Mechanisms of signal transduction and treatment implications in childhood leukaemia

Abstract

Acute lymphoblastic leukaemia (ALL) is the most common form of paediatric malignancy and while the prognosis for patients has dramatically improved, specific molecular subtypes are still associated with inferior outcomes. Comprehensive sequencing studies of large cohorts have identified an expanding number of recurrent drivers associated with B-cell ALL (B-ALL), notably gene fusions, and has resulted in the characterisation of 23 subtypes. Associated risk of relapse has been established for many of these molecular subtypes and is incorporated into clinical risk stratification algorithms. However, molecular features of T-cell ALL (T-ALL) are yet to be incorporated into these algorithms. For a subset of B-ALL patients and the majority of T-ALL patients there is a great need to expand of molecular diagnostic testing to identify recurrent molecular features or potentially actionable lesions.

In this thesis, we aimed to identify novel and rare fusion genes in paediatric ALL using RNA sequencing (RNA-seq) and understand the mechanisms by which these fusions function to drive leukaemia. We performed RNA-seq on 126 ALL patients diagnosed at the Royal Children’s Hospital (RCH) to test the utility of implementing RNA-seq into standard of care diagnostic pipelines. We showed that RNA-seq reliably detected gene fusions identified by clinical diagnostics but has limitation to detect gene fusions in samples with low tumour burden, lowly expressed fusion transcripts (KMT2A rearrangements), and rearrangements involving promotor or enhancer regions (IGH rearrangements). We additionally developed analysis tools to identify IKZF1 deletions and a gene expression classifier to predict Ph-like, ETV6-RUNX1+, and ERG-deleted/DUX4 rearranged ALL, and showed that we could use RNA-seq data to molecularly classify patients that did not express standard lesions. This chapter of the thesis is accompanied by supporting material contained in "Supplementary_table_1.xlsx", which details transcript information for fusion genes identified by the fusion-finding algorithm, JAFFA, in this patient cohort.

Using RNA-seq data, we identified a number of rare and novel tyrosine-kinase activating fusion genes, which are potentially therapeutically targetable with tyrosine kinase inhibitors (TKIs). We cloned and validated the transforming capacity of these fusions in cytokine-dependent cell lines and tested response to therapies. These included a rare CNTRL-FGFR1 fusion that we identified in two patients with biphenotypic leukaemia, and two rare ABL1 fusions, SFPQ-ABL1 and SNX2-ABL1, identified in two patients with B-ALL.

We showed that the structure of the CNTRL-FGFR1 fusion differed from what was previously described, and that the fusion is transcribed from exon one of CNTRL and undergoes alternative splicing. We confirmed that cells expressing the full-length form of CNTRL-FGFR1 were sensitive to TKIs that targeted FGFR1, notably ponatinib. In addition, we designed a Droplet Digital PCR (ddPCR) assay that detects CNTRL-FGFR1, using transcript-specific primers and probes, down to approximately 1 cell in 100,000 (0.001%). This level of sensitivity suggests that this assay could be utilised clinically for minimal residual disease (MRD) monitoring.

Given little is known about the biological function of rare ABL1 fusions, including SFPQ-ABL1 and SNX2-ABL1, we utilised cell viability and proliferation assays to compare the function of these fusions to BCR-ABL1. Cells expressing SFPQ-ABL1 and SNX2-ABL1 have reduced proliferative capacity, reduced sensitivity to TKIs targeting ABL1, and different subcellular localisation, compared to BCR-ABL1. We performed phosphoproteomics to understand signalling differences between SFPQ-ABL1 and BCR-ABL1. We showed that proteins involved in transcriptional regulation and spliceosome were enriched in SFPQ-ABL1, suggesting that this fusion may in part function by altering gene transcription and RNA-splicing.

In this thesis we show that RNA-seq is a valuable tool for the identification of molecular drivers of ALL including gene fusions and IKZF1 deletions. Our data suggests that RNA-seq is most valuable for the molecular classification of patients that do not harbour standard lesions. Our functional studies show that the 5’ fusion partner of tyrosine kinase fusion genes may play a role beyond facilitating dimerisation and activation of the kinase and may mediate subcellular localisation and contribute to signal transduction. In all, the fusion identification and biology pipeline we have developed has the potential to inform clinical management of patients, to both tailor treatment and design molecular assays to track MRD.