Three milliliters of bone marrow cells were obtained from normal donors aged between 20 and 50 years old who had provided written consent under the Ethics Committee of Sun Yat-sen University, which approved the protocol according to the Declaration of Helsinki. To isolate MSC, bone marrow mononuclear cells were isolated via density gradient centrifugation over Percoll (Amersham Pharmacia Biotech, Little Chalfont, UK). Immediately after centrifugation, isolated mononuclear cells were resuspended at 5×10³ cells/cm² in α-modified eagle's medium (αMEM) with high glucose concentration (Gibco, UK), 10% fetal bovine serum (FBS; Lonza, UK), and 1% penicillin-streptomycin (Gibco). Cells were cultured at 37°C in a humidified atmosphere containing 5% carbon dioxide. After 72 h, non-adherent cells were removed. When the cells were 70–80% confluent, adherent cells were trypsinized and expanded for 3–5 weeks. Before use in experiments, the identity of MSC was verified by checking for positivity of CD105, CD106, CD73,CD90,CD144, HLA-class I, and the lack of expression of CD45, CD34, CD133.

For co-culture experiments, mesenchymal stem cells were seeded the day before the experiment onto 6-well plates (Corning Life Sciences) at a concentrationof 1×10⁵ cells/well and incubated at 37°C in 5% CO₂. After confirming the confluence of the stromal layer by phase contrast microscopy, K562 or NB4 cells were added onto the MSC layers separately at a ratio of 20:1. For assessment of MSC-derived drug resistance, cells were treated with As₂O₃ at the indicated time points. Next, K562 or NB4 cells were collected from the top layer leaving the adherent stromal layer intact. Then, K562 and NB4 cells were assayed for cell viability. MSC contamination, assessed by FACS as the fraction of CD19-negative cells, was always <1%. For experiments with AMD3100 and As₂O₃ treatment, K562 cells were pre-incubated for 4 h with AMD3100 and seeded onto confluent marrow stromal cell layers. After 4 h, As₂O₃ was added for a further 24 h. Then, the K562 cell layer was vigorously washed off and collected as previously described.

A short hairpin RNA (shRNAs) plasmid against human RUNX3 and control shRNA were purchased from Joekai Biotechonology LLC (Shanghai, China) and transduced into cells by electroporation. Electroporation was performed with a Gene Pulser Xcell Unit (Bio-Rad Laboratories, Inc., Hercules, CA, USA) under conditions of 250 V, 1200 µF, and exponential wave. Clones were selected with G418 (500 µg/ml). The RUNX3 expression plasmid PCDNA3.1/RUNX3 was constructed using full-length RUNX3 cDNA, as described previously (37). K562 cells were transfected using Lipofectamine LTX from Invitrogen. Empty vectors were transfected as controls.

K562 cells expressing scrambled or RUNX3-targeting shRNA were incubated with As₂O₃. After incubation, the cells were harvested and resuspended in 500 µl of cold PBS and cold ethanol (1.5 ml) for 2 h at 4°C, followed by incubation with 0.1% sodium citrate containing propidium iodide (PI) 0.05 mg and 50 µg RNase for 30 min. Finally, the cell cycle analysis was detected by flow cytometry (FC500; Beckman Coulter, Miami, FL, USA), and the proportion of cells within the G0/G1, S, and G2/M phases of the cell cycle were analyzed using the MultiCycle AV DNA Analysis software (Phoenix Flow Systems).

Total RNA was extracted from each sample using the Total RNA Extraction kit (Invitrogen), following the manufacturer's instructions. The concentration of RNA was measured by spectrophotometry. Total RNA was reverse transcribed to cDNA with reverse transcriptase reagents (Takara, Nanjing, China), according to the manufacturer's protocol. Specific primers for RUNX3 and GAPDH genes were designed based on sequence data from the GenBank database. The primers were purchased from Invitrogen Biological Engineering Technology & Services Co., Ltd. For human RUNX3 transcripts, 5′-AGGCATTGCGCAGCTCAGCGGAGTA-3′ was used as a sense primer and 5′-TCTGCTCCGTGCTGCCCTCGCACTG-3′ as an antisense primer. For the human GAPDH transcripts, 5′-TCACAGGGGTTCCCTTCTCTC-3′ was used as a sense primer and 5′-CACTTCTTTGTGCCATCCATGG-3′ as an antisense primer. Conditions for all PCRs were optimised on an iCycler iQ (Bio-Rad Laboratories, Inc.), and the optimum annealing temperature was 60°C.

