Biomarker LEPRE1 induces pelitinib-specific drug responsiveness by regulating ABCG2 expression and tumor transition states in human leukemia and lung cancer

Biomarkers for treatment sensitivity or drug resistance used in precision medicine include prognostic and predictive molecules, critical factors in selecting appropriate treatment protocols and improving survival rates. However, identification of accurate biomarkers remain challenging due to the high risk of false-positive findings and lack of functional validation results for each biomarker. Here, we discovered a mechanical correlation between leucine proline-enriched proteoglycan 1 (LEPRE1) and pelitinib drug sensitivity using in silico statistical methods and confirmed the correlation in acute myeloid leukemia (AML) and A549 lung cancer cells. We determined that high LEPRE1 levels induce protein kinase B activation, overexpression of ATP-binding cassette superfamily G member 2 (ABCG2) and E-cadherin, and cell colonization, resulting in a cancer stem cell-like phenotype. Sensitivity to pelitinib increases in LEPRE1-overexpressing cells due to the reversing effect of ABCG2 upregulation. LEPRE1 silencing induces pelitinib resistance and promotes epithelial-to-mesenchymal transition through actin rearrangement via a series of Src/ERK/cofilin cascades. The in silico results identified a mechanistic relationship between LEPRE1 and pelitinib drug sensitivity, confirmed in two cancer types. This study demonstrates the potential of LEPRE1 as a biomarker in cancer through in-silico prediction and in vitro experiments supporting the clinical development of personalized medicine strategies based on bioinformatics findings.


Results
Biomarkers in blood cancer cells screened for six EGFR signaling pathway drugs. To select candidate genes in blood cancer cells associated with EGFR signaling pathway drugs, data on 167 blood cancer cell lines from GDSC1 and six EGFR signaling pathway drugs were analyzed. We found that LEPRE1 or CCRL2 overexpression was associated with pelitinib sensitivity, and TAX1BP3 overexpression was associated with afatinib sensitivity (Fig. 1a). A positive IC 50 value indicated that a particular gene was associated with increased drug resistance, whereas a negative value indicated increased drug sensitivity.
Although statistical analyses of the entire samples are generally reliable, it is also important to consider the drug responsiveness of differentially expressed samples. We used differentially expressed samples to prioritize the candidates. Figure 1b shows the difference in drug responsiveness between the two groups of cells where a gene was overexpressed in one group (z-score > 1) and under-expressed in the other (z-score < − 1). Under-expression of IL21R or RUNX3 was associated with afatinib resistance. The association between LEPRE1 and pelitinib, which exhibited the highest IC 50 , was further evaluated using drug responsiveness and mechanism studies. LEPRE1 expression and pelitinib sensitivity differ in AML cell lines. To examine LEPRE1 expression in AML cells, we evaluated expression levels in four different AML-derived cell lines, THP-1, U937, KG-1, and HL60 (Fig. 1c). LEPRE1 was strongly expressed in THP-1 and U937 cells, however, a weaker expression was detected in KG-1 and HL60 cells. We then confirmed protein expression levels of EGFR in the AML cell lines. EGFR protein was observed in only THP-1 and KG-1 cells (Fig. 1d).
Recently, there have been several reports on the ability of EGFR inhibitors and EGFR-TKIs to induce differentiation or death of AML cells, which is generally considered an off-target effect 25,26 . To determine whether pelitinib affected AML cell viability, we exposed the cells to pelitinib at concentrations ranging from 0 to 10 μM for 72 h. A dose-dependent decrease in cell viability was observed for the EGFR-positive cell lines THP-1 and KG-1, but not for U937 or HL60 cell lines (Fig. 1e). When U937 and HL60 cells were treated with pelitinib, differentiated cell morphology was observed (Fig. 1f). From these results, we selected the KG-1 and THP-1 cell lines for gain-of-function and loss-of-function evaluation, respectively.

Verification of bioinformatics prediction regarding LEPRE1 and pelitinib drug responsiveness in AML.
