CD34+ Cell Selection Is Required to Accurately Measure HOXA9 Levels by Quantitative RT-PCR in Patients with Myelodysplastic Syndrome.

In recent years, it has become increasingly clear that overexpression of specific major HOX genes is linked to transformation of the malignant hematopoietic stem cells underlying human acute myeloid leukemias (AML) and myelodysplastic syndromes (MDS). For example, several recent studies indicate that overexpression of HOXA9 is associated with a poor outcome in AML and MDS. As we initiated our own studies of HOXA9 expression in MDS, we were concerned about the validity of studying whole bone marrow (BM) cell RNA obtained at diagnosis, due to the diversity of cell types and low blast cell numbers in the BM that characterize MDS, and prior studies of G. Savageau and K. Humphries demonstrating that HOXA9 expression in normal BM cells is limited to the CD34+ fraction, which is enriched for hematopoietic stem and progenitor cells. Thus, we initiated this study to investigate whether valid HOXA9 expression levels can be determined from whole BM cell RNAs of MDS patients or whether CD34+ cell separation is required prior to analysis. BM aspirate samples were collected at diagnosis from 9 patients with MDS and subjected to Ficoll-Hypaque light density cell separation, followed by magnetic bead separation to remove CD3+/CD19+ mature lymphoid cells. The CD3−/CD19− fraction was further fractionated into CD34+ and CD34− populations. RNA was extracted by standard methods and two-step quantitative real-time RT-PCR was performed using SYBR Green to measure HOXA9 and control housekeeping gene expression levels. RNAs from peripheral blood CD34+ cells from 8 healthy donors served as controls. Relative HOXA9 expression levels were obtained by normalization to three housekeeping genes (GAPDH, YWHAZ, and RPL4) for each fraction. Our results conclusively show that HOXA9 levels of CD34+ cells from MDS patients are on average three-fold higher than those of CD34− cells from the same patient (p=0.004, Wilcoxon signed rank). Thus, as is the case for normal BM cells, HOXA9 expression in MDS is highest in the stem and progenitor cell fraction. Our results indicate that CD34+ cell selection must be carried out to accurately assess the expression of HOXA9 in the malignant clone of MDS patients. Unselected BM cell measurements include variable fractions of CD34− and CD34+ cells, which can have significant variability, even in samples obtained from the same patient, thus confounding attempts to accurately measure HOXA9 levels and determine the contribution of this oncogene to MDS. Moreover, we detected levels of HOXA9 expression in the CD34+ cells of MDS patients that were on average 4.3 times those of normal controls (p=0.004, Wilcoxon rank sum). Our studies indicate that in order to advance understanding of the molecular pathogenesis of MDS, the expression levels of specific HOX genes and their co-regulators will need to be carefully evaluated in purified CD34+ BM cells from patients with this disease.