The differentiation of hematopoietic stem cells to mature cells is essential for the function of the hematopoietic system. Disturbance of this process can lead to the emergence of Acute Myeloid Leukemia (AML). AML is characterized by an accumulation of immature, malignant blasts, which disturb the function of the “normal” hematopoietic cells. The differentiation to myeloid cells is regulated among others by transcription factors. Growth factor independence 1 (GFI1) is such a hematopoietic transcription factor regulating the differentiation of myeloid cells.

We sought to investigate whether different Gfi1 levels are causative for emergence of AML and how different levels of Gfi1 might influence the prognosis of patients. Using published expression array data, we observed that Gfi1 is expressed at a lower expression level in blast cells and in leukemic stem cells compared to the control non-malignant cells and stem cells.

We then correlated Gfi1 expression level in blast cells of patients from different centers in with the event free survival. In Essen and Dresden (Germany), low expression levels in blast cells were (n=39) associated with an inferior prognosis (EFS 9 months for low expression compared to 42 months; p=0.0095).

We confirmed our observation with an independent cohort from Rotterdam and Nijmegen. Patients with low Gfi1 expression (n=32) had an inferior event free survival (9 months) compared to patients with higher Gfi1 levels (n=144; 17 months; p=0.02). To further investigate how different levels of Gfi1 might influence initiation and progression of leukemia, we used mice expressing Gfi1 at different levels, i.e. Gfi1 deficient mice (Gfi1KO), mice heterozygous for Gfi1 (Gfi1 het) or mice expressing Gfi1 only at 20% of normal Gfi1 expression levels (Gfi1KD). We used different murine AML models to examine the role of Gfi1 in AML development. First we crossed these mice with Nup98HoxD13 mice that recapitulate MDS disease course. We observed that knockdown of Gfi1 (Gfi1KD n=15, P=0.05) and heterozygosity of Gfi1 (Gfi1 het) (n=12) accelerated AML development and were associated with higher blast cell number compared to Gfi1 wt mice (n=16). Interestingly, complete loss of Gfi1 (Gfi1KO, n=16) inhibited leukemia development.

To confirm our findings, we used an independent approach. It has been shown previously that enforced retroviral expression of certain onco-fusion proteins such as MLL-AF9 or AML1-ETO9a or proteins such as MN1 can cause AML in mice. To this end we transduced lineage negative (Lin-) cells from the different mouse strains with retroviruses overexpressing these different proteins and plated the Lin-cells in methycellulose media. Similar to above, transduced Gfi1 KD cells generated more colonies and proliferated at higher levels than wt or Gfi1 KO cells (ranging between 1,5 to 4 fold KD compared to wt, depending on oncofusionprotein or oncogene, p=0.05). We also transplanted these cells into lethally irradiated mice. Again, mice transplanted with MLL-AF9 transduced Gfi1 KD cells (n=6) developed leukemia faster than mice transplanted with transduced wt (n=8; p= 0.05). We hypothesized that if lower expression of Gfi1 promotes leukemia formation, then overexpression of Gfi1 should inhibit leukemia formation.

To this end we transduced Lin neg cells simultaneously with retroviruses overexpressing either MLL-AF9 or AMLETO9a and either Gfi1 or an empty vector. The cells were then plated in methycellulose and cells overexpressing Gfi1 generated fewer colonies (between 3-5 fold less, depending on oncogene, p=0.01) than cells transduced with an empty control vector.

Thus, up to now our data suggests that decreased levels of Gfi1 influence prognosis of AML development and are involved in the pathogenesis of AML. On a molecular level, we found that knock-down and complete loss of Gfi1 leads to deregulation of genes in AML development such as HoxA9. However, whereas Gfi1KD cells show a normal response to apoptotic events, complete loss of Gfi1 is associated with a very high level of spontaneous apoptosis, possibly explaining why knock-down but not complete loss of Gfi1 accelerates AML development. In summary we demonstrate that Gfi1 plays a crucial role in AML development depending on the expression level, which in turn might explain the role of Gfi1 in human AML.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.