High BMP4 expression in low/intermediate risk BCP-ALL identifies children with poor outcomes

Abstract

Pediatric B-cell precursor acute lymphoblastic leukemia (BCP-ALL) outcome has improved in the last decades, but leukemic relapses are still one of the main problems of this disease. Bone morphogenetic protein 4 (BMP4) was investigated as a new candidate biomarker with potential prognostic relevance, and its pathogenic role was assessed in the development of disease. A retrospective study was performed with 115 pediatric patients with BCP-ALL, and BMP4 expression was analyzed by quantitative reverse transcription polymerase chain reaction in leukemic blasts at the time of diagnosis. BMP4 mRNA expression levels in the third (upper) quartile were associated with a higher cumulative incidence of relapse as well as a worse 5-year event-free survival and central nervous system (CNS) involvement. Importantly, this association was also evident among children classified as having a nonhigh risk of relapse. A validation cohort of 236 patients with BCP-ALL supported these data. Furthermore, high BMP4 expression promoted engraftment and rapid disease progression in an NSG mouse xenograft model with CNS involvement. Pharmacological blockade of the canonical BMP signaling pathway significantly decreased CNS infiltration and consistently resulted in amelioration of clinical parameters, including neurological score. Mechanistically, BMP4 favored chemoresistance, enhanced adhesion and migration through brain vascular endothelial cells, and promoted a proinflammatory microenvironment and CNS angiogenesis. These data provide evidence that BMP4 expression levels in leukemic cells could be a useful biomarker to identify children with poor outcomes in the low-/intermediate-risk groups of BCP-ALL and that BMP4 could be a new therapeutic target to blockade leukemic CNS disease.

Graphical abstract

Figure 1

Clinical data show that elevated BMP4 mRNA expression on leukemic blasts at diagnosis is associated with poor outcomes. (A) Using a cutpoint of 18 AU for the entire cohort (n = 114; EFS values from 1 BMP4^low patient was not available), 5-year EFS was reduced in patients with high BMP4 expression levels on leukemic cells (P = .087; log-rank test). (B) Kaplan-Meier survival curve shows that, for low/intermediate risk patients with BCP-ALL (n = 94), the cutpoint of 18 AU identifies patients with significantly reduced EFS (P = .026; log-rank test). Significant differences between both survival curves were observed from day 730 (shown by an arrow; P = .05; test of proportions). (C) Kaplan-Meier survival curves for the entire validation cohort (n = 236) and (D) low/intermediate-risk patients with BCP-ALL (n = 210) show similar results.

Figure 2

BMP4^high BCP-ALL blasts exhibit an increased leukemogenic activity. (A) Outline of the in vivo experimental design for panels B-E and H. BMP4^high and BMP4^low leukemic blasts were IV-transplanted into NSG mice at day 0. Recipient mice were monitored for disease symptoms and the presence of blasts in blood and were euthanized upon the appearance of clinical signs. (B) Nine primary samples of pediatric BCP-ALL patients were chosen according to BMP4 mRNA expression (4 BMP4^high primary samples [patients #1, #2, #3, #8] and 5 BMP4^low primary samples [patients #4, #5, #6, #7, #9] as shown in supplemental Table 1) and xenografted into 3 to 5 mice. Mice were followed for 4 months or were euthanized when they showed clinical signs. Kaplan-Meier curves show reduced survival in mice xenografted with BMP4^high primary samples (P = .0001; log-rank test). (C) Nalm6 or Nalm6-BMP4 (0.5 x 10⁵ cells per mouse) were xenografted into NSG mice. Reduced survival rate was observed in mice injected with Nalm6-BMP4 ALL blasts (P = .007; log-rank test). (D) IL-6 serum levels in NSG mice transplanted with Nalm6 or Nalm6-BMP4 cells and in control healthy mice (n = 5-6 mice per group). Each dot represents a single transplanted mouse, and lines represent the mean level of IL-6 in each group (**P ≤ .01 and #P ≤ .05 represent statistical significances relative to control healthy mice [*] or Nalm6 mice [#]). (E) Percentages (mean ± standard deviation [SD]) and absolute numbers of leukemic blasts recovered from the spleen of mice xenografted with either Nalm6 (n = 7) or Nalm6-BMP4 (n = 10) cells. Each dot represents a transplanted mouse, and bars and lines represent the mean percentages and absolute numbers (**P ≤ .01 and ***P ≤ .001). (F) Proliferation rate of Nalm6 and Nalm6-BMP4 leukemic cells recovered from the spleen of xenografted mice (n = 6), determined by 7-AAD staining and analyzed by flow cytometry. Results represent the mean ± SD of cells in G₀/G₁ phases (**P ≤ .01). Representative flow cytometry histograms are shown in the right panel. (G) Relative viability of Nalm6-BMP4 leukemic cells with respect to Nalm6 cells recovered from spleens of xenografted mice (n = 4) and cultured during 48 hours with different concentrations of methotrexate (MTX) and cytarabine (Ara-C) (*P ≤ .05). (H) Percentage of CD19⁺ leukemic blasts infiltrated in CNS in either Nalm6 (n = 7) or Nalm6-BMP4 (n = 10) mice. Lines represent average percentages of CNS-infiltrating CD19⁺ cells for each cohort (P = .058). Data in (D-H) panels were compared using 2-tailed Mann-Whitney U tests.

