Influence of EGFR-activating mutations on sensitivity to tyrosine kinase inhibitors in a KRAS mutant non-small cell lung cancer cell line

In non-small cell lung cancer (NSCLC), oncogenic driver mutations including those in KRAS and EGFR are typically mutually exclusive. However, recent reports indicate that multiple driver mutations are found in a certain percentage of cancers, and that the therapeutic responses of such cases with co-mutations of driver genes are largely unclear. Here, using CRISPR-Cas9-mediated genome editing, we generated isogenic cell lines harboring one or two copies of an EGFR-activating mutation from the human NSCLC cell line A549, which is known to harbor a homozygous KRAS gene mutation. In comparison with parent cells with KRAS mutation alone, cells with concomitant EGFR mutation exhibited higher sensitivity to EGFR-tyrosine kinase inhibitors (TKIs) but not to conventional anti-cancer drugs. In particular, cells with two copies of EGFR mutation were markedly more sensitive to EGFR-TKIs compared with parent cells. Thus, the presence of concomitant EGFR mutation can affect the TKI response of KRAS-mutated cells, implying that EGFR-TKI may represent an effective treatment option against NSCLC with EGFR/KRAS co-mutation.


Introduction
Lung cancer is the leading cause of cancer-related death worldwide, and non-small cell lung cancer (NSCLC) is the most common type of lung cancer [1,2]. Lung adenocarcinomas, which account for approximately 50% of NSCLC, are molecularly subclassified and subjected to different therapeutic strategies according to the presence of distinct alterations in genes such as EGFR, ALK, and KRAS [3]. Among such mutations, EGFR-activating mutation is that most frequently identified in Asian patients, accounting for 30%-50% of cases [4], while in Western patients, KRAS mutation is the most common, accounting for 25% of cases [3].
In NSCLC prior to treatment, the majority of oncogenic driver mutations are mutually exclusive with other mutations [5,6]. However, recent studies have shown that additional driver mutations such as KRAS, ALK, and PI3K mutations co-exist with EGFR-mutations in a certain percentage of lung cancers [5,[7][8][9][10][11][12][13][14]. Of note, these driver mutations have been found not only in different cell populations in tumors, but also within the same cell population [15,16]. However, little is known about the therapeutic responses of cancer cells harboring multiple driver mutations. Cancer cells harboring EGFR-activating mutations have been shown to be highly sensitive to the EGFR tyrosine kinase inhibitor (TKI) gefitinib, which was first approved in Japan in 2002 [17,18]. In addition to gefitinib, several EGFR-TKIs including afatinib and osimertinib have since been developed and approved for first-line treatment of advanced NSCLC harboring EGFR-activating mutations [17]. Unlike EGFR, however, effective drugs targeting KRAS have not been developed to date. Indeed, small molecules targeting the KRAS pathway, including farnesyl transferase inhibitors and downstream MEK inhibitors, have failed to show a beneficial effect, while KRAS mutant allelic imbalance is likely to influence the efficacy of MEK inhibitors [19]. Thus, there remains a significant unmet clinical need to address KRASmutated NSCLC, given the high frequency of cases having this mutation.
In the present study, we aimed to assess the response to EGFR-TKIs of cancer cells with EGFR/KRAS co-mutations. Using CRISPR/Cas9 genome editing, we successfully generated isogenic cell lines with one or two copies of L858R EGFR mutation, one of the most common types of EGFR mutation [4], from A549 cells, a well-established cell line harboring homozygous G12S KRAS mutation but wild-type EGFR [20][21][22]. Comparison of TKI sensitivity among these cell lines showed that EGFR/KRAS co-mutated cells were more sensitive than parent cells, indicating that EGFR-TKIs may represent a treatment option for cancers harboring EGFR/KRAS co-mutation.

Cell culture
The human NSCLC cell line A549 was obtained from the National Cancer Institute (Frederick, MD) and isogenic cell lines were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and 100 μg/mL of kanamycin (Sigma-Aldrich, St. Louis, MO, US).

