Bladder-preserving trimodality therapy (TMT) incorporating concurrent chemoradiotherapy (CRT) provides a curative-intent alternative to radical cystectomy for muscle-invasive bladder cancer (MIBC), yet its efficacy is frequently limited by intrinsic treatment resistance, the molecular basis of which remains poorly defined. To address this, we performed bulk transcriptomic profiling of pretreatment tumors from 179 patients uniformly treated with bladder-preserving CRT and systematically integrated gene expression data with tumor immune features and clinical outcomes. We identified a ferroptosis-suppressive transcriptional signature (FSS) associated with a distinct resistance-associated tumor state that independently stratified radiographic progression-free survival and overall survival following CRT. FSS-high tumors were characterized by inferior outcomes, enrichment of basal/squamous and immune-excluded phenotypes, and reduced intratumoral immune infiltration, whereas FSS-low tumors preferentially exhibited luminal unstable and immune-inflamed features. Consistent with clinical observations, a ferroptosis-suppressive transcriptional program was recapitulated in CRT-resistant bladder cancer cell line models. Genome-wide CRISPR/Cas9 loss-of-function screening further identified core ferroptosis suppressors as functionally relevant dependencies specifically under irradiation stress, and pharmacologic induction of ferroptosis effectively restored radiosensitivity in otherwise resistant cells. Together, these findings support ferroptosis suppression as a biologically relevant resistance-associated state that mechanistically links tumor-intrinsic transcriptional programs to immune contexture and therapeutic vulnerability and provide a translational framework for improved risk stratification and future treatment refinement in bladder-preserving therapy for MIBC.

4574 Background: The EV-302 trial established enfortumab vedotin plus pembrolizumab (EV/P) as first-line therapy for metastatic urothelial carcinoma (UC). Although EV/P is administered without NECTIN4 testing based on the assumption of uniform NECTIN4 expression, real-world data have identified NECTIN4-low tumors associated with inferior responses to EV, highlighting the need for NECTIN4-based precision therapy. We aimed to define the molecular and immunologic roles of NECTIN4, identify predictive biomarkers of EV/P outcomes, and explore rational therapeutic strategies for NECTIN4-low UCs. Methods: We performed multi-omics analyses of 817 UC cases integrating whole-exome sequencing, RNA-seq, multiplex immunofluorescence, and spatial single-cell profiling (CODEX). We evaluated associations among NECTIN4 status, molecular subtype, tumor microenvironment (TME), survival, and response to immune checkpoint inhibitors. Basal bladder cancer (BC) cell lines and orthotopic MB49 bladder cancer models with NECTIN4 overexpression (OE) or control were used to assess tumor biology and therapeutic response. Results: Clustering by NECTIN4 H-score and molecular subtype revealed that basal/squamous (Ba/Sq) muscle-invasive BCs with low NECTIN4 had significantly worse survival outcomes than NECTIN4-high tumors in both our cohort (P=0.001) and the IMvigor210 trial (P=0.019). In vitro, NECTIN4 OE in basal BC cells suppressed proliferation and migration through downregulation of inflammation-related signaling pathways. CIBERSORTx and EcoTyper analysis of RNA-seq data indicated a more immune-inflamed TME in NECTIN4-high Ba/Sq tumors. Spatial single-cell profiling showed increased CD8+ T cells, particularly CD8+/TIGIT+ cells, in NECTIN4-high Ba/Sq UCs. Moreover, in orthotopic mouse models, NECTIN4-OE tumors exhibited reduced growth and increased CD8+ T-cell infiltration. Clinically, NECTIN4-high Ba/Sq tumors showed higher objective response rate (ORR) to pembrolizumab/avelumab than NECTIN4-low tumors (55.5% vs 0%, P=0.002) in our cohort; in IMvigor210, NECTIN4-high Ba/Sq patients also had superior ORR (38.5% vs 15.6%, P=0.025). Additionally, in orthotopic mouse models, EV/P showed limited efficacy in NECTIN4-low tumors, whereas NECTIN4-OE tumors responded well. Transcriptomic analyses identified histone deacetylases (HDAC) as potential targets in NECTIN4-low tumors; accordingly, HDAC inhibitors demonstrated greater efficacy in NECTIN4-low basal BCs in vitro and vivo. Conclusions: NECTIN4 influences tumor progression and shapes the TME, thereby impacting immunotherapy efficacy. NECTIN4 is a potential predictive biomarker for survival and for outcomes with immunotherapy and EV/P in Ba/Sq UCs. NECTIN4-low Ba/Sq tumors may require alternative strategies, and HDAC inhibition represents a rational therapeutic approach for NECTIN4-low Ba/Sq UC patients.