Figure 3

Inhibition of canonical BMP4 signaling reduces ALL CNS infiltration and neurological clinical symptoms. (A) Outline of the in vivo experimental design for panels (B-F). Mice were transplanted with BMP4^high primary samples at day 0 (samples from patients #1, #2, and #3, which were shown to engraft in 100% animals in previous experiments, were used). Leukemia development was monitored, and 8 days after injection, mice transplanted with each sample (6-8 animals per sample) were randomly divided into 2 groups (n = 11 mice per group) and treated with DMH1 or control DMSO. All animals were euthanized upon detection of hind limb paralysis in DMSO control mice. (B) Both control DMSO- and DMH1-treated mice (n = 11 mice per group) showed similar engraftment in peripheral organs. Bars show mean percentages (±standard deviation [SD]) of CD19⁺ leukemic blasts in BM, spleen, and PB. (C) CNS leukemia infiltration was reduced in DMH1- vs control DMSO-treated animals (n = 11 mice per group). Each data point represents a single mouse transplanted with blasts from patients #1, #2, or #3 (shown as circles, squares, or triangles), and lines represent the mean percentage of CD19 leukemic blasts in each group (**P ≤ .01; 2-tailed Mann-Whitney U test). (D) Bars represent average serum IL-6 levels measured in DMSO- and DMH1-treated mice at euthanasia (n = 4 mice per group; *P ≤ .05; 2-tailed Mann-Whitney U test). (E) Olfactory habituation–dishabituation test of healthy (green triangles), DMSO- (blue circles), and DMH1-treated (red squares) mice (n = 4 mice per group). Data represent the exploration time (mean ± SD) of successive cotton swabs soaked in octanal, heptanal, or anisole. Two-way ANOVA followed by Bonferroni correction showed significant differences for heptanal and anisole odors: *P ≤ .05 and **P ≤ .01 show significant differences between vehicle-DMSO and healthy control mice; #P ≤ .05 shows significant differences between DMSO- and DMH1-treated mice. (F) The incidence of hind limb paralysis in DMH1-treated mice was notably reduced with respect to the DMSO control group.

Figure 4

High levels of BMP4 induce changes in adhesion molecule expression favoring leukemia cell migration. (A) Comparison of the migration ability of Nalm6 and Nalm6-BMP4 leukemic cells across monolayers of human brain microvascular endothelial cells in a transwell culture system for 4 hours. Results represent the mean ± standard deviation (SD) of 9 independent experiments (***P ≤ .001; 2-tailed Student t test). (B) The migration ability of Nalm6-BMP4 cells was similarly assayed in the presence of BMP signaling pathway inhibitors, BMPRIA-Fc chimera protein (IA-Fc), and DMH1 (*P ≤ .05; 2-tailed Student t test). (C) qRT-PCR quantification of mRNA levels of VCAM-1 and VE-cadherin in human brain microvascular endothelial cells in the presence or absence of rhBMP4 with or without ALL cells. Results represent the mean ± SD of 4 independent experiments (*P ≤ .05; 2-tailed Mann-Whitney U test). (D) qRT-PCR quantification of mRNA levels of VLA-4 and LFA-1 integrins and ADAM10 in leukemic cells recovered from spleens of mice (n = 5) transplanted with Nalm6 or Nalm6-BMP4 cells (*P ≤ .05; 2-tailed Mann-Whitney U test).

Figure 5

High levels of BMP4 increase VEGF expression in leukemic cells and exacerbate angiogenesis in CNS of xenografted mice. (A) Representative images of CD31⁺ endothelial cells (green) and GFAP⁺ astrocytes (red) in sagittal brain cryosections of healthy, Nalm6, and Nalm6-BMP4 mice (Scale bars, 50 μm). (B) Quantitation of microvessel density using CD31 immunofluorescence. Three random cortical areas per section were imaged at magnification ×40. Fiji software was used for computerized quantification of immunostained vascular structures, and positive pixels were quantified and expressed as a percentage (±standard deviation) of CD31⁺ area per tissue total area (4 cryosections/3 mice per group; #P ≤ .05 and ***P ≤ .001; 2-tailed Student t test). (C) VEGFα production in Nalm6 and Naml6-BMP4 cell cultures (n = 4 independent experiments). Results are shown as increments relative to Nalm6 control cultures (*P ≤ .05; 2-tailed Mann-Whitney U test). (D) VEGFA mRNA expression in leukemic cells recovered from spleens of Nalm6 (n = 4) and Nalm6-BMP4 (n = 6) mice. Results are presented as increments relative to Nalm6 control cells (**P ≤ .01; 2-tailed Mann-Whitney U test). (E) Pearson's correlation coefficient (r) and P value between VEGFA and BMP4 mRNA expression in primary BCP-ALL samples at diagnosis are shown (n = 57).