The RING domain in the anti-apoptotic protein XIAP stabilizes c-Myc protein and preserves anchorage-independent growth of bladder cancer cells

X-linked inhibitor of apoptosis protein (XIAP) suppresses apoptosis and plays key roles in the development, growth, migration, and invasion of cancer cells. Therefore, XIAP has recently attracted much attention as a potential antineoplastic therapeutic target, requiring elucidation of the molecular mechanisms underlying its biological activities. Here, using shRNA-mediated gene silencing, immunoblotting, quantitative RT-PCR, anchorage-independent growth assay, and invasive assay, we found that XIAP's RING domain, but not its BIR domain, is crucial for XIAP-mediated up-regulation of c-Myc protein expression in human bladder cancer (BC) cells. Mechanistically, we observed that the RING domain stabilizes c-Myc by inhibiting its phosphorylation at Thr-58 and that this inhibition is due to activated ERK1/2-mediated phosphorylation of glycogen synthase kinase-3β (GSK-3β) at Ser-9. Functional studies further revealed that c-Myc protein promotes anchorage-independent growth and invasion stimulated by the XIAP RING domain in human BC cells. Collectively, the findings in our study uncover that the RING domain of XIAP supports c-Myc protein stability, providing insight into the molecular mechanism and role of c-Myc overexpression in cancer progression. Our observations support the notion of targeting XIAP's RING domain and c-Myc in cancer therapy.

The incidence and mortality rate of bladder cancer (BC) 4 rank in the first place in urologic malignancies and have con-tinued to rise in recent years. There are more than 429,000 new cases of BC diagnosed every year worldwide (1), and about 81,190 patients were diagnosed in the United States in 2018 (2). Treatment based on the grade and stage of BC ranges from transurethral resection to radical cystectomy to systemic chemotherapy. However, BC is not sensitive to radiotherapy and chemotherapy. Currently, the overall therapeutic effects on muscle-invasive BC are limited, and the 5-year survival rate has remained at a low level (3). Thus, further exploration of genetic regulatory networks involved in BC progression and development of precise strategies are of great significance.
X-linked inhibitor of apoptosis protein (XIAP), a member of the inhibitor of apoptosis (IAP) family, contains three baculoviral IAP repeat (BIR) domains and a ubiquitin-associated RING domain (4). XIAP not only functions as a suppressor of apoptosis, but also plays key roles in development, anchorageindependent growth, migration, and invasion of cancer cells, as well as mediating resistance of cancer cells to chemotherapeutic drugs and radiotherapy (5). Recently, we found that XIAP promoted urothelial transformation through its C-terminal RING domain-initiated miR-4295 expression and downstream reduction of p63␣ protein translation (6). On the other hand, we previously have reported that the RING domain of XIAP enhances BC cell anchorage-independent growth through upregulating cyclin D1 expression (7), and we also have revealed that the XIAP N-terminal BIR domain increases EGFR translation and promotes growth of BC cells (8). Moreover, we recently discovered that XIAP promoted BC invasion in vitro and lung metastasis in vivo through enhancing nucleolin-mediated Rho-GDI␤ mRNA stability (9). Judging from the multiple roles of XIAP in tumor formation and progression, it may serve as a promising target for BC therapy.
The oncoprotein c-Myc plays key roles in formation and progression of various cancers, and it is reported that c-Myc could regulate expression of about 15% of all genes through binding to their promoters and enhancers (10). The expression of activation targets of c-Myc, an embryonic stem signature, was observed more frequently in poorly differentiated BC than in well-differentiated ones (11). When human BC cell lines with low versus high metastatic potential are compared, c-myc is found to be up-regulated in the highly metastatic tumor cell lines (12). In the present study, we explore the mechanism underlying increased c-Myc expression in muscle-invasive bladder cancer cells and focus on the correlation between upregulation of c-Myc and overexpression of XIAP. We found that the XIAP RING domain could stabilize c-Myc protein through inhibition of its phosphorylation at Thr-58, which is mediated by increased phosphorylation of glycogen synthase kinase-3␤ (GSK-3␤) at Ser-9 due to ERK1/2 activation. Moreover, c-Myc is an effector for promotion of anchorage-independent growth and invasion by XIAP RING domain in BC cells.