RNA therapeutics, including antisense oligonucleotides (ASOs), have emerged as a promising class of drugs, with several already approved for clinical use. To date, most approved ASO-based RNA therapies target non-malignant disorders such as neurodegenerative diseases, and only a single therapy in this class has been approved for cancer. Notably, nearly half of existing RNA therapeutics act by modulating RNA splicing. Given the growing evidence implicating aberrant RNA splicing in cancer pathogenesis, the development of ASO-based therapeutics for oncologic indications is expected to accelerate. More than 250 clinical trials have evaluated oligonucleotide agents targeting diverse cancer-associated molecules, with several showing encouraging early results. In this review, we summarize recent advances in understanding cancer biology relevant to ASO-based therapies and highlight ongoing progress in the development of RNA-targeted approaches for cancer treatment.

ABSTRACT Background Tumor‐associated cell surface proteins are frequently proposed as circulating biomarkers for colorectal cancer (CRC) based on their high tumor expression. However, many candidates identified through tissue‐based analyses fail to translate into clinically useful biomarkers. We investigated the translational gap between tissue‐level expression and circulating detectability in CRC, focusing on molecular subtypes defined by caudal‐type homeobox 2 (CDX2) expression. Methods Transcriptomic data from The Cancer Genome Atlas (TCGA) were analyzed to identify cell surface markers differentially expressed between CDX2‐Low and CDX2‐High CRCs. A clinical cohort of right‐sided CRC patients was evaluated using paired tumor tissue and preoperative plasma samples. CDX2 expression was assessed by immunohistochemistry, and circulating concentrations of selected cell surface proteins were quantified using a multiplex ELISA platform. Results Several tumor‐associated cell surface markers exhibited marked CDX2‐dependent differences in tissue expression. However, for most markers, circulating plasma levels did not mirror tissue‐level patterns. CEACAM1 was the sole marker demonstrating concordant CDX2‐dependent differences in both tumor tissue and plasma, with significantly lower levels in CDX2‐Low CRCs. In contrast, CEACAM5 showed a dissociation between tissue expression and circulating levels, despite analytical validation against serum carcinoembryonic antigen (CEA). Conclusions Our findings demonstrate that tumor overexpression of cell surface markers does not necessarily translate into detectable circulating biomarkers. This translational disconnect underscores limitations of biomarker selection strategies based solely on tissue expression and highlights the importance of integrating systemic biology into biomarker development. While some tumor‐associated proteins may lack utility as circulating biomarkers, they may still represent viable therapeutic targets in CRC.

Genes affecting DNA methylation (DNAme) are frequently comutated with splicing factors in acute myeloid leukemia (AML) and associate with more aggressive phenotypes. To elucidate the underlying molecular mechanisms, we deeply profiled wild-type and IDH2R140Q/SRSF2P95 single- or double-mutant AMLs. We find a unique set of mis-spliced genes and differentially methylated CpGs in double mutants. Mis-spliced exons are enriched in CCNG splicing enhancers and in the corresponding DNAme changes. Using a machine learning model, we can accurately predict exon inclusion levels from proximal CpGs. These CpGs are more likely to overlap footprints of RNA binding and chromatin-modifying complexes but not transcription factors. We also report unique gene expression profiles associated with each genotype; however, the differentially expressed genes do not overlap with mis-spliced transcripts. Instead, the mis-spliced genes encode for proteins that interact with the complexes regulating these differentially expressed genes. Thus, aberrant DNAme and splicing lead to the mis-splicing of key regulatory complexes, resulting in the aberrant gene expression profiles characteristic of these AMLs.

