Fig 1.
BM-MSCs induce ALDH activity in AML cells and enhance engraftment in mice.
(A, B) One million firefly luciferase-expressing AML cells (Molm13) were implanted subcutaneously, with or without 1 million BM-MSCs and 100 μL of Matrigel, into NOD/SCID mice. Bioluminescence imaging was performed at 1 and 2 weeks to check leukemia engraftment and growth rate. (C) AML cell lines and patient samples were cultured with or without BM-MSCs for 3–5 days. ALDH activity in AML cells was measured using ALDEFLUOR® assay. Gene expression analysis was performed by RNA sequencing. (D) Fluorescence-activated cell sorting of OCI-AML3 cells was performed based on phenotype (CD45+, CD90-) to distinguish them from BM-MSCs (CD90+, CD45-). (E) OCI-AML3 cells were cultured with or without BM-MSCs for 3 or 5 days. Cells were stained with ALDEFLUOR®, CD45, and CD90. During FACS analysis, MSCs (CD90+, CD45-) were gated and ALDH activity was measured in OCI-AML3 cells by flow cytometry. Data were analyzed on FlowJo software. (F) Histogram representation of the percentage of ALDH+ OCI-AML3 cells corresponding to the experiment done in E. (G) The same experiment was done as in E, using HL60 AML cells. (H) Histogram representation showing the percentage of ALDH+ HL60 cells corresponding to the experiment done in G. Data are plotted as the mean value with error bars representing standard error. For B, F, and H a linear model with interaction term was used to evaluate the significance. N = 3 for each group.
Fig 2.
Differential expression of ALDH isoforms in AML cells co-cultured with BM-MSCs.
(A, B) Patient-derived primary AML cells from bone marrow and peripheral blood samples were cultured with or without MSCs for 3 days and ALDH activity was measured using ALDEFLUOR® assay by flow cytometry. Data show a graphical representation of ALDH mean fluorescence intensity (MFI) and percentage of ALDH+ AML patient samples when co-cultured with stromal cells. (C) Total RNA was extracted from 5 million AML cells. RT-PCR was performed to analyze gene expression of different ALDH isoforms using primers listed in S2 Table. All samples were run in triplicate. The relative fold increase of specific RNA was calculated by the comparative cycle of threshold detection method, and values were normalized to GAPDH. Fold changes in gene expression were calculated using the 2-ddC method after normalization to GAPDH. Data are plotted as mean values with error bars representing standard error. (D) AML expression data for ALDH isoforms and Kaplan-Meier survival analysis were obtained from the TCGA dataset. For A and B, the paired t test was used. For C, the Mann-Whitney U test/Student’s t test was used (N = 3). For D, the log-rank test was used to test the significance.
Fig 3.
Activation of TGF-β1 gene signature in OCI-AML3 cells co-cultured with BM-MSCs.
(A, B) OCI-AML3 cells were cultured with or without BM-MSC cells for 3 days. OCI-AML3 cells were FACS sorted to separate them from BM-MSCs and the gene expression analysis was performed by RNA sequencing. Samples were sequenced on the HiSeq Sequencing System. Sequence reads were mapped to human genomics (build hg19) with bowtie2 aligner using RSEM software. R software was used to compare the differential expression between MSC co-culture samples and OCI-AML3 controls. Genes with adjusted p values less than 0.05 and absolute fold changes larger than 2 were considered significant. Analysis of differentially expressed genes using the Ingenuity® pathway analysis tool revealed activation of a TGF-β1-associated gene signature. (C) Real-time PCR analysis was performed to analyze the expressions of the indicated genes that are differentially regulated by TGF-β1 in OCI-AML3 cells co-cultured with BM-MSCs compared to OCI-AML3 controls. (D) TGF-β1 knockdown MSCs were generated by transfection with plasmid-containing lentiviruses and co-cultured with OCI-AML3 cells for 3 days. ALDH activity was measured by flow cytometry in OCI-AML3 cells cultured with TGF- β1 knockdown MSCs compared to OCI-AML3 cells cultured alone or with control BM-MSCs. Data are plotted as mean values with error bars representing standard error. For C, the Mann-Whitney U test or Student’s t test was used. For D, one-way ANOVA with Tukey's HSD post-hoc test was used.
