in this work, independently KRAS(G12C) expressing clones generated in different isogenic system (NCI-H1299 and NIH-3T3) showed a similar degree of cisplatin resistance when compared with KRAS(wt) clones, thus indicating that this finding is not restricted to a single cell line.
Analysis of the signaling pathways downstream of KRAS, such as involving MAPK and PI3K, [29] did not provide a clue for the peculiar response of KRAS(G12C) expressing cells to cisplatin. Similarly, another factor previously reported to be associated with resistance to cisplatin, namely increased detoxification through GST/ GSH, [34,35] did not play a significant role in this context. Reduced uptake or export of cisplatin can also underlie cisplatin resistance [36]. In our experimental model, cells expressing mutant compared to KRAS(wt) displayed a slight reduction in intracellular platinum but, as this was true for all three mutant clones, this aspect was unlikely to account for KRAS(G12C) clone-specific resistance to cisplatin. Given a similar intracellular cisplatin level and comparable adduct formation on DNA at early timepoints, downstream cellular activities had to account for the resistance of the KRAS(G12C) clone. Several pieces of evidence pointed to an impaired DNA damage response: i) analysis of the cell cycle perturbation induced by cisplatin treatment indicated only a weak G2/M phase block in KRAS(G12C) cells and this result was far different from all the other clones; ii) γH2AX foci formation following treatment was significantly reduced in the KRAS(G12C) clone compared to the others despite functionality of damage detection in these cells as demonstrated by experiments in which X-ray treatments induced γH2AX foci formation in a comparable way in the different clones and the KRAS(G12C) expressing clone also showed less DNA damage according to analyses of 53BP1-foci numbers and ATM activation; iii) finally, platinum bound to DNA disappeared completely within 24 h in KRAS(G12C) cells.
Altogether these data suggested that the KRAS(G12C) mutation stimulates a DNA repair mechanism promoting platinum removal from DNA before intra-and inter-strand crosslinks, avoiding cell growth arrest and/or death. Because levels of both DSB markers γH2AX and 53BP1 were reduced, this decisive DNA repair process might either be active before the formation of platinum cross-links which require cleavage and DSB formation at stalled replication forks or accelerate DSB repair itself [17,20,37].
To identify the molecular cause underlying the reduced drug response of KRAS(G12C) cells, we investigated different DNA repair mechanisms known to play a role in the response to cisplatin. Clones expressing mutated compared to KRAS(wt) showed reduced capacity of the more error-free DSB repair activities HR and canonical NHEJ as well as of the error-prone pathways SSA and MMEJ. As it applied to all three KRAS mutant clones, this feature was unlikely to explain cisplatin resistance in KRAS(G12C) expressing cells. These findings were corroborated by the analysis of the expression of proteins of interest involved in the different DSB repair pathways.
Other DNA repair mechanisms that were demonstrated to play a role in cisplatin damage repair, and thus were candidate systems for the removal of adducts from DNA were NER and the FA pathway [17,20,35,38]. However, our expression and functional data addressing these pathways excluded their involvement in the cisplatin resistance mechanism of KRAS(G12C) expressing clones. Of note, the KRAS(G12C) clone was, among the different clones tested, the most susceptible to UV light suggesting, if any, a less active NER system. Even mismatch repair has been implicated in the cisplatin response, namely in mediating cytotoxicity [39]. Notably, the mismatch repair protein MLH1 had previously been found to become downregulated by promoter methylation in a significant fraction of NSCLC [40] but comparable levels of MLH1 protein were detected in different KRAS expressing clones in our work.
Here, we provide several pieces of evidence supporting the idea that short-patch BER involving Polβ could be involved in the different behavior of KRAS(G12C) mutant expressing cells once treated with cisplatin.
Higher Polβ levels detected in the KRAS(G12C) compared with the other cells may stimulate BER activity resulting in a fast platinum adducts removal from the DNA. This feature would prevent intra-and inter-strand crosslink allowing these cells to growth and survive. Restoration of cisplatin sensitivity in the resistant KRAS(G12C) clone after BER/Polβ specific inhibitors treatment supported the idea that BER and in particular Polβ activity may account for cisplatin resistance in these cells. Interestingly, co-treatment with BER or Polβ specific inhibitors was also able to rescue the altered (compared to the other KRAS mutants and wt cells) phenotype of KRAS(G12C) cells, restoring apoptosis, cell cycle perturbation, γH2AX foci formation and ATM activation.
Elevated levels of Polβ have been associated with resistance to cisplatin treatment in colon cancer [41] and high levels of Polβ have been found in many cancers including breast, colon, ovarian and prostate cancers [42][43][44][45]. Metabolic changes in tumors lead to oxidative stress [46] and thus oxidatively damaged DNA requiring higher BER activities and elevated levels of the limiting component Polβ for survival, which concomitantly result in resistance to therapy.
We established a link between KRAS(G12C) and Polβ, showing that cells expressing this specific mutant have an increased expression of Polβ likely due to an increased transcription of the gene. Polβ gene transcription is regulated by several transfactors (including ATF/CREB family members) and mutations in the binding sites present in its promoter affect its transcription [47,48]. Very recently a direct link between CREB and KRAS has been postulated [49] thus making possible the hypothesis that a different interaction between different KRAS mutants and CREB could be responsible for the increased transcription observed in KRAS(G12C) mutants cells. This hypothesis will be actively pursued in the future.
At present, there is little information regarding Polβ expression in KRAS mutated lung cancer, so that this biomarker will have to be assessed in this type of cancer in the future. Importantly, when testing sensitivity to different inhibitors of PARP, which has been proposed to be involved in BER and to target HR-defective cells, [50] we observed increased sensitivity rather than resistance of mutant KRAS including KRAS(G12C) expressing cells (data not shown). Thus, PARP inhibition may overcome excessive BER in KRAS(G12C) expressing cells and at the same time target HR deficiency in all mutant KRAS cells. Therefore, the possibility remains that PARP inhibition could represent an additional therapeutic option for NSCLC in combination treatment approaches including combined cisplatin and PARP inhibitor regimens [51].
In conclusion our data demonstrate and confirm that different KRAS mutations have a different impact on cisplatin sensitivity. This information will have to be taken into account when designing new clinical studies aiming at the evaluation of the role of KRAS as prognostic and predictive marker in NSCLC. Classification of tumors solely by the presence of a mutation in KRAS, without defining the specific mutation, might not be enough to identify patients with a different response to therapy in cancers harboring a KRAS mutation. Our preclinical models represent an important tool to test new therapeutic strategies which could represent the starting point for the design of new trials in the clinical setting.

