Gibco™ Dulbecco's modified Eagle's medium (DMEM), RPMI-1640, penicillin-streptomycin solution, and L-glutamine were purchased from Thermo Fisher Scientific, Inc. (Waltham, MA, USA). Hyclone™ fetal bovine serum (FBS) was obtained from GE Healthcare Life Sciences (Logan, UT, USA). Human cancer cell lines including breast (MDA-MB-231 and MCF-7), colorectal (SW480 and SW620), and liver (HepG2 and SK-Hep-1) were purchased from the American Type Culture Collection (ATCC; Rockville, MD, USA). MDA-MB-231, MCF-7 and HepG2 cells were cultured in DMEM supplemented with 10% FBS. SW480, SW620 and SK-Hep-1 cells were maintained in RPMI-1640 medium supplemented with 10% FBS. Human normal mammary epithelial cells (HMECs) and its medium (Mammary Epithelial Cell Growth Medium, MEGM) were purchased from Lonza (Walkersville, MD, USA) and cultured as recommended by the manufacturer. Human normal colon epithelial cells (CCD 841 CoN) and human normal liver epithelial cells (THLE-3), from ATCC, were provided by Dr Jutamaad Satayavivad, Chulabhorn Research Institute, Thailand. CCD 841 CoN was cultured in DMEM supplemented with 10% FBS, and 1% L-glutamine while THLE-3 was cultured in DMEM supplemented with 10% FBS and 25 mmol/l HEPES (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany). All culture media were supplemented with 100 U/ml penicillin and 100 mg/ml streptomycin. The cells were maintained in humidified atmosphere of 95% air and 5% CO₂ at 37°C.

The levels of O-GlcNAc, OGT and β-actin were determined by immunoblotting using monoclonal mouse anti-human O-GlcNAc antibody, RL2 (ab2739, Abcam, Cambridge, MA, UK), monoclonal rabbit anti-human OGT antibody (O6264, Sigma-Aldrich; Merck KGaA) and monoclonal mouse anti-human β-actin antibody (mAb3700, Cell Signaling Technology, Beverly, MA, USA) as previously described (13). Briefly, cells were lysed in RIPA buffer containing 1% protease inhibitor cocktail (Sigma-Aldrich; Merck KGaA) and 20 µmol/l Thiamet-G (Sigma-Aldrich; Merck KGaA). Protein samples (20 µg) were separated by 10% SDS-PAGE and transferred to PVDF membranes (Millipore, Bedford, MA, USA). Duplicate gels with equal amount protein loading were performed; one for O-GlcNAc modified proteins and another for OGT and β-actin. The membranes were probed with specific primary antibodies at 1:1,000. Immunoblots were developed with WesternBright™ ECL (Advansta, Menlo Park, CA, USA). The signals were captured and measured using an image analysis program (ImageQuant LAS4000; GE Healthcare, Marlborough, MA, USA). β-actin was used to compare protein loading of cell lysates.

siRNA oligonucleotides of O-GlcNAc transferase (OGT) (sense, 5′-UAAUCAUUUCAAUAACUGCUUCUGC-3′ and antisense, 5′-GCAGAAGCAGUUAUUGAAAUGAUUA-3′) and scramble negative control medium GC duplex were designed and purchased from Invitrogen; Thermo Fisher Scientific, Inc. For monolayer cultures, transfection of the siRNA oligos into cancer cells was carried out using Invitrogen™ Lipofectamine 2000™ in a forward transfection mode as described previously (11). For soft agar and anoikis resistance cultures, cells were transfected using a reverse transfection mode as described according to the manufacturer's instructions (Invitrogen; Thermo Fisher Scientific, Inc).

Cancer cell growth was assessed by monitoring cell viability throughout 5 days using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay in monolayer cultures as previously described (14). Briefly, cell suspensions were seeded into 96-well plates, cultured for 1 day, and transfected with RNA interference. Transfected cells were further cultured for 1–5 days. Then, the wells were replaced and incubated with fresh culture media containing 0.5 mg/ml MTT (Sigma-Aldrich; Merck KGaA) for 2 h at 37°C. Finally, the medium was removed and replaced with DMSO (100 µl/well). Absorbance was measured at 550 nm and subtracted with the absorbance at 650 nm, using a microplate reader. The number of viable cells was determined from the absorbance. Cultures with at least three independent wells were studied. The number of viable cells in each day was normalized by those of the Scramble siRNA at day 1, and reported as the percent relative cell growth.