Cells were incubated on poly-L-lysine-coated slides for 30 min, washed gently with PBS and fixed in 10% methanol at room temperature. After washing with PBS, the cells were permeabilized with 0.1% Triton in PBS for 60 min at room temperature. The cells were then incubated with the rabbit anti-RUNX3 antibodies (BD Biosciences) overnight at 4°C. After washing with PBS, the slides were incubated with the secondary antibody, Alexa 555-conjugated goat anti-rabbit immunoglobulin G (IgG; Molecular Probes), for 2 h and then washed with PBS. Nuclei were stained with 10 nM DAPI, and cells were then subjected to fluorescence microscopy.

Morphological evaluation of apoptotic cell death was performed using the Hoechst 33258 kit (KeyGen Biotech), according to the manufacturer's instructions. Cells were cytospun, stained with Hoechst 33258, and observed with light microscopy.

After culturing mesenchymal stem cells in 24-well plates for 24 h, medium was collected and concentrated by ultrafiltration. CXCL12 levels in the medium were assayed with a Human CXCL12/SDF-1 α immunoassay kit (R&D Systems, Abingdon, UK) according to the manufacturer's instructions. Samples were run in triplicate.

Monoclonal antibodies against human CXCR4-PE (BD Pharmingen, DakoCytomation) were used for flow cytometry analysis. PE-conjugated IgG1 and IgG2a control monoclonal antibodies were from Cell Signaling Technology, Inc.

For quantitative transmigration assays, K562 cells (5×10⁴ cells) were placed in the upper chamber of a Transwell system (24-well plate, 5 µM-pore filter; Corning Life Sciences) in a total volume of 150 µl DMEM medium. Next, 600 ml of the supernatant was concentrated 10-fold, and placed in the lower chamber of the Transwell. Further control wells assessed the chemotactic activity of recombinant CXCL12 (rCXCL12) at concentrations of 100 ng/ml. Cells were harvested from the lower chamber after 3 h and counted using a hemocytometer.

Apoptosis was measured using the Calbiochem assay kit (Calbiochem, San Diego, CA, USA) according to the manufacturer's instructions. K562 cells expressing scrambled or RUNX3-targeting shRNA were incubated with As₂O₃. After incubation, cells were washed with phosphate-buffered saline (PBS) and stained using phycoerythrin-labeled Annexin V. Numbers of both RFP- and Annexin V-positive cells were determined by flow cytometry. In some cases, the Annexin V-propidium iodide staining kit (Calbiochem) was used instead, according to the manufacturer's protocol.

Cells were cultured with the indicated concentrations of As₂O₃ for the specified times, harvested, washed, and lysed using RIPA buffer containing protease inhibitors. After normalization for total protein content (60 µg/lane), the samples were subjected to 12% SDS-PAGE and then transferred to a PVDF membrane. After blocking with 5% non-fat dry milk and 0.1% Tween-20 in Tris-buffered saline, the membranes were incubated with the following primary antibodies: Mcl-1 (Cell Signaling Technology, Inc.), RUNX3 (BD Pharmingen), p21, Bax, Bcl-2, cleaved caspase-3, cleaved caspase-9, and β-actin (all from Cell Signaling Technology, Inc.). After extensive rinsing with TBST, the membranes were incubated with HRP-conjugated anti-mouse or rabbit secondary antibodies (Cell Signaling Technology, Inc.), and proteins were detected with the use of an enhanced chemiluminescence system (EMD Millipore, Billerica, MA, USA) and captured on a light-sensitive imaging film.

