A genome-wide genetic screen reveals the P2Y2-integrin axis as a stabilizer of EGFR mutants in non–small cell lung cancer (NSCLC)

Abstract

Activating mutations in the epidermal growth factor receptor (EGFR) gene drive non–small cell lung cancer (NSCLC). Oncogenic EGFR mutants are ligand-independent and more stable, but the underlying mechanism remains unclear. We hypothesized that EGFR mutants selectively leverage cellular stabilizers to evade degradation. Genome-wide RNA interference screens identified genes (encoding for stabilizers) responsible for mutant EGFR stability, with P2Y2 receptor (P2Y2) emerging as a bona fide stabilizer. Mechanistically, high extracellular adenosine triphosphate (ATP) levels transactivate EGFR mutants via P2Y2 activation, previously shown to signal through Src kinase–dependent EGFR phosphorylation. Our study reveals that ATP-driven P2Y2 activation stabilizes EGFR mutants by forming a P2Y2-integrin β1-EGFR complex enriched in endosomes. Targeting this axis destabilizes EGFR mutants and offers a strategy against drug resistance. Elevated P2Y2 and integrin β1 expression in patients with NSCLC implies clinical relevance. Our results provide previously unidentified insight that EGFR mutants enhance extracellular ATP levels to activate P2Y2-integrin for enhanced stability of EGFR mutants to drive the oncogenic program.

The ATP, P2Y2, and integrin signaling axis helps EGFR mutants survive in lung cancer, pointing to a target to tackle resistance.

INTRODUCTION

Lung cancer is the leading cause of cancer-related deaths globally (1). Around 85% of all lung cancer cases are attributed to non–small cell lung cancer (NSCLC). Activating mutations in the epidermal growth factor receptor (EGFR) gene are detected in up to 40 to 60% of lung adenocarcinoma (a type of NSCLC) in Southeast Asia (2, 3). The primary EGFR mutations, 19del and L858R, make up around 90% of all EGFR mutation–positive NSCLC tumors (4) and respond well to EGFR tyrosine kinase inhibitors (TKIs) (5–7). Although there have been notable breakthroughs in the development of TKI treatments, the emergence of acquired resistance to EGFR TKIs attributed to secondary and tertiary mutations, such as the T790M mutation (8, 9) and C797X (10), respectively, contributes to treatment failure (11). Although several mechanisms can lead to acquired drug resistance to TKIs in NSCLC (12–15), the most common is resistant mutations within the EGFR gene (8–10, 16–18). New EGFR TKIs are under development to counter drug resistance associated with EGFR mutations (10, 19, 20). This underscores the compelling need for ongoing efforts to create innovative therapies that can effectively tackle EGFR mutations and evolving drug resistance.

For alternative and new treatment options to target EGFR, it is essential to explore the uncharted properties of EGFR-mutant proteins. Our investigations, along with others, have revealed that activating mutations (such as 19del and L858R) impede their degradation, resulting in enhanced stability and the sustenance of oncogenic signaling by the mutant EGFR (21–23). We hypothesize that EGFR mutations have an inherent property to evade degradation by relying on cellular mechanisms for their enhanced stability. Understanding the molecular mechanism responsible for the enhanced stability of mutant EGFR is critical for the development of alternative treatment options. Here, we established a genome-wide small interfering RNA (siRNA) screen pipeline to identify genes encoding candidate proteins (referred to as stabilizers) crucial for stabilizing the EGFR mutant.

Among the candidate stabilizers, we focused our functional and mechanistic studies on P2Y2, a member of purinergic receptors that transduces cellular signaling upon binding to adenosine triphosphate (ATP) or uridine triphosphate (UTP) (24). P2Y receptors, a subgroup of purinergic G protein (heterotrimeric guanine nucleotide–binding protein)–coupled receptors (GPCRs), are activated through the binding of extracellular nucleotides to initiate cellular signaling to regulate proliferation, differentiation, adhesion, and migration (25). Previous studies have shown that P2Y2 contributes to tumor progression through multiple mechanisms, including its interaction with integrins and its ability to transactivate EGFR via Src-dependent phosphorylation (26–30). Our results demonstrate that P2Y2 is highly expressed in patients with NSCLC and plays a critical role in sustaining mutant EGFR by preventing its degradation, a process driven by the selective elevation of extracellular ATP in EGFR-mutant cells. P2Y2 stabilizes EGFR through its interaction partner integrin β1 by forming the EGFR-P2Y2-integrin complex. Targeting the EGFR-integrin β1-P2Y2 signaling axis presents a promising therapeutic strategy, achievable either through small-molecule inhibition of P2Y2 or by disrupting integrin-mediated signaling pathways. We also show that small-molecule P2Y2 antagonists effectively down-regulate the protein levels of oncogenic EGFR variants and reduce the tumorigenic characteristics of NSCLC both in vitro and in vivo (29, 31, 32). Our results establish P2Y2 as a stabilizer of EGFR mutants, and targeting the EGFR-P2Y2-integrin β1 axis can offer potential therapeutic benefit in overcoming drug resistance resulting from EGFR mutations in NSCLC.

RESULTS

RNAi screen to identify the candidate genes in stabilizing the EGFR-activating mutant

Aberrations in sorting EGFR to lysosomes for degradation, particularly in the presence of constitutively active receptors, could play a key role in the initiation and progression of tumors (33). It has been shown earlier that the EGFR mutant avoids degradation, but the wild-type EGFR (EGFR-WT) is efficiently degraded (Fig. 1A) (21–23). To identify genes required for stabilizing EGFR-19del, we have engineered H1299 cells, which have low levels of endogenous EGFR, stably expressing EGFR-19del fused with green fluorescent protein (GFP) at the C terminus. The intensity of GFP reflects the levels of EGFR-19del. Our objective was to identify genes that, upon silencing, result in the decrease in the GFP signal as a measure of EGFR-19del protein levels (Fig. 1B). We carried out a whole genome-wide RNA interference (RNAi) screen using a siRNA library targeting 21,026 human genes in arrayed SMARTpool format. The H1299 cells stably expressing EGFR-19del-GFP were reverse transfected using siRNAs in 384-well plates. After 72 hours, the cells were fixed, imaged for nucleus and GFP fluorescence, and subjected to data analysis. We derived a cytoplasmic mask using a proximal ring relative to the nucleus position, which includes EGFR at endosomes and the plasma membrane. The mean EGFR-19del GFP pixel intensity was then extracted. A normalization of mean GFP pixel intensity for each tested well was applied relative to plate key controls and produced a transformed distribution of tested siRNA at “0” for nontargeting control (siControl) level to “−1” for the siEGFR level with maximal suppression of signal (Fig. 1C). The distribution of genes in a genome-wide screen is presented in Fig. 1D. The Z′ analysis score for siControl and siEGFR controls in each plate was >0.7, suggesting good consistency for a cell-based assay. The RNAi screen was conducted in duplicate and produced a Pearson correlation coefficient r = 0.72 and a coefficient of determination R² = 0.52, suggesting a reproducible outcome between the two independent screens. We focused our hit selection on reproduced positive hits from both screens with a loss of EGFR-GFP signal above 70% (score ≤−0.7) and recovered a list of 110 genes selected (Fig. 1D, inset).

Fig. 1. Genome-wide RNAi screen to identify candidate stabilizers of EGFR mutants.

(A) EGFR degradation assay using H1299 stable cells constantly expressing EGFR-WT and EGFR-19del mutation. Immunoblot of EGFR and actin, which serves as a loading control. (B) Schema to identify genes encoding potential EGFR stabilizers using an RNAi-based approach. (C) Schematic workflow of the high-throughput genome-wide siRNA screen with the H1299-EGFR 19del cells. Specific siRNAs printed on 384-well plates were reverse transfected to the cells, fixed, and analyzed for measuring the EGFR-GFP signal. (D) For the primary screen, 21,026 genes were analyzed. siRNA screening was performed in duplicate. EGFR-GFP intensity data for each gene across the genome from the second screen are displayed in a dot plot. The y axis represents the normalized GFP intensity scores of the EGFR-19del-GFP of a gene, and the x axis represents individual genes from the human genome. Nontargeting siRNA control (siControl), positive control (siEGFR), test siRNAs, and selected hits are represented by the red, cyan, gray, and green dots, respectively. Individual genes with an average score <−0.7 on both screens are highlighted in green as selected siRNA hits. The inset shows the correlation between screen 1 and screen 2.

To bolster confidence in the primary genome-wide screen, we conducted a secondary validation screen of the 110 candidate genes. Each gene’s pooled siRNAs were deconvoluted into four individual siRNAs, and the EGFR-19del GFP cells were subjected to analytic screen again. Candidates were considered “on-target” siRNA hits only if they showed suppression of EGFR from at least two independent siRNAs (fig. S1A). Intriguingly, through the secondary screen, we were able to confirm at least 98 genes encoding candidate stabilizers of EGFR-19del of the initial 110 potential genes. With an 89.09% positive hit rate, this outcome validates the reliability of our screening approach. The Reactome pathway analysis of confirmed EGFR-19del candidate stabilizers unveiled enrichment in pathways associated with diverse cellular processes, including extracellular matrix organization, protein metabolism, and chromatin organization, indicating that the stability of mutant EGFR is regulated by a variety of biological processes (fig. S1B).

P2Y2 as a bona fide EGFR stabilizer: Implications on proliferation and tumorigenicity in NSCLC cells

Among our list of candidate stabilizers for the EGFR mutant, P2Y2 stood out as one of the top candidates because of its consistent high scores from our pilot RNAi screens to the genome-wide screens. In addition, P2Y2 is known to transactivate EGFR through well-characterized mechanisms. One pathway involves the recruitment of Src kinase to SH3-binding motifs within the cytoplasmic domains of the ATP-activated P2Y2 receptor, enabling Src-dependent phosphorylation and activation of EGFR, which promote downstream proliferative signaling (30). Another mechanism involves P2Y2-mediated activation of metalloproteases that cleave membrane-tethered EGFR ligands, such as HB-EGF (heparin-binding epidermal growth factor–like growth factor), facilitating EGFR activation through ligand binding (34). Moreover, as a cell surface GPCR, P2Y2 is an accessible and attractive target for interventions using small molecules or biologics. Therefore, we prioritized validating P2Y2 as a representative candidate in NSCLC cells. The loss of P2Y2 had a notable impact on mutant EGFR, comparable to the effect of siRNA EGFR itself. P2Y2, a member of the purinergic receptor family, has been implicated in various cellular processes, including cell proliferation, migration, and differentiation (26, 27, 35–37). P2Y2 has been shown to transactivate EGFR, leading to an elevation in both mitogen-activated protein kinase and phosphoinositide 3-kinase activities to regulate proliferation of various cancer cells (38–41). However, the role of P2Y2 in EGFR protein stability is not reported to the best of our knowledge. We found a strong loss of EGFR-19del GFP signal in our two independent genome-wide siRNA screens (in addition to our pilot screen), closer to the positive control (siEGFR) (Fig. 2A). The average GFP intensities per cell in the screens are significantly lower in siP2Y2 compared to siControl (P < 2.2 × 10⁻¹⁶) and comparable to siEGFR levels (Fig. 2B).

Fig. 2. Identification of P2Y2 as a stabilizer of EGFR.

