T-cell leukemogenesis involves ectopic activation of proto-oncogenes which often regulate physiological development. In this study, comparison of gene expression profiles of NK/T-cell lines identified the apparently aberrant expression of developmental homeobox genes, HOXA9 and HOXA10, a bHLH gene, ID2, and RUNX2 in T-ALL cell line LOUCY. These T-ALL oncogenes are physiologically expressed in hematopoietic progenitor cells and regulate the differentiation lineage of the closely related NK-cells [8
]. Due to the high regulatory impact of these genes we speculate that they contribute significantly to the separate positioning of LOUCY after cluster analysis of T-ALL cell lines.
We showed that HOXA9 and HOXA10 activated ID2 in T-cells (and HELA) while RUNX2 was not regulated by these homeodomain proteins. Coexpression of HOXA9/HOXA10 and ID2 was detected in cell lines expressing MLL-fusion proteins which have been shown to activate particular HOXA genes, including HOXA9 and HOXA10 [16
]. Thus, our results highlight ID2 as an indirect target of MLL fusion proteins. However, phosphorylation of HOXA9 and HOXA10 proteins by protein kinase C reduces their DNA-binding activity [52
]. This modification may explain the limited correlation between endogenous expression levels of HOXA10 and ID2 as observed in JURKAT cells.
ID2 belongs to the basic helix-loop-helix family of transcription factors. Physiologically, ID2 is expressed in progenitors of NK-cells and contributes essentially to the differentiation of this lymphocytic lineage [8
]. Together with family members TAL1 and LYL1, ID2 confers the potential to suppress the activity of bHLH proteins TCF3 and TCF12. Their activity in turn has fundamental importance for T-cell development as demonstrated by differentiation arrest upon suppression [5
]. Furthermore, TCF3 activates pro-apoptotic BIM, enhancing sensitivity to cell death during selection processes [38
]. We showed that the expression levels of BIM were reduced in HOXA9/HOXA10 and ID2 overexpressing cells, presumably mediated by ID2. Accordingly, regulation of BIM by ID2 has been also described in colorectal carcinoma [54
]. Finally, the NOTCH-pathway is frequently activated via various mutations in T-ALL, and ID proteins release the negative autoregulation of the NOTCH-target HES1 [55
]. Therefore, deregulation of ID2 via HOXA9/HOXA10 may be a crucial step in thymocytic transformation by perturbing T-cell differentiation and apoptosis.
HOXA genes are physiologically expressed in lymphocytic progenitor cells and become downregulated during T-cell development [14
]. During differentiation of NK-cells HOXA10 remains active, indicating a crucial role in this related lymphocytic lineage [33
]. HOXA5, HOXA9 and HOXA10 have been identified as oncogenic targets in T-ALL samples, supporting their dominant role in lymphocyte development [30
]. Clustered homeobox genes, including HOXA genes, are regulated by competing chromatin complexes which comprise activating HMTs like MLL or SET1 proteins and repressing polycomb-complexes [19
]. By expression profiling we identified two genes coding for components of PRC2, namely EZH2 and HOP, and demonstrated their involvement in regulation of HOXA9/HOXA10 expression in NK/T-cells.
