靶向产生eRNA的超级增强子可将不可成药的调控枢纽转化为癌相关成纤维细胞中的治疗靶点

Cancer-associated fibroblasts (CAFs) are central orchestrators of tumor progression, immune evasion, and extracellular matrix remodeling, yet their extreme heterogeneity has hindered effective therapeutic targeting. Super-enhancers (SEs) represent ideal conceptual targets, as they maintain cell identity and pathological transcriptional programs. However, SEs are broad noncoding regulatory elements and are therefore intrinsically undruggable by conventional small molecules or antibodies. In contrast, enhancer RNAs (eRNAs) transcribed from SEs provide a molecularly addressable output of these regulatory hubs, offering a druggable interface through antisense and transcriptional inhibition strategies. Here, we present a recently developed weight-maximization framework that integrates large-scale transcriptomic and epigenomic data to systematically prioritize therapeutically actionable eRNA-producing SEs in CAFs. Using recently generated total RNA-seq profiles from 250 primary CAF samples, together with ATAC-seq and H3K27ac ChIP-seq from representative CAF subsets (n=5), we define active SEs using ROSE and identify eRNAs based on enhancer-localized, non-spliced transcription. For each eRNA-SE, we compute four independent quantitative dimensions: (i) CAF specificity and prevalence across the cohort, (ii) regulatory causality inferred from enhancer-gene coupling (ABC-informed and correlation-based), (iii) predicted safety based on separation from normal tissue programs, and (iv) therapeutic tractability reflecting sensitivity to eRNA perturbation and suitability for antisense or transcriptional inhibition, supported by preliminary perturbation analyses of prioritized hubs.Rather than heuristically combining these metrics, we learn their weights in a data-driven manner such that the linear combination of eRNA-SE scores maximally separates pathological CAF states (e.g., inflammatory versus matrix-remodeling CAFs) across the cohort. In this optimization, CAF specificity and regulatory causality emerge as dominant discriminative features, while safety and tractability act as corrective terms that favor clinically executable targets among equivalently potent regulators. This process converges on a compact set of high-confidence eRNA-SE hubs, recently identified as key regulators of secreted cytokine, chemokine, and extracellular matrix programs. Together, this framework transforms CAF heterogeneity into a quantitative regulatory landscape and converts previously undruggable SEs into immediately actionable, eRNA-defined therapeutic entry points. These findings establish a generalizable paradigm for uncovering druggable regulatory interfaces in complex tumor microenvironments.