Cell cultures and drug treatments
The NIH3T3, NCI-H1299 and NCI-H358 were purchased by ATCC. The NIH-3T3 was grown in DMEM, the NCI-H1299 and NCI-H358 were grown in RPMI-1640 medium. Clones were obtained by transfecting the NCI-H1299 and NIH-3T3 cell line with the expression plasmids encoding for the different mutations (G12C, G12D and G12V) and the wt KRAS as a control. All clones were grown in medium including 500 ug/ml of G418 (Gibco). Cells are routinely tested for mycoplasma contamination by PCR and authenticated with the PowerPlex 16 HS System (Promega) every 6 months by comparing the STR profiles to which deposited in ATCC and/or DSMZ databases. Cisplatin, melphalan, PA, MA, MMS (Sigma Aldrich), isoquinolinediol (Calbiochem) and NU1025 (Enzo Life Sciences) were dissolved in medium just before use. Treatments, unless otherwise specified, were performed at 5 uM for 2 h or 1.5 uM for 24 h for cisplatin, 200 uM for MA and 100 uM for PA. The MTS assays (Promega) were performed as described in [12]. Clonogenic assays were performed as described in [52]. Survival curves, unless otherwise specified, were plotted as percentages of untreated controls, consisted of at least 6 replicates for each time point and represented the average mean and SD of at least 3 independent experiments.

Western blotting analysis
Proteins were extracted and visualized as reported in [52,53]. Immunoblotting was carried with the antibodies reported in the Supplementary Information. Protein bands of interest were quantified using Image Lab 4.1 (Bio-Rad Laboratories) or ImageJ 1.48 (NIH) softwares and corrected with the values obtained for the loading control each.

KRAS activation assay
The active form of KRAS was measured with KRAS Activation assay Kit (Cell Biolabs) according to the manufacturer's instructions.

Immunofluorescence microscopy and foci count
Immunofluorescence microscopy were performed as described in [53] and [54]. Primary antibodies used are reported in the Supplementary Information. Immunolabeled foci were scored by automated quantification and mean numbers of foci per cellular nucleus in the total cell population calculated from 4 slides analyzing 200 nuclei in total.

Real time PCR
Total RNA was reverse transcribed with High-Capacity cDNA Kit (Life Technologies) and amplified by 7900 HT Sequence Detection System (Life Technologies). Actin was used as internal control. Primers and TaqMan probes were purchased for all genes as ready-to-use solutions (Life Technologies). Two samples which showed at least 2-fold differences were considered differently expressed.

In vivo activity
Procedures involving animals were described in [52] and their care are reported in the Supplementary Information section.
For these specific experiments female athymic NCr-nu/nu mice, seven weeks old, obtained from Harlan Laboratories were inoculated s.c. with 200 ul of cell suspension containing 10 7 cells. When the average of the tumor weights reached about 200 mg (excluding animals with tumors <100 mg or >400 mg in weight), mice were randomized. Cisplatin was given intravenously at the dose of 5 mg/kg, every 7d for three times (q7dx3). Each group comprised 8 mice. The investigator who performed the in vivo studies was not informed about the in vitro results regarding cisplatin citotoxicity. A T/C < 42% is considered the minimum level for activity [55]. www.impactjournals.com/oncotarget

Caspases 3 and 7 activity assay
Twenty-four hours after cells plating, drug treatment was performed and 24 h or 48 h later caspase activity was assessed using the Caspase-Glo 3/7 Assay (Promega) according to the manufacturer's instructions.

Monoparametric staining of DNA content
Sample preparation and monoparametric DNA histograms analysis were performed as described in [56].

Measurement of platinum content
Equal numbers of cells were resuspended in 200 ul of PBS and 400 ul of HNO 3 :HCl (1:3). After 12 h at RT, 600 ul of water were added, vortexed and centrifuged at 16,000 × g for 10 min at 4°C. The supernatant was injected into Analyst 600 (Perkin Elmer). A calibration curve with platinum standard (Sigma Aldrich) was generated.
Platinum bound to DNA was determined with DRC-ICP-MS using an ELAN DRC (Perkin Elmer) equipped with cyclonic spray chamber and a Meinhard type concentric nebulizer. The uncertainty of measurements was evaluated as suggested by international bodies (ISO and EURACHEM/CITAC).

Determination of DSB repair frequencies
Fluorescence-based DSB repair measurements were performed for different DSB repair pathways following targeted cleavage by I-Sce I-meganuclease as described in [57,58].

Statistical analyses
The Statistical analyses were performed using GraphpadPrism version 5.01. Specific tests used to analyze specific experiments are indicated in the Supplementary Table S1. Differences between groups were considered statistically significant when the p-values were ≤ 0.05.