All experiments were repeated in triplicate. The data are reported as the mean ± SEM. Statistical analyses were performed using Student's t-test or one-way ANOVA. Probability (P) value of 0.05 was considered to be significant (SPSS v13.0 for windows).

To determine whether the expression of RUNX3 is induced by As₂O₃ treatment, western blot analysis was performed to detect the level of total RUNX3 and the RUNX3 target gene, p21. As shown in (Fig. 1A), As₂O₃ treatment resulted in the elevation of total RUNX3 in K562 and NB4 cells in a dose-dependent manner. Elevation of RUNX3 levels was observed at 3 h post-treatment with 5 µM As₂O₃ in NB4 cells, and expression levels reached a peak at 9 h post-treatment. An increase in RUNX3 expression was detected at 1 h following treatment with 3 µM As₂O₃ in K562 cells, and expression levels reached a peak at 4 h post-treatment (Fig. 1B). Similarly, we also observed that the expression of RUNX3 mRNA was enhanced after As₂O₃ treatment in both K562 and NB4 cell lines. Protein levels of p21 were elevated in a dose- and time-dependent manner and corresponded to the elevation of RUNX3 observed following As₂O₃ treatment. Immunofluorescence staining revealed that RUNX3 was located mainly in the cytoplasm before As₂O₃ treatment and accumulated in the nucleus following As₂O₃ treatment (Fig. 1C). Consistent with the results obtained using NB4 cells, our analysis of K562 cells and primary APL blast cells from three patients revealed that RUNX3 localized in the cytoplasm prior to As₂O₃ treatment and translocated into the nucleus following As₂O₃ treatment. It has been previously reported that As₂O₃ treatment resulted in elevation in cellular ROS stores and production. In our studies, treatment of cells with As₂O₃ also resulted in the generation of ROS (Fig. 1D). This ROS induction seems to be necessary for As₂O₃-dependent elevation of RUNX3 because pre-treatment of cells with NAC, a scavenger of ROS, resulted in the inhibition of As₂O₃-induced expression of RUNX3 (Fig. 1E).

We next determined the role of RUNX3 in As₂O₃-induced apoptosis using RNA interference to reduce RUNX3 mRNA. We established two stable clones expressing shRNA specific to RUNX3, designated A2 and B4, and a scramble shRNA-expressing clone, designated SCR. Western blot analysis with anti-RUNX3 antibody revealed that endogenous RUNX3 protein levels were markedly diminished in A2 clones and significantly reduced in the B4 clone compared with SCR (Fig. 2A). Cell growth assays showed that A2 cells exhibited significantly enhanced growth compared with SCR cells. As₂O₃ inhibited the growth of SCR cells, but this growth inhibition was compromised in the RUNX3 shRNA-transduced cells treated with As₂O₃ (Fig. 2B). Next, we examined the involvement of RUNX3 in As₂O₃-induced apoptosis and cell cycle arrest. As shown in (Fig. 3A), the ratio of Annexin V and propidium iodide-positive cells was increased in a time-dependent manner, and 21.6% of the SCR cells were positive on the second day after As₂O₃ treatment. By contrast, apoptosis was significantly reduced during the observation periods up to day 2 in RUNX3 knockdown clones (Fig. 3A and B). Similar results were obtained from our Hoechst 33258 staining analysis (Fig. 2C and D). Fig. 6H shows that re-expression of RUNX3 enhanced sensitivity of K562 cells to As₂O₃ in a dose-dependent manner. Transduction of control constructs alone did not alter cellular response to As₂O₃. The percentages of cells in G1, S, and G2/M phase are illustrated in Fig. 3C and D. In SCR cells, treatment with As₂O₃ increased the fraction of cells in G2/M progressively from 8.95 to 27.61%. However, suppression of RUNX3 expression by RUNX3 shRNA attenuated As₂O₃-induced cell cycle arrest. Consistent with these results, p21 expression was suppressed in the A2 and B4 clones after As₂O₃ treatment. A decrease in the Mcl-1 protein was observed after treatment with As₂O₃ in SCR clones, though this reduction in Mcl-1 levels was compromised in A2 and B4 clones (Fig. 3E).

