IGFBP7 activates retinoid acid-induced responses in acute myeloid leukemia stem and progenitor cells

Abstract

Treatment of acute promyelocytic leukemia (APL) with all-trans retinoic acid (ATRA) in combination with low doses of arsenic trioxide or chemotherapy leads to exceptionally high cure rates (>90%). ATRA forces APL cells into differentiation and cell death. Unfortunately, ATRA-based therapy has not been effective among any other acute myeloid leukemia (AML) subtype, and long-term survival rates remain unacceptably low; only 30% of AML patients survive 5 years after diagnosis. Here, we identified insulin-like growth factor binding protein 7 (IGFBP7) as part of ATRA-induced responses in APL cells. Most importantly, we observed that addition of recombinant human IGFBP7 (rhIGFBP7) increased ATRA-driven responses in a subset of non-APL AML samples: those with high RARA expression. In nonpromyelocytic AML, rhIGFBP7 treatment induced a transcriptional program that sensitized AML cells for ATRA-induced differentiation, cell death, and inhibition of leukemic stem/progenitor cell survival. Furthermore, the engraftment of primary AML in mice was significantly reduced following treatment with the combination of rhIGFBP7 and ATRA. Mechanistically, we showed that the synergism of ATRA and rhIGFBP7 is due, at least in part, to reduction of the transcription factor GFI1. Together, these results suggest a potential clinical utility of IGFBP7 and ATRA combination treatment to eliminate primary AML (leukemic stem/progenitor) cells and reduce relapse in AML patients.

Graphical abstract

Figure 1.

Enhanced IGFBP7 expression sensitizes APL cells to physiological concentrations of ATRA, and rhIGFBP7 potentiates ATRA-induced differentiation, inhibition of proliferation, and apoptosis in non-APL AML cells. For all experiments, AML cells were cultured in low serum condition and stimulated with PBS (Control/Ctrl), 300 µg/mL rhIGFBP7, 1.0 µM ATRA, or the combination (Combi) for 96 hours unless stated otherwise. Differentiation (membrane CD11b expression), proliferation, and apoptosis were measured using flow cytometry, and data are plotted as mean ± SEM. (A) Percentage of differentiation in NB4 cells, transduced with IGFBP7 shRNAs (#1 and #2) or scrambled (SCR) shRNA, after stimulation with 0.5 µM ATRA. P values were determined using 1-way ANOVA with post hoc Dunnett multiple-comparison test. (B-C) NB4 cells, transduced with control or IGFBP7-expressing (IGFBP7-OE) vectors, were stimulated with 0.01 µM ATRA. (B) Percentage increase in CD11b expression. (C) CFU assay (in duplicate), normalized against untreated control. (D) CD11b expression in HL60 cells, quantified relative to flow count beads. (E) Morphology of HL60 cells after stimulation for 72 hours, analyzed using a bright-field microscope (upper panels) and by May-Grünwald-Giemsa staining (lower panels). (F) Viability of AML cell lines after stimulation for 72 hours (HL60 and OCI-AML3 cells) or 120 hours (THP1 and KG1a cells) with rhIGFBP7, ATRA, or the combination, normalized against untreated control. (G) AnnexinV⁺ THP1 cells after stimulation for 120 hours. (H) Percentage of apoptotic (AnnexinV⁺ and 7AAD⁺) THP1 cells (left panel) or OCI-AML3 cells (right panel) after stimulation for 120 hours, quantified relative to flow count beads. Graphs are representative of ≥3 independent experiments. *P < .05, **P < .01, ***P < .001, ****P < .0001, 1- or 2-way ANOVA with post hoc Tukey’s multiple-comparison test, unless stated otherwise.

Figure 1.

Enhanced IGFBP7 expression sensitizes APL cells to physiological concentrations of ATRA, and rhIGFBP7 potentiates ATRA-induced differentiation, inhibition of proliferation, and apoptosis in non-APL AML cells. For all experiments, AML cells were cultured in low serum condition and stimulated with PBS (Control/Ctrl), 300 µg/mL rhIGFBP7, 1.0 µM ATRA, or the combination (Combi) for 96 hours unless stated otherwise. Differentiation (membrane CD11b expression), proliferation, and apoptosis were measured using flow cytometry, and data are plotted as mean ± SEM. (A) Percentage of differentiation in NB4 cells, transduced with IGFBP7 shRNAs (#1 and #2) or scrambled (SCR) shRNA, after stimulation with 0.5 µM ATRA. P values were determined using 1-way ANOVA with post hoc Dunnett multiple-comparison test. (B-C) NB4 cells, transduced with control or IGFBP7-expressing (IGFBP7-OE) vectors, were stimulated with 0.01 µM ATRA. (B) Percentage increase in CD11b expression. (C) CFU assay (in duplicate), normalized against untreated control. (D) CD11b expression in HL60 cells, quantified relative to flow count beads. (E) Morphology of HL60 cells after stimulation for 72 hours, analyzed using a bright-field microscope (upper panels) and by May-Grünwald-Giemsa staining (lower panels). (F) Viability of AML cell lines after stimulation for 72 hours (HL60 and OCI-AML3 cells) or 120 hours (THP1 and KG1a cells) with rhIGFBP7, ATRA, or the combination, normalized against untreated control. (G) AnnexinV⁺ THP1 cells after stimulation for 120 hours. (H) Percentage of apoptotic (AnnexinV⁺ and 7AAD⁺) THP1 cells (left panel) or OCI-AML3 cells (right panel) after stimulation for 120 hours, quantified relative to flow count beads. Graphs are representative of ≥3 independent experiments. *P < .05, **P < .01, ***P < .001, ****P < .0001, 1- or 2-way ANOVA with post hoc Tukey’s multiple-comparison test, unless stated otherwise.

