NG2 antigen is involved in leukemia invasiveness and central nervous system infiltration in MLL-rearranged infant B-ALL

Abstract

Mixed-lineage leukemia (MLL)-rearranged (MLLr) infant B-cell acute lymphoblastic leukemia (iMLLr-B-ALL) has a dismal prognosis and is associated with a pro-B/mixed phenotype, therapy refractoriness and frequent central nervous system (CNS) disease/relapse. Neuron-glial antigen 2 (NG2) is specifically expressed in MLLr leukemias and is used in leukemia immunophenotyping because of its predictive value for MLLr acute leukemias. NG2 is involved in melanoma metastasis and brain development; however, its role in MLL-mediated leukemogenesis remains elusive. Here we evaluated whether NG2 distinguishes leukemia-initiating/propagating cells (L-ICs) and/or CNS-infiltrating cells (CNS-ICs) in iMLLr-B-ALL. Clinical data from the Interfant cohort of iMLLr-B-ALL demonstrated that high NG2 expression associates with lower event-free survival, higher number of circulating blasts and more frequent CNS disease/relapse. Serial xenotransplantation of primary MLL-AF4⁺ leukemias indicated that NG2 is a malleable marker that does not enrich for L-IC or CNS-IC in iMLLr-B-All. However, NG2 expression was highly upregulated in blasts infiltrating extramedullar hematopoietic sites and CNS, and specific blockage of NG2 resulted in almost complete loss of engraftment. Indeed, gene expression profiling of primary blasts and primografts revealed a migratory signature of NG2⁺ blasts. This study provides new insights on the biology of NG2 in iMLLr-B-ALL and suggests NG2 as a potential therapeutic target to reduce the risk of CNS disease/relapse and to provide safer CNS-directed therapies for iMLLr-B-ALL.

Figure 1

Clinical impact of NG2 expression in MLLr infant B-ALL. (a) The HR of relapse for different cutoffs of NG2 expression was investigated to define NG2^high versus NG2^low patients (n=55). The HR of 1.75 corresponding to the 40% cutoff was used. (b) Five-year EFS in NG2^high and NG2^low patients (n=55). (c) Frequency of patients with immature CD10^neg immunophenotype in NG2^high and NG2^low subgroups (left panel), WBC count at diagnostic (middle panel) and frequency of CNS disease (right panel) in NG2^high and NG2^low patients (n=55).

Figure 2

Leukemia development and phenotype in primografts of NG2⁺ and NG2⁻ blast populations. (a) Representative NG2 immunophenotype of diagnostic MA4⁺ BM samples and high-purity FACS-sorting of NG2⁺ and NG2⁻ populations. More than 95% of NG2⁺ and NG2⁻ blasts carry the t(4;11)/MA4 by fluorescence in situ hybridization (FISH). (b) Outline of the in vivo experimental design. NG2⁺ or NG2⁻ blasts were IBM-transplanted into NSG mice at day 0. The health of the mice and leukemia development were monitored over 20 weeks. Mice were killed when disease was evident, when leukemic cells were 10% in PB or at day 140 (in the absence of symptoms or PB engraftment). (c) Kaplan–Meier survival curves for EFS according to decreasing cell doses (200k to 1k) for NG2⁺ and NG2⁻ transplanted mice (n=245). (d) Estimated frequency (and 95% confidence interval) of L-IC in NG2⁺ and NG2⁻ primografts calculated for cell doses <200k or <50k. (e) Representative flow cytometric analysis of leukemic mice. The human graft, identified as CD45⁺HLA-ABC⁺, reproduces the pro-B phenotype (CD34⁺CD19⁺CD10⁻) seen in patients. Engrafted leukemias always re-establish NG2 variable expression and carry the t(4;11)/MA4 as detected by dual-fusion or break-apart FISH. (f) Level of leukemia engraftment in hematopoietic tissues from mice transplanted with NG2⁺ and NG2⁻ blasts. Each dot represents a transplanted mouse and bars represent mean level of engraftment. (g) Both NG2⁺ and NG2⁻ transplanted mice consistently displayed splenomegaly, high WBC counts and a skewed granulocytic-to-lymphoid cell representation in PB. Control group includes non-engrafted mice. *P<0.05.