Fig 4.
TGF-β1 non-canonical pathway induces expression of ALDH2 and other downstream targets in AML cells.
(A) OCI-AML3 and HL60 cells were treated with recombinant TGF-β1 (5 ng/mL) for 3 days. Cells were stained with ALDEFLUOR® and ALDH activity was measured. (B) Histogram representation of the percentages of ALDH+ cells in OCI-AML3 and HL60 AML cells treated with recombinant TGF-β1 in A. (C) OCI-AML3 cells were treated with recombinant TGF-β1 (5 ng/mL). RT-PCR was performed to analyze mRNA expression of ALDH and the downstream TGF-β targets WNT5A, CTGF1, and COL1A1. Relative fold increase values in gene expression were normalized to GAPDH. Data are plotted as mean values with error bars representing standard error. (D) OCI-AML3 cells were treated with recombinant TGF-β1 (5 ng/mL) for 3 days. Protein lysates were harvested and Western blotting was performed to analyze ALDH2 expression in treated cells compared to untreated controls. (E) OCI-AML3 cells were cultured in RPMI overnight. MSC cells were centrifuged and the MSC supernatant and TGF-β1 (5 ng/mL) were added to stimulate OCI-AML3 cells. Western blot analysis was performed at the indicated time points to analyze protein expression of downstream targets of the TGF-β canonical and non-canonical signaling pathways. (F) OCI-AML3 cells were treated with the p38 MAPK inhibitor SB-203580 (5 μM) in the presence or absence of recombinant TGF-β1 (5 ng/mL). Protein lysates were harvested and p38 and ALDH2 expression was analyzed by Western blotting. Due to the similarity in molecular weight of several proteins tested, samples were run on different gels. Original raw western blot images are available as S1 Raw images. All membranes were scanned using Odyssey Western blot scanner (LI-COR Biosciences®) and images were obtained using the corresponding software. For B and C, the Mann-Whitney U test and Student’s t test were used, respectively (N = 3).
Fig 5.
p38 MAPK inhibitor decreases ALDH2 expression in AML cells in the presence of TGF-β1 or stromal cells.
(A). OCI-AML3 cells were treated with the p38 MAPK inhibitor SB-23580 (5 μM) in the presence or absence of recombinant TGF-β1 (5 ng/mL). ALDH activity was measured by flow cytometry using ALDEFLUOR® assay in comparison to untreated controls. (B) OCI-AML3 cells were cultured with or without BM-MSCs and treated with p38 MAPK as in A. ALDH activity was measured in treated cells in comparison with untreated controls. Data are plotted as mean values with error bars representing standard error via one-way ANOVA with Tukey's HSD post-hoc test (N = 3).
Fig 6.
ALDH2 inhibitors decrease stroma-induced ALDH activity in AML cells and sensitize them to chemotherapy.
(A) OCI-AML3 cells were cultured and treated with increasing concentrations of the ALDH2 inhibitor diadzin. ALDH activity was measured by ALDEFLUOR® assay using flow cytometry. OCI-AML3 cells were co-cultured with MSCs (200 000 cells/well density) and treated with diadzin (5 μM) for 3 days in the presence or absence of recombinant TGF-β1 (5 ng/mL). ALDH activity was measured as described above. (B) The same experiments were performed as in A, using CVT-10216 (at increasing concentrations and 1 μM) instead of diadzin. (C) OCI-AML3 cells were cultured alone or with MSCs (100 000 cells/well density) and treated with cytarabine (2.5 μM) alone or in combination with CVT-10216 (2 μM) for 48 hours. Cell samples were stained with Annexin V and analyzed by flow cytometry to measure treatment-induced cell death. Data are plotted as mean values with error bars representing standard error. (D) Simplified schematic representation of the interaction between stromal cells and AML cells leading to ALDH2 expression. BM-MSCs secrete TGF- β1, which induces ALDH2 expression in AML cells through the non-canonical/p38-dependent pathway, thereby promoting leukemogenesis. For A, B, and C one-way ANOVA with Tukey's HSD post-hoc test was used (N = 3).