Transcriptional reprogramming through induced proximity has emerged as a powerful strategy for modulating the expression of oncogenic and tumor-suppressive genes. Inspired by transcriptional reprogramming approaches such as transcriptional/epigenetic chemical inducers of proximity (TCIPs) that link BCL6 inhibitors to transcriptional regulators, we sought to develop covalent ligands that rewire BCL6 proximity to selectively suppress MYC transcriptional output while derepressing BCL6 target loci. Through a chemistry-driven and chemoproteomics-enabled design strategy, we generated a panel of BCL6-based electrophile-bearing hybrid ligands and identified a nondegradative molecular glue, ZD-1-186, that potently suppresses MYC and robustly induces CDKN1A (p21) in diffuse large B-cell lymphoma cells. ZD-1-186 downregulates MYC more effectively than BCL6 inhibitors or degraders, while strongly derepressing canonical BCL6 targets, including p21. Through BCL6 pulldown proteomics, ZD-1-186 induced a selective recruitment of the noncanonical BAF complex subunit BRD9 to BCL6 and covalently modified BRD9 at C288. Pharmacologic inhibition or genetic knockdown of BRD9 attenuated ZD-1-186-mediated MYC suppression and blunted p21 induction. Transcriptomic profiling of ZD-1-186 showed simultaneous derepression of BCL6-repressive loci and suppression of MYC transcriptional programs. These findings demonstrated that ZD-1-186 acted as a transcriptional rewiring glue, recruiting BRD9 to BCL6-repressive loci to activate tumor-suppressive transcription, while also potentially redirecting BCL6 to BRD9-bound oncogenic loci. Overall, our work provides a blueprint for the rational discovery and design of electrophile-enabled, nondegradative molecular glues for targeted transcriptional rewiring.
Magnification-Aware Distillation (MAD): A Self-Supervised Framework for Unified Representation Learning in Gigapixel Whole-Slide Images
arXiv:2512.14796v1 Announce Type: cross Abstract: Whole-slide images (WSIs) contain tissue information distributed across multiple magnification levels, yet most self-supervised methods treat these scales as independent


