Differential Expression of Adhesion Molecule Receptors May Influence Bone Marrow Microenvironment-Mediated Protection of Leukemia-Initiating Cells (LICs) in Infant MLL-rearranged (MLL-R) Acute Lymphoblastic Leukemia (ALL)

Background: Compared to older children with ALL who achieve a minimal residual disease (MRD)-negative remission, infants with ALL who achieve an MRD-negative remission have a significantly higher rate of relapse. This suggests that, despite the achievement of MRD negativity, a very small amount of leukemia cells persist and contribute to disease relapse in infant ALL; LICs are likely to represent this persistent population of cells. Previously, we demonstrated that infant MLL-R ALL derive significantly more protection from co-culture with bone marrow (BM) stromal cells than non-MLL-R ALL. Therefore, protection by the BM microenvironment may contribute to LIC persistence in infant MLL-R ALL. We hypothesized that infant MLL-R ALL LICs have increased interactions with the BM microenvironment, through increased surface expression of adhesion molecule receptors. We also hypothesized that co-culture with normal BM stroma would protect LICs from spontaneous and chemotherapy-induced apoptosis, and AMD3100 (plerixafor) would decrease stromal protection through inhibition of CXCR4-CXCL12 signaling.

Methods/Results: We analyzed 9 viably cryopreserved diagnostic samples collected from infants with MLL-R ALL (n=4 MLL-AF4, n=5 MLL-ENL). Using flow cytometry, we identified 2 phenotypically-defined LIC subpopulations, CD34+CD38+CD19+ and CD34-CD19+ (Aoki et al 2015), within the bulk leukemic blast population (CD45+). Consistent with previous findings, 34+38+19+ LICs were predominant in MLL-AF4 samples, while 34-19+ LICs were predominant in MLL-ENLsamples. Next, we measured surface expression of the adhesion molecule receptors CXCR4, CD49d (VLA-4), CXCR7, and CXCR3 (quantified as mean fluorescence intensity, MFI). Overall, surface expression of CXCR4, CD49d, CXCR7, and CXCR3 was higher in the 34+38+19+ LICs, compared to 34-19+ LICs (e.g., CXCR4 MFI 142 in 34+38+19+ vs. 76 in 34-19+, p=0.02).

Next, samples were treated with dose ranges (0-30 µM) of AraC or etoposide (Etop) and cultured for 48 hours in 3 conditions: 1) off stroma, 2) on stroma, or 3) on stroma with the CXCR4 inhibitor AMD3100 (10 μM). Stromal cells were cultured from a healthy BM donor. In vehicle control-treated 34+38+19+ LICs, stromal co-culture led to increased surface expression of CD49d (avg 64% increase, p=0.03) and CXCR7 (avg 67% increase, p=0.02). Conversely, in vehicle control-treated 34-19+ LICs, stromal co-culture did not affect surface expression of the measured adhesion molecule receptors. In all 3 culture conditions, surface expression of CXCR4, CD49d, and CXCR7 was significantly higher in 34+38+19+ LICs compared to 34-19+ LICs (p≤0.01).

We then measured apoptosis by flow cytometry and Annexin V binding. In the absence of chemotherapy, stromal co-culture significantly protected 34+38+19+ LICs from spontaneous apoptosis (p<0.001), but not 34-19+ LICs (p=0.16). Treatment with AMD3100 decreased stromal protection from spontaneous apoptosis in 34+38+19+ LICs (p<0.05), but not 34-19+ LICs (p=0.13). Notably, 34-19+ LICs were more resistant to spontaneous apoptosis even when cultured without stromal support (avg Annexin V-negative fraction 76.5% in 34-19+ vs. 62.8% in 34+38+19+, p=0.005). In spite of differences in stromal-mediated protection from spontaneous apoptosis, stromal co-culture protected both LIC populations from AraC- or Etop-induced apoptosis (p<0.001 for 34+38+19+, p<0.01 for 34-19+). Also, 34-19+ LICs were more resistant to Etop than 34+38+19+ LICs in the presence (p<0.001) or absence (p<0.01) of stroma. Finally, AMD3100 decreased stromal protection from AraC- or Etop-induced apoptosis (p<0.001 for 34+38+19+, p<0.05 for 34-19+).

Conclusions: Our results suggest that the two immunophenotypically distinct LIC populations in infant MLL-R ALL may rely differently on the BM microenvironment. Higher surface expression of adhesion molecule receptors may explain why 34+38+19+ LICs derive more benefit from stromal co-culture, while lower surface expression of adhesion molecule receptors may explain decreased reliance on stroma by 34-19+ LICs. There may also be differences in chemotherapy resistance between the LIC populations. Despite these differences, CXCR4 inhibition enhanced chemosensitivity in both LIC populations. These results support the development of new therapeutic strategies that target leukemia-stroma interactions to improve outcomes for infants with MLL-R ALL.