To elucidate the effects of LEPRE1 expression in AML, siRNAs or plasmid DNA were electroporated into THP-1 and KG-1 cells. First, we investigated the efficacy of siRNAs against endogenous LEPRE1 in THP-1 cells. Based on quantitative immunoblotting results, si2293 suppressed LEPRE1 expression to 83% of the negative control ( Supplementary Fig. 1a). The ectopic expression of LEPRE1/pcDNA3.1 significantly increased the LEPRE1 protein levels in the KG-1 cell line compared to the vector control ( Supplementary Fig. 1b).
To examine the role of LEPRE1 in pelitinib cytotoxicity, cells transfected with LEPRE1 (KG-1_pcDNA3.1, KG-1_LEPRE1, THP-1_NC, and THP-1_siLEPRE1) were treated with pelitinib. Analyses of pelitinib cytotoxicity and expression of cleavage-PARP (c-PARP) revealed that LEPRE1 overexpression resulted in increased pelitinib sensitivity compared with that of control cells receiving the empty vector (Fig. 2a,b). In contrast, THP-1_ siLEPRE1 transformed cells showed resistance to pelitinib compared to the negative control cells (Fig. 2c,d).
We evaluated the influence of LEPRE1 on the EGFR signaling pathway in KG-1 and THP-1 cells for functional analyses and found that p-ERK in the LEPRE1-overexpression group shared a repressive tendency compared with that of KG-1 control cells transformed with empty vector. In the siLEPRE1 group, p-ERK levels were increased  www.nature.com/scientificreports/ compared to the THP-1 negative control cells (Fig. 2e). These data demonstrated that LEPRE1 might be related to the ERK signaling pathway. Next, we assessed the effect of pelitinib on the EGFR signaling pathways in these cell lines. Whole-cell lysates from KG-1 and THP-1 cells were evaluated for total EGFR, AKT, and ERK levels by western blotting after 24 h of exposure to pelitinib. In KG-1_LEPRE1 cells, p-ERK levels were reduced in a pelitinib dose-dependent manner (Fig. 2f). The resistance of p-ERK was observed in THP-1_LEPRE1 cells (Fig. 2g). These results indicate that LEPRE1 inhibition promoted pelitinib resistance, whereas LEPRE1-overexpression induced high pelitinib sensitivity.
LEPRE1 expression is associated with pelitinib-specific sensitivity/resistance in AML. We then examined the chemosensitizing effect of various EGFR-TKIs (afatinib, erlotinib, gefitinib, lapatinib, and pelitinib) in AML cell lines and investigated the association between LEPRE1 expression levels and sensitivity to EGFR-TKI treatment ( Supplementary Fig. 2). Pelitinib produced the most dramatic inhibitory effect on KG-1_LEPRE1 cells, which had an IC 50 value of 0.329 μM compared to KG-1_pcDNA3.1 cells. Afatinib and lapatinib both inhibited the proliferation of KG-1_LEPRE1 cells, although they were less potent than pelitinib (Fig. 3a). Erlotinib and gefitinib exhibited IC 50 values of more than 10 μM and failed to inhibit proliferation of either pcDNA3.1-transfected or LEPRE1-transfected KG-1 cells. Only THP-1_ siLEPRE1 showed resistance to pelitinib compared to negative controls (Fig. 3b). Erlotinib, gefitinib, and lapatinib had IC 50 values exceeding 10 μM in both negative control and LEPRE1 knockdown THP-1 cells and failed to inhibit proliferation.
Immunoblotting was performed to determine whether sensitivity to EGFR-TKIs was mediated through EGFR signaling inhibition. Consistent with the proliferation results of the WST-1 assays, pelitinib-treated KG-1_LEPRE1 cells demonstrated decreased p-ERK and increased c-PARP levels (Fig. 3c). Meanwhile, THP-1_ siLEPRE1 cells showed increased p-ERK expression and decreased c-PARP compared with the THP-1_Negative control cells (Fig. 3d). These results suggest that LEPRE1 expression was associated with pelitinib-specific sensitivity or resistance in AML cells, consistent with the bioinformatics data shown in Fig. 1a,b.

LEPRE1 contributes to pelitinib responsiveness in A549 cells via EGFR-independent signaling.
