Differential Gene Expression in CLL Cells from Bone Marrow and Peripheral Blood Suggests a Role of Bone Marrow Stroma in Leukemic Cell Proliferation.

Chronic lymphocytic leukemia (CLL) has been described as an accumulative disease of mature lymphocytes. In the peripheral blood (PB) CLL cells are in Go/G1-phase, and gene expression is most consistent with resting B-cells. However, a recent study demonstrated that a surprisingly large fraction of CLL cells are constantly turned over and that proliferation contributes significantly to the expansion of the clone. The sites where proliferation occurs are not well defined but likely include the bone marrow (BM) and/or lymphoid organs. Likewise, the signals governing proliferation of the leukemic cells are ill defined but there appears to be a role for CLL extrinsic factors including stroma cell interactions or antigen stimulation. We hypothesized that gene expression in BM-CLL cells differs from that in PB-CLL cells. Here we report our analysis of 8 pairs of matched CLL samples derived from 7 individuals in whom we simultaneously obtained PB and BM. All patients were untreated, 3 female, 4 male; in 2 the CLL cells expressed unmutated IgVH genes, and 4 were ZAP70 positive. After informed consent, we obtained a dedicated research aspirate. CD19 selection resulted in >98% purity in all samples. PB-CLL cells express high levels of CXCR4, the receptor for SDF-1. As SDF-1 is expressed by BM stroma cells and CLL cells internalize CXCR4 after binding SDF-1 we measured CXCR4 expression by flow cytometry as an indicator of recent contact between leukemic cells and stroma. In PB, a mean of 63% (range 29–90%) of CLL cells expressed CXCR4 above isotype as compared to 30% (3–62%) of cells from the BM (p=0.007). Conversely, we followed CD69 expression as a marker of activation. In 5 of the 8 pairs CD69 was more highly expressed in BM than in PB derived CD19+ cells (p=0.005) indicating activation of the leukemic cells in the BM microenvironment. We performed gene expression analysis of total mRNA of all matched pairs on Affymetrix U133A 2.0 arrays according to standard protocols. We considered all genes with present calls in either PB-CLL or BM-CLL. The samples were normalized to correct for the individual to individual variance by first normalizing each individual’s PB and BM expression values by their PB value, and then averaging over the 8 individuals using GeneSpring software (Agilent). There were 543 genes with at least 1.5x higher expression in BM vs PB and 192 genes with at least 1.5x higher expression in PB versus BM at p<0.05. Genes more highly expressed in the BM derived CD19-selected cells included topoisomerase II alpha, several cyclins, including cyclin D1, signal transduction components such as PI-3 kinase and components of the Wnt pathway, transcription factors and enhancers such as c-Fos and Sox-4 and several chemokines and chemokine receptors. Of note, there was no difference in the expression of ZAP70, LPL, ADAM29, Bcl-2 and Mcl-1 between the two sites. Our findings are consistent with a model in which CLL cells migrate along an SDF-1 chemokine gradient to the BM where they are stimulated in contact with BM-stroma cells. The higher expression of cell cycle genes in the BM resident CLL cells supports recent findings of a sizeable proliferating fraction of CLL cells and suggests that at least part of this proliferating pool resides in the BM. Ongoing analysis is directed at identifying signaling pathways contributing such proliferation signals.