Hi-C Allows Genome-Wide Characterization of Enhancer-Driven Oncogene Rearrangements in Formalin-Fixed Lymphoid Cancer Biopsies

Introduction: Genomic rearrangements that juxtapose oncogenes to distal enhancers are recurrent lymphoid cancer drivers. Fluorescence in situ hybridization (FISH), the current diagnostic standard, is low-throughput and cannot identify or characterize uncommon partner loci. Hi-C can both identify genomic rearrangements and characterize topological interactions genome-wide, including across rearrangement breakpoints. We explored the utility of formalin-fixed, paraffin-embedded (FFPE) tissue-compatible Hi-C for identifying enhancer-hijacking rearrangements around oncogenes in lymphoid cancer biopsies.

Methods: We performed Hi-C on 44 FFPE biopsies comprising B- and T-lymphoblastic lymphomas, peripheral T-cell and mature B-cell lymphomas, and plasma cell neoplasms. Our largest comparison groups were GCB DLBCL/HGBL (n=11, enriched for FISH-detected rearrangements), and unselected primary CNS DLBCL (n=6) and plasma cell myeloma/plasmacytomas (n=13). We generated Hi-C contact matrices and called topological features (compartments, TADs and loops) using HiCUP, Juicer Tools and TADLib. We used Hi-C Breakfinder and EagleC to identify genomic rearrangements. Potential enhancer hijacking events were nominated by NeoLoopFinder or as local maxima of interactions between oncogenes near breakpoints (<1Mbp) and partner locus enhancers present in reference H3K27ac datasets.

Results: We identified consistent differences in topological features between systemic GCB-DLBCL/HGBL, primary CNS DLBCL and plasmacytoma samples around genes with regulatory significance during lymphocyte development, including differential compartment states, TAD boundaries, and loops at the MME (CD10), EBF1 and LMO2 loci respectively. Hi-C readily detected genomic rearrangements, including expected gene fusions such as ETV6::RUNX1 and NPM1::ALK1, and unanticipated driver fusions such as IGH::RHOH and DYRK1A::TP63. Hi-C contact maxima across rearrangement breakpoints supported new enhancer-promoter interactions in the two most frequent rearrangements in our dataset, IGH::BCL2 (n=8) and IGH::CCND1 (n=8). Hi-C identified functionally similar variants undetected by clinical FISH assays, IGL::BCL2 (n=1) and IGH::CCND2 (n=1), and rearrangement of MYCN to ARHGAP24 locus enhancers in a mantle cell lymphoma, a lesion with homology to MYC-activating events. Hi-C also identified rearrangements pairing LMO2 with TRA and RAG2 enhancers in our two T-ALL samples, and rearrangements pairing CD274 (PD-L1) and PDCD1LG2 (PD-L2) with the IGH 3’ regulatory region and a PAX5/ZCCH7 enhancer in large B-cell lymphoma samples which could indicate dependency on targetable PD-1 immune checkpoint function.

Hi-C-derived topology aided the interpretation of BCL6 and MYC rearrangements, which are diverse in partner locus and function. Hi-C distinguished between distinct functional classes of BCL6 locus rearrangement, including BCL6 gene-activating rearrangements linking the first intron of BCL6 with a distant promoter or promoter-like immunoglobulin switch region (n=3), a possible BCL6-activating rearrangement with looping over a breakpoint to a BCL11A locus enhancer (n=1), and enhancer-hijacking rearrangements linking the BCL6 distal enhancer complex to MYC (n=2). MYC rearrangements detected by Hi-C (n=11) mostly involved previously described partner loci such as IGH, BCL6, PAX5/ZCCH7, IRAG2, and KYNU, which show active compartment states and new MYC interactions with partner locus enhancers. However, two cases showed MYC rearrangements to previously undescribed enhancer-poor loci, one occurring in the setting of chr8 chromothripsis. These events may represent non-functional “passenger” rearrangements as these cases showed preserved interactions between MYC and native MYC locus enhancers that were absent in most MYC-rearranged biopsies.

Conclusion: Our results show that FFPE-compatible Hi-C detects coding and non-coding oncogene rearrangements in routine lymphoid cancer biopsies and provides topological information supporting the interpretation of enhancer-hijacking events. These findings establish the feasibility of Hi-C as a diagnostic tool to identify and characterize drivers that would otherwise be missed by targeted approaches.