Altered Expression of Hoxa1 Perturbs Haematopoiesis in a Dose-Dependent Manner and Induces Myelodysplasia in Mice

Homeobox A1 (Hoxa1) is a highly conserved transcription factor. Through pre-mRNA splicing within exon 1, wildtype Hoxa1 (WT-Hoxa1) expresses two transcript variants: a full-length Hoxa1 (Hoxa1-FL) containing a homeobox-binding domain, and a truncated homeoboxless Hoxa1 (Hoxa1-T). We have demonstrated that the two spliceforms have opposing proliferative potential, suggesting a dominant negative role exerted by Hoxa1-T.

Using in silico analysis, we found that 50% of patients with myelodysplastic syndromes (MDS) have upregulated expression of HOXA1 in CD34+ cells. Delineating this further by quantitative real-time PCR, we observed that a significant proportion of patients with high-risk MDS also lacked expression of HOXA1-T. To explore how Hoxa1 alters haematopoiesis and whether the magnitude of Hoxa1-FL expressionand/or loss of Hoxa1-T results in a more severe MDS phenotype, we generated conditional knock-in mice with altered Hoxa1 expression.

Two strains of transgenic mice were created: WT-Hoxa1 (which express both Hoxa1-FL and Hoxa1-T) and mutated Hoxa1 (MUT-Hoxa1, which can only express Hoxa1-FL due to oligomutagenesis at the acceptor splice site). They were crossed to tamoxifen-inducible haematopoietic-specific Cre (hSclCre^ERT) mice to express WT-Hoxa1 or MUT-Hoxa1 in a heterozygous (WT-Hoxa1^ki/+, MUT-Hoxa1^ki/+)or homozygous (WT-Hoxa1^ki/ki, MUT-Hoxa1^ki/ki)manner at the Rosa26 locus.

Four weeks after induction of the knock-in, WT-Hoxa1^ki/+ and MUT-Hoxa1^ki/+ mice displayed significant thrombocytopenia (mean platelet count (x10^6/ml): controls=848; WT-Hoxa1^ki/+=646; MUT-Hoxa1^ki/+=566, p<0.01 and p<0.0001 for WT-Hoxa1^ki/+ and MUT-Hoxa1^ki/+ respectively vs controls). By 4 months, heterozygous knock-in mice showed relative pancytopenia compared to controls, with associated myeloid bias. There was no difference in bone marrow cellularity, and mature and immature progenitors were present in similar proportions. However, knock-in mice showed significant reductions in Lin^-cKit⁺Sca1⁺ (LKS⁺) cells (WT-Hoxa1^ki/+: p<0.001, MUT-Hoxa1^ki/+: p<0.0001, vs controls) with a trend to reduction in the proportions of multipotent progenitors (MPPs), accompanied by increased apoptosis.

Homozygous knock-in mice showed more dramatic haematopoietic changes. Four weeks after induction of the knock-in, WT-Hoxa1^ki/ki and MUT-Hoxa1^ki/ki mice were profoundly thrombocytopenic (mean platelet count (x10^6/ml): controls=804; WT-Hoxa1^ki/ki=238; MUT-Hoxa1^ki/ki=50, p<0.0001 for WT-Hoxa1^ki/ki and MUT-Hoxa1^ki/ki vs controls; p<0.05 for WT-Hoxa1^ki/ki vs MUT-Hoxa1^ki/ki). This was accompanied by myeloid bias at 4 months (mean granulocytes (x10^6/ml): controls=0.91; WT-Hoxa1^ki/ki=1.88; MUT-Hoxa1^ki/ki=2.21, p<0.05 and p<0.001 for WT-Hoxa1^ki/ki and MUT-Hoxa1^ki/ki respectively vs controls). The mice also demonstrated reduced LKS⁺ cells (mean LKS⁺ (%): controls=4.04; WT-Hoxa1^ki/ki=0.43; MUT-Hoxa1^ki/ki=0.11, p<0.0001, vs controls) with significant reductions in MPPs (WT-Hoxa1^ki/ki: p<0.0001, MUT-Hoxa1^ki/ki: p<0.05, vs controls). Strikingly, there was skewed lineage commitment towards granulopoiesis at the expense of B-lymphopoiesis in the bone marrow.

These phenotypes were transplantable, with the majority of recipients of bone marrow from WT-Hoxa1^ki/+ and MUT-Hoxa1^ki/+ mice developing MDS at 12 to 15 months (WT-Hoxa1^ki/+: 60%; MUT-Hoxa1^ki/+: 71%). They displayed normocytic or macrocytic anaemia with or without thrombocytopenia. Transplant recipients of knock-in cells showed near complete exhaustion of LKS⁺ cells (mean LKS⁺ (%): controls=2.2; WT-Hoxa1^ki/+=0.2; MUT-Hoxa1^ki/+=0.7, p<0.01, vs controls). Morphologic evidence of trilineage dysplasia was present without an increase in the blast count. Transplantation studies of homozygous knock-in bone marrow cells have been performed, and the mice are being monitored for MDS and leukaemia.

In summary, knock-in mouse models of WT-Hoxa1 and MUT-Hoxa1 recapitulated features of human MDS, with findings reminiscent of MDS with multilineage dysplasia. Bone marrow function was more severely compromised in homozygous mice than in heterozygous mice in keeping with a dose-dependent phenotype. Collectively, these models are clinically tractable, making them valuable pre-clinical platforms for understanding MDS and developing novel therapies.