The Leukemic Stem Cell (LSC) in Ph⁺ CML Is a CD34⁺/CD38⁻/Lin⁻ Cell That Co-Expresses Dipeptidylpeptidase IV (CD26) and Disrupts LSC-Niche Interactions by Degrading the CXCR4 Ligand SDF-1α

Abstract 961

In Philadelphia-positive (Ph⁺) chronic myeloid leukemia (CML), leukemic stem cells (LSC) supposedly reside in a CD34⁺/CD38⁻/Lin⁻ fraction of the leukemic clone. However, little is known about phenotypic properties of LSC in CML. We screened for novel LSC markers and targets in CML by gene chip studies and extensive flow cytometry analyses using monoclonal antibodies against various surface antigens (n=50). A total number of 240 bone marrow or peripheral blood samples (CML, n=95; AML, n=103; CMML, n=10, control marrow, n=32) were examined. In common with normal SC, CD34⁺/CD38⁻ CML LSC were found to co-express the homing-receptor CD44, G-CSF-R (CD114), KIT (CD117), FLT3 (CD135), and CXCR4 (CD184). Similar to LSC in AML and CMML, CML LSC were found to display higher levels of Siglec-3 (CD33) and IL-3RA (CD123). Most significantly, however, we found that in contrast to normal CD34⁺/CD38⁻ stem cells, CD34⁺/CD38⁻ CML LSC aberrantly express IL-2RA (CD25), dipeptidylpeptidase IV (DPPIV=CD26), and IL-1RAP. In other myeloid leukemias (AML, CMML), CD34⁺/CD38⁻ LSC also co-expressed CD25, but usually did not express CD26 or IL-1RAP. Whereas CD26 was expressed almost invariably on CD34⁺/CD38⁻ cells in all CML patients tested, the surface enzyme was neither detectable in more mature CD34⁺/CD38⁺ progenitor cells nor on CD34⁺/CD38⁻ stem cells in reactive bone marrow or healthy controls. During successful treatment with imatinib or nilotinib (patients examined at CCyR and/or MMR), CD34⁺/CD38⁻ stem cells invariably showed a ‘normal' phenotype (CD25⁻, CD26⁻, IL-1RAP⁻), whereas in relapsing CML, CD34⁺/CD38⁻ cells were again found to co-express CD25 and CD26. Sorted Lin⁻/CD26⁻ stem cells obtained from CML patients (at diagnosis) engrafted irradiated NOD-SCID IL-2Rγ^−/− (NSG) mice with normal multilineage BCR/ABL1⁻ hematopoiesis, whereas Lin⁻/CD26⁺ stem cells were found to engraft NSG mice with BCR/ABL⁺ cells. We next examined the regulation of expression of CD25 and CD26 on CML LSC. Whereas expression of CD25 was found to depend on BCR/ABL1 and STAT5-activity, CD26 expression was found to be expressed independent of BCR/ABL1 and independent of STAT5-signaling. In a next step, we examined the potential function of CD26 on CML LSC. In these studies, CD26 was identified as a target-enzyme disrupting the niche-related SDF-1α/CXCR4 axis by degrading SDF-1α. Correspondingly, CD26-targeting gliptins (sitagliptin, 1 μM; vildagliptin, 1 μM) were found to revert recombinant DPPIV/CD26-induced or cellular CD26-induced inhibition of SDF-1α-mediated in vitro migration of CD26⁺ leukemic cells. Finally, we found that in a CML patient treated with nilotinib, in whom uncontrolled diabetes mellitus required therapy with saxagliptin, BCR/ABL1 levels (in percent of ABL according to IS) that were found to increase before the start of saxagliptin (IS before saxagliptin: 1.6 [-4 months], 2.3 [-3 months], and 2.4 [at therapy-start]), decreased over time during saxagliptin-therapy (IS: 1.0 [+1 month], 1.0 [+3 months], 0.8 [+5 months]). Together, the CML-initiating LSC is a CD34⁺/CD38⁻ cell that exhibits aberrant expression of IL-1RAP, CD25, and DPPIV/CD26. All three markers may be useful for purification of CML LSC. DPPIV/CD26 appears to be a functionally and pathogenetically relevant antigen that may facilitate niche-independent uncontrolled redistribution and thus extramedullary spread of LSC and LSC-derived progenitor cells in CML. Whether CD26 can be developed as a novel therapeutic target in CML is currently under investigation.