(A) Micrographs of EGFR-19del-GFP stable cells transfected with siRNAs targeting control (siControl), EGFR (siEGFR), or P2Y2 (siP2Y2) from two independent screens. (B) Quantification of EGFR-19del-GFP intensity in individual cells from both screens. (C) Immunohistochemistry staining of P2Y2 in tumor specimens from 29 patients with NSCLC, including EGFR-WT and EGFR-mutant cases. Representative immunohistochemistry images shown; scale bar, 75 μm. (D) P2Y2 protein expression quantified using the H-score and analyzed by a one-way ANOVA. (E) P2Y2 mRNA levels (normalized to actin) measured in NSCLC cell lines, grouped by EGFR status (WT: blue; mutant: red). Statistical analysis via a one-way ANOVA with Tukey’s test (P = 0.0006). H1299, A549, H1299-19del, PC9, H1975, and H1650 cells were transfected with siControl or siP2Y2. Knockdown efficiency was confirmed by quantitative PCR (F), and EGFR, phospho-AKT, phospho-ERK1/2, and actin levels were assessed by immunoblotting (G). EGFR mRNA levels were also quantified (H). (I) H1299-EGFR-19del cells were transfected with siControl or siP2Y2, serum starved, and treated with BafA1 (500 nM) or MG132 (10 μM) for 6 hours. EGFR and actin levels were assessed by immunoblotting. (J) EGFR-19del-GFP internalization following EGF stimulation in siControl or siP2Y2 cells. Cells were fixed at indicated time points and stained for Lamp1 and DAPI; images are from a confocal microscope. Matrigel invasion (K) and cell proliferation (M) assays were performed in NSCLC cells transfected with siControl or siP2Y2. Invaded area quantified using ImageJ (L); proliferation measured over 96 hours using CyQUANT assay (M). Statistical analysis: one-way ANOVA with Tukey’s post hoc test. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001. Scale bar, 10 μm. a.u., arbitrary units. OD, optical density. ns represents nonsignificant differences.

To gain clinical relevance of P2Y2 expression in NSCLC and its potential role in stabilizing mutant EGFR, we initially evaluated P2Y2 expression via immunohistochemistry in tumor and adjacent normal tissue samples from patients of Southeast Asian descent with NSCLC harboring either EGFR-WT (13 patients) or mutant EGFR (16 patients) with stages 1A, 1B, 2B, 3A, 3B, and 4 tested. Sixteen patients were treated with chemotherapy, and no one was treated with targeted therapy or immunotherapy. Five of the patients had distant metastases (table S1). P2Y2 expression was detected in bronchiolar epithelial and alveolar macrophages and lymphocytes in the normal tissues. P2Y2 expression in macrophages and lymphocytes has been reported previously (42–44). For our analysis, we considered expression in normal bronchiolar epithelial cells and tumor cells excluding blood cells. Intriguingly, our analysis unveiled a higher P2Y2 expression in the tumor region compared to adjacent normal tissue (bronchiolar epithelial cells) in both patients who are EGFR-WT and EGFR-mutant (Fig. 2C). Moreover, P2Y2 expression levels were similar between patients who are EGFR-mutant and EGFR-WT, with mean [standard deviation (SD)] values of 77.500 (SD of 51.259) and 67.038 (SD of 43.099), respectively (Fig. 2D). This difference was not statistically significant, which may be attributed to the limited sample size.

In our effort to assess P2Y2 expression in NSCLC cell lines, we were unable to specifically detect endogenous P2Y2 protein using several commercially available antibodies for immunoblotting or immunofluorescence. Consequently, we turned to reverse transcription polymerase chain reaction (RT-PCR) analysis to assess P2Y2 mRNA expression. Our results unveiled a significant (P value: 0.0006) difference in P2Y2 mRNA expression levels in EGFR-mutant cell lines (red bars) compared to EGFR-WT cells (blue bars) (Fig. 2E). Subsequently, we investigated the role of P2Y2 in stabilizing endogenous EGFR mutants by using NSCLC cells encompassing EGFR-WT (H1299 and A549), EGFR-19del (PC9 and H1650), and EGFR-L858R-T790M (H1975) mutations, alongside H1299-19del-GFP stable cells. We confirmed the efficient knockdown of P2Y2 using specific siRNAs by RT-PCR experiments (Fig. 2F). Immunoblots of EGFR in siP2Y2 cells revealed a more pronounced reduction in EGFR protein levels and downstream activity of AKT (pAKT) levels in mutant EGFR cells (both 19del and L858R-T790M), while it had a moderate effect on EGFR-WT levels (Fig. 2G and fig. S2A). Extracellular signal–regulated kinase (ERK) signaling remains largely unaffected upon the knockdown of P2Y2 across the cell lines tested. We also tested EGFR mRNA levels in NSCLC using RT-PCR. We found no significant differences in EGFR mRNA levels upon the knockdown of P2Y2 (siP2Y2), indicating that the loss of EGFR is attributed to a reduction in protein levels, therefore due to decreased stability (Fig. 2H). EGFR undergoes ubiquitination and is subsequently degraded through either the lysosomal or proteasomal pathway (45, 46). To determine whether the loss of P2Y2 promotes EGFR degradation through lysosomal or proteasomal pathways, we serum-starved and treated H1299-EGFR-19del cells with the lysosomal inhibitor bafilomycin A1 (BafA1; 500 nM) or the proteasomal inhibitor MG132 (10 μM) for 6 hours under siControl or siP2Y2 conditions and analyzed EGFR protein levels (Fig. 2I). The loss of P2Y2 led to EGFR degradation, which was rescued by lysosomal inhibition with BafA1 but not by proteasomal inhibition with MG132 (Fig. 2I, lanes 3 and 4, respectively). To further confirm that mutant EGFR is degraded in lysosomes upon P2Y2 loss, we conducted an EGFR degradation assay. H1299-EGFR-19del cells expressing GFP-tagged EGFR were treated with EGF and fixed at 0, 30, 60, and 90 min. Immunofluorescence staining for lysosomal-associated membrane protein 1 (Lamp1), a lysosomal marker, was performed, and EGFR colocalization with lysosomes was examined using confocal microscopy (Fig. 2J). Starting at 30 min post–EGF treatment, we observed increased colocalization of EGFR with lysosomes in P2Y2-deficient cells, indicating enhanced lysosomal degradation of mutant EGFR (Fig. 2J, right). These results demonstrate that the loss of P2Y2 promotes the accumulation and degradation of mutant EGFR in lysosomes, suggesting that P2Y2 depletion specifically triggers lysosomal degradation of mutant EGFR.

We next investigated whether EGFR degradation resulting from P2Y2 loss affects the invasive potential of NSCLC cells (Fig. 2K). Using transwell Matrigel invasion chambers, we observed a more significant reduction in invasiveness in NSCLC cells harboring mutant EGFR compared to those with EGFR-WT following P2Y2 knockdown (Fig. 2L). In addition, we evaluated cell proliferation in NSCLC cells with P2Y2 knockdown versus control cells. While P2Y2 loss led to an overall decrease in cell proliferation, the reduction was significantly more pronounced in cells with mutant EGFR (PC9 and H1975) compared to EGFR-WT cells (H1299 and A549) (Fig. 2M).

Loss of P2Y2 decreases EGFR levels and inhibits EGFR signaling in cells harboring TKI-resistant mutations

We used the CRISPR-Cas9 system to create a knockout (KO) of P2Y2 (P2Y2-KO1 and P2Y2-KO2 cells representing two different knockout clones) in H1975 cells, which harbor the EGFR secondary mutant (L858R-T790M) resistant to the first-generation TKIs erlotinib and gefitinib (47). KO of P2Y2 led to an almost complete loss of EGFR protein levels (Fig. 3A). However, this effect was not mirrored at the mRNA levels (fig. S2B). In addition, downstream signaling events, such as phospho-AKT (pSer⁴⁷³) and phospho-Erk1/2 (pThr²⁰²/Tyr²⁰⁴), were also affected upon the loss of P2Y2 (Fig. 3A). KO of P2Y2 resulted in a notable increase in apoptotic signals, as quantified by TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) staining in H1975, P2Y2-KO1, and P2Y2-KO2 cells (Fig. 3B). The impact of P2Y2 loss on cell proliferation and clonal expansion was evident in the significant decrease in colony formation and spheroid formation experiments (Fig. 3C and fig. S2C). To validate the impact on tumorigenesis in the P2Y2 KO cells in vivo, we subcutaneously injected 3 × 10⁶ H1975 cells and P2Y2 KO cells (P2Y2-KO1) into nonobese diabetic-severe combined immunodeficient (NOD-SCID) mice. Tumor volumes were monitored every 3 to 4 days for 46 days postinjection. Notably, all xenograft tumors injected with the P2Y2-KO1 clone were notably smaller than those injected with WT H1975 cells (Fig. 3D). The tumor volumes of P2Y2-KO1 cells subcutaneously injected showed a significant difference (P < 0.0001) with H1975 cell–injected tumors (Fig. 3E). This was evaluated using repeated measures of two-way analysis of variance (ANOVA) with days as the within-the-subject factor and P2Y2-KO1 as the between-subject factor. The knockdown and KO data collectively suggest that the loss of P2Y2 significantly affects the signaling and function of the EGFR mutant by destabilizing the mutants both in vitro and in vivo.

Fig. 3. Loss of P2Y2 destabilizes multiple activating mutants of EGFR.

(A) Immunoblotting analysis on H1975 cells, with control and with P2Y2 knocked out via CRISPR-Cas9 (P2Y2-KO1 and P2Y2-KO2). EGFR and its downstream signaling events (pAKT and pERK) along with total AKT and total ERK were analyzed. Actin serves as the loading control. Quantification of (B) apoptotic cells using TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling) assay and (C) colony-forming ability using clonogenic assay was performed on H1975 cells and P2Y2 KO cells. (D) Xenograft tumor growth assay was performed on H1975 and P2Y2-KO1 cells by subcutaneously injecting 3 × 10⁶ cells, and (E) monitoring of the tumor volume for up to 46 days was carried out to analyze their tumorigenicity in vivo. The tumor volume of H1975 compared to H1975 P2Y2-KO1 shows a statistically significant difference using an ANOVA test, with P < 0.0001. (F) Immunoblot analysis of the EGFR and actin levels of EGFR-WT and EGFR mutants (19del, 19del-T790M, and 19del-T790M-C797S) upon siRNA-based knockdown of P2Y2 (siP2Y2) or control (siControl) in H1299 stable cells (left). Analysis of mRNA expression of P2Y2, EGFR, and actin using RT-PCR (right). (G) Quantification of the EGFR protein levels normalized to actin in three independent experiments. Statistical analyses were performed using a ANOVA, and Tukey post hoc test was selected. *P ≤ 0.05, ***P ≤ 0.001, and ****P ≤ 0.0001.

We next tested whether the P2Y2-based stabilization of EGFR-19del can be extrapolated to other clinically important activating mutations of EGFR in NSCLC such as T790M and C797S. We constructed cells stably expressing EGFR-WT, EGFR-19del, EGFR-19del-T790M, and EGFR-19del-T790M-C797S in H1299 cells; performed a knockdown of P2Y2; and subjected cells to RT-PCR and immunoblotting for measuring the mRNA levels of P2Y2 and EGFR and the protein level of EGFR (with actin as a control) (fig. S3, A and B, and Fig. 3F). As shown, the knockdown of P2Y2 resulted in a stronger loss of the EGFR mutant than EGFR-WT (Fig. 3F). Quantification of the EGFR protein levels normalized to actin indicates a stronger loss of levels of EGFR mutants (19del: 55.24 ± 19.96%; 19del-T790M: 50.06 ± 22.25%; 19del-T790M-C797S: 57.33 ± 25.51%) than that of EGFR-WT (73.63 ± 15.95%) (Fig. 3G). However, the EGFR mRNA levels remain unchanged in all the siControl and siP2Y2 H1299 stable cells (Fig. 3F, right, and fig. S3B). To eliminate any potential discrepancy related to the nature of the EGFR mutation (19del), we have also used H1299 cells that stably express other EGFR mutants such as L858R, L858R-T790M, and L858R-T790M-C797S. As seen with EGFR-19del stable cells, the H1299-EGFR-L858R stable cells showed a stronger loss of EGFR mutants and downstream signaling events (pAKT) than the EGFR-WT cells (fig. S3C) upon P2Y2 suppression. The EGFR mRNA levels remain similar in all the siControl and siP2Y2 H1299 stable cells (fig. S3C, bottom).

