DNA Microarray Analysis of Stage Progression Mechanism in Adult T-Cell Leukemia.

Abstract

Adult T-cell leukemia (ATL) is an intractable malignancy of peripheral CD4-positive T cells, in which human T-cell leukemia virus type I plays a pathological role. Clinical course of ATL can be subdivided into several phases; indolent “smoldering” and “chronic” stages, and aggressive “acute” stage. Individuals at the former usually undergo a stage progression toward the latter within several years, and ATL cells at the acute stage are highly refractory to current chemotherapeutic reagents. Molecular mechanisms underlying this stage progression are poorly understood yet. DNA microarray enables us to quantitate the mRNA amount for tens of thousands of genes simultaneously, and may provide novel insights into the pathogenesis as well as the stage progression mechanism of ATL. However, since the proportion of ATL cells within mononuclear cells (MNCs) of peripheral blood (PB) varies among different stages, it would be desirable to purify and directly compare ATL cells for an accurate profiling of gene expression. Given the fact that most PB MNCs are occupied by ATL cells in both chronic and acute stages, we purified CD4-positive T cells from PB of ATL individuals at chronic (n = 19) or acute (n = 22) stage. As a normal control, surface marker-matched T cells were purified from PB of healthy volunteers, and stimulated in vitro or not with PHA (n = 3 for each). A total of 47 specimens were thus subject to DNA microarray analysis with Affymetrix HGU133 A&B array sets, measuring the transcriptional level of ~33,000 human genes. Unsupervised hierarchical clustering based on the whole genes indicated that normal CD4-positive T cells, irrespective of PHA stimulation, had a molecular signature distinct from that of CD4-positive ATL cells, while the acute and chronic stages of ATL were not clearly separated from each other. Combination of a t-test (Welch's ANOVA, P<0.001) and an effect size selection (>150 U) identified 46 genes, expression of which contrasted the two clinical stages of ATL. Correspondence analysis of such stage-associated genes also demonstrated visually that both phases of ATL have a distinct molecular signature. Additionally, a very high accuracy was obtained by an artificial neural network in a trial of gene expression-based stage diagnosis. Further, we tried to screen, from our data set, novel “acute stage-specific gene markers”, resulting in the identification of a gene for a cytokine receptor. Intriguingly, the serum concentration of its ligand protein was elevated in ATL patients, especially at the acute stage. This autocrine loop for cell growth would be an interesting candidate for the transforming events which triggers the stage-progression in ATL, and also be a candidate for the targets of novel ATL therapies.