Figure 2.

rhIGFBP7 activates ATRA-driven responses in primary non-APL AML cells. For all ex vivo experiments, cells were stimulated with PBS (Control/Ctrl), 100 µg/mL rhIGFBP7, 0.5 µM ATRA, or the combination (Combi) for 7 days. Percentages of CD33⁺CD11b⁺, viable CD45^dim and CD45^dimCD34⁺ cells were measured using flow cytometry, quantified relative to flow count beads, normalized against untreated control cells, and plotted as mean ± SEM. Patient sample characteristics are summarized in supplemental Table 1. (A) CD11b membrane expression on CD33⁺ blasts relative to the untreated control sample in AML13. (B) Percentage of viable CD45^dim cells (blue; upper panels) and CD45^dimCD34⁺ cells (red; lower panels) in AML7. (C) Heat map of clinical and genetic features of 28 primary AML samples. Responsive AML cases were defined as >5% increase in CD11b-expressing myeloid CD33⁺ blasts and/or >5% reduction in CD45^dim blast survival upon ATRA-rhIGFBP7 combination treatment compared with the single treatments. The fold increase in CD11b and fold decrease in CD45^dim cells represent the ratio of CD11b-expressing CD33⁺ blasts and the ratio of reduction in CD45^dim blast survival following combination treatment relative to untreated control or single therapies. Percentages are shown in supplemental Table 2. (D) Induction of differentiation in 8 primary AML samples responsive to rhIGFBP7-ATRA. Percentage of viable CD45^dim cells (E) and CD45^dimCD34⁺ cells (F) in 10 primary AML samples responding to rhIGFBP7-ATRA. (G) Schematic overview of the experiment (left panel). After injection of T-cell–depleted primary AML cells, NSG mice were treated with ATRA (10 mg, 21-day-release pellet) in week 2 and/or rhIGFBP7 (12 mg/kg) in week 5 (days 1-3). At week 16, the bone marrow cells of the mice were analyzed for the presence of hCD45⁺ cells (middle panel) and myeloid hCD45⁺CD33⁺ cells (right panel) using flow cytometry. *P < .05, **P < .01, ***P < .001, ****P < .0001, 1-way ANOVA with post hoc Tukey’s multiple-comparison test. BM, bone marrow; N/A, not available; PB, peripheral blood.

Figure 3.

Enhanced IGFBP7 expression or treatment with rhIGFBP7 induces sensitivity to ATRA in primary AML stem and progenitor cells. For all ex vivo experiments, cells were incubated with PBS (Control/Ctrl), 100 µg/mL rhIGFBP7, 0.5 µM ATRA, or the combination (Combi). For CFU progenitor and long-term liquid culture (stem cell) assays, samples (in duplicate) were incubated for 7 days or 4 weeks, respectively, and normalized against untreated controls. Data are plotted as mean ± SEM, unless stated otherwise. Patient sample characteristics are summarized in supplemental Table 1. (A-C) Primary AML samples were lentivirally transduced with control or IGFBP7-OE vectors. (A) CFU progenitor assays of primary AML samples, plotted as mean ± SD. (B) CFU plates, containing all colonies, were harvested, and flow cytometric analysis of AML cells derived from the colonies was performed, showing leukemic CD45^dim blasts in the transduced cell population for AML9 (LAIP = CD15; left panels) and AML29 (LAIP = CD56; right panels). (C) Percentage of CD45^dimLAIP⁺ blasts (left panel) and CD45^highLAIP⁺ blasts (middle panel) derived from the colonies, quantified, and normalized against untreated controls. Percentage of lymphocytes (right panel), measured using flow cytometry and plotted as mean ± SD. (D-E) CFU progenitor assays of primary AML samples, with AML cases not responding to combination therapies shown in the right lighter bars. Samples were incubated with rhIGFBP7 and ATRA (D) or with TCP (10 µM) and ATRA (E). (F) CFU stem cell assay (long-term liquid culture) of primary AML samples incubated with rhIGFBP7 and ATRA. (G) Schematic overview of the experiment (left panel). After injection of T-cell–depleted primary AML cells, NSG mice were treated with rhIGFBP7 (10 mg/kg) in week 4 (day 1) and week 8 (day 1-3) and/or ATRA (10 mg, 21-day-release pellet) in week 8 (day 3). Subsequently, equal numbers of human myeloid hCD45⁺CD33⁺ cells derived from the first transplant were injected into secondary recipients. At week 16, bone marrow cells from the mice were analyzed for the presence of hCD45⁺ cells (middle panel) and myeloid hCD45⁺CD33⁺ cells (right panel). *P < .05, **P < .01, ***P < .001, 1- or 2-way ANOVA with post hoc Tukey’s multiple-comparison test, unless stated otherwise.

