Abstract Approximately 20% of patients with metastatic non-small cell lung cancer (NSCLC) harbor an EGFR-mutation (EGFRm). Osimertinib, a third generation EGFR inhibitor, has dramatically improved patient outcomes, but invariably, residual disease or drug-tolerant “persister” (DTP) cells survive treatment, eventually giving way to resistant or progressive disease. Eradicating DTPs remains a key challenge. Although DTPs have been characterized at the transcriptional level, a major gap in our understanding is how the tyrosine kinome is perturbed. We hypothesized that defining the phospho-tyrosine landscape of DTPs would uncover kinases that could be exploited therapeutically. We generated DTPs by treating EGFRm cell lines with osimertinib (IC90) for 14 days and performed tyrosine-enriched, phospho-proteomics. Analysis using Kinase Library revealed that FER, a non-receptor tyrosine kinase that has been implicated in regulating a variety of pathways, is highly active in DTPs. To validate our findings, we confirmed that FER is auto-phosphorylated in DTPs. We then demonstrated that FER and its known substrates are phosphorylated within hours of osimertinib treatment, suggesting FER is activated early and throughout treatment. Next, we tested the functional impact of FER. FER knockdown had no effect on baseline phenotypes such as morphology or proliferation, but FER knockdown markedly decreased DTPs after 14 days of osimertinib treatment. We thus evaluated inhibiting FER as a therapeutic strategy. Alectinib, a well-tolerated FDA approved drug designed to target the ALK kinase, potently inhibits FER. We hypothesized that inhibiting FER using alectinib would eliminate DTPs and prevent osimertinib resistance. We first confirmed that 1) alectinib inhibits FER kinase activity and 2) ALK is not expressed in EGFRm NSCLC, consistent with prior reports. We subsequently treated cells with alectinib at 500nM, the physiologic concentration achievable using low dose alectinib in patients. Combining alectinib with osimertinib dramatically reduced DTPs in four EGFRm cell lines. Phospho-proteomics and RNA-sequencing revealed that alectinib and osimertinib suppressed FER phosphorylation and RHO GTPase signaling, a pathway known to drive DTP survival. Combining alectinib with osimertinib also significantly enhanced osimertinib sensitivity in patient-derived, EGFRm organoids. To test the strategy in vivo, we treated EGFRm xenografts with osimertinib or osimertinib and alectinib for only 21 days. After stopping treatment, 100% of the osimertinib-only treated mice had to be euthanized while 33% of mice treated with combination osimertinib and alectinib were disease-free after 150 days of observation. Collectively, our data show that FER is a key tyrosine kinase that is activated in DTPs and that targeting it with alectinib, an FDA approved drug, could be an effective strategy to enable durable remission in patients with EGFRm NSCLC. Citation Format: Bobak Parang, Rabia Khan, Ariana Kupai, Yuyun Huang, Sungyun Cho, Michal J. Nagiec, Eric E. Gardner, Florencia M. Rowdo, Benjamin D. Hopkins, Qin Fu, Sheng Zheng, Timothy F. Burns, John Blenis. FER drives drug-tolerant persister cell survival in EGFR-mutant lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 7039.

The adaptive nature of cancer is a major obstacle limiting durable treatment responses. Histological transformation (HT) is a process whereby one cancer changes into a categorically different tumor type, often following treatment with targeted therapy. Best characterized in lung and prostate adenocarcinomas, HT is particularly disconcerting because the resultant cancer no longer depends on the initial oncogenic driver program, is therapeutically recalcitrant, and is often highly metastatic. Partly because HT is technically difficult to study, this process remains poorly described. As newer therapies broaden the scope of oncogenic drivers that can be targeted, HT may become more prevalent, highlighting the need for further research to dissect these phenomena. We propose that modern experimental and analytical tools present an opportunity to advance our understanding and improve our clinical management of histologically transforming cancers.