To rule out any disparity resulting from the background of the cells, we constructed stable cells expressing EGFR-WT, EGFR-19del, EGFR-19del-T790M, and EGFR-19del-T790M-C797S in human embryonic kidney (HEK) cells (HEK293 or HEK) because these cells have negligible levels of endogenous EGFR (48, 49). The HEK stable cells expressing EGFR species were subjected to scrutiny to find whether they qualify for tumorigenic experiments. First, we checked for the activation of EGFR (pY1068) and downstream signaling events such as pAKT/AKT and pERK1/2/ERK1/2 levels compared to EGFR-WT cells. All the mutant EGFR stable cells show a higher EGFR activation and subsequent pAKT/pERK levels (fig. S4A). Cell proliferation, spheroid, and clonogenic assays showed a significant increase in the EGFR-mutant cells than the EGFR-WT cells, confirming the oncogenic nature of the EGFR mutants in HEK stable cells (fig. S4, B to D). We have also found a stronger loss of EGFR downstream signaling events such as pAKT in the EGFR mutants compared to EGFR-WT upon P2Y2 knockdown in HEK stable cells (fig. S4E). Besides immunoblotting, we also confirmed the loss of the EGFR mutant using immunofluorescence of EGFR in HEK stable cells (fig. S4F). In summary, these findings suggest that P2Y2 functions as a stabilizer for multiple activating EGFR mutants regardless of the nature of mutations, and the absence of P2Y2 results in a substantial decrease in EGFR-mutant proteins and their downstream signaling as compared to EGFR-WT.

Increased extracellular ATP levels cause P2Y2-mediated stabilization of the EGFR mutant

To define the molecular mechanism underlying the selective stabilization of EGFR mutants by P2Y2 as compared to WT, we tested the hypothesis that EGFR mutants may preferentially activate P2Y2 (by increasing levels of its ligand –ATP) to enhance their stability. Extracellular ATP has been suggested to play key roles in signaling under multiple physiological and pathological conditions (50). Extracellular ATP is one of the key biochemical components of the tumor microenvironment (TME) and has been shown to be increased to promote tumorigenesis in multiple tumors (51–53). The extracellular ATP levels in the tumor interstitial space are estimated to be 10⁴ times higher than those in normal tissues and induce TKI resistance (54, 55). We checked whether there was any difference in the levels of ATP, one of the P2Y2 ligands, between EGFR-WT and EGFR mutants in the stable cells. We opted to assess extracellular ATP levels in the HEK stable cells because these cells exhibit a notable difference in oncogenic properties between EGFR mutants and WT. In addition, HEK stable cells, being isogenic, eliminate any variations specific to the cell lines and illustrate the distinct difference between EGFR-WT and EGFR mutants. We used a genetically encoded single-wavelength sensor for imaging cell surface ATP in the HEK-EGFR stable cells (56). The ATP sensor uses circularly permuted superfolder GFP in the epsilon subunit of F₀F₁-ATPase of Bacillus PS3 extracellularly displayed on the plasma membrane (iATPSnFR1.0) and responds to increased ATP concentrations with quick increases in fluorescence (56). We transfected the cell surface–targeted iATPSnFR1.0 sensor into HEK stable cells with similar expression and serum starved the cells to avoid background ATP noise from fetal bovine serum (FBS) (Fig. 4A). Notably, we find that the fluorescence from the sensor in the EGFR-mutant cells was significantly stronger than that in cells expressing EGFR-WT (Fig. 4B). Quantification of the GFP signal from the sensor showed a robust and almost threefold increase in the ATP levels in the EGFR-mutant cells than the WT cells (Fig. 4B, graph). To corroborate the possible higher extracellular ATP levels in the EGFR-mutant cells, we used a different approach based on extracellular ATP measurement with a bioluminescence assay designed for monitoring of ATP levels. Similarly, we observed a significant increase in the ATP levels of the EGFR-mutant cells (fig. S5A). We next examined whether extracellular ATP levels in NSCLC cells follow a similar trend. Using the bioluminescence assay, we observed overall elevated extracellular ATP levels in EGFR-mutant cells (H1975, HCC827, and HCC4006), whereas EGFR-WT cells generally exhibited lower levels. Notably, A549 cells, despite harboring EGFR-WT, showed elevated extracellular ATP levels (fig. S5B). It is important to note that A549 cells carry the KRAS (G12S) mutation, which results in constitutive activation of KRAS, a mutation that mimics the activated EGFR pathway similar to the EGFR mutant (57, 58). Given the diverse mutational backgrounds of NSCLC cell lines beyond EGFR, isogenic cell models stably expressing either WT or EGFR mutants provide a more accurate framework for studying extracellular ATP levels specifically in relation to EGFR status. Together, these results suggest that cells with activating EGFR mutants have elevated ATP levels in the extracellular environment.

Fig. 4. Extracellular ATP stabilizes EGFR protein in a P2Y2-dependent way.

(A) Immunoblot of the expression levels of the iATPSnFR1.0 probe (using an anti-GFP antibody), EGFR, and actin in HEK293 cells stably expressing EGFR-WT and EGFR mutants (19del, 19del-T790M, and 19del-T790M-C797S) using a GFP-based single-wavelength ATP sensor transfected with cell surface ATP probe—iATPSnFR1.0. (B) Micrographs of GFP fluorescence representing extracellular ATP levels in HEK stable cells with bright-field images in the bottom panel. The bar graph represents quantification of the GFP intensity of the ATP sensor. Immunoblots using EGFR and actin antibodies on H1975 (C) and H1299 (D) cells that were treated with ATP (100 μM). Graphs on the bottom are the quantification of EGFR protein from the immunoblot normalized to actin. hrs, hours. (E) Immunoblots of H1975 and H1299 cells treated with apyrase (10 U/ml) and then harvested at the specified time points. The bottom panels represent the EGFR and actin mRNA expression analysis using RT-PCR. Graphs on the bottom are the quantification of EGFR protein from the immunoblot normalized to actin. (F) siRNA-based knockdown of P2Y2 (siP2Y2) or control (siControl) in H1975 cells treated with or without ATP (100 μM). Cells were harvested after 1 hour and immunoblotted for EGFR or actin (left), quantification of the EGFR protein levels normalized to actin in three independent experiments (middle) was performed, and mRNA expression of EGFR and actin using RT-PCR (right) was determined. Actin serves as the loading control. Statistical analyses were performed using an ANOVA, and Tukey post hoc test was selected; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

We next tested whether this elevated extracellular ATP level has a role in stabilizing EGFR or not. We first analyzed cells expressing an EGFR mutant (H1975) and an EGFR-WT (H1299). After 16 hours of serum starvation, ATP (100 μM) was added to the culture media, and cell lysates and total RNA were harvested at a 30-min interval until 2 hours. Cells were then subjected to immunoblotting or RT-PCR to assess EGFR and actin levels. Starting from 30 min, we found that ATP treatment significantly stabilized both EGFR-mutant and EGFR-WT proteins (Fig. 4, C and D, graphs), but there was no noticeable change in the EGFR mRNA levels (fig. S5C). To confirm independently that extracellular ATP specifically enhanced the stability of EGFR, we treated cells with apyrase (10 U/ml) that catalyzes the hydrolysis of ATP to AMP (adenosine monophosphate). Upon scavenging of ATP by apyrase treatment, EGFR protein was significantly lost in both H1975 and H1299 cells in 1 to 2 hours (Fig. 4E, graphs). We then tested more lung cancer cell lines (H226, H460, A4549, HCC4006, PC9, and H1650) and validated both the increased stability of EGFR upon ATP treatment and degradation of EGFR upon scavenging of ATP by apyrase treatment (fig. S5, D and E). To find whether ATP stabilizes EGFR via a P2Y2-dependent mechanism, we treated siP2Y2- or siControl-transfected H1975 cells with ATP and harvested cell lysates or mRNA after 1 hour. We found a significant increase in the EGFR levels in the siControl cells after ATP treatment but not in siP2Y2 cells (Fig. 4F), suggesting that the ATP-induced stability of EGFR is dependent on P2Y2.

We hypothesize that elevated ATP levels in EGFR-mutant tumors may increase the expression of genes involved in ATP release and utilization in patients with NSCLC. To explore this in a clinical context, we analyzed RNA sequencing (RNA-seq) datasets from two published NSCLC studies. The dataset from Chen et al. (59) included 98 EGFR-WT and 71 EGFR-mutant tumors, while the dataset from Huang et al. (60) comprised 27 EGFR-WT and 21 EGFR-mutant tumors. We curated a list of 86 genes involved in ATP release and utilization and compared their expression between EGFR-mutant and EGFR-WT tumors (61). In both datasets, ATP-using and signaling genes were significantly up-regulated in EGFR-mutant tumors compared to EGFR-WT tumors (fig. S5, F and G). Notably, ATP-binding cassette transporters exhibited markedly higher expression in EGFR mutants (fig. S5, H and I). These results support our hypothesis that EGFR-mutant tumors have elevated extracellular ATP levels. We next examined whether extracellular UTP, another ligand of P2Y2, could also stabilize EGFR, similar to ATP. H1299 and H1975 cells were treated with 0, 5, 10, and 30 μM UTP under similar experimental conditions used for ATP. Even at 5 μM, UTP was sufficient to stabilize EGFR, and this EGFR stability increased in a dose-dependent manner (fig. S5J). These findings confirm that the activation of P2Y2 by either ATP or UTP promotes EGFR stabilization. Together, these results suggest that high extracellular ATP/UTP levels in the EGFR-mutant cells cause a preferential P2Y2-dependent stability of EGFR. The selective increase in extracellular ATP/UTP level in cells expressing the EGFR mutant is likely the underlying basis for preferential dependence on P2Y2 for the enhanced stability of EGFR mutants. We further use ATP in our assays as it has been very well established that extracellular ATP is one of the major components of TME (51).

Integrins play a key role in ATP-P2Y2–dependent stabilization of EGFR

P2Y2 contains an integrin-binding motif arginine-glycine-aspartic acid (RGD) in its first extracellular loop and has been shown to associate with integrins to mediate downstream signaling (Fig. 5A) (26, 27, 62). To identify whether there is any role of integrins in the stabilization of EGFR, we analyzed our genome-wide screen data and checked whether there is loss of EGFR upon the loss of any integrins (Fig. 5B). From both screens, there are a few integrins (β5, β1, and β4) that, upon loss, caused a clear loss of EGFR-19del-GFP (Fig. 5, B and C). In addition, upon ATP treatment, the integrin downstream signaling molecule focal adhesion kinase (FAK) is activated (phosphorylation at Tyr³⁹⁷; phospho-FAK) in H1975 cells (Fig. 5D). Conversely, FAK is inactivated upon apyrase treatment (Fig. 5E), indicating activation of the integrin signaling pathway by extracellular ATP. To find whether the activation of the integrin signaling pathway is required for the stability of EGFR, we pretreated cells with the RGD peptide or dimethyl sulfoxide (DMSO) and added ATP (100 μM) to the medium of H1975 cells. ATP treatment caused an activation of FAK, and the RGD treatment strongly inhibited FAK activation, indicating a strong attenuation of integrin signaling upon RGD treatment (Fig. 5F). ATP treatment along with blocking of integrin using RGD caused no change in EGFR level, in contrast to the increased levels of EGFR caused by ATP without RGD (Fig. 5F). AIIB2 is an integrin β1 function–blocking antibody (63). We treated H1975 cells with different concentrations of AIIB2 antibody (1, 5, and 10 μg/ml) or control rat immunoglobulin G (IgG; 10 μg/ml) for 16 hours, and the immunoblot analysis of the antibody-treated cell lysates showed a dose-dependent decrease in the EGFR levels (Fig. 5G), substantiating the role of integrins in the ATP-P2Y2–based stabilization of EGFR.

Fig. 5. Concomitant role of integrins along with P2Y2 in stabilizing EGFR.

