Hematopoietic Stem Cells Are Primarily Involved in Pathogenesis of Chronic Lymphocytic Leukemia

Chronic lymphocytic leukemia (CLL) is characterized by consistent expansion of B cells in peripheral lymphoid organs. CLL B cells frequently express CD5 antigen, and have clonal rearrangement of immunoglobulin heavy chain (IGH) gene with restricted usage of V1, V3 and V4 of the variant region. CLL has thus been believed to represent retention or proliferation of abnormal B cell clones presumably with anti-apoptotic potential, or with deregulated response to auto-antigens. In this study, we extensively search for CLL-initiating cells by utilizing the NOD/SCID/IL2rgnull (NOG) xenogeneic transplantation system, in which human hematopoietic stem cells (HSCs) can normally develop multi-lineage cells including polyclonal B and T cells. In the NOG xenotransplant system, neither CD34-CD19+ circulating B cells, nor CD34+CD38+ bone marrow (BM) progenitor populations from 12 CLL patients engrafted even after injection of >106 cells. We then transplanted CD34+CD38− BM HSC population from 7 CLL patients into 13 NOG mice. Injection of as few as 103 cells of the CD34+CD38− BM population resulted in multi-lineage reconstitution. Most of these mice, however, died within 12–24 weeks after xenotransplantation. In 9 mice analyzed by multi-color FACS, 7 mice possessed both CD5+ and CD5− B cell populations, and the remaining 2 mice had only CD5− B cells. These CD5+ or CD5− human B cell populations were purified separately by FACS, and tested for the IGH gene rearrangement. Strikingly, 16 out of 20 B cell populations were clonal with single IGH rearrangement irrespective of their CD5 expression, by multiplex PCR analysis. In contrast, CD34+CD38− HSC populations in CLL patients never had IGH rearrangement. We then directly sequenced PCR products of IGH gene in each B cell clones as well as those in the original CLL cells purified directly from patients’ blood. Surprisingly, VDJ recombination in B cell clones developed in NOG mice were different from that of the original CLL clones in all 7 CLL cases. Interestingly, all of these clones used only V1, V3 and V4 regions for their VDJ recombination like primary CLL cells. Furthermore, when the CD34+CD38− BM HSC fraction from single CLL patients was transplanted into a set of 3 mice simultaneously, each mouse developed independent B cell clones with different VDJ recombination in all 3 experiments. The fact that CD34+CD38− HSCs from CLL patients but not those from normal individuals give rise to clonal B cell population in our xenograft model strongly suggests that some genetic abnormality for CLL progression is acquired already at the HSC level in CLL patients. HSCs in CLL patients are multipotent, but once they commit to the B cell lineage, they use preferentially the V1, V3 and V4 regions for IGH recombination. Our hypothesis is that such B cell clones may already be abnormal in that they clonally expand in response, for example, to auto-antigens (xeno-antigens in NOG mice), and they may possibly sequentially receive additional mutations to become clinical CLL. Although this xenograft model may not recapitulate full picture of CLL progression, our data clearly show that primary leukemogenic event occurs at the multipotent HSC stage in human CLL.