Abstract Background: EGFR mutant lung cancers initially respond to frontline targeted therapy. However, resistance inevitably develops through diverse mechanisms. Resistant tumors universally emerge from drug tolerant persister cells (DTPCs) that survive initial treatment. Despite this knowledge, the biology of EGFR mutant DTPCs is not well understood. This study was conducted to interrogate DTPC phenotypes that enable evolution of therapy resistance. Methods: DTPC models were generated by treating EGFR mutant cell lines with >IC90 concentrations of osimertinib for 14 days. After 14 days, cells were profiled with 10X 3’ single cell RNA sequencing (scRNAseq). Seurat was used to integrate data, normalize data, cluster cells, and conduct differential expression analyses. UCell was used to score cells for activity of biology pathways. scanpy was used to analyze scRNAseq data from the Gardner et al. ERPMT mouse model. Human patient samples were collected from lung cancer patients treated at MD Anderson Cancer Center using IRB approved protocols. Samples were profiled with 10X 5’ scRNAseq. scanpy was used to integrate data, normalize data, and cluster cells. Palantir was used to perform trajectory analyses. Results: Differential expression (DE) analysis comparing DTPCs and control cells were conducted for cell line pairs. Using DE results, we generated consensus gene signatures capturing genes universally upregulated or downregulated across DTPC models. Review of these genes identified that KRT17 was highly upregulated in osimertinib DTPCs. We found KRT17+ DTPCs upregulate epithelial-to-mesenchymal (EMT), stemness, and aberrant basaloid signatures compared to KRT17- DTPCs. Subclustering showed that KRT17 DTPCs are heterogeneous and label non-overlapping EMT, proliferative, and MET expressing populations. To validate our in vitro findings, we analyzed scRNAseq datasets from multiple in vivo sources. In GEMM models, Krt17+ cells populate minimal residual disease (MRD) following suppression of mutant EGFR. Krt17+ cells were a stem-like population that emerged as residual tumor cells lost alveolar epithelial identity prior to neuroendocrine transformation. Lastly, we confirmed the clinical relevance of our findings by investigating human patient clinical specimens. Compared to treatment naïve samples, KRT17+ cells were significantly enriched in osimertinib-treated MRD samples. Trajectory analyses demonstrated that KRT17 expression characterized cell populations that preceded development of resistance through several known mechanisms, including MET amplification and histologic transformation (HT). Conclusions: Our data demonstrate that EGFR-mutant cells surviving initial treatment converge onto a novel KRT17+ DTPC state which serves as a multipotent progenitor population capable of leveraging diverse mechanisms to resist therapy, including HT. Citation Format: Benjamin B. Morris, Monique B. Nilsson, Ethan Earlie, Eric E. Gardner, Hong Chen, Santiago G. Trevino, Alexa J. Halliday, Yuanxin Xi, Jing Wang, Natalie Vokes, Don Gibbons, Jianjun Zhang, Ashley M. Laughney, Yasir Y. Elamin, Xiuning Le, John V. Heymach. Multi-lineage evolution of drug resistance via a novel keratin 17+ drug tolerant persister population in EGFR-mutant NSCLC [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 7027.

Recent studies show that genetic sequencing can not fully explain drug resistance in non-small cell lung cancer (NSCLC), suggesting undiscovered non-genetic mechanisms that can enable cancer cell survival. Propionate metabolism is the pathway by which odd-chain fatty acids, branched chain amino acids, and cholesterol are metabolized. We have previously shown that methylmalonic acid (MMA), a byproduct of propionate metabolism that accumulates when the pathway is disrupted, can activate epithelial-to-mesenchymal transition (EMT) in cell lines. But the clinical significance of propionate metabolism in cancer patients is not known. Here we show, for the first time, that propionate metabolism is dysregulated in patients with non-small cell lung cancer. MMA is elevated in lung tumors and in the serum of patients with metastatic NSCLC. Metabolism of cobalamin associated B (MMAB), a key regulatory gene of propionate metabolism, is downregulated in NSCLC and drug-tolerant persister cells, leading to MMA accumulation and EMT activation. We show that restoring expression of MMAB in NSCLC enhances targeted therapy and suppresses TGFβ signaling. These findings reveal propionate metabolism dysregulation as a non-genetic mechanism of drug resistance and highlight propionate metabolism as a potential therapeutic target.

