Human gastric cancer cell lines Kato III, AGS, N87, MKN45, and normal human lung fibroblast cell line WI-38 were purchased from American Type Culture Collection and cultured in DMEM (HyClone, Logan, UT), supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT). miRNA miR-34a, b, c mimics, antagonists, and negative control miRNA mimic (NC mimic) were obtained from Dharmacon (Chicago, IL) with the sequences for hsa-miR-34a: 5'-uggcagugucuuagcugguugu-3'; hsa-miR-34b: 5'-caaucacuaacuccacugccau-3'; hsa-miR-34c: 5'-aggcaguguaguuagcugauugc-3'. Bcl-2 3'UTR luciferase reporter plasmid or its mutant were kindly provided by Dr. Eric Fearon of the University of Michigan 10.

Gastric cancer cells were transfected 24 hours after being seeded in 6-well plates. miRNA mimics (100 pmol) in 200 μl of serum-free, antibiotic-free, medium were mixed with 5 μl of Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA) dissolved in 200 μl of the same medium and allowed to stand at room temperature for 20 min. The resulting 400 μl transfection solutions were then added to each well containing 1.6 ml of medium. Six hours later, the cultures were replaced with 2 ml fresh medium supplemented with 10% FBS and antibiotics. For Western blot, cells were collected after an additional 48 hours.

The feline immunodeficiency virus (FIV) lentiviral system expressing miR-34a (miR-34a-MIF) or vector control (MIF), as well as their lentiviral packaging system, were purchased from System Biosciences (SBI, Mountain View, CA). Gastric cancer cells were infected with the FIV lentiviral system expressing miR-34a (miR-34a-MIF) or vector control (MIF), according to the manufacturer's instructions, and stable cells were obtained by antibiotic selection (Zeocin 50 μg/mL, Invitrogen).

Kato III cells were transfected in 6-well plates with 2 ug of Bcl-2 3'UTR luciferase reporter plasmid or its mutant, and 2 ug of the control β-galactosidase plasmid per well, using Lipofectamine 2000 (Invitrogen). Cells in each well were also co-transfected with 100 pmol of each miR-34 mimics or NC mimic as indicated, using Lipofectamine 2000. Luciferase assays were performed 24 hrs after transfection using Bright-Glo Luciferase Assay System (Promega). Luciferase activity was normalized relative to β-galactosidase activity detected by the β-galactosidase Assay System (Promega). In each case, Mutant Bcl-2 3'UTR indicates the introduction of alterations into the seed complementary sites of Bcl-2 3'UTR 10. Luciferase activity was normalized relative to β-gal activity (Promega).

To determine the levels of protein expression, cells were harvested and lysed in a RIPA lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% SDS, 1% NP-40, 0.25% Sodium deoxycholate and 1 mM EDTA) with freshly added protease inhibitor cocktail (Roche) for 15 min on ice, then centrifuged at 13,000 rpm for 10 min. Whole cell extract was measured for total protein concentration using Bradford reagent (Bio-Rad, Hercules, CA), and proteins were resolved by SDS-PAGE (Bio-Rad). After electrophoresis, the proteins were electrotransferred to nitrocellulose membranes (Bio-Rad), blocked with 5% skimmed milk, probed with the relevant primary antibody followed by HRP (horseradish peroxidase) conjugated secondary antibody (Pierce, Rockford, IL), and detected with the SuperSignal West Pico chemiluminescence substrate (Pierce). Intensity of the desired bands was analyzed using TotalLab software (Nonlinear Dynamics, Durham, NC).

Quantitative real-time PCR was performed to determine the expression levels of potential miR-34 target genes. 24 hours after miR-34 mimic transfection of Kato III cells (100 pmol per well in 6-well plates), potential target genes' mRNA levels were measured by qRT-PCR with TaqMan SYBR Green PCR System (Applied Biosystems). Briefly, total RNA was extracted from the transfected cells using TRIZOL (Invitrogen) according to the manufacturer's instructions. Reverse transcription was performed using a TaqMan Reverse Transcription Kit (Applied Biosystems). For qRT-PCR, 1 μl of gene primers with SYBR Green (Applied Biosystems) in 20 μl of reaction volume was applied. Primers were designed as:

Bcl-2

HMGA2

Notch1

Notch2

Notch3

Notch4

β-actin

For cell cycle analysis by flow cytometry, Kato III cells were transfected with miR-34 mimics or NC mimic in 6-well plates, trypsinized 24 hours later and washed with phosphate-buffered saline, and fixed in 70% ethanol on ice. After centrifugation, cells were stained with 50 μg/ml propidium iodide and 0.1 μg/ml RNase A, and analyzed by flow cytometry using a FACStar Plus™. Each histogram was constructed with data from at least 5,000 events. Data were analyzed to calculate the percentage of cell population in each phase using CellQuest software (Becton Dickinson).