Aberrant pre-mRNA splicing is a pervasive feature of cancer and an emerging therapeutic vulnerability. Recurrent mutations in core spliceosomal components, including SF3B1, SRSF2, U2AF1, and ZRSR2, are common in myeloid malignancies, while dysregulated splicing regulators and cis-acting splice-site alterations shape cancer-relevant isoform programs across solid tumors. Together with high transcriptional output, rapid proliferation, and oncogene-driven RNA-processing demand, these alterations can reduce the capacity of cancer cells to tolerate additional splicing perturbation, creating a therapeutic window for pharmacological splicing modulation. Multiple strategies are under investigation, including SF3B complex modulators, splicing kinase inhibitors, RBM39-directed molecular glues, PRMT/arginine-methylation-directed approaches, and selected splice-switching strategies. Early clinical experience indicates that pharmacodynamic modulation of splicing is achievable in patients, yet objective clinical benefit has been inconsistent. This reflects narrow therapeutic windows, incomplete concordance between peripheral-blood pharmacodynamic markers and tumor-tissue splicing perturbation, and the limited predictive value of mutation status alone. Rational combinations with apoptosis-targeted agents, oncogene-directed therapies, DNA-damaging agents, PARP inhibitors, and immunotherapies may offer a more effective route to clinical translation than maximal single-agent splicing inhibition. Continued progress will require more selective splicing-directed modalities, pharmacodynamic biomarkers that measure splicing perturbation in the relevant tumor or blood compartment, longitudinal mapping of genetic and tumor cell-state plasticity-driven resistance, and biomarker-defined combination trials to support expansion from hematologic malignancies into solid tumors.

195 Background: Bladder-preserving chemoradiotherapy (CRT) is a curative alternative to radical cystectomy for muscle-invasive bladder cancer (MIBC), yet durable disease control is limited by intrinsic resistance whose basis remains unclear. Methods: We performed transcriptomic profiling of pretreatment tumors from 179 patients uniformly treated with CRT and integrated gene expression with clinical outcomes, molecular subtype features, and immune contexture. Functional validation used CRT-resistant bladder cancer models and genome-wide CRISPR/Cas9 knockout screening under irradiation (IR). Results: We identified a ferroptosis-suppressive transcriptional signature (FSS) that defines a distinct resistance state and independently predicts radiographic progression-free and overall survival after CRT. High-FSS tumors showed inferior outcomes and were enriched for basal/squamous and immune-excluded phenotypes with reduced immune infiltration. This state was recapitulated in experimentally derived CRT-resistant models. CRISPR screening under IR identified core ferroptosis suppressor genes as essential survival dependencies and radiosensitizers, and pharmacologic ferroptosis induction restored radiosensitivity in resistant cells. Conclusions: A ferroptosis-suppressive state represents a mechanistically defined resistance program linking tumor-intrinsic transcriptional circuitry, immune microenvironment architecture, and therapeutic vulnerability. These findings establish ferroptosis regulation as a clinically actionable axis for risk stratification and therapeutic intensification in bladder-preserving MIBC. Multivariable Cox models evaluating clinicopathologic factors and FSS for radiographic progression-free survival. Multivariable Cox for rPFS HR (95% CI) P value Multivariable Cox for OS HR (95% CI) P value FSS High / Low 2.58 (1.64-4.05) <0.001*** FSS High / Low 2.79 (1.66-4.67) <0.001*** N1 / N0 2.44 (1.46-4.08) 0.001** LVI (+) / (-) 2.18 (1.14-4.16) 0.018* Ba/Sq / Others 2.06 (1.26-3.36) 0.004** With variant / pure UC 1.69 (0.89-3.19) 0.109 LVI (+) / (-) 1.71 (1.00-2.92) 0.048* N1 / N0 1.64 (0.92-2.95) 0.095 With variant / pure UC 1.52 (0.84-2.77) 0.168 Ba/Sq / Others 1.40 (0.82-2.40) 0.214 Grade high / low 1.38 (0.49-3.88) 0.539 cT stage > T2 / 1.16 (0.70-1.92) 0.561 cT stage > T2 / 1.33 (0.85-2.08) 0.219 Grade high / low 1.04 (0.33-3.30) 0.942 CIS (+) / (-) 0.57 (0.30-1.07) 0.081 Age ≥ 75 / <75 0.70 (0.38-1.28) 0.245 Age ≥ 75 / <75 0.55 (0.32-0.96) 0.036* CIS (+) / (-) 0.41 (0.18-0.97) 0.041* rPFS, radiographic progression-free survival; OS, overall survival (OS); HR, Hazard ratios; CI, confidence intervals; FSS, ferroptosis-suppressive signature; LVI, lymphovascular invasion; Ba/Sq, basal/squamous subtype; UC, urothelial carcinoma; CIS, carcinoma in situ. * P < 0.05; ** P < 0.01; *** P < 0.001.