(A) Schematic showing P2Y2 embedded in the plasma membrane, highlighting the RGD motif in green. Created in BioRender. G. T. K. Boopathy (2025); https://biorender.com/yklytxg. (B) Dot plot of normalized EGFR-19del intensity scores for 27 integrin genes from two replicates of the genome-wide RNAi screen. Blue, red, and green dots represent siControl, siEGFR, and siP2Y2, respectively; black dots indicate individual integrin genes, with selected gene names labeled. (C) Micrographs of EGFR-19del-GFP cells treated with siRNAs targeting β1 (ITGB1), β4 (ITGB4), β5 (ITGB5), and EGFR (siEGFR) from the RNAi screen. H1975 cells treated with (D) ATP (100 μM) or (E) apyrase (10 U/ml) and harvested at indicated time points. Lysates were immunoblotted for phospho-Y397-FAK (pFAK), total FAK, and actin. (F) H1975 cells pretreated with or without the RGD peptide and stimulated with or without ATP for 30 min or 1 hour. Lysates were immunoblotted for EGFR, pFAK, total FAK, and actin. (G) H1975 cells treated with an integrin β1–blocking antibody (AIIB2) at indicated concentrations for 16 hours. Lysates were immunoblotted for EGFR, pFAK, total FAK, and actin. (H) EGFR, integrin β1, and actin levels in H1299 EGFR-WT, H1299 EGFR-19del, H1299, and H1975 cells following integrin β1 knockdown (siβ1) or siControl. After 72 hours, cells were serum starved for 16 hours and lysed. EGFR band intensities were normalized to actin and indicated on the blot. (I) Representative immunohistochemistry images of integrin β1 (ITGB1) in NSCLC tissues from 29 patients (13 EGFR-WT and 16 EGFR-mutant). (J) Quantification of ITGB1 protein expression (H-score) from (I). Statistical analysis: one-way ANOVA; *P < 0.05. Scale bar, 75 μm.

To determine whether integrins preferentially stabilize the EGFR mutant compared to EGFR-WT, we evaluated EGFR protein levels following the knockdown of integrin β1 (siβ1) or control (siControl) in H1299 cells expressing either EGFR-WT or EGFR-19del (Fig. 5H, top panels). Knockdown of integrin β1 resulted in a 15% reduction in EGFR-19del levels (Fig. 5H, top right; 0.85-fold) compared to no change in EGFR-WT levels (Fig. 5H, top left; 1.19-fold). To further validate this observation, we conducted a similar experiment in NSCLC cell lines H1299 and H1975. Notably, integrin β1 knockdown led to a 40% reduction in EGFR-mutant levels (Fig. 5H, bottom right; 0.62-fold), while EGFR-WT levels remained largely unaffected (Fig. 5H, bottom left; 0.97-fold). These results demonstrate that integrin β1 preferentially stabilizes the EGFR mutant over EGFR-WT, highlighting its critical role in maintaining EGFR-mutant stability.

We next assessed integrin β1 expression in tumor and adjacent normal tissue samples from the same set of patients with NSCLC mentioned previously. Our analysis revealed a predominant expression of integrin β1 in the tumor region compared to adjacent normal tissue in both patients who are EGFR-WT and EGFR-mutant (Fig. 5I). In addition, we observed a slight increase in integrin β1 expression level in patients who are EGFR-mutant compared to patients who are EGFR-WT, with values of 29.375 ± 22.426 and 16.346 ± 22.141, respectively (Fig. 5J); this difference is not statistically significant, maybe due to the small sample size. Together, these results support a clinical relevance of integrins in the ATP-P2Y2–based stabilization of EGFR in NSCLC.

P2Y2 colocalizes and forms a complex with EGFR-WT and its mutants

P2Y2 is known to colocalize with and transactivate EGFR-WT (30), but the localization and role of P2Y2 in EGFR mutants in NSCLC are currently unknown. We investigated the subcellular distribution of P2Y2 in comparison to EGFR-WT and the EGFR mutants. EGFR constructs for expressing GFP-fused EGFR-WT, 19del, 19del-T790M, and 19del-T790M-C797S were individually cotransfected with the P2Y2 plasmid (with a C-terminal FLAG tag) into H1299 cells. After 24 hours, the cells were fixed and processed for immunofluorescence using an anti-FLAG antibody. The micrographs of the H1299 cells with EGFR-WT exhibit a strong colocalization with P2Y2 at the plasma membrane (fig. S6A, top row). Mutant EGFR species, 19del, 19del-T790M, and 19del-T790M-C797S, demonstrated substantial colocalization with P2Y2 at both the plasma membrane and intracellular vesicles (fig. S6A, rows 2 to 4). The dotted white line in the micrographs represents a snapshot illustrating the colocalization of P2Y2 and EGFR, as depicted in the adjacent graphs (fig. S6A, graphs). With these results, we hypothesized that P2Y2 might form a protein complex with both EGFR-WT and EGFR mutants. To investigate this, we performed coimmunoprecipitation experiments to assess whether P2Y2 coprecipitates with the different EGFR species. Individually, we transfected EGFR-WT and EGFR mutants along with P2Y2 in H1299 cells and performed immunoprecipitation (IP) of P2Y2 using an anti-FLAG antibody (fig. S6B). The immunoprecipitates were subjected to immunoblotting for EGFR using an anti-GFP antibody. P2Y2 can be efficiently immunoprecipitated from the lysates as cells transfected with the vector alone did not show bands of P2Y2, while lysates upon transfection of P2Y2 (FLAG) did (fig. S6B, row 3 from the top, lanes 1 and 2). As hypothesized, EGFR-WT and all the mutants can be specifically coprecipitated with P2Y2 from the lysates, as the cells transfected with the vector and P2Y2 alone did not show any band of EGFR (fig. S6B, bottom row, lanes 1 and 2). Our observations collectively indicate that both EGFR-WT and EGFR mutants colocalize and form a protein complex with P2Y2.

The P2Y2-integrin-EGFR signaling axis is required for the stability of the EGFR-19del mutant

Given that our results indicate that integrins play a role in the ATP-P2Y2–based stabilization of EGFR, we hypothesized that ATP-activated P2Y2 may form a protein complex with integrins and EGFR to carry out its stabilizing role. We took integrin β1 as the representative because it is a commonly found subunit in integrin heterodimers and shown to pair with a variety of α subunits to form 12 different integrin heterodimers (64). Also, the overexpression of the integrin β1 subunit in lung adenocarcinoma cells compared to the normal lung promotes EGFR signaling and the development of lung adenocarcinoma (65–67). To test the formation of the P2Y2-integrin β1-EGFR-19del complex, we transfected the vector alone, GFP, or EGFR-19del-GFP along with P2Y2-FLAG in H1299 cells. A distinctive band of 44 kDa was detected in H1299 cells expressing P2Y2-FLAG but not in cells expressing the vector alone, indicating that P2Y2-FLAG migrated in the SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gel as a 44-kDa band (fig. S6C). A second and more diffuse band of 55 to 75 kDa representing more heavily glycosylated forms of the P2Y2 was also seen. A similar expression pattern for P2Y2 was reported earlier with the fluorescent protein tagged at the C terminus (68). IP of EGFR-19del with an anti-GFP antibody followed by immunoblotting with an anti-FLAG antibody for P2Y2 and an anti-integrin β1 antibody (endogenous protein) showed that the GFP-transfected control did not show any reactivity with P2Y2 and integrin β1 (fig. S6C, left lane), but EGFR-19del–transfected cells displayed an interaction with P2Y2 and integrin β1 (fig. S6C, bottom, right lane). This experiment reveals a physical association among EGFR-19del, P2Y2, and endogenous integrin β1. A similar experiment—IP of P2Y2 with an anti-FLAG antibody followed by immunoblotting with an anti-GFP antibody (for EGFR-19del) and an anti-integrin β1 antibody—further confirms the association among P2Y2, integrin β1, and EGFR-19del (fig. S6D). We validated the existence of the P2Y2-integrin β1-EGFR protein complex through IP of endogenous EGFR with tagged P2Y2 in H1975 and H1299 cells. IP with an anti–EGFR-specific antibody in H1975 cell lysates transfected with P2Y2 efficiently pulled down EGFR, integrin β1, and P2Y2 (Fig. 6A, right lane of bottom IP panels). Similarly, pulldown of P2Y2 using anti-FLAG antibodies coprecipitated EGFR and integrin β1, further confirming the presence of the P2Y2-integrin β1-EGFR mutant complex (Fig. 6B, right lane of bottom IP panels). In H1299 cells transfected with P2Y2, coimmunoprecipitation with an anti-EGFR antibody also efficiently pulled down integrin β1 and tagged P2Y2 (fig. S6E, right lane of bottom IP panels). Likewise, pulldown using anti-FLAG antibodies in these cells coprecipitated EGFR and integrin β1, demonstrating the existence of the P2Y2-integrin β1-EGFR-WT complex (fig. S6F, right lane of bottom panels). These results suggest the existence of the P2Y2-integrin β1-EGFR complex in both H1299 and H1975 cells. Notably, we observed that IP with an anti-EGFR antibody specifically precipitated a distinct species of P2Y2, likely representing a glycosylated or posttranslationally modified form of the protein, while IP with anti-FLAG-M2 agarose pulled down all available P2Y2 forms.

Fig. 6. Integrin β1 and P2Y2 form a protein complex with EGFR, which can be targeted therapeutically.

Coimmunoprecipitation using (A) an anti-IgG or anti-EGFR antibody and (B) an anti-P2Y2 (anti–FLAG-M-agarose) antibody in H1975 cell lysates transfected with P2Y2-FLAG and immunoblotted with anti-integrin β1, anti-FLAG, and anti-EGFR antibodies. Coimmunoprecipitation using an anti-IgG or anti-EGFR antibody on (C) H1299 cell lysates treated with and without ATP (100 μM; 5 min) and (D) H1975 cell lysates with ATP (100 μM; 1 hour) with or without apyrase (10 U/ml; 1 hour) and immunoblotted with anti-integrin β1, anti-FLAG, anti-EGFR, and actin antibodies. Band intensities of EGFR, integrin β1, and P2Y2 were quantified and normalized to immunoprecipitated EGFR, with normalized values indicated below the respective blots. (E) EGF-induced internalization of EGFR-WT-GFP or EGFR-19del-GFP, tdTomato-integrin β1, and P2Y2 (FLAG immunofluorescence) was detected at the indicated time points in the presence of ATP and demonstrates the colocalization of complex members during EGFR signaling. DAPI stains the nucleus. (F) Coimmunoprecipitation of EGFR-19del-GFP with P2Y2 in H1299 cells treated with either DMSO or the RGD (50 μM) peptide for 16 hours. The cell lysates and immunoprecipitates were blotted for GFP (EGFR), FLAG (P2Y2), pFAK, FAK, and actin. (G) Coimmunoprecipitation of EGFR-19del-GFP with P2Y2-FLAG in H1299 cells treated with DMSO or kaempferol (50 μM) for 16 hours. The cell lysates and immunoprecipitates were blotted for GFP (EGFR), FLAG (P2Y2), and actin. Scale bar, 10 μm.

We next evaluated whether endogenous EGFR-WT or EGFR mutant preferentially interacts with integrin β1 and P2Y2. To assess this, we performed IP of endogenous EGFR from H1299 and H1975 cells following P2Y2-FLAG transfection and analyzed the interactions in a single immunoblot (fig. S6G). Our results confirmed the presence of the P2Y2-integrin β1-EGFR protein complex in both cell lines. However, a stronger association was observed between the EGFR mutant, P2Y2, and integrin β1 compared to EGFR-WT. Quantification of band intensities for integrin β1 and P2Y2, normalized to immunoprecipitated EGFR, revealed that EGFR-WT (H1299) coprecipitated 1.07-fold integrin β1 and 0.76-fold P2Y2 (fig. S6G, left IP panels), whereas the EGFR mutant (H1975) coprecipitated 4.59-fold integrin β1 and 4.78-fold P2Y2 (fig. S6G, right IP panels). These findings suggest that the EGFR mutant forms a preferential P2Y2-integrin β1-EGFR complex compared to EGFR-WT, underscoring its crucial role in stabilizing the EGFR mutant.