Anchorage-independent cell growth assay in vitro was performed using soft agar cultures as described previously (11). Briefly, 1×104 cells (OGT knockdown and scramble cells using a reverse transfection mode) were suspended in 1 ml top agar medium (the complete medium with 0.4% agar). The cell suspension was then overlaid onto 1.5 ml bottom agar medium (the complete medium with 0.8% agar) in 6-well culture plates in triplicate. Fresh complete medium was added to plates every 3 days as a feeder layer. Once colonies were propagated (MDA-MB-231 and SW620 for 21 days, MCF-7, SK Hep-1 and HepG2 for 18 days, and SW480 for 28 days), they were stained with 0.005% crystal violet in 50% methanol for 1 h, and images of the stained wells were captured. Colony number counts and average sizes were determined by ImageJ software version 1.42I (National Institutes of Health, Bethesda, MD, USA). All images were saved in an 8 bit format. The measured area was selected by elliptical selection and the threshold image was set using the threshold tool. The mode of analyzing particles was used with parameters of size: 1-infinity and circularity: 0.00–1.00. Cultures with at least two independent wells were studied. The results were reported as the average ± standard deviation of colony number and size in OGT-knockdown cells normalized by those in the siScramble control.

Anoikis resistance of cancer cells was determined by culturing the cells in polyHEMA-coated plates as descried previously with some modifications (2). Briefly, 2×10⁵ cells (OGT-knockdown and scramble cells using a reverse transfection mode) were cultured in polyHEMA-coated plates. The polyHEMA-coated plates were prepared by soaking 30 mg/ml poly-HEMA (Sigma-Aldrich; Merck KGaA) in 95% ethanol and putting this onto the plates and drying at 37°C in an incubator, followed by extensive washing with water and UV sterilization. After culturing for 3 days, cells were photographed using an inverted phase-contrast microscope at ×10 magnification and harvested by centrifugation at 1,000 × g for 5 min. After harvesting, the cell pellets were resuspended in PBS buffer, and incubated with trypsin-EDTA solution (Gibco; Thermo Fisher Scientific, Inc.) for 10 min. The dissociated cells were collected for measurement of OGT and O-GlcNAc levels as described above. Aliquots of the collected cells were further determined for viability of anoikis-resistant cells using trypan blue dye exclusion assay. Trypsinized cells (20 µl) were mixed with 20 µl of 0.04% trypan blue. Viable and non-viable cells from four microscopic viewing areas were counted using a hemocytometer. Cultures with at least two independent wells were studied. The results were reported as the average ± standard deviation of anoikis-resistant cells (viable cells) in the OGT-knockdown cells normalized by those in the siScramble control.

2-DE was performed to measure the global protein alteration between OGT-knockdown and scramble control groups. Briefly, cultured cells (2×10⁶) were lysed with 2-D lysis buffer, 1% protease inhibitor cocktail (Sigma-Aldrich; Merck KGaA), homogenized and harvested as described previously (11). Protein samples (100 µg) were electro-separated by isoelectric focusing (IEF) in immobilized pH gradient (IPG) strips, pH 3.0–10.0, followed by 10% SDS-PAGE. After electrophoresis was performed, proteins on the gels were stained using 0.1% Coomassie brilliant blue R-250 (CBB). Protein spots were scanned using ImageScanner (Amersham Biosciences; GE Healthcare, Chicago, IL, USA) and quantitatively measured by ImageMaster 2-DE Platinum 7.0 software (GE Healthcare). Three independent experiments were performed. The relative intensity of each protein spot showing statistically significant difference was selected and reported as a fold change between two groups (OGT knockdown vs. Scramble control).

Protein spots with statistical difference in expression levels were excised from the gel, destained and enzymatically digested by trypsin (Promega, Madison, WI, USA). The digested peptides were then identified using Nanoflow liquid chromatography coupled with the amaZon speed ion trap mass spectrometry (Bruker, Billerica, MA, USA) as previously described (13). MASCOT search with NCBInr version 20130630 sequence databases (http://www.matrixscience.com) was performed in order to identify the protein spots. Search parameters were set as follows: peptide mass tolerance was 1.2 Da, MS/MS ion mass tolerance was 0.6 Da, allowance was set to 1 missed cleavage, enzyme set as trypsin, the limit of peptide charges was 1+, 2+ and 3+. Decoy was marked. Proteins with molecular weight (MW) and pI consistent to the spot on 2-DE gel and total ion scores >80 units (P-value <0.05) were considered positively identified.

