T-cell acute lymphoblastic leukemia (T-ALL) is mainly a child and adolescent blood malignancy. T-ALL patients present an increased risk of Central Nervous System (CNS) relapse defined by leukemic cell infiltration in cerebrospinal fluid and brain. Using transgenic mice and T-ALL cell lines previous works have shown that T-ALL migration in CNS depends on CCR7 chemokine receptor expression (S. Buonamici et al., Nature, 2009). VE-cadherin and CD31/PECAM1 also seem implicated, as it has been shown in vitro (S. M. Akers et al., Exp Hematol, 2010). In patients, high level of IL-15 at diagnosis predicts current CNS invasion and sometimes at relapse (G. Cario et al., J Clin Oncol, 2007). So far no study has investigated mechanisms involved in CNS infiltration using T-ALL patient samples in vivo.

In the present study we developed a mouse model of CNS infiltration using leukemic cells isolated from patients and transplanted into NOD/SCID IL2Rуc-/- (NSG) mice. Proper conditioning of NSG mice and different routes of injection were tested to define a protocol avoiding non-specific CNS infiltration of leukemic cells. Also bone marrow (BM) engraftment levels of leukemia between 60 to 100% were used to set up the excision time of hematopoietic tissues and brain. Leukemic blasts from 8 patients with or 9 patients without CNS invasion were grafted and brain infiltration was followed up using standard histology and immunohistochemistry techniques.

Our data indicate that (1) under specific experimental procedures, leukemic cells from patients with CNS invasion did infiltrate mouse CNS (8/8 samples) whereas the majority of cells from “non-infiltrated” patients did not (7/9 samples), (2) leukemic cells recovered from NSG brain and BM were similar in terms of brain and/or BM infiltration in secondary transplant experiments. Moreover, T-Leukemia Initiating Cell frequency was the same whatever the BM or CNS origin of blasts in the primary recipient. Interestingly, analysis of blasts at diagnosis showed that surface expression of adhesion molecules can not discriminate CNS+ or CNS- leukemic cells. However, blocking of CD31 decreased in vitro migration of blasts from CNS+ compared to CNS- patients through endothelial layer derived from blood brain barrier cells. Pioneered in vivo experiments show that CNS+ blasts pre-treated with CD31 antibody and injected in NSG are less prone to colonize mouse brain. Moreover, knocking down CD31 in CNS+ T-ALL by lentiviral shRNA strategy impairs leukemia development in mice, further decreasing CNS infiltration, whatever injection routes is used including intrafemoral injection.

In conclusion, T-ALL xenografts in NSG mice mimic CNS invasion in patients. CD31 is a major player in blast cells migration in vitro and brain infiltration in vivo. This new model opens a new area of investigation to improve our knowledge of the molecular mechanisms of CNS infiltration in T-ALL.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.