To assess whether MSC could protect leukemia cells from As₂O₃-induced apoptosis, K562 and NB4 cells were exposed to a range of As₂O₃ concentrations in the presence or absence of MSC. After 24 and 48 h, cells were analyzed by probing for Annexin V and propidium iodide (Fig. 4A). While exposure of NB4 cells to As₂O₃ for 24 or 48 h in control cultures induced apoptosis in a dose-dependent fashion (6.5±2, 9.2±3.5 and 13.5±3.7% of Annexin V/propidium iodide⁺ cells for 24 h; 8±2.8, 15.9±4.1 and 38.1±5.4% for 48 h for As₂O₃ at 3, 6 and 9 µM, respectively), the proportion of apoptotic NB4 cells in the presence of MSC was significantly reduced (6.7±1.8, 7.1±3 and 8.1±2.6% for 24 h; 7.7±3.3, 8.4±3.5 and 16±4.2% for 48 h, with As₂O₃ at 3, 6 and 9 µM, respectively; P=0.003, 0.015 and 0.024, NB4 vs. NB4 + MSC with 3, 6 and 9 µM As₂O₃, respectively) (Fig. 4A). As shown in Fig. 4A, the protective effect of MSC was also observed for K562 cells. No significant protection was observed when CML cells were cultured and treated on a monolayer of endothelial cells.

We also determined whether MSC affect As₂O₃-induced cell cycle arrest. As shown in (Fig. 4B and C), As₂O₃ treatment of K562 cells resulted in a dose-dependent accumulation of cells with 4N and <2N sub-G1 DNA content, which suggests induction of G2/M-phase cell cycle arrest and apoptosis. MSC robustly reduced the proportion of G2/M-phase and sub-G1 cells. MSC did not significantly affect cell cycle arrest induced by As₂O₃ in NB4 cells. To determine if MSC modulate the level of ROS, K562 and NB4 cells incubated with or without As₂O₃, either in the presence or absence of MSC, were analyzed for ROS by DCFDA staining. As shown in Fig. 1D, the level of ROS in NB4 cells increased after treatment with 6 µM As₂O₃ in a time-dependent manner; however, MSC were able to attenuate the quantity of ROS produced following As₂O₃ treatment.

The intrinsic, mitochondrial pathway is the principal mechanism of As₂O₃-induced apoptosis. In the mitochondrial apoptotic pathway, the ratio of expression of pro-apoptotic Bax to that of anti-apoptotic Bcl-2 ultimately results in either cell death or survival. We determined the effect of MSC on the Bax/Bcl-2 ratio. Upregulation of Bax expression, as well as downregulation of Bcl-2 expression induced by As₂O₃, was blocked in the presence of MSC (Fig. 5D).

MSCs have been found to protect NB4 and K562 cells from As₂O₃-induced apoptosis, and this protection mechanism is related to the expression of apoptotic proteins. We measured protein levels of cleaved caspase-3 and cleaved caspase-8. Consistent with the previous report, As₂O₃ treatment induced the activation of caspase-8, whereas this modulation was not noted in the presence of MSC. As₂O₃-induced death of cells is associated with strong activation of caspase-3, which was significantly inhibited when cells were treated in the presence of MSC (Fig. 5D). Additionally, there were low levels of Mcl-1 protein in NB4 and K562 cells treated with As₂O₃. However, in the presence of MSC, the decrease in Mcl-1 levels was compromised (Fig. 5D).