Pelitinib is a potent irreversible EGFR-TKI that has been evaluated in clinical trials for the treatment of lung cancer 27,28 . As there are many studies regarding EGFR-TKI targeting NSCLC, we verified the drug responsiveness of pelitinib relative to LEPRE1 levels in EGFR-expressing human lung cancer-derived A549 cells.
To evaluate pelitinib cytotoxicity, LEPRE1 overexpression or knockdown A549 cells were treated with pelitinib. Pelitinib-induced sensitivity in the A549 cells was consistent with the AML-based results (Fig. 2a,c). LEPRE1 overexpression correlated with greater pelitinib sensitivity (Fig. 4a). In contrast, LEPRE1-knockdown A549 cells showed greater resistance to pelitinib compared to the negative control cells.
Expression of downstream signaling molecules in A549 cells changed according to the level of LEPRE1 expression (Fig. 4b). Similar to the AML results, LEPRE1 regulated AKT/ERK and EGFR activity in A549 cells. To confirm the EGFR pathway inhibition by pelitinib treatment, transfected A549 cells were treated with hEGF after 24 h pre-treatment with pelitinib. EGF treatment stimulated EGFR phosphorylation and increased AKT and ERK phosphorylation. However, when cells were pre-treated with pelitinib, EGFR phosphorylation was inhibited in both A549_LEPRE1 and A549_siLEPRE1 cells in a dose-dependent manner (Fig. 4c,d). In addition, reduced p-ERK was detected in LEPRE1-overexpressing A549 cells in pelitinib dose-dependently (Fig. 4c). However, a decrease in p-ERK relative to pelitinib concentration was not observed in LEPRE1-knockdown cells (Fig. 4d). These results support that LEPRE1 regulated ERK activation via an EGFR-independent pathway. In conclusion, the loss of LEPRE1 expression decreased pelitinib-induced apoptosis and promoted pelitinib resistance in A549 cells via the activation of ERK. Furthermore, the therapeutic efficacy of pelitinib increased in A549 cells in conjunction with higher LEPRE1 expression. Pelitinib-specific drug sensitivity was also observed in A549 cells ( Supplementary Fig. 3).

LEPRE1 affects cell motility and cell morphology of A549 cells.
To further investigate the role of LEPRE1 in pelitinib responsiveness, we evaluated the cellular morphology and migration of A549 cells. Cell proliferation was comparable among the parental, LEPRE1-overexpression, vector control cells, and LEPRE1knockdown and negative control cells (Fig. 5a). In terms of cell morphology, LEPRE1-overexpression cells were more likely to adhere to each other and typically formed cell clusters compared to vector control cells (Fig. 5b, upper images). In contrast, LEPRE1-knockdown A549 cells displayed a mesenchymal phenotype, including decreased cell-cell contacts and increased pseudopodia formation (Fig. 5b, lower images). These characteristics are consistent with the typical cellular structure of EMT.
EMT is a physiological process used by tumor cells to acquire critical oncogenic features, such as migration/invasion, stemness, and drug resistance 29 . Therefore, we then examined whether LEPRE1 levels affected cell migration, invasion, and colony formation. The rate of cell growth was not affected by LEPRE1 (Fig. 5b). Accordingly, we concluded that the differences observed in the migration experiments were mainly due to the influence of LEPRE1 on cell motility. The results showed that LEPRE1-overexpression suppressed the migration and invasion ability of A549 cells, while LEPRE1-knockdown promoted migration and invasion (Fig. 5c,d). Based on colony formation assays, ectopic LEPRE1 expression promoted, and LEPRE1 knockdown suppressed colony formation in soft agar (Fig. 5e). Considering these results, we suggest that LEPRE1 modulated the cell fate of clonogenic properties or EMT by regulating cell morphology and migration/invasion. Moreover, we further investigated the growth of each clone in the presence of pelitinib (0, 0.1, 1, and 10 μM) using clonogenic survival assays. Pelitinib treatment significantly reduced the number of colonies in a dosedependent manner. We also observed that the overexpression of LEPRE1 resulted in greater pelitinib sensitivity.