XIAP RING domain, but not BIR domain, is crucial for XIAPmediated up-regulation of c-Myc protein expression
To investigate the correlation between up-regulation of c-Myc and overexpression of XIAP, shRNAs specifically targeting human XIAP RING domain (shXIAP77) and specifically targeting human XIAP BIR domain (shXIAP78) mRNA coding areas were used to knockdown endogenous XIAP in two different invasive bladder cancer cell lines, T24T and UMUC3, respectively (Fig. 1, A and B). XIAP RING domain-negative (⌬RING) and its scramble vector control (pEEB) plasmids were then co-transfected with shXIAP77, and XIAP BIR domainnegative (⌬BIR) and its scramble vector control (pEEB) plasmids were co-transfected with shXIAP78 (thus avoiding knockdown of the exogenous RING domain of XIAP by its shRNA) to rescue the expression of the BIR and RING domains, respectively, in these knockdown transfectants. The stable transfectants were established and analyzed for c-Myc expression. As shown in Fig. 1 (C and D), knockdown of XIAP by shXIAP77 or shXIAP78 in both cell lines resulted in a profound reduction of c-Myc protein levels in both T24T and UMUC3 cells, which were effectively reversed by ectopic expression of RING (⌬BIR) domain, but not BIR (⌬RING) domain. However, the mRNA levels of c-Myc were not affected in these stable transfectants (Fig. 1, E and F). We also used XIAP RING domain deletion knock in (⌬RING) and WT mice to establish both WT mouse embryonic fibroblasts (MEFs) and ⌬RING MEFs. The results obtained from comparison of c-Myc expression between ⌬RING MEFs and WT MEFs indicated that c-Myc protein expression in XIAP RING domain-deficient MEFs(⌬RING) was impaired as compared with that in WT-MEFs (Fig. 1G). Moreover, when RING domain was constitutively expressed in ⌬RING MEFs, c-Myc protein level was also restored (Fig. 1H). Similar to what was observed in BC cells, neither RING domain deletion nor its ectopic expression affected c-Myc mRNA levels in MEFs (Fig. 1, I-L). These results indicate that the XIAP RING domain, but not the BIR domain, can positively regulate the expression of c-Myc protein, which is beyond mRNA level.

XIAP RING domain stabilization of c-Myc XIAP RING domain stabilizes c-Myc protein through inhibiting its phosphorylation at Thr-58
To define whether XIAP regulates c-Myc protein via proteasome-mediated degradation, UMUC3 cells stably transfected with control nonsense or shXIAP78 were pretreated with or without the proteasome inhibitor MG132 and subsequently treated with cycloheximide (CHX) to detect c-Myc protein degradation rates. As shown in Fig. 2A, knockdown of XIAP effectively promoted c-Myc protein degradation in a time-dependent manner, whereas the levels of ␤-catenin and hexokinase II, which were reported to be regulated by proteasome-mediated degradation (13,14), were not affected. In line with this, MG-132 could reverse c-Myc protein degradation upon CHX treatment (Fig. 2B) and rescued c-Myc protein levels in XIAPknockdown cells back to the levels in control cells (Fig. 2C). To further determine the role of the XIAP RING domain in regulation of c-Myc protein degradation, the paired UMUC3 cells expressing shXIAP78/pEBB and shXIAP78/⌬BIR were also used for a protein degradation assay under the same conditions. The results revealed that restoration of the RING domain could stabilize c-Myc protein in XIAP knockdown cells (Fig. 2D). It has been verified that Thr-58 and Ser-62 of c-Myc protein are two key sites that are phosphorylated to be recognized by E3 ubiquitin ligases, such as Fbw7, to execute its proteasome-mediated degradation (15). Thus, we then detected Thr-58 and Ser-62 phosphorylation levels of c-Myc, as well as expression of Fbw7, in the indicated stable transfectants where XIAP was knocked down and where the BIR or RING domain, respectively, was resumed by ectopic overexpression. As shown in Fig.  2E, knockdown of XIAP by shXIAP77 or shXIAP78 in UMUC3 cells caused significant increases of c-Myc phosphorylation at Thr-58, but not Ser-62, whereas expression of Fbw7 was not affected. Importantly, restoration of the RING domain, but not the BIR domain, profoundly inhibited c-Myc phosphorylation at Thr-58 in XIAP knockdown cells (Fig. 2E). Consistent with these, the level of c-Myc phosphorylation at Thr-58 was enhanced in XIAP RING domain-deficient MEFs (Fig. 2F) and was suppressed after RING domain was restored (Fig. 2G), whereas its phosphorylation at Ser-62 was not changed (Fig. 2, F and G). Given that phosphorylation of c-Myc protein at Thr-58 is associated with its protein degradation, it is anticipated that XIAP RING domain could stabilize c-Myc protein through inhibiting its phosphorylation at Thr-58.