We next investigated whether elevated extracellular ATP levels promote the formation of the P2Y2-integrin β1-EGFR complex. To test this, we used H1299 cells to assess the effect of increased extracellular ATP on complex formation. IP of endogenous EGFR in H1299 cells transfected with P2Y2-FLAG revealed increased binding to P2Y2 (1.23-fold) and integrin β1 (1.54-fold) following treatment with 100 μM ATP for 5 min (Fig. 6C). To confirm that this complex formation was specifically driven by ATP, cells were treated with apyrase (10 U/ml) alongside ATP (100 μM), followed by coimmunoprecipitation of EGFR. In apyrase-treated cells, we observed a marked reduction in EGFR’s affinity for integrin β1 (0.68-fold) and P2Y2 (0.78-fold) compared to cells treated with ATP alone (Fig. 6D). These results suggest that elevated extracellular ATP levels enhance the physical association between EGFR, P2Y2, and integrin β1 by activating P2Y2 to protect EGFR from degradation.

To determine the extracellular ATP concentration required for EGFR-mutant stabilization and compare it with the half-maximal effective concentration (EC₅₀) for P2Y2 activation, we measured ATP-induced calcium responses in H1975 cells using Cal-520 AM, a calcium-sensitive fluorescent dye (fig. S6H, top). The EC₅₀ for ATP-induced calcium signaling was ~5.43 μM, consistent with previous reports in other cell types (69–71). We then examined ATP-induced EGFR stabilization in a dose-dependent manner (0 to 1000 μM) in H1975 cells with or without integrin β1 (fig. S6H, bottom). Densitometric analysis revealed an EC₅₀ of ~2.16 μM for ATP-mediated EGFR stabilization, with maximal stabilization occurring between 50 and 100 μM ATP. The EC₅₀ values for ATP-induced calcium signaling and EGFR stabilization are within the similar range, supporting a correlation between P2Y2 activation and its role in EGFR stabilization. Loss of integrin β1 markedly reduced EGFR stabilization, with only minimal EGFR stability observed at very high ATP concentrations (1000 μM). These results indicate that integrin β1 is essential for activated P2Y2 to mediate the stabilization of the EGFR mutant by forming a P2Y2-integrin-EGFR complex enriched in the endosome.

Endocytosis is a critical step in EGFR degradation. To investigate differences in the behavior of the P2Y2-integrin β1-EGFR complex during endocytosis, we compared EGFR-WT and EGFR-19del upon treatment with EGF (50 ng/ml) and ATP (100 μM). H1299 cells were cotransfected with EGFR-WT or EGFR-19del-GFP, P2Y2-FLAG, and integrin β1-tdTomato (C terminus). Serum-starved cells were treated with EGF and ATP; fixed at 0, 30, 60, and 120 min with paraformaldehyde (PFA); stained with an anti-FLAG antibody; and analyzed by confocal fluorescence microscopy (Fig. 6E). In EGFR-WT cells, EGFR endocytosis was observed between 30 to 60 min posttreatment, with EGFR localizing predominantly to endosomal-lysosomal compartments (perinuclear area). However, EGFR-WT did not colocalize with P2Y2 or integrin β1 during this process (Fig. 6E, left panel, top rows). The majority of P2Y2 and integrin β1 largely remained at the plasma membrane at all time points tested (Fig. 6E, left panel, middle rows). In contrast, EGFR-19del exhibited distinct behavior. At 0 min, EGFR-19del, P2Y2, and integrin β1 were colocalized at the plasma membrane. Following EGF stimulation (in the presence of ATP), EGFR-19del underwent endocytosis, and this colocalization with P2Y2 and integrin β1 persisted within endosomes at 30, 60, and 120 min (Fig. 6E, right panel, middle rows). While EGFR-WT undergoes rapid endocytosis and degradation within 60 min, even in the presence of ATP, the P2Y2-integrin β1 complex remained consistently associated with EGFR-19del throughout the endocytic process. Even under ATP-stabilized conditions, EGFR-WT undergoes endocytosis and is subsequently degraded following EGF stimulation, indicating that EGF can override ATP-mediated stability in EGFR-WT cells. In contrast, the EGFR-19del mutant remained resistant to degradation. These findings suggest that the P2Y2-integrin β1-EGFR-19del complex is highly stable, which likely contributes to the sustained activity and enhanced stability of the EGFR mutant.

Multiple studies have established that the RGD motif in P2Y2 plays a critical role in its binding to α and β integrins, enabling a direct interaction with RGD-binding integrins such as αvβ3, αvβ5, and α5β1 (26–28, 72, 73). This interaction initiates downstream signaling cascades involving small guanosine triphosphatases—including Rho, Rac, Vav2, and Cdc42—through G proteins such as G_o and Gα₁₂, thereby modulating cytoskeletal dynamics and cellular signaling pathways (26–28, 72, 73). In this study, we further validated the functional importance of the RGD motif in mediating the interaction between P2Y2 and integrins to enhance the formation of P2Y2-integrin and EGFR mutants in NSCLC cells. To test this, we assessed the impact of mutating the RGD motif on the formation of the P2Y2-integrin β1-EGFR complex. We mutated the RGD motif to arginine-glycine-glutamic acid (RGE) and generated a FLAG-tagged P2Y2-RGE construct. H1299 cells expressing EGFR-19del were transfected with either the vector, WT P2Y2 (P2Y2-RGD), or P2Y2 mutant (P2Y2-RGE). Coimmunoprecipitation of P2Y2 was performed using an anti-FLAG antibody (fig. S6I). As hypothesized, P2Y2-RGD efficiently coprecipitated EGFR-19del and integrin β1. In contrast, P2Y2-RGE showed only partial precipitation of EGFR-19del and failed to pull down integrin β1 (fig. S6I, bottom panels, right lane). These results support that the RGD motif in P2Y2 is essential for the formation of the P2Y2-integrin β1-EGFR complex. In addition, the observation that IP of the P2Y2-RGE mutant partially pulls down EGFR-19del provides evidence for an integrin-independent interaction between the P2Y2 receptor and EGFR. This finding is consistent with a previous study showing that Src-dependent colocalization of EGFR and P2Y2, along with ATP-induced transactivation of EGFR, enhances MAP kinase signaling and promotes cell proliferation (30). We next investigated whether we could destabilize the EGFR mutant by disrupting this protein complex using the RGD peptide or small-molecule inhibitors. H1299 cells were transfected with EGFR-19del-GFP and P2Y2-FLAG plasmids or control vectors. The cells were treated with DMSO or the RGD peptide for 16 hours (Fig. 6F). The IP experiment showed that the RGD peptide resulted in the loss of interaction between EGFR-19del and P2Y2 (0.48-fold) (Fig. 6F, bottom, right lane). Kaempferol is a P2Y2 antagonist and has been shown to inhibit P2Y2-specific signaling (74–76). We also treated the cells with kaempferol (50 μM) or DMSO for 16 hours and immunoprecipitated EGFR-19del-GFP using an anti-GFP antibody and immunoblotted for an anti-FLAG antibody to detect coimmunoprecipitated P2Y2-FLAG (Fig. 6G). Inhibition of P2Y2 using kaempferol strongly suppressed the interaction of EGFR-19del with P2Y2 (0.51-fold). Together, these results suggest that P2Y2-integrin-EGFR is in a stable protein complex upon activation of P2Y2 by ATP and inhibition of P2Y2 or integrins abrogated the protein complex.

Inhibition of P2Y2 degrades the EGFR mutant and prevents tumorigenesis in vitro and in vivo

The P2Y2 inhibitor kaempferol prevented the P2Y2-integrin-EGFR complex formation, so we next tested whether the inhibition of this complex using kaempferol would down-regulate EGFR-mutant levels. For this study, we used H1975 (EGFR-L858R-T790M), H1650 (EGFR-19del with TKI resistance), and H1299 (EGFR-WT) cells. Kaempferol was previously used to treat primary fibroblasts at concentrations between 20 and 100 μM (76–80), so we used a similar concentration range of 0 to 100 μM. H1975, H1650, and H1299 cells were treated with 0 to 100 μM kaempferol for 24 hours, harvested, and immunoblotted for EGFR, pAKT, pERK1/2, and actin (Fig. 7A). In H1975 cells, we find that 10 to 20 μM concentrations of kaempferol significantly down-regulated the EGFR mutant and downstream AKT signaling (Fig. 7A, top panel and graph). H1650 cells require a much higher concentration (at least 25 μM) to significantly down-regulate the EGFR mutant (Fig. 7A, middle panel and graph). Even with the 100 μM concentration of kaempferol, there is no change in EGFR-WT levels in H1299 cells (Fig. 7A, bottom panel and graph). AR-C118925XX is a specific competitive antagonist of P2Y2 receptor (81). To rule out the possibility of nonspecificity of kaempferol, we have further confirmed these results using AR-C118925XX. Our results on H1650 (fig. S7A), H1975 (fig. S7B), and H1299 (fig. S7C) have shown that a lower concentration of AR-C118925XX than kaempferol is enough to down-regulate EGFR and the downstream signaling events of the EGFR mutant in H1975 and H1650 cells but AR-C118925XX had no effect on EGFR-WT–expressing H1299 cells. Both knockdown of P2Y2 in Fig. 2G and inhibition of P2Y2 using kaempferol in Fig. 7A caused a significant loss of AKT activity but not ERK1/2 activity.

Fig. 7. Kaempferol inhibits tumor formation in a P2Y2-dependent manner in vitro and in a mouse xenograft.

(A) EGFR-WT (H1299) and EGFR-mutant (H1650 and H1975) NSCLC cells were treated with increasing concentrations of kaempferol for 16 hours. Cell lysates were harvested and immunoblotted for EGFR, phospho-AKT, phospho-ERK1/2, and actin. (B) Schema of the mouse xenograft assay in NSG mice using H1299 and H1975 cells and the kaempferol treatment regimen. Created in BioRender. G. T. K. Boopathy (2025); https://biorender.com/yklytxg. Human NSCLC xenograft generated from the subcutaneous injection of H1299 (C) and H1975 (D) cells (5 million cells each). After 10 days, both H1299 cell– and H1975 cell–bearing mice were grouped into vehicle or kaempferol-treated mice. The kaempferol-treated mouse group received a daily dose of kaempferol (150 mg/kg) orally for 24 days. The tumor volume of post–kaempferol treatment or vehicle is recorded and plotted for (E) H1299 cells and (F) H1975 cells. The data represented are the means ± standard error (n = 5). Statistical significance was calculated between control and treatment groups by a two-way ANOVA followed by Bonferroni’s multiple comparison test. (G) Fluorescence-activated cell sorting (FACS) of annexin V–stained kaempferol-induced apoptosis using representative tumors from H1299 and H1975 xenografts with and without kaempferol. The percentage of apoptotic cells in each category is represented in the top right corner of each graph. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

We further evaluated the proliferation of several NSCLC cell lines, comparing EGFR-WT cells (H1299, A549, and H460) with EGFR-mutant cells (PC9, H1975, and HCC4006) following treatment with either kaempferol or the FAK inhibitor PF-562271, versus control cells (fig. S7, D and E). Kaempferol treatment led to a general reduction in cell proliferation across all tested lines, but this effect was significantly more pronounced in EGFR-mutant cells (fig. S7D; PC9, HCC4006, and H1975) compared to EGFR-WT cells (fig. S7D; H1299 and H460). In contrast, inhibition of FAK with PF-562271 caused a dose-dependent reduction in cell proliferation in all NSCLC cell lines, regardless of EGFR mutation status (fig. S7E). This may be due to the central role of FAK in normal and cancer cells, where it is important for multiple critical signaling pathways (82).