Transfection
Plasmids were transfected into cells by electroporation using the Neon Transfection System according to the manufacturer's instructions (Thermo Fisher Scientific, Waltham, MA, US). In brief, 5 μg of pSpCas9n-(BB)-2A-GFP plasmids (2.5 μg of each plasmid with a different sgRNA sequence; set 1 and 2) were transfected into 5 × 10 5 cells at 1550 V in a triple 10-ms pulse. For knock-in, ssODNs were dissolved in filtered water and transfected at 0.01 nmol per reaction together with 5 μg of the plasmids. After transfection, cells were incubated in the presence of 1 μM SCR-7, a non-homologous end joining inhibitor (Xcessbio, San Diego, CA, US), for 4 days.

Positive clone selection using L858R mutation-specific primers
To isolate L858R knock-in clones, transfected cells were seeded at 1 cell/well into 96-well plates. After cells reached 50%-90% confluency, cell suspensions collected from each well were pooled into one tube (16 wells/tube), giving a total of 90 cell mixtures from 15 96-well plates. PCR analysis amplified a band corresponding to the expected size (177 bp) in 22/90 mixtures. For the second screening, each candidate clone was scaled up to growth in 24-well plates. A total of 21 candidate clones were obtained and further screened by immunoblotting using an L858R EGFR antibody (Cell Signaling Technology, Danvers, MA, US) and by sequencing around the target region.

DNA extraction
Genomic DNA was extracted from cells by washing them with 200 μl phosphate-buffered saline and incubating them in 50 μl of 25 mM NaOH/0.2 mM EDTA solution for 10 min at 98˚C. After cooling to room temperature, the lysates were mixed with 50 μl of 40 mM Tris-HCl. The lysate was then centrifuged at 4000 × g for 3 min prior to PCR genotyping using 2 μl of supernatant as a template. For DNA sequencing, genomic DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen, Germantown, MD, US).

Copy number analysis
Genomic DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen). Quantitative real-time PCR was performed with an ABI Prism 7300 sequence detection system (Applied Biosystems, Foster City, CA, US) using SYBR GreenER (Invitrogen) with EGFR-specific primers: EGFR_ex27 forward, 5 0 -GCCTTGACTGAGGACAGCAT-3 0 and EGFR_ex27 reverse, 5 0 -TACGCCCTTCACTGTGTCTG-3 0 . The relative copy number to the control in TIG-3 (diploid) and A549 (triploid) cells was calculated using the comparative Ct method.

Immunoblotting
Cells were lysed with sodium dodecyl sulfate (SDS) lysis buffer (62.5 mM Tris-HCl, pH 7.2, 10% glycerol, 1% SDS) and immediately boiled for 10 min to obtain clear lysates. The protein concentration was measured by the Bradford method (BIO-RAD, Hercules, CA, US), and lysates containing an equal amount of proteins were separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes (Millipore, Darmstadt, Germany) for western blot analysis using appropriate antibodies. Immunoreactive proteins were visualized using the Immobilon Western Chemiluminescent HRP substrate (Millipore) or Clarity Western ECL substrate (BIO-RAD), and light emission was quantified with a LAS-3000 lumino-image analyzer (Fuji, Tokyo, Japan). The antibodies used in this study were: anti-EGFR, phospho-EGFR (Y1068), L858R EGFR, AKT, and phospho-AKT(S473) (Cell Signaling Technology, Danvers, MA, US) and anti-β-actin (Sigma-Aldrich). Band intensity was quantified using Image J software.

Colony formation assay
Cells were plated in 6-well plates at 200 cells per well. After 1 day, they were incubated in the presence or absence of the indicated concentration of TKIs gefitinib and afatinib. Fresh media containing each TKI was added every 3 days. After 10 days, cells were fixed in methanol and stained with 0.5% crystal violet. The number of foci >1 mm was automatically determined using Image J software.