We have demonstrated that RUNX3 is required for As₂O₃-induced apoptosis. To determine whether MSC modulate the expression of RUNX3, western blot analysis was used to detect the level of total RUNX3 in K562 and NB4 cells treated with As₂O₃ in the presence or absence of MSC (Fig. 5A). The presence of MSC reduced RUNX3 protein expression of cells treated with different concentrations of As₂O₃ compared to the cultures in which the leukemic cells were exposed to As₂O₃ without MSC. The expression of p21 induced by As₂O₃ was also inhibited in the presence of MSC. We next studied the subcellular distribution of RUNX3 by performing a western blot analysis on nuclear extracts from NB4 and K562 cells treated with As₂O₃ for different time intervals in the presence or in the absence of MSC. In K562 cells, the expression of RUNX3 was induced in 2 h, peaking at 4 h. In NB4 cells, RUNX3 expression was induced later and peaked at 9 h. In the presence of MSC, the level of RUNX3 in the nuclear fraction was significantly reduced at the indicated time in NB4 and K562 cells (Fig. 5B and C).

CXCL12/CXCR4 signaling is an important mediator for leukemia-stroma interaction, and this interaction has consequences on tumor survival. Therefore, we determined whether CXCL12/CXCR4 modulates the expression of transcription factor RUNX3 in K562 cells. The expression level of CXCR4 was analyzed in K562 cells by flow cytometry (Fig. 6A). K562 cells showed higher expression of CXCR4 when co-cultured with MSC. Furthermore, this increase in CXCR4 expression was enhanced in the presence of As₂O₃. The expression of CXCR4 peaked 10 min post-treatment with As₂O₃ when K562 cells were co-cultured with MSC. Second, the production of CXCL12 by MSC was analyzed in MSC supernatants by western blot analysis (Fig. 6B). To determine whether the CXCL12/CXCR4 interaction had functional activity, we investigated the chemotactic properties of MSC on K562 cells in a transmigration assay. The K562 cells migrated toward a CXCL12 gradient, demonstrating that CXCR4 was fully functional. As shown in Fig. 6C, MSC supernatant exhibited potent chemoattractive activity on K562 cells, as 24±3.2% of K562 cells migrated toward undiluted supernatant compared to 0.95±0.06% of cells that migrated in the presence of media (P=0.013). Blocking CXCR4 led to a significant reduction in chemoattractive activity in response to MSC, as well as the chemokine CXCL12 (P=0.014; Fig. 6C). All of these data indicate the presence of a functional CXCR4/CXCL12 signaling between CML cells and MSC and that this signaling is important for promoting migration.

To assess whether CXCR4/CXCL12 could modulate the expression of RUNX3, the expression of RUNX3 was analyzed in the presence of CXCL12 by western blot analysis. As shown in Fig. 6D, CXCL12 induced a significant reduction of RUNX3 mRNA expression in K562 cells. Remarkably, treatment of K562 cells with the CXCR4 antagonist AMD3100 induced RUNX3 expression (Fig. 6E). Moreover, the expression of RUNX3 was upregulated after treatment with AMD3100 in the presence of MSC, indicating that the inhibition of CXCR4/CXCL12 signaling was rescuing the expression of RUNX3.

Mesenchymal stromal cells have been found to protect leukemia cells from As₂O₃-induced apoptosis. We tested the hypothesis that the overexpression of RUNX3 could restore the effect of As₂O₃ on K562 cells. K562 cells were transfected with a RUNX3 expression vector or an empty control vector. The expression of RUNX3 in these cells was confirmed by western blot analysis (Fig. 6F). When K562 cells transfected with RUNX3 were co-cultured with MSC, the Annexin V and PI-positive cells treated with increasing concentrations of As₂O₃ was significantly increased compared to the cultures in which leukemic cells transfected with empty vector were exposed to As₂O₃ in the presence of MSC. Similar results were observed from APL primary blast patients (Fig. 6H). In addition, we found that p21 protein expression was significantly increased after exposure to As₂O₃ in cells transfected with the RUNX3 expression vector in the presence of MSC, but not for cells transfected with control vector (Fig. 6G). The influence of overexpression of RUNX3 on the expression of anti-apoptotic proteins in the presence of the MSC was determined by immunoblotting. Mcl-1 protein upregulation was observed in K562 cells grown with stromal cells. However, cells transfected with RUNX3 downregulated stroma-induced Mcl-1 levels in K562 cells (Fig. 6G).