T-ALL Oncogenes Function in the Context of Pre- and Post T-Cell Commitment Programs

Human T-cell development is less well understood than its murine counterpart. We previously described the transcriptional landscape of 11 immature human T-cell developmental stages developed from hematopoietic stem cells on the OP9-DL1 system that match those of ex vivo flow-sorted murine and human thymocyte subsets. We defined a gene signature comprising of 547 genes that distinguishes pre- from post- αβ T-cell commitment stages. Commitment is marked by loss of dim CD44 expression of early CD7⁺CD5⁺CD45^dim cells, before acquiring CD1a surface expression. Unlike uncommitted CD44^dimCD1a^- thymocytes,CD44^-CD1a^- post-committed thymocytes undergo first TCR rearrangements while having lost myeloid-, B- or NK-lineage potential.¹We used this gene signature to study whether genetic or transcription-based subtypes of human T-cell acute lymphoblastic leukemia are connected to pre- or post-commitment T-cell development.

Clustering of human T-ALL using our 547 pre- and post-commitment gene signature distinguished four T-ALL clusters that were highly similar to the four subgroups that we identified before using an unsupervised cluster analysis. This previous analysis distinguished immature/ETP ALL, TLX, Proliferative and TALLMO subtypes, strongly correlated with MEF2C or MEF2C -activating abnormalities, HOXA -activating or TLX3 rearrangements, NKX2-1 or TLX1 rearrangements, and aberrations affecting members of the TAL1 and/or LMO2 gene families, respectively.² Further inspection of the original unsupervised gene expression signature revealed that nearly half of all genes overlapped with genes that were expressed at pre- or post-commitment. This implies strong dependency or collaboration of specific oncogene-driven pathogenic programs and pre- or post-commitment programs. In fact, Immature/ETP-ALL and TLX subtypes are each strongly characterized by expressing a pre-commitment program, whereas the Proliferative and TALLMO subtypes are associated with the post-commitment program. Interestingly, the TLX subgroup that has preferentially been associated with γδ-lineage arrest implies that normal γδ T-cell development does not transit through the αβ T-cell commitment point before branching off to the γδ-lineage. Consistent with this, rearrangements of the BCL11B locus that impact on critical levels of the haploinsufficient BCL11B transcription factor for T-cell commitment function are nearly exclusively associated with these 'pre-commitment' ETP-ALL and TLX subgroups. In contrast, TCR-driven chromosomal aberrations were almost exclusively found in the post-commitment Proliferative and TALLMO subtypes.

We then further explored whether particular networks or signaling routes as captured by specific Hallmark gene sets are activated in pre- or post commitment programs and associated with particular T-ALL subtypes. We identified hallmark gene sets connected to interferon gamma and inflammatory responses that are activated in ETP-ALL, as well as signaling of the KRAS, the IL2-STAT5 and the IL6-JAK-STAT3 pathways including CD44 . For the TLX subgroup, hallmark gene sets included NOTCH signaling, coagulation, angiogenesis and myogenesis. These data are in agreement with activating mutations in IL7R-JAK-STAT5B and RAS pathways that are most prevalent in ETP ALL and TLX subtypes. The TLX subtype is particularly associated with strongly-activating NOTCH mutations in line with the requirement of strong NOTCH signaling for γδ T-cell development. Hallmark gene sets for the Proliferative group include E2F and MYC targets, G2M checkpoint and glycolysis. Surprisingly, we did not identify specific Hallmark gene sets for the TALLMO subtype, implying that it may include patients that have arrested at several distinct stages of αβ development.

In conclusion, our data demonstrate that T-ALL oncogenes can exert their pathogenic effects only in the context of particular T-cell developmental programs.

¹Canté-Barrett, K. et al. Front Immunol 2017; 8:32. doi: 10.3389/fimmu.2017.00032

²Homminga, I. et al. Cancer Cell 2011; 19(4). doi: 10.1016/j.ccr.2011.02.008