Figure S4. Members of the salt Inducible kinase family phosphorylate TPI1.

The discovery of EGFR mutations two decades ago launched an era of rapid development and clinical application of targeted therapies in NSCLC. Today, increasing numbers of targeted therapies against somatic aberrations involving nine different genes have become available for treating patients with lung cancer and have improved their outcomes. However, acquired resistance and tumor tolerance to these therapies remains one of the biggest challenges in lung cancer treatment today. Most, if not all, targeted therapies have limited durability, which we now recognize is due to both genetic and non-genetic mechanisms of resistance. The state of our current understanding of resistance and new approaches to prevent or overcome resistance were recently presented at the International Association for the Study of Lung Cancer Hot Topics Meeting. Here, we summarize and discuss the emerging concepts and new strategies for combating drug tolerance and resistance in targeted therapies, including our understanding of the role of genetics, drug-tolerant persister cells, tumor plasticity and lineage transformation, spatial and temporal heterogeneity, microenvironmental influence, and novel therapeutic approaches.

We previously described our initial efforts to develop a model for small cell lung cancer (SCLC) derived from human embryonic stem cells (hESCs) that were differentiated to form pulmonary neuroendocrine cells (PNECs), a putative cell of origin for neuroendocrine-positive SCLC. Although reduced expression of the tumor suppressor genes TP53 and RB1 allowed the induced PNECs to form subcutaneous growths in immune-deficient mice, the tumors did not display the aggressive characteristics of SCLC seen in human patients. Here, we report that the additional, doxycycline-regulated expression of a transgene encoding wild-type or mutant MYC protein promotes rapid growth, invasion, and metastasis of these hESC-derived cells after injection into the renal capsule. Similar to others, we find that the addition of MYC encourages the formation of the SCLC-N subtype, marked by high levels of NEUROD1 RNA. Using paired primary and metastatic samples for RNA-sequencing, we observe that the subtype of SCLC does not change upon metastatic spread and that production of NEUROD1 is maintained. We also describe histological features of these malignant, SCLC-like tumors derived from hESCs and discuss potential uses of this model in efforts to control and better understand this recalcitrant neoplasm.

Lineage plasticity, the ability of cells to transition to an alternative phenotype as a means for adaptation, is an increasingly recognized mechanism of tumor evolution and a driver of resistance to anticancer therapies. The most extensively described clinical settings impacted by such molecular phenomena include neuroendocrine transformation in androgen receptor-dependent prostate adenocarcinoma, and adenocarcinoma-to-neuroendocrine and adenocarcinoma-to-squamous transdifferentiation in epidermal growth factor receptor-driven lung adenocarcinoma, affecting 10%–20% of patients treated with targeted therapy. Recent analyses of human tumor samples and in vivo models of histological transformation have led to insights into the biology of lineage plasticity, including biomarkers predictive of high risk of transformation. However, no clinically available therapies aimed to prevent or revert plasticity are currently available. In the present review, we will provide a biological and therapeutic overview of the current understanding of common and divergent molecular drivers of neuroendocrine and squamous transdifferentiation in tumors from different origins, including descriptive analysis of previously known and recently described molecular events associated with histological transformation, and propose evidence-based alternative models of transdifferentiation. A clear definition of the commonalities and differences of transforming tumors in different organs and to different histological fates will be important to translate molecular findings to the clinical setting.

Figure S7. Tpi1 is required for tumor growth in KP and KPL LUAD models.

