Multi-Omics Integration to Elucidate the Cis-Regulatory Network Re-Wiring upon NSD2 Degradation in t(4;14) Multiple Myeloma

Introduction. The t(4;14) translocation activates nuclear receptor binding SET domain protein 2 (NSD2) to define a molecularly distinct subset of multiple myeloma patients with especially unfavorable prognosis. Despite the clear delineation of t(4;14) as a segment with NSD2 activation as the driver, an incomplete understanding of the mechanistic cascade towards myelomagenesis in these patients has been a major hurdle. We thus developed a tool CRL4^CRBN E3 ubiquitin ligase-dependent ligand-directed degrader (LDD) that rapidly catalyzes proximity-induced degradation of NSD2 to facilitate the multi-omics characterization of NSD2’s role in myeloma, uncovering direct impacts on both the local activity and long-range organization of cis-regulatory elements governing critical programs.

Methods. t(4;14) and non-t(4;14) multiple myeloma preclinical models, at baseline and with NSD2-LDD treatment, were profiled using: CUT&RUN/ChIP-Seq (CTCF, H3K27ac/me3, H3K36me2/3, H3K4me1/3), Hi-C, ATAC-seq, RNA-seq, histone mass spectrometry. Soft agar colony formation, fibronectin cell adhesion, and FBS chemotaxis migration assays were used to validate the phenotypic consequences in vitro, whereas cell-line derived xenograft models were used to assess tumor control in vivo. Additionally, a co-culture system of myeloma cells with bone marrow stromal cells was used to evaluate both paracrine signaling and cell adhesion-mediated drug resistance.

Results. NSD2 degradation-dependent therapeutic benefit was demonstrated across multiple in vivo models: up to 73% tumor volume reduction in mice bearing subcutaneous KMS34 tumors as well as intrafemoral engraftment of luciferase-labelled KMS34 cells that not only saw significant survival benefit (p = 0.034) but also absence of extra-nodal growth – contrasting CNS-resident and contra-lateral metastases in the vehicle group. Seeing that the in vivo efficacy of NSD2 degradation coincided with deep NSD2 degradation (-94%) as well as loss of its direct catalytic product H3K36me2 (-74%), we turned to in vitro models to elucidate NSD2’s trans-omics mechanistic cascade.

Consistent with H3K36me2 loss (9x) and H3K27me3 gain (3x) being the most significant changes among the 40 histone post-translational modifications detected by mass spectrometry, cells treated with NSD2-LDD exhibit global restoral of transcriptional control across multiple modalities with: (1) 3x excess number of down-regulated genes based on RNA-seq, (2) 6x more sites with reduced accessibility based on ATAC-seq, (3) 2x surplus of genomic regions with significantly more heterochromatic compartment score based on Hi-C. Besides changes to local activity, comparison of genome-wide Hi-C contact probability further revealed a specific gain of long-range interactions around the scale of 2-megabase upon NSD2 degradation; accordingly, we observed an 20% increase in the strength of heterochromatic long-range interaction. We then integrated 7 different epigenetic signals to categorize 110k candidate cis-regulatory regions into 6 baseline chromatin states such as boundaries, promoters, enhancers, polycomb regions, etc. to determine the predictors of genes responsive to NSD2 modulation. Ultimately, we pinpointed H3K36me2 at enhancers coincident with H3K27me3 repressive looping at promoters to primarily underlie transcriptional down- and up-regulation upon treatment, respectively influencing tumor-extrinsic (e.g., cell-cell junction: CD44, JAM2, TJP1) and intrinsic (e.g., B cell markers: LAIR1, POU2AF1, IL12RB1) pathways. These effects were subsequently reflected in not only dose-dependent reduction in growth, adhesion, and migration, but also elevated sensitivity to dexamethasone (monoculture) and bortezomib (HS-5 co-culture) following NSD2-LDD pre-treatment.

Conclusions. We leverage targeted NSD2 degradation to clarify the epigenetic signaling circuitry driving t(4;14) multiple myeloma, implicating repressive polycomb regions and activating enhancers in balancing normal programming versus disease-associated processes. With evidence across in vitro and in vivo models, we affirm the effectiveness of targeted NSD2 degradation against a t(4;14)-specific chromatin-based vulnerability not addressed by existing approaches and rationalizes the combination of NSD2 degraders with standard of care regimens for increased efficacy.