Abstract Alternative pre-mRNA splicing generates extensive transcript diversity, yet the regulatory code that determines how splicing decisions are encoded across the transcriptome remains poorly defined. Splicing outcomes are controlled by combinatorial RNA-binding protein (RBP) interactions and positional context, but how these features are integrated at the transcriptome scale remains unclear. CLIP-based approaches have mapped RBP binding, but directly comparable endogenous maps across multiple RBPs are lacking, limiting inference of global regulatory principles. Here we introduce SCALE-CLIP, an endogenous CLIP framework that integrates CRISPR-Cas9-mediated epitope tagging with long-read-guided read attribution to generate directly comparable RBP binding maps across splicing-regulatory factors. Applied to 23 RBPs, SCALE-CLIP expanded endogenous RBP coverage and, across benchmarked shared factors, increased peak recovery by a median of 12.2-fold relative to ENCODE eCLIP while preserving specificity and reproducibility. We define a transcriptome-wide positional and combinatorial code for alternative splicing, in which binding position is a primary determinant of regulatory outcome: SRSF binding within alternative exons promotes inclusion, whereas binding on flanking exons drives exon skipping. Higher-order SRSF occupancy further tunes this code, buffering exon inclusion when centered on alternative exons but reinforcing repression when distributed across flanking exons. We also show that m 6 A provides an epitranscriptomic layer that locally enhances SRSF binding and is associated with increased exon inclusion. Together, these results establish a multi-layered RNA-binding logic in which binding position, combinatorial RBP architecture and RNA modification jointly shape splicing outcomes, providing a framework for rational interpretation and modulation of alternative splicing.

Histiocytoses are clonal hematopoietic disorders frequently driven by mutations mapping to the BRAF and MEK1 and MEK2 kinases. Currently, however, the developmental origins of histiocytoses in patients are not well understood, and clinically meaningful therapeutic targets outside of BRAF and MEK are undefined. In this study, we uncovered activating mutations in CSF1R and rearrangements in RET and ALK that conferred dramatic responses to selective inhibition of RET (selpercatinib) and crizotinib, respectively, in patients with histiocytosis.