Abstract Purpose: Small cell lung cancer (SCLC) is a highly aggressive and deadly malignancy. Two major factors contributing to the high mortality of SCLC are early metastasis and rapid development of therapy resistance. Recent research suggests upregulation of the epithelial-mesenchymal transition (EMT) program and the EMT transcription factor Twist1 correlated with accelerated tumor progression and chemoradiation (CRT) resistance in SCLC. However, a causal relationship between Twist1 and these aspects of SCLC biology has not been rigorously studied. Here, we investigated whether Twist1 upregulation could promote SCLC tumorigenesis, metastasis, and resistance to CRT. Materials and Methods: We analyzed transcriptomic data from the IMpower133 phase 3 trial of extensive stage SCLC stratified by median TWIST1 levels using Kaplan-Meir statistic. To investigate the roles of Twist1 directly in SCLC biology, we have generated a novel genetically engineered mouse model (GEMM) termed RPGT (Rb1Flox; Trp53Flox; ROSA26LSL-rtTA-IRES-EGFP; Twist1-TetO7-Luc). The RPGT GEMM enables the generation of autochthonous SCLC tumors after induction with Cre recombinase adenovirus. By withdrawing or providing doxycycline to mice, we can control Twist1 expression to activate or inactivate EMT in tumors. By characterizing mouse tumor samples with histological, immunostaining, and transcriptomic profiling analyses, we evaluated the impact of Twist1 and EMT upregulation on SCLC tumor biology and plasticity as well as metastatic spread and outgrowth. To evaluate the impact of EMT activation on SCLC sensitivity to CRT, we performed in vitro viability assays on primary tumor cell lines established from RPGT tumors. We also treated RPGT mice with vs. without Twist1 overexpression with CRT to validate our in vitro data. Results: In the chemotherapy alone control arm of IMpower133 there was a trend towards a difference in overall survival (OS) between TWIST1-high or -low patients (p=025 by log-rank) and in the NMF1/NEUROD1 subtype, patients with TWIST1-high had statistically significantly inferior OS (p=0.01). We then observed that both control (no Twist1 overexpression) and Twist1-overexpressing RPGT mice developed neuroendocrine SCLC tumors. While SCLC-ASCL1 was the predominant subtype in both cohorts, RPGT tumors exhibited more plasticity, with features associated with SCLC-NEUROD1 subtype. Furthermore, Twist1 overexpression dramatically increased the metastatic incidence in RPGT compared to control animals, indicating an important role of EMT in SCLC dissemination. Transcriptomic profiling of primary tumors and matching metastases in RPGT mice also revealed downregulation of Twist1 and EMT in metastases, suggesting that EMT suppression was necessary for metastatic outgrowth. Furthermore, we found that repressing Twist1 expression enhanced SCLC susceptibility to CRT both in vitro and in vivo. Conclusions: Overall, our data suggest that TWIST1/Twist1 plays an important role in promoting SCLC plasticity, metastasis, and treatment resistance. Citation Format: Triet Nguyen, Jinhee Chang, Kathleen Gabrielson, Amol Shetty, Yang Song, Apaala Chatterjee, Audrey Lafargue, Yoo Sun Kim, Danielle Council, Aaron Chan, Dipanwita Dutta Chowdhury, Muhammad Ajmal Khan, Nick Connis, Daniel Sforza, Eric Gardner, Christopher McFarland, Mohammad Rezaee, Nitin Roper, Christine Hann, Phuoc T. Tran. TWIST1 Associates with resistance to treatment and Twist1 drives tumor progression in vivo for small cell lung cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Translating Targeted Therapies in Combination with Radiotherapy; 2025 Jan 26-29; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(2_Suppl):Abstract nr B029.

Figure S3. TPI1 phosphorylation regulates triose phosphate levels and metabolic flux.

Figure S6. TPI1 allelic variants effects central carbon metabolic flux.

Figure S8. Tpi1 inactivation in KL autochthonous tumor model alters histology and pathology.

Figure S2. Phosphorylation of human TPI1 is LKB1-dependent.