Activation of ERK1/2 by the XIAP RING domain mediated phosphorylation of GSK-3␤ at Ser-9 and in turn inhibited c-Myc phosphorylation at Thr-58
Because GSK-3 activity has been reported to play a key role in inhibiting phosphorylation of c-Myc specifically at Thr-58 and to increase c-Myc stability (16), we next tested whether changes in the activity of GSK-3 were involved in inhibition of c-Myc phosphorylation at Thr-58 by the XIAP RING domain. As shown in Fig. 3A, knockdown of XIAP by shXIAP77 or shXIAP78 in UMUC3 cells effectively suppressed GSK-3␤ phosphorylation at Ser-9 with no effect on total protein level of GSK-3␤, whereas both the phosphorylation status and total protein level of GSK-3␣ were not changed. Given that the phosphorylation level of GSK-3␤ at Ser-9 is negatively associated with its kinase activity (17), suppression of phosphorylation of GSK-3␤ at Ser-9 revealed that GSK-3␤ activity was increased in XIAP knockdown transfectants. Moreover, restoration of the RING domain, rather than the BIR domain, resulted in a marked increase of GSK-3␤ phosphorylation at Ser-9 in XIAP knockdown cells (Fig. 3A). In line with these findings, compared with WT MEFs, phosphorylation of GSK-3␤ at Ser-9 was decreased in XIAP RING domain-deficient MEFs (Fig. 3B) and was up-regulated upon restoration of the RING domain (Fig.  3C). To further confirm the role of GSK-3␤ in XIAP RING domain regulation of c-Myc phosphorylation at Thr-58, we transfected the constitutive active mutant of GSK-3␤ (GSK-3␤ S9A) into RING domain-restored UNUC3 (shXIAP78/⌬BIR) cells, and the stable transfectants were identified, as shown in Fig. 3D. The dominant active effect was proved by the promotion of a GSK-3␤-regulated phosphorylation of ␤-catenin at Ser-33, Ser-37, or Thr-41 (Fig. 3D), as reported in previous studies (18). We found that the active form of GSK-3␤ could effectively enhance phosphorylation of c-Myc at Thr-58 to down-regulate its total protein expression, whereas phosphorylation at Ser-62 was also not affected (Fig. 3D).
Our previous study has demonstrated that extracellular signal-regulated protein kinase (ERK1/2) activation could be promoted by XIAP in T24T cells (9), and ERK1/2 has been reported to be the upstream kinase of GSK-3␤ (19). We therefore determined the potential effect of ERK1/2 on the regulation of phosphorylation of c-Myc. As shown in Fig. 4A, knockdown of XIAP by shXIAP77 or shXIAP78 in T24T cells effectively inhibited ERK1/2 phosphorylation without affecting its total protein levels. Similarly, phosphorylation of ERK1/2 was down-regulated in XIAP RING domain-deficient MEFs (Fig. 4B) and was increased upon restoration of RING domain expression (Fig. 4C). To further address the role of ERK1/2 in XIAP RING domain regulation of c-Myc phosphorylation at Thr-58, we transfected the dominant negative form of ERK2 (DN-ERK2) into RING domain-restored UNUC3 (shXIAP78/ ⌬BIR) cells, and the stable transfectants were identified, as shown in Fig. 4D. The dominant negative effect was observed

