Abstract 385

Epigenetic modifications influence chromatin accessibility, impacting on cell fate decisions, such as stem cell self-renewal and differentiation, in both normal and leukemic stem cells (LSC). To investigate the putative role of histone demethylases (HDM) in modulating primary hematopoietic stem cell (HSC) fate, an in vivo functional screen was performed, using an RNAi based strategy, involving 25 members of the Jumonji (JmjC) domain protein family.

As a first step, expression profile studies of these gene candidates were undertaken. Transcripts of all these enzymes were detected in isolated HSC populations (frequency 1:2) from fetal liver (n=1) and bone marrow (n=2), except for Hairless. As compared to unsorted bone marrow (BM), stem cells harboured higher expression of Jarid1b (relative-fold enrichment (RQ) of 3.9±1.7), Jmjd2d (RQ3.8±1.9), and Jhdm1b (3.1±1.7). Next, 5shRNA were designed against each of the 25 JmjC containing proteins, and cloned into a retroviral LMP vector encoding GFP to permit tracking of transduced cells in vivo. HSC-enriched CD150+CD48Lincells (∼60 LT-HSC) were infected over 5 days by co-culture with retroviral producer cells in an arrayed 96-well format, with one shRNA per well. Directly after infection, the in vivo reconstituting potential of ¼ of each well was evaluated through duplicate competitive repopulation assays involving the co-transplantation of 1.5 × 105 congenic BM competitor cells into irradiated recipients. The remaining cell fraction served to asses gene transfer by GFP epifluorescence measurements, and RNA isolated from sorted GFP+ cells was used to evaluate gene knockdown levels by Q-RT-PCR analysis. Blood reconstitution was evaluated at an early (4wks) and late time point (16–20wks), tracking the contribution of the donor CD45.1+ transduced (GFP+) cells to recipient hematopoiesis over time. As baseline references, sh-RNA to Luciferase (no effect) and the histone acetyl transferase Myst3 (stem cell loss) were used, as well as Hoxb4 over-expression (stem cell expansion). The primary screen, followed by validation experiments, unravelled one positive (Jhdm1f/Phf8) and two negative (Jarid1b, Hif1an) regulators of HSC activity. The strongest impact was seen with Jarid1b knockdown, and the resulting gain in HSC activity. As a confirmation step, cells were kept in culture for one week, to better contrast an increase in HSC activity, compared to control HSC. After 7 days in vitro, 1/8 equivalents of single well cultures were transplanted into 3 mice, and blood reconstitution levels serially assessed. Cells transduced with sh-RNA against Jarid1b contributed more significantly to host hematopoiesis than sh-RNA Luciferase transduced cells (58±16% vs 26±3% GFP), or Hoxb4 over-expressing cells (37±2% GFP), at comparable gene transfer rates, at the 16 week time point and beyond (3 independent experiments). Long-term HSC frequencies were evaluated from these cultures, and found to be 6–10 fold increased in shJari1d1b-cell cultures. In long-term recipients, differentiation potential of these cells was preserved, as evidenced by CD4+CD8+ thymic cells, B220+ splenic cells and CD11b+ bone marrow cells in the GFP positive contingent. Clonality studies on DNA isolated from these sorted populations confirmed oligoclonality of the stem cell expansion, and HSC pluripotency. There were no cases of leukemic transformation in all of the transplant recipients (n>30). As assessed by Q-RT-PCR, levels of HoxA5, HoxA9, HoxA10 and CxCl5 were increased in day7 sh3Jarib1b-cells (vs ctl), while the levels of the tumor suppressors Cav1, Sash1 and Egr1 were decreased. A more detailed assessment of the HoxA cluster revealed predominant expression of 5' cluster genes in expanding shJarib1b-cells, from HoxA5 to HoxA11, with a concomitant increase in the level of H3K4 tri-methylation, as assessed by ChIP-CHIP.

In conclusion, HDM of the JmjC family can modulate HSC activity, both positively and negatively. These data suggest that the H3K4 demethylase Jarid1b (KDM5b) restrains stem cell self-renewal, acting as a co-repressor, possibly via epigenetic regulation of the HoxA gene cluster, among other target genes. This observation could be further exploited as an HSC expansion strategy.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.