Abstract

Notch activation can both inhibit myeloid differentiation and enhance self-renewal of haematopoietic progenitors, both key features of acute myeloid leukaemia (AML) pathogenesis. Activating mutations of Notch-1, frequent in T-cell acute lymphoblastic leukaemia, are rare in AML, although a subgroup of patients over-express the Notch ligand Jagged-1 and Notch-1 itself. Once activated, intracellular Notch binds to and releases the transcription factor CSL (RBPJ-K) from its co-repressor complex, whereupon it recruits the co-activator MAML1 and initiates downstream transcription of various genes including Hes-1. By RT-PCR, we found CSL was expressed in normal T-lymphocytes, neutrophils, CD34+ cells and primary AML samples. Interestingly, most AML patients expressed an alternatively spliced isoform of CSL where the last 78bp of exon 10 had been spliced out at a cryptic GC donor site (termed CSL-TREX TRuncated Exon X) coding for a portion of the beta-trephoil domain. Comparison with the crystal structure suggests this would not alter DNA-binding, but potentially affect complex formation in either its co-repressor or co-activating state. When primary AML blasts were stimulated in vitro to differentiate using IL-3, G-CSF and GM-CSF, CSL-TREX levels markedly decreased, median CSL-TREX % (as a % of total CSL) 36% and 8% respectively (n=6, P.0003), suggesting it was a feature of undifferentiated cells. Although the levels of CSL-TREX were significantly higher in AML blasts (median relative % of CSL-TREX 58%, range 0–100%) than in normal CD34+ cells (P.0004, n=10, median 22%, range 4–41%), they were very low in normal neutrophils (n=6, median 3%, range 0–8%) and absent in T-cells (n=6). In order to assess its biological relevance, we quantified CSL-TREX levels in 236 young adult patients treated on the UK MRC AML trials. There was no association between CSL-TREX level with AML subtype, age or sex, but high CSL-TREX levels were significantly correlated with lower presenting white cell count (P.001, Pearson test for correlation). There were no obvious cut off values and therefore clinical data was analysed for groups divided into quartiles. Median follow-up was 40 months. The overall remission rate was 86%, and was not significantly associated with CSL-TREX levels (P.16). However, patients with higher proportions of CSL-TREX had improved disease free survival (DFS) (highest quartile vs lowest quartile for CSL-TREX, 63% vs 51% at 5 years, HR across all quartiles 0.77 [CI 0.62–0.96], P.02), lower relapse rates (25% vs 40% HR 0.74 [CI 0.58–0.95], P.02) and improved overall survival (OS) (65% Vs 44%, HR 0.77 [CI 0.63–0.94], P.009). When adjusted for WCC, age, performance status and cytogenetics there was a trend for improved OS (P.04). There was no association between levels of CSL-TREX and either FLT3-ITD positivity (P.62) or NPM mutations (P.80). To assess its ability to activate transcription, CSL-TREX and full length (FL) CSL were cloned into an MSCV vector and transiently expressed in U20S cells together with a CSL-reporter (luciferase cloned downstream of 10xCSL-binding sites). CSL-TREX increased luciferase activity with similar potency to full-length CSL, and did so in a Notch-dependent manner, demonstrating it is able to bind both DNA and Notch. Furthermore, CSL-TREX was able to upregulate Hes-1 expression as determined by qPCR. Immunofluorescence in HEK-293T cells showed CSL-TREX was localised to both the nuclear and cytoplasmic compartments, similar to full-length CSL. In summary, we have identified a novel functional alternatively spliced isoform of CSL that constitutes the predominant isoform in many AML patients, where it associated with low presenting WCC and improved OS.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author