In this report, we demonstrate the direct involvement of HoxA9 and its cofactors Meis1, Pbx1 and Pbx2 in the transcriptional regulation of c-myb in HPCs and HoxA9/Meis1-transformed myeloblastic cellular models. In the HPC cell line, we show that Pbx1 acts in synergy with HoxA9 and Meis1 to induce c-myb expression, although Pbx2 displays a repressive activity. In contrast, in the leukemic model, Pbx2 acts as an activator of c-myb transcription and seems to substitute for Pbx1. In those cells, Pbx2 colocates with HoxA9 and Meis1 upstream of c-myb exon 2 in a region that appears to display promoter properties.
Noticeably, following previous work on the CCRF-CEM lymphoid leukemia cell line29
and the cloning of a c-myb cDNA, which 5' suggested alternative transcription initiation within c-myb intron 1, the human hortologue region was described by Jacobs and colleague as an alternative promoter relating to several transcription start sites.28
Specific cooperative interactions between Pbx and Hox proteins have been shown to modulate the properties of Hox proteins and influence their DNA-binding specificities.30
Amongst the Hox proteins, HoxA9 and HoxA10 paralogs exhibit the capacity to interact with either Meis or Pbx proteins to form DNA-binding complexes with respective affinities for HoxA/TALE consensus sequences.19
Meis1 can also bind both HoxA9 and Pbx, which allows for the formation of HoxA9/Meis1 or Meis1/Pbx dimeric complexes31
and HoxA9/Meis1/Pbx trimeric complexes.32
We have observed that HoxA9, Meis1 and Pbx bind the c-myb
locus on different HoxA/TALE consensus sequences in HPC and leukemic cell lines. The differences of binding sites, in particular in the intronic HS-D region, could reflect the formation of different functional complexes in leukemic cells that express high levels of HoxA9 and Meis1. Interestingly, our result contrasts with the recent work of Huang et al.
which did not identify these binding, suggesting that the recruitment of HoxA9 and Meis1 may be subject to external signals and culture conditions.
Imbalances in the expression of HoxA9 and Meis1 are observed in over 50% of human AML and nearly all ALL-containing translocations of the MLL gene.34, 35, 36, 37
In addition, both proteins cooperate in inducing AML-like leukemia in mice.23
Similarly, functional studies have highlighted the collaborative role of Pbx, whose synergistic function with HoxA9 was shown to be crucial for myeloid transformation.38
However, the different Pbx family members appear to participate in leukemogenesis to different extents: Pbx1 exhibits no synergistic effect with HoxA9 and Meis1 in transforming primary bone marrow cells and inducing leukemia in a murine transplantable AML model.23
A possible explanation for this can be found in the work of Shen and colleagues who showed that HoxA9 and Meis1 preferentially form complexes with Pbx2 in a myeloid environment.32
Accordingly, Pbx2 or Pbx3 were found to be crucial to MLL transformation, although Pbx1, which is expressed to a lesser extent in myeloid cells, proved to be dispensable.39
It is therefore perhaps not surprising that we find that Pbx2, but not Pbx1, has a profound effect on the regulation of c-myb
expression in myeloblastic cells. Adding to the fact that c-myb
expression was reported to be a crucial downstream event of MLL transformation,16
our findings add to the growing body of evidence linking Pbx2 activity to leukemogenesis. However, a key role of Pbx2 in normal or deregulated hematopoiesis contrasts with the apparent lack of phenotype observed in Pbx2
gene knockout mice,40
although this lack of effect could be due to redundancy with the other member of the Pbx family. Further studies involving Pbx2 conditional knockout mice models would unveil the detailed role of Pbx2 in hematopoiesis.
Remarkably, although the repressive activity of Pbx2 on c-myb
expression in HPCs is associated with its recruitment to the promoter and the mid-intronic region, its activating function in leukemic myeloid cells correlates with its binding to the HS-D regulatory module, together with HoxA9, Meis1 and RNA PolII. Contrasting with its persistent association with the c-myb
promoter in normal differentiated myeloid cells derived from the HPC7 cells, the switch in RNA PolII occupancy to the HS-D region seems to reflect a leukemia-related phenomenon. Such a feature could be a direct consequence of the inappropriate expression of HoxA9 and Meis1 in the committed cells, which is likely to create an environment that favors Pbx2 and RNA PolII recruitment leading to transcriptional activation of the c-myb
gene despite an attenuation mechanism. In this respect, our findings could be a first hint regarding the molecular mechanisms underlying the deregulation of the c-myb
gene in certain types of myeloid leukemia. In the transformed cells, HoxA9, Meis1, Pbx and RNA PolII binding to the HS-D domain correlate with the repositioning of promoter-related epigenetic modifications immediately upstream of c-myb
exon 2. Incidentally, this highly conserved region corresponds to a second c-myb
promoter and cluster of transcription start sites that has been characterized in human cells.28
Positioned downstream of the c-myb
elongation attenuation domain,5
this alternative promoter could drive c-myb
expression in the FMH9 cells although overriding the normal downregulatory mechanism associated with the differentiation process. Although the transformed cells display a promyeloblast phenotype, it is probable that earlier progenitors constituted the population initially targeted by HoxA9 transforming activity. Using the HPC7 as a model of nontransformed early progenitors, we propose that, within the transformation process, HoxA9 overexpression participates in counteracting c-Myb downregulation associated with further commitment to the myeloid lineage rather than activating c-myb expression as reported by others in different systems.16, 33, 41
However, the precise mechanisms underlying the abnormal expression of c-myb
observed in several types of leukemia could have different origins. Indeed, considering the central role of c-myb
during the different stages of the hematopoiesis, it would not be surprising to find that various transcriptional and posttranscriptional mechanisms are involved in the control of its expression. Further studies using in vitro
and in vivo
leukemic models, as well as other models of hematopoietic diseases would be required to address these questions.