CITED2 Cooperates with Low PU.1 and DNMT3A to Maintain Self-Renewal in Hematopoietic Stem Cells

CITED2 has a conserved role in the maintenance of normal hematopoiesis. We have recently shown that ~70% of acute myeloid leukemia (AML) patients display enhanced CITED2 expression levels. Interfering with CITED2 expression is detrimental for leukemia maintenance in vitro and in vivo, demonstrating that CITED2 is critically important for the survival of leukemic stem cells (LSCs). Ectopic expression of CITED2 in normal CD34⁺ stem and progenitor cells (HSPCs) led to significantly better human engraftment in transplanted NSG mice, consistent with the maintenance of very primitive lin^-CD34⁺CD38^-CD90⁺CD45RA^- HSCs within the bone marrow 28 weeks after transplantation. Although the CITED2-engrafted mice displayed enlarged spleens, blood development appeared normal, as measured through myeloid, B and T cell staining. This indicates that CITED2 as a single hit is not sufficient to transform human CD34⁺ cells.

CITED2 expression frequently coincides with low expression of the myeloid transcription factor PU.1, suggesting that combined effects, rather than single events are important during AML development. To investigate this, we combined lentiviral downregulation of PU.1 with overexpression of CITED2 (PU.1^Low-CITED2^High) and studied hematopoietic development. CITED2 increased the percentage of immature CD34⁺CD38^- cells 5-fold, which was not further increased by the additional downregulation of PU.1. However, functional analysis through limiting dilution LTC-iC assays indicated that combining PU.1 down-, with CITED2 upregulation led to a synergistic 8.5-fold increase in LTC-iC frequency, whereas only changing PU.1 or CITED2 induced a respective 1.4 to 3-fold change in HSC frequency. To more stringently assess self-renewal, we cultured transduced cells for 4 weeks on MS5 cells under myeloid differentiating conditions (G-CSF, IL3 and TPO) and subsequently performed CFC assays. Whereas after 4 weeks all groups displayed similar colony numbers, secondary and tertiary replatings demonstrated that self-renewal could only be maintained for more than 10 weeks when CITED2 upregulation was combined with PU.1 downregulation. This replating capacity of PU.1^Low-CITED2^High cells was limited to CD34⁺CD38^- HSCs, as replating of CD34⁺CD38⁺ progenitor-derived colonies did not yield new CFCs.  

In order to investigate the underlying mechanisms, we performed transcriptome analysis on human HSCPs after knockdown of PU.1, overexpression of CITED2 or the combination of both. PU.1^Low-CITED2^High cells displayed a gene expression pattern different from the PU.1^Low or CITED2^High only cells, suggesting that the two events have synergistic effects. Some genes, like HLX and SF3B1 have been shown to cause or are mutated in AML, demonstrating that the synergistic changes are related to AML. When comparing the differentially regulated genes in the PU.1^Low-CITED2^High cells to the gene expression in the Hemaexplorer database, a similar pattern was observed, when compared between AML and normal cells.

In order to investigate the effects of the PU.1^lowCITED2^high combination on AML development, we resorted to a PU.1-dependent mouse model of AML development. CITED2 expression in BM cells from PU.1^KD/KD mice (in which deletion of an Upstream Regulatory Element leads to an 80% downregulation of PU.1), led to a steady increase of GFP⁺ cells over time as compared to control cells and demonstrated a dramatic expansion of Gr-1⁺Mac-1⁺ cells, a hallmark of AML in these mice. This suggests that CITED2 contributes to a faster progression towards AML upon lowering of PU.1.

To identify if our model corresponds to AMLs with a specific subset of mutations, we clustered publically available AML data (TCGA), based on the gene expression changes in the PU.1^Low-CITED2^High cells. The majority of AMLs clustered together in 2 groups, in which FLT3, p53 and DNMT3A mutations were most prevalent. FLT3 mutations, through its activation of STAT5, are consistent with high CITED2 expression, whereas p53 mutations are consistent with our data indicating that CITED2 loss regulates HSCs in a p53-dependent manner. The presence of DNMT3A mutations suggests that DNA methylation changes collaborate with high CITED2 and low PU.1 during leukemogenesis. This is currently under investigation.

In summary, our data imply that CITED2, low PU.1 and potentially changes in DNA methylation all contribute to maintenance of self-renewal and leukemogenesis.