Abrahamzon, Jennifer L.

Abstract Background: Normal and tumor stem cells are characterized by high activity of multidrug resistance (MDR) transporters. One of these, ABCG2 (ATP-binding cassette, sub-family G member 2 protein), is an ATP dependent transporter and putative stem cell marker responsible for verapamil sensitive Hoechst efflux. While ABCG2 is known to be expressed in normal and leukemic stem cells, as well as a small population of normal lymphocytes and some B-cell malignancies, its expression in chronic lymphocytic leukemia (CLL) is unknown. It has been postulated that leukemic stem cells due to their quiescent nature and expression of MDR transporters represent a population resistant to therapy and that this residual population is critical for tumor persistence and recurrence. Hypothesis: We hypothesized that ABCG2 is expressed in a small percentage of primary CLL B cells; gene expression profiles of ABCG2 positive versus ABCG2 negative CLL B cells differ in respect to expression of self renewal and lymphoid development genes; the frequency of ABCG2+ CLL B cells increases after treatment in patients responding to therapy. Methods: We analyzed ABCG2 expression by primary CD5+, CD19+ CLL-B cells from untreated CLL patients of all Rai stages by flow cytometry. In a subset of patients we used fluorescence activated cell sorting (FACS) to sort CD19+, CD5+ ABCG2+ and CD19+, CD5+ ABCG2- cells. Gene expression profiling was then performed using the U133 plus 2.0 Affymetrix microarray platform. In a separate cohort of patients treated in a clinical trial of pentostatin, cyclophosphamide and rituximab (PCR), the percentage of ABCG2+, CD19+, CD5+, CD79b dim cells at baseline and then two months after completion of 6 cycles of PCR therapy where patients had minimal residual disease (MRD) was assessed and correlated with clinical response. Results: ABCG2+ CD19+, CD5+ detectable populations were seen in all 20 CLL assessed patients (median percentage 0.6%; range 0.08%–3.8%). There was no difference in percentage of ABCG2+ cells based on Rai stage, IGVH mutational status, Zap70 or CD38 expression. Preliminary analysis of the gene expression profiling of ABCG2 positive versus negative CLL B cells from four randomly selected patients revealed significantly higher expression of genes associated with self-renewal, cell cycle and early B-cell development including: cyclin-dependent kinase inhibitor 1C (CDKN1C, p=0.034), transcription factor 7-like 2 (TCF7L2, involved in WNT pathway regulation, p=0.016), beta-catenin (p=0.034) and pre-B-cell colony enhancing factor 1 (PBEF-1, p=0.037). Flow based assessment of the levels of ABCG2 positive populations at baseline and after therapy with PCR in patients with minimal residual disease showed a dramatic increase in frequency of ABCG2 positive CLL B cells. The percentage of ABCG2+ cells went from a median level of 0.19% (range 0.04%–0.19%) prior to therapy to a median level of 10.93% (range 0.15%–25.12%), p<0.001. In contrast two patients who did not reach MRD (partial responses by NCI-WG criteria) had no significant increase in percentage of ABCG2 positive cells (0.14%; 0.23% and 0.16%; 0.21% prior and after therapy, respectively, p=0.68). Conclusion: Our data indicate that ABCG2 positive CLL B-cells constitute 0.1–3.8% of circulating CLL B-cells in untreated patients. The frequency of ABCG2+ CLL B-cells appears to dramatically increase after therapy in the MRD state; this could be related to their relative resistance to therapy and/or a shift from extravascular compartments post therapy. Since ABCG2 positive CLL B-cells demonstrate expression of early B-cell development and self-renewal genes we believe that that this population could represent a putative self renewing CLL B-cell compartment. Further studies to characterize features of ABCG2 CLL-B –cells in relation to their capacity to be self renewing and resistance to therapy are warranted.

Abstract It is believed that malignant cells “condition” the microenvironment to facilitate tumor cell survival. We hypothesized that crosstalk between CLL B-cells and marrow stromal cells impacts both cell types bi-directionally and ultimately contributes to leukemic cell apoptotic resistance. To test this hypotheses, bone marrow stromal cells from core bone biopsies from CLL patients were isolated and cultured using methods we have previously described (Leuk Res 2007 31(7):899). Subsequently, we determined the impact of co-culture on CLL B-cell features including apoptosis and CD38 expression. In addition, we evaluated the release of angiogenic cytokines on co-culture and signal events in the stromal cells. Immunophenotyping demonstrated that cultured bone biopsy derived stromal cells were CD73+, CD105+, CD146+, CD14−, CD45−, CD34−, HLA-DR-, suggesting they were mesenchymal stem cells (MSC). Co-culture of these MSC with CLL B-cells protected CLL B-cells from both spontaneous apoptosis (SA) and drug-induced (fludarabine and chlorambucil) apoptosis (DA). For SA, the mean survival of CLL B-cells with or without co-culture of MSC for 5 days were 56.9 ± 10.0 and 7.7 ±3.7 (p<0.05), respectively. When CLL B cells were treated with fludarabine or chlorambucil, the fraction of CLL cells tightly adherent to MSC (TA-CLL) showed higher survival than a less adherent but viable fraction of CLL B-cells. The mean survival of TA-CLL cells treated with 10 μM of fludarabine for 48 hours in the presence of MSC were 67.5 ± 3.6 vs 29.8 ± 11.1 without MSC (P<0.05), respectively. When CLL cells with evidence for CD38 expression were co-cultured with MSC, both the percentage of CD38 positive cells and level of expression of CD38 per cell were up-regulated (mean fold change: CD38 percentage, 2.7, p<0.05; CD38 MFI, 1.9, p<0.05) after 2 weeks. In contrast, the CD38 percentage and expression were not changed in cells with minimal CD38 expression when these CLL B-cells were co-cultured with MSC. In addition, co-culture of MSC with CLL cells induced rapid ERK and AKT phosphorylation (within 30 min) in the MSC on immunoblot analysis. When CLL B cells and MSCs were cultured in transwells, the activation of ERK and AKT in MSC occurred at similar levels, indicating that activation of MSC was mediated by soluble factors. In addition, co-culture led to increased secretion of vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF) as well as a decrease of thrombospondin-1 (TSP-1) in the culture medium. These findings confirm that co-culture of CLL B-cells and MSC culminates in “angiogenic switch.” Taken together, these results strongly suggest interactions between MSC and CLL B cells are a bi-directional process. In leukemic cells, the interaction not only protects against spontaneous and drug induced apoptosis but also leads to an increase in CD38 expression consistent with an activated status. In MSC, the interaction leads to activation of ERK and AKT. Co-culture also facilitates angiogenic switching. These results underscore the dynamic and complex nature of the interactions between bone marrow stromal cells and CLL B-cells. Further studies are needed to dissect how crosstalk between CLL B-cells and MSC relates to disease progression, and determines whether these interactions can be targeted with therapeutic intent.