LEPRE1 regulates EMT/mesenchymal-epithelial transition (MET) state via EGFR-independent signaling pathways.
To further investigate the mechanism of LEPRE1 affecting pelitinib responsiveness in A549 cells, we focused on the EGFR signaling pathway, multidrug resistance (MDR), actin rearrangement, and EMT. In LEPRE1-knockdown cells, EGFR and p-EGFR expression were decreased compared with those in the www.nature.com/scientificreports/ negative control cells. However, the p-EGFR/EGFR ratio was increased in LEPRE1-knockdown and decreased in LEPRE1-overexpressing A549 cells (Fig. 6a).
To determine the cause of drug sensitivity/resistance according to the level of LEPRE1 expression, we evaluated the expression of ABC transporters ABCB1 and ABCG2. ABCG2 expression levels in LEPRE1-knockdown cells were lower than those in the negative control cells (Fig. 6b). In contrast, ABCG2 expression was increased in LEPRE1-overexpressing A549 cells (Fig. 6b).
According to a previous study, cofilin expression is increased in primary fibroblasts of osteogenesis imperfecta patients with mutations in the LEPRE1 gene 30 . To evaluate cofilin expression in transfected A549 cells, we performed western blotting for cofilin and p-cofilin. The levels of cofilin and p-cofilin and the p-cofilin/cofilin ratios were increased in LEPRE1-knockdown cells compared with those in the negative controls. In contrast, the expression of cofilin and p-cofilin and the p-cofilin/cofilin ratio were decreased in LEPRE1-overexpressing A549 cells (Fig. 6c).
To determine whether LEPRE1 induced EMT in A549 cells, we performed western blot analysis to evaluate the levels of EMT-related proteins, including vimentin and E-cadherin. As shown in Fig. 6d, E-cadherin levels were decreased, and vimentin increased when LEPRE1 was knocked down. In contrast, increased E-cadherin levels and decreased vimentin levels were observed in LEPRE1-overexpressing cells. These results indicated that LEPRE1 might be involved in EMT/MET regulation and regulate the tumor state via either EGFR-dependent or EGFR-independent pathways.
EMT is a crucial step of metastasis, it is therefore essential to determine whether LEPRE1 expression is associated with the clinical outcome of cancer patients. We performed Kaplan-Meier survival analyses on lung squamous cell carcinoma (LUSC) from human protein atlas (https:// www. prote inatl as. org/, Supplementary  Fig. 4), and found low expression of LEPRE1 to be consistently associated with poor clinical outcomes during the early stages of cancer; meanwhile, LEPRE1 overexpression is associated with poor clinical outcomes during the late stages of cancer. This suggests LEPRE1 as a novel therapeutic target against cancer metastasis at various cancer stages.

LEPRE1 induces pelitinib-specific drug responsiveness by regulating ABCG2 expression.
To confirm the relationship between LEPRE1 and ABCG2 role in pelitinib-induced responsiveness, LEPRE1 or ABCG2 plasmid was transfected with single or co-transfection (Fig. 7a). In ABCG2-overexpressing A549 cell, the LEPRE1 expression was increased, along with a slight increase in the p-ERK levels, compared to pcDNA3.1 vector control. The reduced p-ERK expression was detected in LEPRE1 and ABCG2 co-transfected cell, and was more reduced compared to LEPRE1-overexpressing cells.
To examine the role of LEPRE1 and ABCG2 in pelitinib cytotoxicity, transfected cells were treated with pelitinib (Fig. 7a). Pelitinib produced the most significant inhibitory effect on LEPRE1 + ABCG2 co-transfected cells, with an IC 50 value of 1.139 μM compared to A549_pcDNA3.1 cells. In LEPRE1-or ABCG2-expressing cells, pelitinib inhibited proliferation, although with less potency than co-transfected cells (Fig. 7d).