Inhibition of EZH2 by DZNep or siRNA consistently increased expression of HOXA10 while its activation by rapamycin decreased HOXA10 levels. The latter result may influence the treatment of HOXA-positive T-ALL patients because of frequent aberrant activity of the AKT-pathway in this disease [58
]. T-ALL cell line LOUCY lacks EZH2 expression and, coherently, showed no effects after rapamycin or DZNep treatments. These data suggested that LOUCY cells have no or only restricted function of PRC2 due to absence of EZH2. Accordingly, reduction of PRC2 component EED in embryonic stem cells results in loss of repressor activity as well and subsequently in increased expression of HOXA genes [40
]. In mammalian cells knockdown of PRC2 components EZH2 or PHF1 led to upregulated HOXA gene expression [59
]. Furthermore, in Drosophila
mutations in each of ten tested different PRC components, including the EZH2-homologue, results in ectopic HOX-gene expression, supporting the necessity of a complete complex for its gene repression activity [60
By examining the cause of EZH2 silencing in LOUCY we identified reduced TFDP1 expression. Increased expression of TFDP1 in LOUCY via lentiviral transduction of an expression construct restored EZH2 expression. TFDP1 is a cofactor for E2F transcription factors, acting as heterodimers, which have been shown to regulate EZH2 transcriptional expression [50
]. Therefore, TFDP1 may together with T-cell expressed E2F1 physiologically contribute to EZH2 gene activity. E2F1 regulates proliferation and apoptosis during T-cell development. Reduced E2F1 activity enhances survival of both thymocytes and T-ALL cells [29
]. Therefore, we suggest that low expression levels of TFDP1 may additionally restrict E2F1-mediated apoptosis in T-ALL cells. Interestingly, TFDP1 has also been shown to interact via E2F6 with EPC1 [62
], demonstrating the lack of another component of PRC2 in this T-ALL cell line. Recently, deep sequencing analysis discovered mutated and consequently inactive EZH2, identifying EZH2 as a tumor suppressor gene in difffuse large B-cell lymphoma [63
HOP has also been shown to interact with PRC2 component EPC1 [47
]. Forced expression of HOP in JURKAT cells and reduced expression in NK-cell lines decreased and increased expression of HOXA10, respectively, demonstrating an inhibitory effect in T-cells and NK-cells. Therefore, these results indicated that HOP mediated repression of HOXA10 by modulation of PRC2 activity. Accordingly, NK-cell line NK-92 expressed higher levels of HOP and lower levels of HOXA9/HOXA10 in comparison to YT cells, supporting a repressive role of HOP. This mode of HOX gene regulation seems to be maintained during the development of NK-cells and deactivated in that of T-cells because HSCs were HOP positive. Surprisingly, HOP-overexpressing T-cells were significantly more sensitive to DZNep-treatment. This observation confirmed an effect of HOP on PRC2 and may suggest that HOP mediated conformational changes within this complex, resulting in enhanced repressive activity concomitantly facilitated by an increased accessibility of DZNep to its binding site.
Activating chromatin complexes contain HMT activity mediated by closely related MLL/SET1 proteins [19
]. Chromosomal translocations of MLL are frequent aberrations in both, AML and ALL, including T-ALL. Their most prominent oncogenic targets are HOXA genes [16
]. Furthermore, Van Vlierberghe and colleagues identified recently a microdeletion in LOUCY, generating the fusion gene SET-NUP214 [21
]. This fusion protein in addition to CALM-AF10, which has also been detected in T-ALL, interacts with HMT DOTL1, performing aberrant H3-methylation and concomitant HOXA deregulation [21
]. Another fusion protein in T-ALL, SETBP1-NUP98, comprises the SET-interaction partner SETBP1 and, therefore, may also recruit HMT activity [65
]. However, downstream effects of SETBP1-NUP98 have not yet been analyzed. Finally, several direct interactions have been described between activating MLL/SET1, SET and repressing PRC complexes, respectively, representing a large and complex network for regulation of HOXA in addition to other developmental genes [66
]. Those regulative interactions together with results obtained in this study are summarized in Fig. . Taken together, certain components of gene regulating chromatin complexes, including MLL, SET, SETBP1 and EZH2, are oncogenic targets, representing a key mechanism of leukemic HOXA-deregulation in T-ALL. In this context, LOUCY cells may serve as a useful model system to investigate the function of PRC2 and its component EZH2 in T-ALL.
Network of HOXA-regulation. The diagram depicts major chromatin complexes, including PRC1, PRC2, MLL and SET with selected components or mutated fusion proteins which regulate the transcription of HOXA genes.