ABSTRACT:Multiple myeloma (MM) is a plasma cell neoplasm that depends on the bone marrow (BM) microenvironment; however, the underlying mechanisms of epigenetic contribution to the pathogenesis of MM are incompletely understood. Here, we delineate epigenetically driven transcriptional and splicing regulation crucial for MM. We recharacterized the transcriptional program induced by interleukin 6 (IL-6)/Janus kinase (JAK)/signal transducer and activator of transcription 3 (STAT3) pathway by integrating chromatin immunoprecipitation sequencing, transcriptomic analyses, and CRISPR knockout screening, identifying the B-cell lineage factors POU2AF1 and ELL2, as crucial IL-6/JAK/STAT3 targets essential for MM cell growth and survival. Genetic depletion of these factors significantly suppressed MM cell growth in vitro and in the xenograft model of IL-6 humanized mice. Mechanistically, POU2AF1 and ELL2 form an autoregulatory loop with IRF4 and establish an MM-distinct transcriptional program representing cellular immaturity. The IL-6/JAK/STAT3 pathway augments this program by upregulating and recruiting these factors to MM signature genes. Furthermore, POU2AF1 and ELL2 are essential in the regulation of IL-6-dependent alternative RNA splicing. Immunocytochemical and proteomic analyses revealed that POU2AF1 colocalizes and facilitates formation of nuclear speckles, where it interacts with trans-acting splicing factors required for MM cell growth. These findings suggest dual roles of POU2AF1 and ELL2 in coordinating transcription and RNA splicing to generate MM-associated mRNA isoforms. Finally, we showed that gapmer antisense oligonucleotides targeting POU2AF1 reduced MM cell growth in the presence of soluble BM stromal cell factors, including IL-6. Our data demonstrate that IL-6-driven B-cell lineage factors are the vulnerability of MM cells and may represent novel therapeutic targets for this incurable tumor.

Myelodysplastic syndromes and acute myeloid leukemia (MDS/AML) with both inversion/translocation of chromosome 3 (inv(3)/t(3;3)) and monosomy 7 (-7) is an extremely poor prognostic entity. To explore potential therapeutic target of MDS/AML harboring both inv(3)/t(3;3) and -7, we performed drug screen using YCU-AML1, a high-risk MDS/AML cell line harboring t(3;3) and -7 (Kunimoto et al. Hemasphere 2020), as well as OCI-AML20, another AML cell line with inv(3) and -7, and found that both YCU-AML1 and OCI-AML20 showed high response to EZH2 inhibitors valemetostat and tazemetostat. Previous study has shown that EVI1, an oncogenic transcription factor highly expressed in MDS/AML with inv(3)/t(3;3), directly binds to EZH2 and thereby recruits PRC2 complex to PTEN locus, leading to epigenetic silencing of PTEN expression and activation of PI3K/AKT/mTOR pathway in leukemia with 3q rearrangement (Yoshimi et al. Blood 2011). Together with the fact that EZH2 locus is on chromosome 7q, we hypothesized that the survival of MDS/AML cells with inv(3)/t(3;3) and -7 may be highly dependent on residual allelic EZH2-mediated silencing of specific targets which drive cell death. We further validated in colony-forming unit and cell growth assays that YCU-AML1 and OCI-AML20 are both highly sensitive to valemetostat and tazemetostat, whereas FKH-1, Kasumi-3 and SKM-1, MDS/AML cell line with -7, 3q rearrangement, and complex karyotype without chromosome 3 and 7 abnormalities respectively, are resistant to these drugs. Apoptosis analysis revealed that valemetostat and tazemetostat efficiently induced apoptosis in YCU-AML1 and OCI-AML20 but not in FKH-1, Kasumi-3 and SKM-1. To seek molecular basis of EZH2 inhibitor-mediated apoptosis induction, we performed CUT&Tag sequence for H3K27me3 using vehicle or valemetostat-treated cells. Strikingly, promoter region of GADD45 γ was the most robustly and significantly decreased annotated peak locus of H3K27me3 in valemetostat-treated OCI-AML20. H3K27me3 peak in GADD45 γ locus was also significantly decreased in valemetostat-treated YCU-AML1 but not in FKH-1 and Kasumi-3. As expected, valemetostat treatment induced increased expression of GADD45 γ in OCI-AML20. Moreover, transcriptomic analysis also demonstrated GADD45 γ as the most robustly upregulated gene in valemetostat-treated YCU-AML1 compared to vehicle-treated cells. GADD45γ is known to be an upstream regulator of stress-activated protein kinases such as p38 and JNK in which signaling pathways activation induce apoptosis upon various cellular stresses. GADD45 γ promoter region possessed putative EVI1 binding site predicted by rVista2.0 software, indicating that EVI1 may directly bind and recruit PRC2 complex to GADD45 γ locus. We further confirmed phosphorylation of Thr180/Tyr182 (T180/Y182) residues of p38α in valemetostat-treated YCU-AML1 and OCI-AML20 but not in SKM-1 and Kasumi-3. p38 is known to phosphorylate Ser33 (S33) and Ser46 residues of TP53, leading to apoptosis in lung cancer cells (Yogosawa et al. Cancer Sci. 2018). In line with this finding, valemetostat treatment induced phosphorylation of Ser33 residue of TP53 in YCU-AML1 and OCI-AML20. Phosphorylations of p38α (T180/Y182) and TP53 (S33) were also noted in tazemetostat-treated YCU-AML1 and OCI-AML20, suggesting that EZH2 inhibition directly activates GADD45γ-p38α-TP53 axis leading to apoptosis preferentially in MDS/AML cells with inv(3)/t(3;3) and -7. As a proof of concept, p38MAPK inhibitor SB203580 restored valemetostat-induced colony growth inhibition as well as apoptosis induction in YCU-AML1 and OCI-AML20. Importantly, valemetostat treatment significantly reduced leukemic burden and improved overall survival in xenotransplant mouse model of YCU-AML1. Moreover, primary MDS/AML patient bone marrow (BM) sample harboring inv(3) and -7 exhibited preferential sensitivity to valemetostat compared to BM samples derived from healthy control or MDS/AML patients with or without -7 in vitro. Taken together, our study unraveled PRC2-mediated inactivation of GADD45γ-p38α-TP53 axis as a molecular basis for evasion of apoptosis in MDS/AML with inv(3)/t(3;3) and -7, which can be preferentially abrogated by EZH2 inhibition leading to efficient induction of apoptosis in this high-risk MDS/AML.