Then, we investigated the efficacy of siRNAs against endogenous ABCG2 in A549 cells. Based on quantitative immunoblotting results, si1176 suppressed ABCG2 expression to 67% of the negative control (Fig. 7b). To assess the knockdown effect of ABCG2 in pelitinib induced drug sensitivity, we transfected siRNA against ABCG2 and LEPRE1 single or co-transfection (Fig. 7c). In ABCG2 and LEPRE1 double knockdown cells, the p-ERK density was increased along with the IC 50 value compared to negative control cells. The siABCG2 single transfected cells have similar IC 50 value to that of A549_siLEPRE1 cells following pelitinib treatment (Fig. 7e).
These results suggest that LEPRE1 and ABCG2 regulate pelitinib responsiveness by regulating p-ERK expression.

Discussion
In this study, we performed a bioinformatics meta-analysis based on the GDSC and COSMIC databases, focusing on EGFR signaling pathway drugs, including pelitinib. We found that increased LEPRE1 expression was related to pelitinib sensitivity. In-silico result-based predictions were experimentally validated in AML and A549 cell lines using drug response and mechanism studies. Notably, we first report LEPRE1 as a putative biomarker for the selection of pelitinib to treat AML and lung cancer.
LEPRE1 is associated with type VIII osteogenesis imperfecta and has been extensively investigated because of its protective function in cell homeostasis and its correlation with osteogenesis imperfecta 31 . Interestingly, cancer is rarely concurrently seen in patients with osteogenesis imperfecta 32 , and it has been suggested that these patients experience some cancer protection; however, there is currently no direct evidence to support this hypothesis 33 . It has also been reported that an increase in EGFR induces a decrease in LEPRE1 expression in breast cancer 34 . Recently, the possibility of direct binding between EGFR and LEPRE1 has been suggested 35 . We confirmed the binding of LEPRE1 to EGFR using immunoprecipitation experiments (data not shown). Thus, a mechanistic molecular relationship exists between EGFR and LEPRE1 in tumor development.
Pelitinib (EKB-569) is an irreversible EGFR-TKI that covalently binds and inhibits EGFR 23 . Interestingly, pelitinib decreases p-AKT and p-ERK protein levels in hepatocellular carcinoma, where EGFR-targeted therapy is ineffective 22 . These reports indicate that pelitinib has various targets and can exert its drug activity in an EGFRindependent manner. Thus, pelitinib drug responsiveness relative to LEPRE1 expression levels was effective in AML (EGFR-low) and A549 (EGFR-high) cell lines.
We also found that high LEPRE1 expression in A549 cells induced AKT activation, the overexpression of ABCG2 and E-cadherin, and cell colonization. In LEPRE1-overexpressing cells, EGFR expression was increased, but no significant change was observed in the p-EGFR/EGFR ratio compared with control cells. Endogenous LEPRE1 in A549 cells was expressed in the ER, peri-nucleus regions, and vesicles (data not shown). It has also  www.nature.com/scientificreports/ been previously reported that membrane protein trafficking is regulated by PI3K/AKT signaling 36 . Accordingly, we speculated that LEPRE1 might regulate the trafficking of membrane proteins EGFR and ABCG2 via the AKTdependent signaling pathway. In many kinds of human tumor cells, ABCG2 expression on the plasma membrane contributes to MDR during chemotherapy, and ABCG2 is also known to be a marker for identifying cancer stem cells (CSCs) in lung cancers 37 . According to Wang et al. 38 , ABCG2 expression and AKT activation were positively associated with CSC-like phenotypes. Increased ABCG2 expression causes an increase in E-cadherin, inhibits cell migration, induces MET progression, and promotes higher lung metastatic colonization 37 . However, the interactions between LEPRE1 and EGFR/AKT/ABCG2 signaling concerning CSC-like phenotypes remain elusive and require further clarification. Pelitinib has been shown to exhibit high-affinity interaction with the ABCG2 transporter, which acts as a competitive inhibitor of ABCB1/ABCG2 to selectively kill CSC-like cells after hyperthermia in lung cancer 27 . According to the "cancer stem cell" hypothesis, these drug-resistant CSCs are responsible for driving tumor re-growth, and ABCG2 is probably a pivotal efflux transporter that contributes to preserving the CSC sanctuary under chemotherapeutic pressure 39 . In ABCG2-expressing tissues where local drug concentrations are high, the drugs can inhibit ABCG2 function and promote intracellular accumulation and cytotoxic activity, thereby reversing MDR 40 , supporting the hypothesis that the reversing effect of pelitinib in LEPRE1-overexpressing cells is due to increased cytotoxicity caused by ABCG2 overexpression. Our results showed that A549_LEPRE1 + ABCG2 cells were more sensitive to pelitinib than LEPRE1 or ABCG2 expressing cells. Meanwhile, A549_siLEPRE1 + siABCG2 cells were more resistance to pelitinib then siLEPRE1 or siABCG2 transfected cells. This suggests that LEPRE1 may regulate ABCG2 expression while pelitinib enhances accumulation, thereby restoring cellular sensitivity toward ABCG2 substrate cytotoxic drugs by directly blocking ABCG2 function.