Caspase activation of transfected Kato III cells was determined following the instructions of a Caspase-3 activation assay kit (BioVision, Mountain View, CA). 24 hours after transfection, cells were lysed and whole cell lysates (20 μg) were incubated with 25 μM fluorogenic substrate DEVD-AFC in a reaction buffer (containing 5 mM DTT) at 37°C for 2 h. Proteolytic release of AFC was monitored at λex = 405 nm and λem = 500 nm using a fluorescence microplate reader (BMG LABTECH, Durham, NC). Relative increase of fluorescence signal was calculated by dividing the normalized signal in each treated sample with that of NC mimic as 100.

For cytotoxicity assay, the water-soluble tetrazolium salt WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt] uptake method was employed, using CCK-8 reagent (Dojindo, Gaithersburg, MD). Briefly, Kato III cells were transfected with miR-34 mimics or NC mimic for 24 h, plated in 96-well plates (5,000 cells/well), and treated with serially diluted chemotherapeutic agents in triplicate. After 96 h incubation, 20 μl/well CCK-8 reagent was added and incubated at 37°C for 1–3 h. Optical density was measured at 450 nm and 650 nm using a microplate reader (BMG LABTECH). Final absorbance was obtained by subtracting the absorbance at 450 nm from that at 650 nm, and cell viability (%) was normalized by dividing final absorbance of treated samples with that of the untreated control. IC₅₀, the drug concentration that inhibits 50% cell growth, was calculated by GraphPad Prism 5.0 (San Diego, CA), as we described previously 15.

Cells were suspended in serum-free culture medium DMEM containing 1% N2 supplement, 2% B27 supplement, 1% antibotic-antimycotic (Invitrogen), 20 ng/ml human FGF-2 (Sigma, Saint Louis, MO), and 100 ng/ml EGF (Invitrogen), and plated in 24-well ultra-low attachment plates (Corning, Corning, NY) at 2,000 cells per well. 7–10 days later, plates were analyzed for tumorsphere formation and were quantified using an inverted microscope (Olympus) at 100×, 200×, and 400× magnification. For subsequent quantification of cell numbers per tumorsphere, tumorspheres were collected and filtered through a 40 um sieve (BD Biosciences, San Jose, CA), and disassociated with 2.5% trypsin, while the viable cells were counted with trypan blue exclusion.

Data were analyzed with Student's two-tailed t-test or one-way ANOVA, using GraphPad Prism 5.0 software (GraphPad Prism, San Diego, CA). P < 0.05 was defined as statistically significant.

We have examined the four human gastric cancer cell lines, Kato III, AGS, N87, and MKN45, for their expression levels of Bcl-2 as well as other members of the Bcl-2 family proteins. As shown in Figure 1, Kato III cells have a high level of Bcl-2, while AGS and MKN-45 cells have very low levels of Bcl-2 (undetectable by Western blot). We next examined these gastric cancer cell lines for the expression level of miR-34 and target genes using qRT-PCR. Kato III cells have the lowest levels of both pri-miR34a and mature miR-34a, and the highest expression levels of target genes Bcl-2, Notch1, and Notch4 (Figure 2). Since miR-34 is a downstream target of the p53 pathway and Bcl-2 is a direct target of miR-34, our data with Kato III are consistent with the cells' p53-mutant status, i.e., Kato III has mutant p53, the lowest level of miR-34, and the highest level of Bcl-2. Therefore, we focused on this cell line for the current study of the effect of miR-34 restoration.

For miR-34 restoration, we transfected the Kato III cells with miR-34 mimics. As shown in Figure 3, Western blot analysis revealed that transfection of miR-34 mimics downregulated target gene Bcl-2 expression at the protein level, but had no obvious effect on Bcl-xL and Mcl-1 expression, indicating that the Bcl-2 knockdown by miR-34 mimics was sequence-specific. As a target of miR-34, Bax was also downregulated by miR-34. Western blot results were validated by qRT-PCR analysis (Figure 4). More importantly, other potential miR-34 target genes were inhibited in addition to Bcl-2. As shown in Figure 4, Notch1 and HMGA2 were inhibited by all three miR-34a, b, c mimics, while miR-34b mimic inhibited Notch2 and 4, and miR-34c mimic inhibited Notch1-4. Notch1-2 knockdown by miR-34 mimics has been confirmed by Western blot (data not shown).

To evaluate whether the transfected miR-34 mimics were functional, we carried out the Bcl-2 3'UTR reporter assay as described by Bommer et al. 10. KATO3 cells were transfected with Bcl-2 3'UTR luciferase reporter plasmid or its mutant, plus the control β-galactosidase plasmid and 100 pmol of each miR-34 mimic or NC mimic. As shown in Figure 5, the transfected miR-34 mimics effectively inhibited luciferase reporter gene expression, which is controlled by Bcl-2 3'UTR in the promoter region. However, mutation in the Bcl-2 3'UTR complimentary to the miR-34 root sequence abolished this effect, indicating that the observed reporter activity is miR-34 sequence-specific. The results demonstrate that the transfected miR-34a, b, c are functional, and confirm that Bcl-2 is a direct target of miR-34, consistent with earlier reports 81016.