Figure 4.

rhIGFBP7 and ATRA combination treatment does not affect healthy NBM cells. NBM cells derived from healthy donors were incubated with PBS (Ctrl), 100 µg/mL rhIGFBP7, 0.5 µM ATRA, or the combination (Combi) for 5 to 7 days. The effects of the treatments were normalized against untreated control cells and plotted as mean ± SEM. NBM samples were analyzed for the percentage of viable cells (A), CD45^dim cells (B), myeloid CD45^dimCD33⁺ cells (C), CD45^dimCD34⁺ cells (D), CD3⁺ T cells (E), and CD19⁺ B cells (F) using flow cytometry and quantified relative to flow count beads. (G) CFU progenitor assays of NBM samples (in duplicate) after 7 days of treatment.

Figure 5.

rhIGFBP7 induces susceptibility to ATRA by reducing GFI1 expression. AML cells were stimulated with 100 µg/mL rhIGFBP7, 0.5 µM ATRA, 1.0 µM TCP, or the combination (Combi) for 4 days (HL60 cells) or 7 days (primary AML cells). Induction of differentiation (membrane CD11b expression) and cell viability were measured using flow cytometry and quantified relative to flow count beads. Data are representative of ≥3 independent experiments, plotted as mean ± SEM. Patient sample characteristics are summarized in supplemental Table 1. (A) Heat map of the top 111 differentially expressed genes in CD45^dim cells from 3 primary AML samples after treatment with rhIGFBP7 for 48 hours. Genes were selected based on their log2-fold change (FC) in rhIGFBP7-stimulated samples compared with untreated controls (cutoff value = log2FC < −0.5 and > +0.5; P < .01). The top 15 downregulated genes are listed. *Position of GFI1. (B) Pathway analysis of the top upregulated and downregulated genes in response to rhIGFBP7 treatment for 48 hours. The minimum number of overrepresented genes was set to 3, and the Kyoto Encyclopedia of Genes and Genomes pathway and Gene Ontology process were used to perform functional pathway analysis using DAVID (v6.8). (C) GFI1 expression (in triplicate) in primary AML samples after treatment with rhIGFBP7, measured relative to GUS expression using qRT-PCR, and normalized against control sample. P values were determined using a paired Student t test. (D) HL60 cells were lentivirally transduced with GFI1-shRNAs (sh#1 and sh#2) or scrambled (scr) shRNA. The percentage of (Venus⁺)CD11b⁺ HL60 cells was measured after 4 days of ATRA treatment, and the percentage of viable cells (blue) for each condition is listed above the graph. P values were determined using 2-way ANOVA with a post hoc Tukey’s multiple-comparison test. (E-H) AML cells were lentivirally transduced with control-YFP (Control) or GFI1-YFP (GFI1-OE)–expressing vectors. (E) Percentage of CD11b⁺ cells after treatment of HL60 cells. The percentage of viable cells (blue) for each condition is listed above the graph. Example of flow cytometric analysis of AML2 (F) and percentage of membrane CD11b in the transduced (YFP⁺)CD45^dim cell population of 2 primary AML samples, measured 14 days posttransduction (G). (H) Percentage of CD11b⁺ cells after treatment of HL60 cells with TCP, ATRA, or the combination (Combi). The percentage of viable cells (blue) for each condition is listed above the graph. *P < .05, **P < .01, ***P < .001, ****P < .0001, 2-way ANOVA with post hoc Bonferroni multiple-comparison test, unless stated otherwise.

Figure 6.

High RARA expression is a biomarker predicting sensitivity for rhIGFBP7-induced ATRA susceptibility in primary AML. Gene expression levels for GFI1 and RARA were measured by qRT-PCR (in triplicate). Patient sample characteristics of responders and nonresponders to rhIGFBP7 and ATRA combination treatment are summarized in supplemental Table 1. (A) Log2-transformed GFI1 gene expression in EVI-1⁻ (n = 50) and EVI-1⁺ (n = 10) AML samples at diagnosis, measured relative to GUS expression. The relative GFI1 expression of 1 of the EVI-1⁺ AML samples was set at 1. IGFBP7 (B), GFI1 (C), and RARA (D) expression in responders and nonresponders to rhIGFBP7 and ATRA combination treatment, relative to HMBS expression, as measured by qRT-PCR. (E) GFI1 and RARA gene expression levels in responders and nonresponders to the combination of rhIGFBP7 and ATRA, relative to HMBS, as by qRT-PCR. Spearman’s correlation coefficients were calculated. *P < .05, **P < .01, Student t test, unless stated otherwise.