Recent studies show that genetic sequencing can not fully explain drug resistance in non-small cell lung cancer (NSCLC), suggesting undiscovered non-genetic mechanisms that can enable cancer cell survival. Propionate metabolism is the pathway by which odd-chain fatty acids, branched chain amino acids, and cholesterol are metabolized. We have previously shown that methylmalonic acid (MMA), a byproduct of propionate metabolism that accumulates when the pathway is disrupted, can activate epithelial-to-mesenchymal transition (EMT) in cell lines. But the clinical significance of propionate metabolism in cancer patients is not known. Here we show, for the first time, that propionate metabolism is dysregulated in patients with non-small cell lung cancer. MMA is elevated in lung tumors and in the serum of patients with metastatic NSCLC. Metabolism of cobalamin associated B (MMAB), a key regulatory gene of propionate metabolism, is downregulated in NSCLC and drug-tolerant persister cells, leading to MMA accumulation and EMT activation. We show that restoring expression of MMAB in NSCLC enhances targeted therapy and suppresses TGFB signaling. These findings reveal propionate metabolism dysregulation as a non-genetic mechanism of drug resistance and highlight propionate metabolism as a potential therapeutic target.

Figure S5. LKB1 regulates the multimeric state of hTPI1 but not mTpi1 and substitution of Cys in hTPI1 causes formation of a covalent TPI1 dimer.

The discovery of EGFR mutations over two decades ago launched an era of rapid development and clinical application of targeted therapies in non-small cell lung cancer. Today, increasing numbers of targeted therapies against somatic aberrations involving nine different genes have become available for treating lung cancer patients and have improved their outcomes. However, acquired resistance and tumor tolerance to these therapies remains one of the biggest challenges in lung cancer treatment today. Most, if not all, targeted therapies have limited durability, which we now recognize is due to both genetic and non-genetic mechanisms of resistance. The state of our current understanding of resistance and new approaches to prevent or overcome resistance were recently presented at the International Association for the Study of Lung Cancer (IASLC) Hot Topics Meeting. Here, we summarize and discuss the emerging concepts and new strategies for combating drug tolerance and resistance in targeted therapies, including our understanding of the role of genetics, drug tolerant persister cells, tumor plasticity and lineage transformation, spatial and temporal heterogeneity, microenvironmental influence, and novel therapeutic approaches.

Figure S1. Differential co-occurrence and effects of KRAS, TP53 and LKB1 mutations on human and mouse LUADs.

Metastatic cancer cells invade tissue, overcome nutrient stress, and survive transit to distant sites. Many of the mechanisms that support these processes are incompatible with proliferation. This study defines cellular transition states in breast epithelial cells undergoing epithelial-mesenchymal transition (EMT) driven by ERK2 and TGF-β signaling. EMT triggers robust endolysosomal system upregulation and metabolic adaptations that balance proliferative and invasive states. Surprisingly, invasive cells rely on scavenging via lysosomes and macropinocytosis to acquire amino acids, rather than plasma membrane transport, even in nutrient-rich conditions. Macropinocytosis increases intracellular amino acid storage, promoting survival during amino acid deprivation. This metabolic shift depends on c-MYC downregulation, an early EMT event. Reintroducing c-MYC suppresses the metabolic switch, endolysosomal induction, macropinocytosis, and the proliferation-to-migration transition. These findings reveal how cells dynamically balance proliferation and invasion, offering insights into transition states difficult to capture in models of breast cancer metastasis.

The tumor suppressor LKB1 is a serine/threonine protein kinase that is frequently mutated in human lung adenocarcinoma (LUAD). LKB1 regulates a complex signaling network that is known to control cell polarity and metabolism; however, the pathways that mediate the tumor-suppressive activity of LKB1 are incompletely defined. To identify mechanisms of LKB1-mediated growth suppression, we developed a spheroid-based cell culture assay to study LKB1-dependent growth. We then performed genome-wide CRISPR screens in spheroidal culture and found that LKB1 suppresses growth, in part, by activating the PIKFYVE lipid kinase. Finally, we used chemical inhibitors and a pH-sensitive reporter to determine that LKB1 impairs growth by promoting the internalization of wild-type EGFR in a PIKFYVE-dependent manner.