Given that EGFR mutants and their downstream signaling responses were suppressed upon kaempferol treatment, we next tested whether the kaempferol treatment would prevent EGFR mutant–driven tumorigenesis. H1975 (EGFR mutant) and H1299 (EGFR-WT) cells were subcutaneously injected into 10 mice each with 5 × 10⁶ cells. After an initial 10 days of tumor formation, we equally but randomly divided the mice into vehicle and kaempferol treatment groups. Kaempferol was orally fed to mice at 150 mg/kg daily for 24 days (Fig. 7B). Tumor formation driven by H1299 cells expressing EGFR-WT did not show any detectable difference among the treated and control groups (Fig. 7C). Of relevance is the observation that tumor formation by H1975 cells harboring the EGFR mutant was significantly suppressed in the kaempferol-treated group (Fig. 7D). The change in the size of subcutaneous tumors did not significantly affect the total body weight of tumor-bearing mice, suggesting no associated cachexia (fig. S8, A and B). A continuous monitoring of the whole tumor volumes every 3 to 4 days after the start of drug treatment for 24 days showed a significant reduction of the tumor volume in the kaempferol-treated H1975 cohort from day 17 onward (day 17: P < 0.01; days 20 and 24: P < 0.0001) but not in the H1299 cohort (Fig. 7, E and F). Also, the final tumor weights derived from H1975 cells are significantly lower (70.2 ± 2.4%) in the kaempferol-treated mice than the control group, similar to the results of the tumor volume, whereas tumor weights derived from H1299 cells show no difference in the two groups (fig. S8, C and D). We took the representative tumors from each of the treatment groups and performed apoptosis analysis using annexin V staining in fluorescence-activated cell sorting (FACS) to test whether the tumor reduction in H1975 xenografts after kaempferol treatment is associated with apoptosis. In H1299 tumors, there was no significant difference between control and kaempferol-treated tumors (vehicle: 0.60% versus kaempferol-treated: 0.21%) (Fig. 7G, top row). We observed that a significant number of kaempferol-treated H1975 xenograft cells were apoptotic compared to the control tumor cells, showing that the shrinking of the tumor is associated with increased apoptosis (vehicle: 0.91% versus kaempferol-treated: 32.0%) (Fig. 7G, bottom row, and fig. S8, E and F). Together, P2Y2 inhibition using kaempferol selectively suppressed tumorigenesis by cells expressing the oncogenic EGFR mutant rather than EGFR-WT.

Targeting both P2Y2 and integrin pathways is a strategy for targeting EGFR with TKI-resistant mutations, such as C797S

The most common on-target resistance mechanism to third-generation EGFR TKIs, such as osimertinib, is the development of the C797S mutation, which has been observed in 10 to 26% of cases following treatment with osimertinib (83, 84). Now, there is no approved drug for the treatment of EGFR-C797S resistance in clinical use (85).

To explore alternative strategies, we investigated whether inhibition of P2Y2 and FAK (a downstream effector of integrin signaling) could selectively down-regulate EGFR-C797S compared to EGFR-WT. We used HEK cells stably expressing either EGFR-WT (HEK-EGFR-WT) or the triple mutant EGFR-19del-T790M-C797S (HEK-EGFR-C797S). HEK-EGFR-C797S cells exhibited stronger oncogenic properties than EGFR-WT cells (fig. S4, A to D), confirming the oncogenic nature of EGFR mutants. Cells were treated for 24 hours with kaempferol or the FAK inhibitor PF-562271 to assess EGFR protein stability. In HEK-EGFR-WT cells, kaempferol (20 and 50 μM) did not alter total EGFR levels, while PF-562271 (1 and 2.5 μM) caused a dose-dependent reduction in phospho-FAK (fig. S9A). In contrast, HEK-EGFR-C797S cells showed a pronounced loss of EGFR protein following kaempferol treatment (fig. S9B, top), and a similar dose-dependent reduction was observed with PF-562271 (fig. S9B, bottom).

We evaluated the impact of these inhibitors on cell proliferation. HEK, HEK-EGFR-WT, and HEK-EGFR-C797S cells were treated with varying concentrations of osimertinib, kaempferol, or PF-562271, and proliferation was monitored over 4 days using CyQUANT. Osimertinib had a minimal effect on proliferation in all cell lines, confirming the insensitivity of EGFR-WT cells and the resistance of EGFR-C797S mutants (Fig. 8A and fig. S9, C and D, top panels). Kaempferol showed limited growth inhibition in parental HEK cells up to 40 μM, with significant suppression only at 50 μM (fig. S9C, middle panel). EGFR-C797S cells were more sensitive to kaempferol, with ~16.56% cell survival at 50 μM by day 4 (Fig. 8B, middle panel) compared to ~43.05% survival in EGFR-WT cells (fig. S9D, middle panel).

Fig. 8. Inhibiting P2Y2, FAK, or a combination of both strongly hampers the cell proliferation of EGFR-19del-T790M-C797S.

HEK-EGFR-19del-T790M-C797S stable cells were treated with the indicated doses of (A) third-generation EGFR TKI (osimertinib), (B) P2Y2 inhibitor (kaempferol), and (C) FAK inhibitor (PF-562271) and analyzed for cell proliferation using the CyQUANT assay for 4 days. The heatmap graph illustrates the percentile scores of EGFR-19del-T790M-C797S cell proliferation using combination therapies: (D) kaempferol with osimertinib, (E) PF-562271 with osimertinib, and (F) kaempferol with PF-562271, presented as average fluorescence intensities. (G) Schematic model of the stability of EGFR-WT and mutants. The left panel depicts ATP levels and EGFR degradation in normal cells, the middle panel represents EGFR-mutant tumor cells, and the right panel outlines strategies to target EGFR stability (represented in red). Ab, antibody. Created in BioRender. G. T. K. Boopathy (2025); https://biorender.com/yklytxg. (H) Schematic representation of the stability of EGFR mutants resulting from the increased extracellular ATP levels, activating P2Y2 in complex with integrins and stabilizing EGFR. Created in BioRender. G. T. K. Boopathy (2025); https://biorender.com/yklytxg. (I) Proposed model for comprehensive targeting of diverse EGFR mutants in NSCLC using a P2Y2 inhibitor, NTPDase, FAK inhibitor, or a combination with TKIs to overcome TKI resistance. Created in BioRender. G. T. K. Boopathy (2025); https://biorender.com/yklytxg. Statistical analysis was performed using an ANOVA. **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001.

To assess potential synergistic effects, we tested combinations of osimertinib with kaempferol or PF-562271 and kaempferol with PF-562271. Treatment of HEK-EGFR-C797S cells with kaempferol or PF-562271 alone significantly reduced cell proliferation by days 3 to 4, with higher concentrations producing stronger inhibition (Fig. 8, B and C). We further evaluated combinatorial effects using percentile scores and heatmap visualization. The combination of osimertinib and kaempferol showed modest synergy (Fig. 8D), while osimertinib and PF-562271 demonstrated mild synergy at higher doses (Fig. 8E). Notably, kaempferol and PF-562271 together exhibited a strong synergistic reduction in cell proliferation (Fig. 8F). Collectively, these findings suggest that the concurrent inhibition of the EGFR mutant and FAK, or FAK and P2Y2, could be an effective treatment for the TKI-resistant C797S mutation.

DISCUSSION

Efficient endocytic uptake, accurate endosomal sorting, and the decision of whether receptors are recycled to the cell surface or targeted for lysosomal degradation are essential signaling activities linked to cellular growth, development, and differentiation (86, 87). Activating EGFR mutations, including 19del and L858R, hinder their degradation, leading to increased stability and the sustenance of oncogenic signaling by these EGFR mutants. In this study, a genome-wide RNAi screen revealed that the stability of activated EGFR mutants in NSCLC is maintained by a diverse set of genes acting through multiple signaling pathways, thereby supporting the persistence of oncogenic signaling. EGFR signaling is one of the critical pathways that regulate growth, survival, and proliferation (58). In addition to other receptors, EGFR signaling is intricately regulated by both positive and negative feedback loops (88–90). Considering EGFR’s crucial regulatory role, the presence of multiple EGFR stabilizer protein candidates in various signaling pathways implies a multilayered mechanism for the down-regulation of the receptor during distinct stages of growth, development, and multiple disorders. Receptor tyrosine kinases (RTKs), like EGFR, are a class of receptors that play a crucial role in the progression of various cancers (91). Further studies into the EGFR stabilizer proteins would uncover diverse cellular processes regulating EGFR stability and potentially extend to other RTKs. Inhibiting RTK stability in cancer is a new concept and could yield a new therapeutic approach to target RTKs that cause oncogenesis resulting from any anomaly.

Our findings on the stabilization of the EGFR mutant by P2Y2 due to the higher levels of ATP in the extracellular milieu of cancer cells identify a previously unidentified mechanism of regulation of EGFR. Cancer cells with elevated ATP levels exhibited stem cell–like properties, drug resistance, and increased cell migration, invasion, and metastasis (92–94). In the EGFR-mutant cells, ATP in the TME could be due to regulated or unregulated passive release (51). How EGFR-mutant cells secrete high ATP levels that activate specific pathways is an intriguing point that needs to be addressed by future study. Elevated extracellular ATP in the TME plays a critical role in determining the fate of tumors, making ATP release pathways suitable targets for cancer therapy (51). From a clinical perspective, our demonstration reveals that depleting ATP using apyrase markedly down-regulated oncogenic EGFR. Accordingly, targeting the human homolog of ectonucleotidase/ecto-nucleoside triphosphate diphosphohydrolase (NTPDase), CD39L3 can be a potential target for treating EGFR-mutated tumors.

In normal EGFR-WT cells, low extracellular ATP levels support efficient EGFR degradation in lysosomes (Fig. 8G, left panel). Notably, even when EGFR-WT cells are exposed to high levels of extracellular ATP, EGF stimulation can override the ATP-mediated stabilization, ultimately driving EGFR degradation. Our findings suggest that in activating EGFR-mutant cells, increased extracellular ATP levels activate P2Y2 in a complex with integrins. Furthermore, in patients with NSCLC, both P2Y2 and integrin β1 exhibit high expression levels. The formation of the EGFR-P2Y2-integrin protein complex stabilizes the ATP-induced, P2Y2-mediated EGFR mutant signaling. Consequently, this complex plays a pivotal role in preventing the degradation of the EGFR mutant, which contributes substantialy to lung tumorigenesis (Fig. 8H). The EGFR-P2Y2-integrin complex prevents the degradation of the EGFR mutant in lysosomes by stabilizing it in endosomes (Fig. 8G, middle panel).

An earlier study has shown that EGFR transactivation by ATP-activated P2Y2 occurs via Src kinase binding to SH3 motifs on P2Y2, enabling Src-dependent EGFR phosphorylation and downstream mitogen-activated protein kinase–driven cell proliferation (30). Alternatively, ATP-activated P2Y2 can stimulate metalloprotease-dependent shedding of EGFR ligands, further contributing to EGFR activation (34, 95). While our data reveal a distinct mechanism involving an ATP-driven P2Y2-integrin-EGFR complex that promotes EGFR stability, these classical pathways may function alongside the integrin-based stabilization mechanism, potentially converging on shared downstream proliferative signaling.

Our findings reveal that ATP enhances the interaction between P2Y2 and EGFR, a process critically dependent on the conserved RGD motif of P2Y2 and its association with integrins (26, 27, 62). Supporting this, treatment with an RGD peptide or the integrin β1–blocking antibody AIIB2 disrupted the P2Y2-EGFR interaction and led to a reduction in EGFR protein levels, highlighting the essential role of integrins as structural components of the P2Y2 complex in stabilizing EGFR. Previous studies have shown that P2Y2 interacts with RGD-binding integrins such as αvβ3 and αvβ5 through its first extracellular loop, independently of extracellular ATP (26, 27). This interaction activates downstream signaling cascades involving small guanosine triphosphatases (Rho, Rac, Vav2, and Cdc42) and G proteins (G_o and Gα₁₂), which regulate cytoskeletal remodeling, cell migration, and invasion (28, 72, 73). In pancreatic ductal adenocarcinoma, for instance, P2Y2-integrin engagement via αvβ3 has been shown to promote cell invasion, and disruption of this interaction effectively suppressed tumor progression (62).

While ATP binding to P2Y2 is required for EGFR transactivation, its association with integrins occurs independently of ATP (26, 27). Extending these observations to NSCLC, our data suggest that ATP stimulation enhances the formation of a P2Y2-integrin-EGFR complex, likely by reinforcing preexisting interactions under active signaling conditions rather than initiating a de novo ATP-dependent assembly. This scaffolding complex appears to be central to sustaining EGFR-mutant stability and may contribute to oncogenic persistence. Further studies will delineate whether the complex formation is dynamic or constitutive and how it contributes to EGFR stability in the context of oncogenic signaling. Nonetheless, the preassembled P2Y2-integrin complex is likely key to the EGFR mutant’s stability by forming the ATP-enhanced P2Y2-integrin-EGFR complex.