LEPRE1 silencing reduced EGFR expression in A549 cells but increased the p-EGFR/EGFR ratio. According to Liu et al. 12 , receptor tyrosine kinases (RTKs) and the abnormal activation of downstream signaling molecules have compensatory functions against the inhibition of EGFR by triggering the PI3K/AKT and MAPK signaling cascades. Activation of bypass mechanisms of resistance to RTK inhibition via other RTKs or alternative downstream signaling molecules also contributes to acquired resistance 41 . From these pieces of evidence, we speculated that the activation signals by the EGFR-independent pathway induced Src and ERK phosphorylation, ultimately affecting the p-cofilin/cofilin ratio and vimentin overexpression.
In addition, knockdown of LEPRE1 in cells resulted in an EMT-like phenotype, and migration and invasion were increased. According to previous reports, Src mediates the migration and invasion of breast cancer and hepatocellular carcinoma cells through the downstream activation of the ERK pathway [42][43][44] . It is also known that Src activation causes actin depolymerization via the MEK/ERK/Cofilin cascade, inducing tumor transformation 45 . Oncogenic transformation is accompanied by morphological changes due to changes in actin dynamics and adhesion activity. Furthermore, cofilin-1 has critical roles in switching from EMT and promoting cell migration and invasion by regulating actin cytoskeleton organization 46 . In addition, an altered cytoskeleton and nuclear lamina organization occur in fibroblast cells of patients with osteogenesis imperfecta carrying mutations in LEPRE1 30 . Thus, LEPRE1 silencing induces EMT through actin reconstitution via a series of Src/ ERK/Cofilin cascades.
Due to morphological and physiological changes found in EMT-induced cells, the effectiveness of EGFR-TKIs is limited by acquired resistance, which represents the classical paradigm of molecular-targeted therapies in NSCLC. Recently, a growing number of studies have suggested a relationship between EMT and drug resistance. Along with the decreased expression of E-cadherin, vimentin expression is elevated, illustrating the emergence of EMT 47 . For example, a patient with lung cancer who developed acquired resistance to erlotinib exhibited EMT properties in the tissue sample. In addition, a gefitinib-resistant subline of HCC827 cells shows phenotypic and molecular changes that are consistent with EMT 48 . Although the mechanisms that regulate EMT are non-linear complex networks 49 , we uncovered one mechanism of EMT induction and drug resistance due to reduced LEPRE1 levels (Fig. 7).
Based on our results, we can suggest a method of administering drugs in which EGFR-TKIs resistance caused by LEPRE1-AKT-ABCG2 overexpression in CFC-like cells can be overcome with pelitinib treatment. Furthermore, maximum drug effects can be expected when pelitinib and other drugs are administered in combination. It may also be possible to overcome pelitinib resistance caused by EMT induction in the absence of LEPRE1 by inhibiting Src or MECK signaling. Blockage of the MAPK pathway is possible using MEK inhibitors 50 , including trametinib (GSK1120212), which is a MEK1/2 inhibitor with a longer half-life than previous MEK inhibitors 51,52 . In addition, a synergistic effect can be induced by co-administering pelitinib and various other anti-cancer drugs to advanced cancer patients that express high levels of LEPRE1 and ABCG2. This finding highlights the possible clinical benefit of combining targeted EGFR inhibitor-based therapies with ABCG2 substrate conventional chemotherapeutic drugs.
