FOXC1 Is Derepressed to Functional Effect in Human Acute Myeloid Leukemia

The identification of genes and cellular pathways that are active in acute myeloid leukemia stem cells (AML LSC) but not normal hematopoietic stem and progenitor cells (HSPC) is essential, both for the understanding of disease biology and also for their evaluation as candidate therapeutic targets. Through in silico analysis, we identified FOXC1 as expressed in approximately 15-30% of cases of human AML in both LSCs and bulk cell populations. FOXC1 is a member of the forkhead box family of transcription factors and has essential roles in mesenchymal differentiation. Reflecting murine knockout phenotypes, in patients with Axenfeld-Rieger syndrome haploinsufficiency of FOXC1due to mutation or deletion causes developmental anterior segment abnormalities of the eye.

By quantitative PCR we confirmed high level expression of FOXC1 in 17% of blast cell samples from patients with AML, and medium level expression of FOXC1 in 24% of samples (cohort size n=29). Critically, FOXC1 expression was not detected in any normal human hematopoietic cell population (including prospectively FACS-purified HSC, MPP, GMP, MEP, as well as defined mature cell populations, all from normal human bone marrow). Thus while FOXC1 is not expressed in normal human hematopoiesis, it is expressed in human leukemic hematopoiesis.

To investigate whether FOXC1 derepression in AML makes a functional contribution to transformation, we initially performed knockdown (KD) experiments in human THP1 AML cells (which exhibit high level FOXC1 expression). FOXC1 KD led to loss of clonogenic potential and induction of morphological and immunophenotypic differentiation and this phenotype could be rescued by forced expression of a KD-resistant version of the gene. By contrast, FOXC1 KD in normal HSPC had no effect. Thus, FOXC1 contributes to the differentiation block in human AML cells.

Forced expression of FOXC1 alone in normal murine HSPC induced a transient enhancement of clonogenic potential and myeloid differentiation block in serial replating assays, and myeloid skewing in in vivotransplantation assays. It did not however result in acute leukemia.

Further quantitative PCR analyses demonstrated that high level FOXC1 expression associated strongly with high level HOXA9 expression in human AML. To determine whether co-expression of HOXA9 and FOXC1 is of functional significance, murine KIT⁺ HSPC were retrovirally infected with either Hoxa9 alone (with empty vector, MTV) or in pairwise combinations with FOXC1 or Meis1 (hereafter referred to as Hoxa9/MTV, Hoxa9/FOXC1 and Hoxa9/Meis1 cells, respectively) and their clonogenic potential was assessed in serial replating assays. As expected, Hoxa9 overexpression strongly augmented the clonogenic potential of BM HSPC, an effect which was enhanced by co-expression of Meis1. Importantly, the co-expression of Hoxa9 and FOXC1 also significantly enhanced the clonogenic potential and myeloid differentiation block of BM HSPC versus cells overexpressing Hoxa9 alone, as determined by immunophenotyping and colony morphology. Thus, FOXC1 and HOXA9 collaborate to enhance clonogenic potential and differentiation block in HSPC.

To determine whether HOXA9 and FOXC1 collaborate to initiate leukemia, Hoxa9/MTV, Hoxa9/FOXC1 and Hoxa9/Meis1 double transduced HSPC were transplanted into irradiated congenic recipients. As expected, recipients of Hoxa9/Meis1 cells developed AML more rapidly than recipients of Hoxa9 cells (median latency 57 days versus 125 days). Strikingly, despite reduced initial engraftment levels, recipients of Hoxa9/FOXC1 cells succumbed to AML significantly earlier than mice receiving Hoxa9cells (median latency 83 days versus 125 days). In every case and in each cohort, autopsy demonstrated splenomegaly and pale BM due to infiltration of donor-derived cells of myeloid immunophenotype, confirming that these animals died from AML. These data demonstrate that FOXC1 functions to accelerate and enhance the development of AML in collaboration with HOXA9.

Our functional studies are consistent with a model whereby lineage-inappropriate derepression of FOXC1 in human AML contributes to oncogenic transformation.