After validating that the transfected miR-34 mimics were functional, we carried out cellular assays to examine the effects of miR-34 restoration on gastric cancer cells. First we evaluated the effect of miR-34 mimics on cell cycle. As shown in Figure 6A, the miR-34 mimics induced an accumulation of Kato III cells in G1 phase and a reduction of cells in S phase, consistent with other reports on miR-34 restoration in various tumor models 481012131617. This effect on cell cycle is similar to that of p53 restoration as we previously reported 181920212223, indicating that miR-34 restoration can restore p53 signalling, at least in part, in the cells lacking a functional p53 pathway. Since part of the p53 tumor-suppressing function is via promoting apoptosis 1920, we next examined the effect of miR-34 restoration on apoptosis. As shown in Figure 6B, transient transfection of miR-34 mimics resulted in significantly increased caspase-3 activation, a key indication of the cells undergoing apoptosis.

Bcl-2 is a key inhibitor of apoptosis and protects cancer cells from apoptosis induced by chemotherapeutic agents 2425. Bcl-2 is a direct target of miR-34, and our data have shown that miR-34 restoration inhibits Bcl-2 expression. We therefore investigated whether miR-34 restoration could sensitize gastric cancer cells with high a level of Bcl-2 to chemotherapy. We chose four chemotherapeutic agents, doxorubicin, cisplatin (CDDP), gemcitabine, and docetaxel, all of which are used in gastrointestinal cancer chemotherapy. As shown in Figure 7A, miR-34 restoration in Kato III cells rendered the cells 2-3-fold more sensitive to the four chemotherapeutic agents, as compared with the cells transfected with the control mimic, based on IC50 data. We have also used siRNA specific to Bcl-2 in these cytotoxicity assays and observed a similar, 2-3-fold chemosensitization in the Kato III cells transfected with Bcl-2 siRNA (data not shown). However, for gastric cancer MKN-45 cells that have a low level of Bcl-2 and a high level of miR-34, miR-34 restoration showed no chemosensitization (Figure 7B). The same results were observed with Bcl-2 siRNA transfection (data not shown). Our data demonstrate that miR-34 restoration can chemosensitize those gastric cancer cells that have high levels of Bcl-2 and low basal levels of miR-34, which are dependent on Bcl-2 for survival and drug resistance.

To evaluate the long-term effects of miR-34 restoration, we have employed a lentiviral system to express miR-34a and have generated stable cells. Briefly, Kato III cells were infected with the feline immunodeficiency virus (FIV) lentiviral system expressing miR-34a (miR-34a-MIF) or vector control (MIF), according the manufacturer's instructions, and stable cells were obtained by antibiotic selection (Zeocin 50 μg/mL, Invitrogen). For cell growth analysis, the stable cells were plated in a 24-well plate with equal cell density. Cells in triplicate were collected by trypsinization, and viable cells were counted by Trypan Blue exclusion at 24 hour intervals over 4 days using a Coulter cell counter (Beckman, Fullterton, CA). As shown in Figure 8, lentiviral restoration of miR-34a in Kato III cells significantly delayed cell growth (P < 0.001, n = 3), a biological activity similar to that of the p53 restoration we reported previously 19202122.

In our attempts to isolate gastric cancer stem cells, the multiple cell surface markers we used did not provide robust, tumorsphere-forming, tumor-initiating cells or tumor stem cells. Since no cellular markers for gastric cancer stem cells have been widely accepted thus far, in the current study we employed tumorsphere culture to explore whether there is any link between miR-34 and tumorsphere-forming cells.

Tumorsphere culture, in which cells grow in suspension under non-adherent culture conditions to become tumoroid spheres, has been widely used to assess the self-renewal potential of stem cells and cancer stem cells 3142627. Briefly, in our study the stable clones from Kato III cells infected with miR-34a-MIF or control vector MIF were plated for tumorsphere culture in ultra-low adhesion plates. 7–10 days later, tumorspheres were observed under a microscope and quantified. As shown in Figure 9, restoration of miR-34 by MIF lentiviral system inhibited Kato III tumorsphere formation and growth; the stable cells with functional miR-34a restoration had significantly fewer tumorspheres, and the formed tumorspheres were significantly smaller, as compared with that of the MIF control (P < 0.001, Student's t-test, n = 3). This result was confirmed by a separate tumorsphere study in a 96-well based single cell tumorsphere culture, in which the tumorspheres were seen to be from single cells, not from cell aggregates. Our data provide the first evidence that miR-34 is able to inhibit tumorsphere formation and growth in p53-mutant gastric cancer cells, implying that miR-34 might play a role in the self-renewal of gastric cancer cells, presumably gastric cancer stem cells.