Clonal MLL Fusions with a Signature of Activation Induced Cytidine Deaminase Activity,

Abstract 3444

The MLL gene, encoding a histone methyl transferase, fuses with multiple different partner genes and is a common finding in patients with AML and ALL. This multiplicity of partners is in stark contrast to other fusion oncogenes such as BCR-ABL in CML or PML-RARA in APL, where fusions between the same genes predominate and which has aided the design of targeted therapy. In order to understand the fusion process in more detail, a hot spot for both MLL cleavage and gene fusion adjacent to exon 12 was examined in a series of breast cancer and lymphoma patients, using inverse PCR (IPCR). All patients received chemotherapy containing drugs targeting Topoisomerase II and samples of blood were examined both before and after therapy. Of the fifty patients enrolled in the study a subset was also examined by IPCR, parallel sequencing and custom bioinformatic analysis. The advantage of this approach is that all possible rearrangements may be examined within a single sample. In addition, the technique also records all sequences at the same location that are uninvolved in a translocation. The blood of three out of four patients examined in this way contained no evidence of MLL rearrangements. However in one patient, with a diagnosis of diffuse large B cell lymphoma, a total of thirteen MLL rearrangements were identified that were present prior to therapy. The majority of the fusions were detected for up to 6 months after the end of chemotherapy indicating they were likely of clonal origin. Of the thirteen fusions, five were predicted to provide functional fusion proteins and these included MLL-MLLT3 (AF9), the most common MLL fusion in myeloid leukemia. The remaining fusions predicted to generate functional proteins involved USP46, FER1L5, CCNJL and NKD1. None of these have been previously identified in clinical specimens of MLL linked disease. However, NKD1 is a negative regulator of the WNT pathway that has been linked to the maintenance of the stem cell phenotype in AML.

In order to understand the fusion process in more detail, each fusion breakpoint was examined. All thirteen MLL fusions contained microhomology at the breakpoint, ranging from 1 to 6 bp, indicating NHEJ as the likely pathway generating the fusions. Though the presence of microhomology masked the precise breakpoint, using the 3' edge of microhomology as a reference, eight of the thirteen rearrangements were clustered within a 5 bp tract at the base of a putative stem-loop structure, as we have reported before. Such a restricted distribution of breakpoints found within a clonal population suggests a common mechanism is involved in either cleavage, and/or fusion, at this location. In order to address the mechanism driving these rearrangements, a screen of all sequenced material was undertaken. It was observed that within the residual sequenced material, a selective increase in C>T transitions were noted for two cytosines located within the 5bp breakpoint hotspot. The level of C>T transition was 50–100 fold higher than anywhere else within the sequence generated by IPCR. In addition, the region of C>T transition contained the WRC (A/T, A/G, C) motif characteristic of Activation-Induced Cytidine Deaminase (AID) attack. AID deaminates cytosine leading to a uracil which may be repaired by uracil-DNA glycosylase and the base excision repair (BER) pathway. AID has been implicated in both DNA breaks, via the BER pathway, and a linked increase in translocations. If the BER pathway fails to execute appropriately, a C>T transition may occur as the aberrantly located uracil undergoes replication at the next division. The presence of extensive C>T transitions therefore is indicative of both AID function and defective repair pathway(s). These factors may explain both the extensive number of rearrangements observed in this patient and the legacy C>T transitions from multiple AID attacks.