Moreover, ATP-activated P2Y2 may amplify integrin signaling via G_o and Gα₁₂, potentially enhancing the oncogenic potential of EGFR mutations. An additional layer of regulation may involve agonist-induced desensitization of P2Y2, followed by receptor internalization and recycling. This process could influence EGFR trafficking and signaling from endosomal compartments through the formation of complexes involving EGFR and integrins (96–98). Together, these findings highlight the P2Y2-integrin-EGFR complex as a critical axis in sustaining EGFR mutant stability and signaling. They also suggest that pharmacological disruption of this complex—either by targeting P2Y2 or its integrin partners—represents a promising strategy to destabilize the EGFR mutant and suppress oncogenic signaling in NSCLC (Fig. 8G, right panel).

Considering the correlation in mRNA expression between P2Y2 and EGFR (fig. S10A), P2Y2 and integrin β1 (fig. S10B), and EGFR and integrin β1 (fig. S10C) in patients with lung adenocarcinoma in The Cancer Genome Atlas data, the P2Y2-integrin-EGFR complex is of clinical relevance in the context of extracellular ATP levels in the TME and is also supported by our immunohistochemical staining of P2Y2 and integrin β1 in cancer samples. In addition, ATP-using and signaling proteins, such as ATP-binding cassette transporters, were significantly up-regulated in EGFR-mutant tumors compared to WT tumors (fig. S5, F and G). From a therapeutic standpoint, strategies such as depleting extracellular ATP using NTPDase, inhibiting P2Y2 with small-molecule inhibitors like kaempferol, targeting integrins or their downstream effectors, or using a combination of these approaches could offer effective means of targeting the EGFR mutants and their drug resistance (Fig. 8G, right panel).

Although activating EGFR mutations confer sensitivity to EGFR TKIs, the EGFR mutants are also a contributing factor to the acquired resistance observed with multiple generations of EGFR TKIs. The prolonged use of EGFR TKIs in lung cancer with genetic heterogeneity leads to clonal selection of TKI-resistant cells. The resistance arises because of the selection of cells that EGFR TKIs cannot effectively target, resulting in the eventual development of TKI-resistant lung tumors (Fig. 8I, right pathway). In this work, we herein propose an alternative strategy that targets the stability of EGFR mutants regardless of the nature of mutations. This involves using an inhibition of P2Y2, blocking integrin signaling, or a combination of these approaches alongside EGFR TKIs (Fig. 8I, left pathway). This strategy aims to target all activating EGFR mutants, irrespective of their specific characteristics. By doing this, we anticipate overcoming the resistance induced by multiple generations of EGFR TKIs. Given that kaempferol is currently being consumed by humans, it is potentially useful to synergize the therapeutic efficacy of TKIs with kaempferol in patients with lung cancer harboring activating mutations, although proper clinical trials will be needed.

MATERIALS AND METHODS

Cell culture, antibodies, plasmids, and reagents

NSCLC cells PC9, H1650, HCC827, HCC4006, and H1975 were generously provided by a lab at the institute. HEK293, H1299, A549, H460, and H226 cells were purchased from the American Type Culture Collection. Cells were cultured in RPMI 1640 medium with Hepes (Thermo Fisher Scientific, cat. no. 22400089) supplemented with 10% FBS (Thermo Fisher Scientific, cat. no. 26140079), penicillin-streptomycin (Thermo Fisher Scientific, cat. no. 15140163), puromycin (InvivoGen, cat. no. ant-pr-1), and G418 (InvivoGen, cat. no. ant-gn-5). Cells were allowed to reach confluence before being used for experiments. The following antibodies were used for immunoblotting and/or IP: pY1068-EGFR (Thermo Fisher Scientific, cat. no. 44-788G); Mouse anti-EGF Receptor [13/EGFR (RUO) BD Biosciences, cat. no. 555996]; anti-GFP (Abcam; ab290); anti-Lamp1 [Cell Signaling Technology (CST), cat. no. 9091]; anti–phospho-AKT (CST, no. 4060); anti-AKT (CST, no. 9272); anti–phospho-ERK1/2 (CST, cat. no. 9102); anti-ERK1/2 (CST, cat. no. 4370); anti–phospho-FAK (Tyr³⁹⁷) (CST, cat. no. 3283); anti-FAK (CST, cat. no. 3285); anti–Integrin Beta1 (Developmental Studies Hybridoma Bank; clone AIIB2); anti-FLAG M2 clone for Western blotting (Sigma-Aldrich, cat. no. F3165); EZview Red ANTI-FLAG M2 Affinity Gel for IP (Sigma-Aldrich, cat. no. F2426); ChromoTek GFP-Trap magnetic agarose (Proteintech, cat. no. gtma); anti–β-Actin-HRP [Santa Cruz Biotechnology (SCBT), sc-47778 HRP]; anti-rabbit IgG-HRP (SCBT, sc-2357) and m-IgGκ BP-HRP (SCBT, sc-516102); recombinant human EGF (Sino Biological, cat. no. 10605-HNAE); ATP (Sigma-Aldrich, cat. no. A1852); apyrase (Sigma-Aldrich, cat. no. A6535); RGDS peptide (Sigma-Aldrich, cat. no. A9041); PF-562271 (Selleck Chemicals, cat. no. S2890); AR-C 118925XX (Tocris Bioscience, cat. no. 4890); osimertinib (MedChemExpress, cat. no. HY-15772); kaempferol (MedChemExpress, cat. no. HY-14590).

Establishment of stable cells

To establish HEK293 stable cells expressing EGFR, lentiviruses were generated using pHAGE-EGFR-WT and pHAGE-EGFR-L747_T751delLREAT (19del) plasmids (Addgene plasmid nos. 116731 and 116270, gifts from Gordon Mills & Kenneth Scott). pHAGE-EGFR-L747_T751del-T790M (19del-T790M) and pHAGE-EGFR-L747_T751del-T790M-C797S (19del-T790M-C797S) were created from Addgene no. 116270 using the QuikChange II Site-directed mutagenesis kit (Agilent Technologies, cat. no. 200523). Following transduction, cells were subjected to puromycin selection at a concentration of 1 μg/ml for 72 hours after 48 hours of transduction. To establish stable H1299 cell lines expressing EGFR variants, the EGFR-WT-GFP (fused at the C terminus) plasmid (Addgene plasmid no. 32751, a gift from A. Sorkin) was used. EGFR-19del (L747_T751), 19del-T790M, 19del-T790M-C797S, L858R, L858R-T790M, and L858R-T790M-C797S, with a C-terminal-GFP tag, were derived from the EGFR-WT-GFP plasmid using the QuikChange II Site-directed mutagenesis kit. Plasmids were introduced via nucleofection using Lonza Nucleofector Kit C. Following 48 hours of nucleofection, cells underwent selection with G418 (1 mg/ml) for 5 days and sorted for GFP-positive cells.

Genome-wide RNAi screen

In the primary screen, 2.5 μl of 500 nM siRNA was dispensed into black-walled 384-well plates (Grenier, cat. no. 781091) using Velocity 11. For secondary screens, siRNA plates were prepared by selecting individual siRNA tubes. Thermo Fisher Scientific provided custom-ordered deconvoluted siRNAs for both sets, and these siRNAs were subsequently robotically printed into 384-well plates, similar to the process used in the primary screen. The reverse siRNA transfection method involved premixing 0.25 μl of Lipofectamine RNAiMAX (Thermo Fisher Scientific, cat. no. 13778075) with 7.25 μl of Opti-MEM per well for 5 min. This mixture was then added to the siRNA for complexation, which was allowed to proceed for 20 min. Subsequently, 3000 H1299-EGFR19del-GFP cells per well were introduced using the Multidrop combi (Thermo Fisher Scientific). ON-TARGETplus SMARTpool siRNAs siNon-Targeting (siControl) (cat. no. D-001810-10-50) was used as the negative control and siEGFR (cat. no. L-003114-00-0010) as the positive control.

Transfection, knockdown, and KO experiments

Knockdowns were conducted using ON-TARGETplus SMARTpool siRNAs for human EGFR (cat. no. L-003114-00-0010), P2Y2 (cat. no. L-003688-00-0010), and a nontargeting control pool (cat. no. D-001810-10-50). Transfection was accomplished using Lipofectamine RNAi Max (Invitrogen) with a final siRNA concentration ranging from 100 to 200 nM. After 72 hours of transfection, the knockdown efficiency was validated through a Western blot analysis of whole-cell lysates.

For the CRISPR-Cas9–based KO of P2Y2 in H1975 cells, a combination of three sgRNA (single guide RNA)–containing plasmids was transfected. This was achieved using the P2RY2 sgRNA CRISPR-Cas9 All-in-One non-viral vector set (target sequences: 5′-CTACAGGTGCCGCTTCAACG-3′, 5′-CTGGTCTATTACTACGCCCG-3′, 5′-GTCGTAAGACGCCCAGACAC-3′) obtained from Applied Biological Materials (cat. no. 358711110591). Following 24 hours of transfection, GFP-positive cells were sorted and subsequently plated in a 15-cm tissue culture dish. Individual colonies were isolated, expanded, and subjected to validation for the KO of P2Y2. This validation involved PCR amplification of P2Y2 cDNA using the following primers (forward: 5′-CCCTGGAATGACACCATCAA-3′; reverse: 5′-CCATGAGCACGTAACAGACA-3′; amplicon size: 646 base pairs).

Cell proliferation assays

For the cell proliferation assay following siRNA knockdown, 5000 cells were plated in 96-well plates 1 day after siRNA transfection and incubated for 72 hours. Cell viability at specified time points was assessed using the CellTiter 96 Aqueous One Solution Cell Proliferation Assay kit (Promega). The assay continued until cells in the control group reached 100% confluence, and cell viability was determined by measuring absorbance at 490 nm with a 96-well plate reader. For the cell proliferation assay involving drug treatment, 5000 cells were plated in 96-well plates, and drugs were added the next day; at specified time points, cells were assessed using the CyQUANT Direct Cell Proliferation Assay Kit (Invitrogen) until cells in the control group reached 100% confluence. Cell viability was determined on the basis of fluorescence readings at 485/528 nm using a 96-well plate reader.

Cell surface ATP measurement

To assess the cell surface ATP levels in HEK293 cells expressing EGFR-WT and its mutants (19del, 19del-T790M, and 19del-T790M-C797S), we seeded 100,000 cells in six-well plates and allowed them to grow for 24 hours. Subsequently, the cells were transfected with the GFP-based single-wavelength ATP sensor pm-iATPSnFR1.0. After an additional 24 hours, cells were serum starved for 16 hours, and cells were imaged for fluorescence using a fluorescence microscope. Simultaneously, we analyzed the GFP intensity. The cells were harvested for Western blot analysis, as described above. Extracellular ATP levels were also recorded using the RealTime-Glo Extracellular ATP Assay (Promega, cat. no. GA5010) as per the manufacturer’s instructions.

Colony formation assay

To assess the role of P2Y2 in colony formation, H1975, P2Y2-KO1, and P2Y2-KO2 cells were seeded in six-well plates at a density of 1000 cells per well in 2 ml of complete RPMI medium. The culture medium was replaced every 3 days with 4 ml of fresh RPMI containing 10% FBS, penicillin, and streptomycin. After 10 to 12 days, the colonies were washed twice with phosphate-buffered saline (PBS), fixed, and stained for 30 min with a solution of 0.1% (w/v) crystal violet and 20% (v/v) ethanol. Excess crystal violet was removed by washing the colonies three times with PBS. Once dried, images of the stained colonies were captured using an EPSON scanner, and the colony numbers were quantified using ImageJ across three independent experiments.