Statistical analysis
Error bars for all data represent standard deviation (s.d.) from the mean. P-values were calculated using Bonferroni correction for multiple comparisons using online statistical analysis tool (https://astatsa.com), as shown in each figure legend.

Design and preparation of tools for CRISPR-Cas9 genome editing
Using the CRISPR Design Tool (http://crispr.mit.edu), we designed sgRNAs targeting leucine 858 in EGFR exon 21 ( Fig 1A). For genome editing, we chose the CRISPR-Cas9D10A nickase (Cas9n) system, which minimizes off-target activity [23,24,26]. The human NSCLC cell line A549, which contains three copies of wild-type EGFR [27], was co-transfected with two EGFRtargeting Cas9n vectors and a ssODN (146 nucleotides, sense strand) donor containing a CTG!CGG mutation (boxed in Fig 1A) flanked by sequences homologous to the EGFR target site. Importantly, we synonymously mutated one nucleotide in the protospacer adjacent motif and five or six nucleotides in the 'seed' region (3 0 region) of the sense or anti-sense sgRNA sequence to prevent repetitive attack by Cas9n after successful integration of the exogenous sequence ( Fig 1A) [26]. Four days after transfection, we confirmed the presence of the L858R mutation within EGFR exon 21 in transfected cells by PCR using L858R mutation-specific primers (Fig 1B). Three clones were shown to harbor the L858R mutation (clones 82-12, 47-3, and 73-4). There was no significant difference in cell growth between mutant clones and parental cells under normal culture conditions (S1 Fig).

Establishment of cell lines harboring the L858R EGFR mutation
To confirm successful integration of the L858R mutation, the targeted genomic region was PCR-amplified and subcloned into a plasmid vector, and 20-30 bacterial colonies were sequenced for each clone. Sequence analysis showed that clone 82-12 contained both the L858R allele (boxed CGG) and the wild-type allele (Fig 2A). Clones 47-3 and 73-4 carried The EGFR copy number was determined by quantitative PCR using the diploid cell TIG3 as a control. E. The L858R EGFR copy number was calculated from the mutation ratio of L858R (C) and EGFR copy number (D). https://doi.org/10.1371/journal.pone.0229712.g002
Next, we investigated the EGFR copy number in each clone by quantitative PCR. Comparison with the normal human diploid cell line TIG-3 indicated that each clone carried nearly four copies of EGFR, while parental A549 cells had three copies as reported previously ( Fig  2D) [27]. From the results of the L858R mutation ratio (Fig 2C) and EGFR copy number ( Fig  2D), clones 82-12, 47-3, and 73-4 were inferred to have one, two, and two copies of the L858R EGFR mutation, respectively (Fig 2E).

Expression of the L858R EGFR mutant in established clones
We examined the expression of the L858R EGFR mutant in each clone at both the mRNA and protein level. First, we measured wild-type EGFR, L858R EGFR, and total EGFR mRNA expression using allele-specific primers (Fig 3A). Consistent with the genomic analysis (Fig  2), wild-type EGFR mRNA was expressed in parental cells and clone 82-12 but not in clones 47-3 or 73-4, while L858R EGFR mRNA was detected in all three genome-edited clones, with higher levels in clones 47-3 and 73-4 than in clone 82-12 (Fig 3A and 3B). The expression of total EGFR mRNA in clone 73-4 was lower than in clone 47-3 (Fig 3A and 3B). Of note, EGFR mRNAs containing a frame-shift (35 nt insertion or 47 nt deletion) in clone 73-4 were barely expressed, while the mRNA containing in-frame 51 nt deletion in clone 47-3 was clearly expressed (S2 Fig). Therefore, the lower total EGFR mRNA expression in clone 73-4 may be attributable to degradation of mutant mRNAs by nonsense-mediated decay (NMD) . Fig 3c and 3d show that the protein level of L858R EGFR in each clone correlated well with L858R EGFR mRNA expression. However, the expression of total EGFR protein in clones 47-3 and 73-4 was lower than in parental cells, suggesting the possibility that the mutant EGFR protein encoded by the indel mutants had been eliminated by a protein quality control system such as the endoplasmic reticulum-associated degradation (ERAD) pathway.