The expression level of Hsp27 in the siOGT and siScramble cells was confirmed by immunoblotting (IB). Protein samples (20 µg) were separated by 10% SDS-PAGE, transferred onto PVDF membranes, and probed using monoclonal mouse anti-human Hsp27 antibody (1:2,000; cat. no. ab2790; Abcam, Cambridge, MA, USA). Immunoblots were developed with WesternBright ECL and the signals were captured as described above. β-actin was used to compare protein loading of cell lysates. The results were reported as the average band intensity ± standard deviation of Hsp27 in the OGT-knockdown cells normalized by those in the siScramble control. At least three independent experiments were performed.

Hsp27 was further confirmed to be O-GlcNAc modified using immunoprecipitation (IP). Briefly, proteins (1,000 µg) in low-salt lysis buffer were incubated with antibodies against Hsp27 (1:250; cat. no. ab2790; Abcam) and RL2 (1:200; cat. no. ab2739; Abcam). The suspensions were mixed gently by shaking in an end-over-end manner at 4°C for overnight. After that, the immune complexes were incubated with Protein A and G Sepharose (GE Healthcare) for 2 h to perform coupling reaction. After the washing steps, the immune complexes were eluted by adding 25 µl of 2X sampling buffer and heating at 95–100°C for 10 min. The eluted samples were loaded onto 12.5% SDS-PAGE and immunoblotted with RL2 and Hsp27 antibodies to determine the levels of O-GlcNAc-Hsp27.

siRNA-mediated gene silencing against OGT (Invitrogen; Thermo Fisher Scientific, Inc.) and Hsp27 (sc-29350, Santa Cruz Biotechnology, Dallas, TX, USA) was performed in MCF-7 cells. Transfections were performed in reverse transfection mode using Lipofectamine 2000™ as described above. The effects of siRNA of Hsp27, OGT and Hsp27/OGT on anchorage-independent cell growth and anoikis-resistant cells were examined using soft agar cultures and trypan blue dye exclusion assay, respectively, as described above.

The statistical analysis was conducted using unpaired Student's t-test to test for the difference between two groups. One-way ANOVA was used where appropriate followed by a Bonferroni's multiple comparison test using Prism 5.0 of GraphPad Software Inc. (GraphPad Software, Inc., San Diego CA, USA). The statistical significance was defined as P<0.05 and P<0.01.

Since several reports revealed that the O-GlcNAcylation level is increased in many types of cancer, we examined the levels of this modification and its catalyzing enzyme, OGT in cancer cell lines with different invasive capability, in comparison to their normal cells including breast (MCF-7 and MDA-MB-231 vs HMEC), colon (SW480 and SW620 vs CCD841 CoN) and liver (SK-Hep1 and HepG2 vs THLE-3). O-GlcNAcylation and OGT levels were increased in all tested cancer cell lines (≥1.5 fold), except SK-Hep1 in which its OGT level was not different but many more O-GlcNAc protein bands were obviously observed when compared to those of THLE-3 (Fig. 1). Of note, both O-GlcNAc modification and OGT levels in breast cancer cell lines appeared to be higher than those in colon and liver cancer cell lines.

Since O-GlcNAylation and OGT levels were upregulated in the cancer cell lines, transient knockdown of O-GlcNAylation by RNA interference against OGT was performed in six cancer cell lines. siOGT treatment of six cancer cell lines showed a marked reduction of both OGT and O-GlcNAcylation levels in comparison with the siScramble control group (Fig. 2A). Indeed, we observed that siOGT suppressed the OGT and O-GlcNAcylation levels in our culture system by more than 70%. However, surprisingly, OGT knockdown had little or no effect on cell viability and growth, even though experiments were performed up to 5 days of transfection (Fig. 2B). At some time points, fluctuation of cell viability was observed in siOGT treatment, but this was not consistent in a time-dependent manner.