In conclusion, we verified the expression of LEPRE1 as a candidate biomarker and the reactivity of the EGFR-TKI drug pelitinib using in vitro experiments executed about computed values derived from bioinformatics. Through this process, we confirmed the reliability of the in-silico prediction results. In addition, we provided a function of LEPRE1 as a new biomarker and its novel mechanism for EGFR-TKI drug responsiveness in various AML-derived and lung cancer-derived cells. Indeed, we found no direct association between LEPRE1 and EGFR in the drug reactivity of pelitinib. However, we find it encouraging that the values calculated by the artificial intelligence computer can consider complex internal molecular pathways rather than predict simple target-drug pairings. There is a need to derive many new biomarkers for the development of customized anticancer drugs and accurate cancer diagnosis, which can be advanced based on in-silico analysis. In addition, it is necessary to investigate the molecular mechanisms of EMT, MET, and tumor metastasis and determine the reaction mechanism of related drugs.  24 . The IC 50 values were determined by fitting curves using non-linear least-squares regression in a sigmoidal dose-response model with a variable slope using Origin 6.1 software (OriginLab Corporation, Northampton, MA, USA). Each assay was performed in triplicate for each drug concentration, and assays were independently repeated thrice.
Proliferation assays. Cellular growth curves were plotted using cellular viability values assessed using the WST-1 method. Cells were seeded into 96-well plates at a density of 3 × 10 3 cells/well in 100 μL of culture medium. At various time points (0, 24, 48, and 72 h) after seeding, the cells in each well were treated with 10 μL of the WST-1 solution for 1 h at 37 °C. Absorbance was subsequently determined using a microplate ELISA SpectraMax iD3 Microplate Reader (Molecular Devices) and analyzed using SoftMax Pro 7.1 software. Wavelengths of 450 nm were used for measurements and 650 nm as reference.
Transwell migration and invasion assays. Transwell migration and invasion assays were performed using 24-well Transwell (8-μm pore size; Corning Life Sciences, Bedford, MA, USA). For migration assays, 5 × 10 4 cells were plated into the upper insert containing 0.1% FBS culture medium with the lower chamber containing 10% FBS culture medium and incubated at 37 °C under 5% CO 2 for 24 h. Cells that migrated through the membrane were fixed and stained with crystal violet (Sigma-Aldrich). Invasion assays were performed in 24-well Matrigel (Corning Life Sciences) coated chambers. Again, 5 × 10 4 cells were plated into the upper insert containing 0.1% FBS culture medium with the lower chamber containing 10% FBS culture medium and incubated for 24 h. Cells that invaded through the Matrigel were fixed and stained with crystal violet. The cells were counted in four randomly selected microscopic visual fields. Each sample was assayed in triplicate, and three independent assays were performed. The mean ± SEM was calculated for each sample.
Soft agar colony-forming assays. Colony formation was assessed using the CytoSelect 96-Well Cell Transformation Assays kit (Cell Biolabs, Inc., San Diego, CA, USA) according to the manufacturer's protocol. Briefly, a base agar layer of 0.6% agar, Dulbecco's modified Eagle's medium, and 10% FBS was poured into each well and covered with a cell agar layer of the same composition containing a density of 3 × 10 3 cells/well plus pelitinib at various concentrations (0, 0.1, 1, and 10 μM). After 7 days, the agar was solubilized, and the mixture was incubated with a CyQuant working solution. Fluorescence was measured at 485/520 nm using a 96-well microplate reader. Growth inhibition rates were calculated relative to untreated control cultures. www.nature.com/scientificreports/ Statistical analysis. Values are presented as mean ± SEM of experiments conducted with replicates as indicated (n). Data were analyzed using Origin 6.1 software (OriginLab, Northampton, MA, USA) and/or Microsoft Excel. Differences between control and treated groups were analyzed using one-way analysis of variance (ANOVA) followed by a post-hoc Dunnett's pairwise comparison or linear trend test to determine a dosedependent effect.

Data availability
All data generated and analyzed during the current study are available from the corresponding author on reasonable request.