The influence of L858R EGFR mutant copy number on EGFR-TKI sensitization
We next examined the response of the cells to EGFR-TKIs (gefitinib and afatinib) and conventional anti-cancer drugs (cisplatin and taxol), which are standard treatment options in NSCLC [28,29]. All three clones were more sensitive to EGFR-TKIs than parental cells (Fig 4A), indicating that one or two copies of the L858R mutation was sufficient to render cells sensitive to EGFR TKIs, as observed in cells with amplified EGFR-mutant genes [18,30,31]. Intriguingly, clones 47-3 and 73-4 with two copies of the L858R mutation showed approximately 2.5-and 7-fold higher sensitivities to gefitinib and 25-and 10-fold higher sensitivities to afatinib than clone 82-12 with only one copy of L858R (Fig 4C). The sensitivity of clones 47-3 and 73-4 was comparable to that of HCC827, which harbors the amplified EGFR-activating mutant allele (del 19), but not to TKI-resistant cell line H1975, suggesting that EGFR-activating mutation contributes to TKI sensitivity even against a background of KRAS mutation (S3A and S3B  Fig). However, each clone and parental cell line exhibited similar sensitivities to the conventional anti-cancer drugs cisplatin and taxol (Fig 4B and 4C).
We further investigated the growth of each clone by the clonogenic survival assay in the presence of low-dose gefitinib (10 nM) and afatinib (0.3 nM) treatment. Afatinib treatment reduced the colony formation of each clone by 15% (parental), 35% (clone 82-12), 80% (clone 47-3), and 80% (clone 73-4) (Fig 5A and 5B). Similarly, strong growth inhibition was observed in gefitinib-treated clones with two copies of the L858R mutation (Fig 5A and 5B). Thus, clones with two copies of the L858R mutation were hypersensitive to TKIs compared with the clone with only one copy of the L858R mutation. L858R EGFR, wild-type EGFR, and total EGFR mRNA expression was analyzed by PCR using specific primer sets. B. mRNA expression was normalized to β-actin. Densitometric analysis of each band was performed by Image J software. For L858R EGFR mRNA, the expression in clone 82-12 was used as a control; for wild-type or total EGFR mRNA, parental expression was used as a control. C. L858R EGFR or total EGFR protein expression was analyzed by immunoblotting with an L858R-specific or anti-EGFR antibody. D. L858R EGFR or total EGFR protein was quantified as shown in B. https://doi.org/10.1371/journal.pone.0229712.g003

Sensitivity of cells with KRAS/EGFR double mutations to tyrosine kinase inhibitors
Finally, we investigated the EGFR phosphorylation status in each clone in the presence of gefitinib or afatinib. Gefitinib treatment decreased EGFR phosphorylation in a dose-dependent manner in each cell (Fig 6A and 6B). EGFR phosphorylation was decreased with a lower dose of gefitinib treatment in clones 47-3 and 73-4 compared with parental cells (Fig 6A and  6B). Similarly, downstream AKT phosphorylation was clearly decreased in clones 47-3 and 73-4. However, no significant difference was found in EGFR phosphorylation between parental cells and clone 82-12 (Fig 6A and 6B). EGFR phosphorylation in clones 47-3 and 73-4, but not in parental cells or clone 82-12, was significantly reduced with afatinib treatment (p<0.01) (Fig 6C and 6D). Thus, EGFR phosphorylation in clones 47-3 and 73-4 harboring two copies of the L858R mutation was markedly inhibited with lower doses of TKIs, which is consistent with the results of growth inhibition shown in Figs 4 and 5.

