Gene Expression Changes in Bone Marrow Cells from Diamond-Blackfan Anemia Patients.

Abstract

Diamond-Blackfan anemia is usually characterized by anemia, absence or insufficiency of erythroid precursors in bone marrow, growth retardation and diverse congenital anomalies that are present in approximately half of patients, indicating that DBA is a broad disorder of development. Mutations of RPS19 are found in approximately 25% of DBA patients. There is good evidence for a second DBA gene, located on chromosome 8, and further genetic heterogeneity of the disease is likely. The aim of this study is to determine the most disturbed molecular pathways in DBA patients, based on gene expression changes in bone marrow cells. Knowing these pathways will possibly enable us to decipher the pathogenic mechanisms of DBA and find other genes involved in the disease. Bone marrow cells from 6 normal individuals and 3 DBA patients with RPS19 mutations, currently in remission, were FACS separated into 3 populations: primitive (P), erythroid (E) and myeloid (M) containing CD34+CD71-CD45RA-, CD34+CD71hiCD45RA- and CD34+CD71lowCD45RA+ cells, respectively. The purity of each sorted population was >97%. As a control for cell sorting accuracy, methylcellulose assay demonstrated that the P populations were highly enriched in primitive BFU-E and CFU-GEMM colonies, the E populations gave rise to BFU-E and CFU-E colonies in more than 90% of the CFCs, while more than 99% colonies from M populations were CFU-G, CFU-M and CFU-GM. RNA targets from these three FACS sorted cellular subsets was hybridized to Affymetrix HG-U133A chips (>22,000 probe sets). The data from all 27 samples were analyzed by hierarchical clustering and Principal Component Analysis, and each cell population was also studied separately. All pairwise comparisons among 27 datasets showed correlations with r=0.86–0.99. Hierarchical clustering identified three major specimen clusters, perfectly overlapping with the three different cell populations under study. Principal Component 1 and 2 separated the three studied subgroups P, E, and M. In each cell population analysis, 3 patient samples were compared to 6 control samples using 1)Significance Analysis of Microarrays with fold change 2 or greater and false discovery rate 1%, 2)Geometric Fold Change analysis and 3)Filter on Fold Change GeneSpring application (arithmetic analysis). All fold change analyses revealed the most significantly changed transcripts in patients vs. control individuals in E (45 upregulated and 184 downregulated) and P populations. The most changed genes in E subgroup were apoptosis related genes, namely TNFRSF10B and TNFRSF6 (CD95/Fas), upregulated in patients 10 and 3 fold, respectively. Other most changed genes were cancer related and genes involved in developmental processes and nucleic acid binding. Additionally, several ribosomal protein genes, namely RPL10L, RPL28, RPL36, RPL13, RPL27a and RPL37a were significantly underexpressed in P and E populations of DBA patients. All three analyses showed that RPL10L, RPL28 and RPL36 are underexpressed in the M population. This finding indicates that ribosomal protein genes are closely co-regulated and that RPS19 protein abnormalities result in downregulation of the additional ribosomal protein genes in both erythroid and nonerythroid cells in DBA patients.