Clonal hematopoiesis (CH) is characterized by the expansion of hematopoietic stem and progenitor cells harboring somatic mutations, which confers an increased risk of hematologic malignancies and cardiovascular disease. Among CH-associated mutations, mutations affecting splicing factors (SFs), including splicing factor 3b subunit 1 (SF3B1), serine/arginine-rich splicing factor 2 (SRSF2), U2 small nuclear RNA auxiliary factor 1 (U2AF1), and zinc finger CCCH-type, RNA binding motif and serine/arginine rich 2 (ZRSR2), play a unique role in promoting clonal expansion and leukemogenesis. In this review, we summarize recent findings on the role of SF mutations in CH progression, their interplay with other mutations (e.g., DNA methyltransferase 3 alpha (DNMT3A), ten-eleven translocation methylcytosine dioxygenase 2 (TET2) and isocitrate dehydrogenase 2 (IDH2)), and their impact on hematopoietic homeostasis. Epidemiological studies have demonstrated that SF-mutant CH exhibits an accelerated clonal expansion compared to other CH clones. Furthermore, murine models suggest that SF mutations alone do not inherently confer a growth advantage for clonal expansion but rather enhance disease phenotypes when co-existing with epigenetic mutations, such as IDH2 and TET2. These findings suggest that SF mutations contribute to CH expansion and malignant transformation through a synergistic interplay with other mutations and external factors such as inflammation. Given the clinical significance of SF mutations, ongoing research is focused on developing targeted therapies that modulate aberrant RNA splicing and prevent CH-driven leukemogenesis. Understanding the mechanisms underlying mutant spliceosome-mediated CH expansion may provide novel insights into early detection, risk stratification, and therapeutic interventions in hematologic malignancies.