Discussion
In NSCLC, oncogenic driver mutations are typically mutually exclusive, although cases of mutations in multiple driver genes are increasingly reported [5,[7][8][9][10][11][12][13][14]. EGFR/KRAS co-mutation is likely to represent a certain proportion of cases of multiple mutations in NSCLC [12]. In the clinical setting, the limited prevalence of EGFR/KRAS co-mutated NSCLC means that its response to EGFR-TKI remains unclear. So far, several studies have reported the therapeutic effects of EGFR-TKI in patients with the co-mutations [9,12,32,33]. Zhang et al. reported that EGFR/KRAS co-mutation was detected in 38 out of 2106 patients (1.8%), and that a larger proportion of patients with the co-mutation obtained a better objective response rate (38.9% vs. 9.5%) and longer progression-free survival (8.0 vs. 1.5 months) following EGFR-TKIs therapy compared with patients having a KRAS mutation alone. However, these differences were not observed in patients treated with platinum-based chemotherapy [33]. Likewise, several other studies have described cases in which EGFR/KRAS co-mutation responded to

PLOS ONE
Sensitivity of cells with KRAS/EGFR double mutations to tyrosine kinase inhibitors EGFR-TKI therapy [9,12,32]. Taken together, these reports indicate that EGFR-TKI therapy may represent an effective treatment strategy against NSCLC with EGFR/KRAS co-mutation.
Clinically, it is difficult to determine whether multiple driver mutations detected in tumors are derived from different cells or from the same cell population. Although both cases have been assumed and reported [15][16][17], the available information is limited, particularly for the latter. In this regard, we have successfully generated the first example of a cancer cell model with double driver mutations. Using the CRISPR/Cas9 system, we generated isogenic cell lines with EGFR/KRAS co-mutations from the KRAS-mutated A549 cell line (Figs 1 and 2), and demonstrated that these co-mutated cell lines were more susceptible to EGFR-TKIs, but not to conventional anti-cancer drugs, than parental cells (Fig 4). This was rather surprising, because KRAS is a signaling mediator downstream of EGFR and KRAS mutation was reported to be involved in TKI-resistance [34][35][36][37]. The dominant role of EGFR mutation over the KRAS mutation on the sensitivity to TKIs was supported by the result that HCC827 harboring EGFR mutation (ex19 dels) alone showed similar IC50 value to EGFR/KRAS co-mutated cells generated from A549 (S3 Fig). It should be noted that EGFR del 19 mutant is prone to be sensitive to TKIs compared to L858R mutant [38,39]. To confirm these results, it is important to compare the sensitivity to EGFR-TKIs of the cell lines harboring EGFR mutation alone, KRAS mutation alone, and EGFR/KRAS co-mutations under the same genetic background.
Interestingly, clones with two copies of EGFR mutation (clones 47-3 and 73-4) harbored three and four copies of KRAS mutation, respectively (S4 Fig) and clone 73-4 with four copy KRAS mutation was prone to be resistant to gefitinib but not afatinib in a clonogenic survival assay (Fig 5B). In line with this, it was recently reported that NSCLCs with KRAS mutation transiently respond to a first-generation TKI, whereas they escape prolonged suppression through the activation of other ERBB family members while the second-generation TKI afatinib persistently suppresses tumor growth via broad inhibitory effects on the ERBB family [40,41]. These observations suggest that although EGFR mutation plays a primary role on the sensitivity to EGFR-TKIs, co-mutation of KRAS also affects the sensitivity to TKIs.
In summary, our findings demonstrated that the coexistence of EGFR mutation in KRASmutated cells enhances TKI sensitivity, especially in the presence of a higher EGFR mutation copy number. Given that no effective targeted therapy against KRAS-mutated NSCLC is currently available, the identification and quantification of druggable mutations occurring concomitantly in tumors may provide an opportunity for the development of alternative treatment options for affected patients.