Culturing cells in soft agar is based on the colony formation in anchorage-independent growth, which is considered the most accurate and stringent in vitro assay for detecting malignant transformation of cells. Since OGT knockdown had no effect on the cell viability and growth in the monolayer culture, we aimed to ascertain whether the reduction of this modification may affect anchorage-independent growth in vitro. Anchorage-independent cell growth assay of siOGT and siScramble cells of six cancer cell lines were performed using soft agar cultures as shown in Fig. 3A. Notably, all cancer cell lines treated with siOGT treatment, except SK-Hep1, displayed a significant reduction in colony number (Fig. 3B) and colony size (Fig. 3C) when compared to those of the siScramble cell groups, respectively. Surprisingly, OGT knockdown was unlikely to affect colony formations in SK-Hep1 cells.

Cancer cells can resist anoikis and thereby survive after detachment from their primary site and can travel through the circulatory systems as circulating cancer cells. In this study, anoikis resistance was assessed in vitro by culturing cells on polyHEMA-coated plates. As shown in Fig. 4A, all cancer cells were able to grow as floating spheroids in the medium culture. Among the six cancer cell lines, OGT knockdown statistically decreased viable cells only in the MCF-7, SW480 and SW620 cell lines while it exhibited a weaker effect on MDA-MB-231 and HepG2 cells and no effect was observed in the SK-Hep1 cells (Fig. 4B). We also confirmed that, under the anoikis resistance culture conditions, OGT and O-GlcNAcylation levels were still reduced following siOGT treatment when compared to levels in the siScramble control group (Fig. 4C).

Since siOGT treatment caused a statistically significant decrease in anoikis resistance of MCF-7 cells, two-dimensional (2-DE) gel electrophoresis and mass spectrometric analysis were used to determine which proteins would be affected when the OGT level was reduced under anoikis-resistant conditions. Image analysis revealed that 7 protein spots were differentially expressed in the siOGT-transfected cells compared to those in the siScramble control group (≥1.5 fold, P<0.05) (Fig. 5). Then, these 7 upregulated proteins were identified by mass spectrometric analysis and the results are shown in Table I. These identified proteins were categorized into 3 groups based on their protein functions. The first was heat shock protein 27 (Hsp27) (spot no. 1 and no. 2) which is involved in chaperone/stress response. The second group was involved in energy metabolism including triosephosphate isomerase (TPI) (spot no. 3), inorganic pyrophosphatase (spot no. 5), PCTP-like protein (spot no. 6) and nucleoside diphosphate kinase A (spot no. 7). The last protein was peroxiredoxin-2 (spot no. 4) which is involved in cellular protection/detoxification.

As determined by 2-DE and mass spectrometric analysis, the expression level of Hsp27 was the most markedly increased upon OGT knockdown. Therefore, Hsp27 which exerts chaperone/stress response functions was selected for further validation by immunoblotting. The results showed that the expression level of Hsp27 was increased by siOGT knockdown in comparison to this level in the siScramble control with significantly higher relative band intensity (P<0.01) (Fig. 6A). Using Hsp27 immunoprecipitation and immunoblot analysis of Hsp27 and O-GlcNAc (RL2) antibodies, we found that Hsp27 was modified by O-GlcNAc (Fig. 6B and C). The level of O-GlcNAc-modified Hsp27 was obviously decreased following siOGT transfection when compared to the level in the siScramble control. The reduction in the O-GlcNAc-modified Hsp27 level following siOGT may be the result of a global decrease in the O-GlcNAcylation level upon OGT knockdown.

According to previous results, it has been suggested that the elevation of Hsp27 may be associated with the suppression of anoikis resistance and anchorage-dependent growth of MCF-7 cells. Therefore, we determine whether the reduction in Hsp27 is capable to regain the growth in anoikis-resistant cultures under an OGT silencing condition. Transient knockdown of Hsp27 was performed to diminish the Hsp27 expression level. Hsp27 immunoblotting revealed a decreased level of Hsp27 in siHsp27 and siOGT/siHsp27 transfected cells whereas O-GlcNAc immunoblotting showed that the O-GlcNAc level was reduced in the siOGT and siOGT/siHsp27 cells when compared to those of siScramble controls (Fig. 7A). Double knockdown of siOGT/siHsp27 (siDouble) reversed the inhibitory effect in anoikis-resistant cultures compared to that of the siOGT-transected cells (Fig. 7B). Moreover, double knockdown of siOGT/siHsp27 markedly restored the growth of MCF-7 cells in soft agar cultures when compared to that of the siOGT-transected alone cells (Fig. 7C).