Figure 3

NG2 expression does not enrich for L-IC capacity in secondary recipients. (a) Kaplan–Meier survival curves for EFS according to different cell doses transplanted into secondary recipients (n=60). Black and gray lines represent secondary recipients transplanted with NG2⁺ and NG2⁻ primary animals, respectively. Dotted line depicts EFS rate of 50%. (b) Percentage of long-term leukemic engraftment in the injected (IT) and contralateral (CL) tibia, liver, spleen and PB of secondary mice. (c) Secondary recipients of cells from either NG2⁺ or NG2⁻ primary mice consistently displayed splenomegaly and high WBC counts.*P<0.05.

Figure 4

NG2 expression is upregulated in extramedullary hematopoietic tissues. (a) Percentage of NG2-expressing blasts (black bar) in primary (top panels) and secondary (bottom panels) engrafted mice at killing (n=245). The following tissues were analyzed: IT, intra-tibia; CL, contralateral tibia; Liv, liver; Sp, spleen; PB, peripheral blood. Left panels, NG2⁺ mice. Right panels, NG2⁻ mice. The initial condition represents the percentage of NG2-expressing cells at the moment of transplantation: 100% for primary NG2⁺ mice (black bar) and 0% for primary NG2⁻ mice (gray bar). (b) RT-qPCR confirming sevenfold higher expression of NG2 in xenografted liver as compared with BM. (c) NG2⁺ and NG2⁻ cells were sorted and IV injected. Mice were weekly monitored by FACS for chimerism and killed after 7–8 weeks. Leukemia engraftment in PB is shown for NG2⁺ (black dots) and NG2⁻ (gray dots) transplanted mice. (d) Threefold higher WBC counts in diagnostic NG2^high versus NG2^low MLL-AF4⁺ infants. Mean of NG2⁺ cells was used as cutoff (Table 1). *P<0.05; NS, not significant.

Figure 5

NG2 is not a prospective marker for CNS-IC but it is expressed in almost all MLLr blasts entering the CNS. (a) Representative H&E staining of mice brains defining xenografts with negative and positive CNS involvement (n=60). S, skull. P, brain parenchyma. Leukemic infiltration is exclusively found in leptomeninges and is marked by a white arrowhead. The right panel (macro) depicts the area magnified on each row. (b) H&E staining and immunohistochemistry for CD19 and CD45 performed on paraffin-embedded skulls from mice transplanted with NG2⁺ and NG2⁻ blasts (n=60). Chimeric spleens and skulls from non-engrafted mice were used as positive and negative controls, respectively. CD45⁺CD19⁺ human blasts are marked with a white arrowhead. (c) Immunofluorescence staining for NG2, endomucin (endothelial vessel marker) and DAPI on chimeric spleen, NG2⁺ and NG2⁻ transplanted mice skull and control skull (non-engrafted mice). Dot lines delimit areas with high density of cells and low density of NG2 marker in the chimeric spleen. Almost a 100% of human blasts in the CNS are consistently NG2⁺, regardless of the NG2 phenotype of the population transplanted. (d) RT-qPCR showing 55-fold higher expression of NG2 in the spinal cord than in BM. *P<0.05.

Figure 6

Specific blockage of NG2 antigen with either chase or distinct clones of anti-NG2 MoAb results in dramatic loss of leukemia engraftment (left panel, n=29 mice from 2 independent patients). Recovered blasts were mainly negative for NG2 (right panel).

Figure 7

Global GEP reveals a migratory signature of NG2⁺ MLLr blasts. (a) Heatmap representation of hierarchical clustering of genes differentially expressed between NG2⁺ and NG2⁻ primary t(4;11)⁺ blasts (n=3 independent leukemias). (b) Statistically significant biological functions identified using IPA on genes differentially expressed in NG2⁺ versus NG2⁻ blasts. They are ranked by z-score. A z-score >2 indicates a predicted activation of that biological function. Biological functions associated with ‘leukemic cell viability and migration/movement’ are shown in black. (c) Biological functions identified using IPA on genes differentially upregulated in NG2⁺ as compared with NG2⁻ circulating (PB/spleen) blasts in xenografted mice. The RT² profiler PCR array specific for epithelial-to-mesenchymal transition/migration genes was used. Biological functions associated with ‘migration/movement’ are shown in black.