Transcriptomic response to differentiation induction

Abstract

Background: Microarrays used for gene expression studies yield large amounts of data. The processing of such data typically leads to lists of differentially-regulated genes. A common terminal data analysis step is to map pathways of potentially interrelated genes.

Methods: We applied a transcriptomics analysis tool to elucidate the underlying pathways of leukocyte maturation at the genomic level in an established cellular model of leukemia by examining time-course data in two subclones of U-937 cells. Leukemias such as Acute Promyelocytic Leukemia (APL) are characterized by a block in the hematopoietic stem cell maturation program at a point when expansion of clones which should be destined to mature into terminally-differentiated effector cells get locked into endless proliferation with few cells reaching maturation. Treatment with retinoic acid, depending on the precise genomic abnormality, often releases the responsible promyelocytes from this blockade but clinically can yield adverse sequellae in terms of potentially lethal side effects, referred to as retinoic acid syndrome.

Results: Briefly, the list of genes for temporal patterns of expression was pasted into the ABCC GRID Promoter TFSite Comparison Page website tool and the outputs for each pattern were examined for possible coordinated regulation by shared regelems (regulatory elements). We found it informative to use this novel web tool for identifying, on a genomic scale, genes regulated by drug treatment.

Conclusion: Improvement is needed in understanding the nature of the mutations responsible for controlling the maturation process and how these genes regulate downstream effects if there is to be better targeting of chemical interventions. Expanded implementation of the techniques and results reported here may better direct future efforts to improve treatment for diseases not restricted to APL.

Figure 1

"Up-Up-Up" Temporal Pattern of Microarray Gene Expression in U-937 Cells. The figure demonstrates the clustering of genes sharing the pattern of steadily increasing expression level for RNA following treatment with ATRA as measured by microarray. Two subclones of U937 cells are shown, "Minus" and "Plus", over the time course of 0, 6, 24, and 48 hrs. [For all eight clusters for both subclones see Additional File 2.]

Figure 2

"Hourglass" Diagram. Differences in gene expression microarray patterns between Plus and Minus subclones (right and left of both spines, respectively) of the U-937 monoblastoid cell line treated with all-trans retinoic acid over time. Eight expression patterns (four on the left panel and four on the right panel) were defined (up-up-up [UUU], up-down-up [UDU], etc.) for genes of interest related to differentiation: HOX genes, nuclear receptor genes, and genes associated with differentiation in neutrophils. Visualization of the connection differences on either side of the "spine" gene list suggests differential regulation of gene expression in the two subclones for each of the gene clusters.

Figure 3

Plots of the relationship between the number of genes in the clusters for this study with the concordance of regelems. Regression analyses demonstrate the correlation between the number of genes and the total number of regulatory sequences detected and that the greater the number of genes, the fewer sequences are shared. In other words, the more genes clustered for regelem analysis, the more total number of regulatory elements are involved and the fewer regulatory elements are likely shared in common with increasing numbers of genes. The points represent the 14 gene clusters of the differentiation data set described in the Results.