Cell invasion assay

Cells were seeded in six-well plates 1 day before siRNA transfection. Following knockdown for 48 hours, cells were trypsinized, and 30,000 cells were replated in each upper chamber. The chemoattractant (typically RPMI + 20% FBS) was added to the bottom of a 24-well plate, which was then placed in the incubator. After 24 hours, the cells were fixed with 4% PFA, washed five times with PBS, stained with a filtered crystal violet solution (1% crystal violet in 2% methanol and 6% glutaraldehyde) for 20 min, and then washed five times in PBS. Cells from the top chamber were removed using cotton buds, and the invaded cells were recorded using a bright-field microscope. The area of invasion was quantified using ImageJ by measuring the invasive regions of cells in five random fields for each experimental condition across three independent experiments.

Spheroid formation assay

Spheroid cultures were generated in 96-well round-bottom plates coated with 0.3% agar (1st BASE, cat. no. BIO-1000-500 g). H1975, P2Y2-KO1, and P2Y2-KO2 cells were seeded at a density of 10,000 cells per well in 100 μl of RPMI media containing 10% FBS, penicillin, and streptomycin. The plates were incubated at 37°C in a humidified cell culture incubator for 3 days. Spheroid formation was monitored, imaged, and measured using a microscope.

ATP and apyrase assays

An equal number of cells were seeded in six-well plates and allowed to grow for 24 hours until they reached nearly 90% confluency. Subsequently, they were serum starved in Opti-MEM for 16 hours. The cells were treated with either ATP (100 μM) or apyrase (10 U/ml) and incubated at 37°C for specific time points mentioned. The cells were washed twice with cold PBS and harvested for Western blot analysis as mentioned above.

Immunofluorescence

The cells underwent fixation using a solution of 4% PFA and 2% sucrose, followed by six washes with PBS-CM [1× PBS containing 1 mM magnesium chloride (MgCl₂) and 1 mM calcium chloride (CaCl₂)]. Subsequently, permeabilization was carried out in PBSCM-T (PBS-CM with 0.1% Triton X-100). For blocking, cells were treated with FDB buffer (5% FBS, 5% goat serum, 2% bovine serum albumin in PBS, 1 mM MgCl₂, and 1 mM CaCl₂) and then incubated with primary antibodies, followed by secondary antibodies, both prepared in FDB buffer, for 1 hour at room temperature. Extensive washing followed each antibody incubation. Coverslips were mounted using Vectashield containing 4′,6-diamidino-2-phenylindole (DAPI) for nuclear staining. Observation and capture of images were performed using a Zeiss LSM 880 confocal laser scanning microscope. The primary antibodies used for staining included LAMP1 (1:50), LC3B (1:100), and Flag M2 (1:300), with secondary antibodies being goat anti-rabbit or mouse IgG labeled with Alexa Fluor 488 (1:200), Alexa Fluor 555 (1:300), and Alexa Fluor 647 (1:50), all obtained from Thermo Fisher Scientific. Colocalization analysis was subsequently calculated using Zeiss Zen software.

EGFR degradation assay

Cells were seeded in six-well plates, grown for 24 hours until they were nearly 60 to 70% confluent, serum starved in Opti-MEM for 16 hours, then treated with EGF (50 ng/ml), and incubated at 37°C for specified time points. Then, the cells were washed with cold PBS twice to stop EGFR internalization and degradation and then harvested for Western blot analysis as mentioned above. For immunofluorescence, cells were seeded, serum starved as above, pretreated with EGF (50 ng/ml) on ice to label the cell surface EGFR for 1 hour, then washed in cold PBS and incubated at 37°C for specified time points, and processed for immunofluorescence.

Reverse transcription PCR (RT-PCR)

Total cellular RNA was extracted using the RNeasy Kit (QIAGEN). Complementary DNA from the total RNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, cat. no. 4368814). TaqMan probes for human EGFR (Hs01076078_m1), human P2Y2 (Hs00175732_m1), and human β-Actin (Hs99999903_m1) were acquired from Applied Biosystems. Real-time PCR was conducted in a 96-well format, and experiments and analyses followed the manufacturer’s guidelines. For standard RT-PCR, 2X KAPA Taq Extra HotStart Ready Mix was used to characterize all the primer pairs according to the length of the products. The primer pairs used were the following: EGFR primers (forward: 5′-AACACCCTGGTCTGGAAGTACG-3′; reverse: 5′-TCGTTGGACAGCCTTCAAGACC-3′), P2RY2 primers (forward: 5′-TGGCTCGGCGACTGCTAAA-3′; reverse: 5′-GCATCTCGGGCAAAGCGT-A-3′), and β-actin primers (forward: 5′-TCATCACCATTGGCAATGAG-3′; reverse: 5′-CACTGTGTTGGCGTACAGGT-3′).

Coimmunoprecipitation

For coimmunoprecipitation experiments, cells were grown in complete medium (unless stated) initially harvested using tris-buffered saline containing EDTA (2 mM)/EGTA (1 mM) buffer (150 mM NaCl and 100 mM tris, pH 7.5). After washing the cells twice with tris-buffered saline, they were snap-frozen in dry ice and subsequently lysed in a specialized GPCR IP buffer (1% digitonin, 100 mM tris-HCl, and 150 mM NaCl, pH 7.5) supplemented with a proteinase inhibitor cocktail and phosphatase inhibitor cocktail from Roche Applied Sciences.

From the lysates, 30 μg was set aside as input, while 1 mg was used for the pulldown process. Specifically, P2Y2 was pulled down using anti–Flag-M2 agarose, and EGFR-GFP was pulled down with anti-GFP magnetic beads; both were performed overnight at 4°C. The pulldown lysates underwent three or four washes with GPCR IP buffer and were then denatured at room temperature for 2 hours with shaking using 2× urea sample buffer (4 M urea, 0.1% bromophenol blue, 4% SDS, 100 mM tris, pH 6.8, 20% glycerol, and 200 mM dithiothreitol). Simultaneously, the input lysates were denatured under similar conditions but with 1× IP urea sample buffer (6 M urea, 0.1% bromophenol blue, 2% SDS, 5 mM tris, pH 6.8, 20% glycerol, and 200 mM dithiothreitol). The eluates were then loaded onto 8% gels and resolved by SDS-PAGE.

Immunoblotting

Cell harvesting was carried out using NP-40 lysis buffer (1% NP-40, 25 mM tris-HCl, pH 7.4, and 150 mM NaCl) supplemented with a protease inhibitor cocktail and phosphatase inhibitor cocktail from Roche Applied Sciences. The cell lysates underwent vortexing and were then centrifuged at 14,000 rpm for 15 to 20 min at 4°C. The protein concentration was determined using the Bradford assay from Thermo Fisher Scientific. Subsequently, 30 μg of protein was denatured and boiled for 5 min in SDS sample buffer before being loaded onto 8% SDS-PAGE gels. The separated proteins were transferred to a 0.45-μm-pore-size polyvinylidene difluoride membrane from Merck. After blocking with 5% milk, the membrane was incubated with primary antibodies (1:1000), followed by secondary antibodies (1:2000). The protein bands were detected using the Pierce ECL Western blotting substrate from Thermo Fisher Scientific and visualized with the ChemiDoc imaging system from Bio-Rad. Densitometric analysis was performed using ImageJ.

Mouse xenograft experiments

All animal care and experimental procedures adhered to the guidelines of the Institutional Animal Care and Utilization Committee; female NOD-SCID and female NSG (NOD-SCID γ) mice (6 to 8 weeks old) were purchased from InVivos, Singapore, and fed with standard laboratory diet and distilled water ad libitum. The animals were kept on a 12-hour light/dark cycle at 22 ± 2°C in the Biological Resource Centre, ASTAR, Singapore. For the xenograft model involving H1975 and H1975 P2Y2-KO1 cells, ~50 μl of a cell suspension containing 3 × 10⁶ cells (either WT H1975, n = 6, or H1975 P2Y2-KO1, n = 6) in 50 μl of Matrigel was injected subcutaneously into the right flank of each mouse. Tumor growth at the inoculation site was monitored periodically, and the tumor volume was measured using a vernier caliper for 46 days, after which the mice were euthanized for further analysis. The tumor volume was calculated using the formula V = a × b² × 0.52, where a is the largest and b is the smallest diameter of the tumor. For statistical analysis, we have performed repeated measures of two-way ANOVA with days as the within-the-subject factor and P2Y2 KO1 as the between-subject factor. For the xenograft analysis involving the treatment of H1299 and H1975 cells with kaempferol or vehicle, 5 × 10⁶ cells/100 μl were suspended in Matrigel in equal proportions. Subsequently, 10 mice for each cell line were subcutaneously injected. After 10 days, tumor engraftment was confirmed by palpation, and the mice were stratified into two groups for each cell line. Group one was dosed via oral gavage (16- to 18-gauge size) with vehicle control, and group two was dosed with kaempferol (150 mg/kg) daily for 24 days using an oral gavage needle for both H1299 and H1975 cell xenograft models. The tumor volume and body weight were measured weekly twice. After 24 days of treatment with kaempferol, the study animals were euthanized, and tumor was harvested and weighed. Tumor volumes were calculated using the formula V = LB²/2 (where V represents the volume of the tumor, L is the length of the tumor, and B is the breadth of the tumor measured in millimeters). All aspects of animal care and handling adhered to the guidelines of the Institutional Animal Care and Use Committee (no. 201572) and Institutional Review Board (IRB reference: 2023-063).

Immunohistochemistry

Tissue specimens of patients with NSCLC were collected at China Medical University Hospital (IRB no. CMUH105-REC2-073). The samples were processed, embedded in paraffin, and sectioned at 3 μm. The paraffin sections were deparaffinized and hydrated using the following steps: three times for 5 min in xylene; 5 min in 100% ethanol, 5 min in 80% ethanol, and 5 min in 75% ethanol; and 3 min in PBS at room temperature. For P2Y2 staining, 1 mM EDTA buffer (pH 9.0) was used for antigen retrieval. Endogenous peroxidase was quenched with 3% hydrogen peroxide for 10 min, followed by blocking with 3% blocking buffer for 5 min. An anti-P2Y2 antibody (Alomone Labs, APR-010) and anti-integrin β1 (D2E5) antibody (CST, no. 9699) were used at 1:3000 and 1:100, respectively. The slides were incubated with the HRP (horseradish peroxidase) polymer reagent for 30 min, then incubated with a 3,3′-diaminobenzidine solution (Leica Biosystem, RE7105), and counterstained with hematoxylin (99, 100). The P2Y2 and integrin β1 immunoreactivity was categorized into four groups (scores 0, 1, 2, and 3) according to a well-established system in which the H-score was generated by the percentage of positive normal bronchiolar epithelial cells and tumor cells. The scores with their indicated percentage of cells are 0 (0%), 1 (less than 30%), 2 (31 to 70%), 3 (more than 70%). All slides were independently viewed and scored by two pathologists. Slides in which there was a scoring discrepancy were reevaluated and reconciled by a two-headed microscope.

RNA-seq data analysis on patients with NSCLC

Raw RNA-seq counts from EGFR-mutant [N = 71 for Chen et al. (59) and N = 21 for Huang et al. (60)] and WT NSCLC datasets (N = 98 for Chen et al. and N = 27 for Huang et al.) were accessed. The EGFR mutation status was determined from the respective studies. The R package DESeq2 (version 1.44.0) (101) was used to analyze differential gene expression between EGFR-mutant and WT samples. The statistical significance of gene expression differences was assessed using DESeq2’s built-in Wald test. An ATP-using and signaling gene list was curated from a published study (61). Volcano plots for this gene list were generated using the EnhancedVolcano R package (version 1.22.0) (102). Significant genes were identified as P value <0.05 and |log₂ fold change| > 0.5.

Statistical data analysis

Results are presented as the means ± SD or means ± standard error of the mean (SEM), mentioned accordingly. Statistical significance was determined between control and treatment groups by a one-way and two-way ANOVA followed by Bonferroni’s or Tukey’s multiple comparison test (as recommended) using GraphPad Prism software. Prism was also used for plotting the graph and calculating the P values (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001). A P value less than 0.05 was considered as significant, and ns represents nonsignificant differences.

Supplementary Materials

This PDF file includes:

Figs. S1 to S10

Table S1