XIAP RING domain stabilization of c-Myc
with the inhibition of an ERK-regulated phosphorylation of p-P90RSK at Thr-359/Ser-363 (Fig. 4D), as reported previously (20). Ectopic expression of DN-ERK2 could effectively inhibit the phosphorylation level of GSK-3␤ at Ser-9 and enhanced phosphorylation of c-Myc at Thr-58 to further down-regulate its total protein expression, whereas phosphorylation at Ser-62 was not affected (Fig. 4, D-G). These results clearly indicate that activation of ERK1/2 by the XIAP RING domain mediates phosphorylation of GSK-3␤ at Ser-9, which further inhibits c-Myc phosphorylation at Thr-58. On the other hand, no direct interaction of XIAP and c-Myc was observed as we used the lysates from WT and ⌬XIAP HCT116 cells that were co-immunoprecipitated with anti-XIAP antibody in our previous studies (21) to do Western blotting for the determination of c-Myc protein (Fig. 4H).

RING domain is involved in preserving the XIAP-regulated anchorage-independent growth and invasion abilities of BC cells
XIAP has been reported to play key multiple roles in development, anchorage-independent growth, migration, and invasion of cancer cells, and we have recently found that the XIAP RING domain can promote urothelial transformation (22). To address the specific functions of the XIAP RING domain in BCs, UMUC3 (shXIAP78/pEBB) and UNUC3 (shXIAP78/ ⌬BIR) were applied to the anchorage-independent growth assay, and T24T (shXIAP78/pEBB) and T24T (shXIAP78/ ⌬BIR) were applied to the transwell migration and invasion assay in the present study. As shown in Fig. 5 (A-D), stable knockdown of XIAP blocked anchorage-independent growth in UMUC3 and T24T cells, which was remarkably reversed by restoration of the RING domain (⌬BIR). Accordingly, both the relative migration and invasion rates of UMUC3 cells were reduced by knockdown of XIAP, which was resumed by restoration of the RING domain (Fig. 5, E-G). Although the migration of T24T cells was even promoted by knockdown of XIAP, the relative invasion rates of T24T cells were also inhibited by knockdown of XIAP and restored by transfection of the RING domain (Fig. 5,  H-J). These results indicate that the RING domain is involved in preserving the XIAP-regulated anchorage-independent growth and invasion abilities of BC cells.

c-Myc is required for promotion of anchorage-independent growth and invasion by the XIAP RING domain in BC cells
To explore the functions of up-regulation of c-Myc by the XIAP RING domain in the promotion of anchorage-independent growth and invasion in BC cells, c-Myc expression vector was constitutively overexpressed in UMUC3 (shXIAP78) and

XIAP RING domain stabilization of c-Myc
T24T (shXIAP78) cells. The stable transfectants were identified as shown in Fig. 6 (A and B) and then applied to the anchorageindependent growth assay and transwell migration and invasion assay. The anchorage-independent growth ability of XIAP knockdown cells was resumed upon c-Myc overexpression in UMUC3 (shXIAP78) and T24T (shXIAP78) cells (Fig. 6, C-F), and the cell migration and invasion abilities were restored by ectopic expression of c-Myc in UMUC3 (shXIAP78) cells (Fig.  6, G-I). These results suggest that a c-Myc defect is responsible for the reduction of anchorage-independent growth and invasion caused by XIAP knockdown. To further test this notion, we knocked down c-Myc in UMUC3 (shXIAP/⌬BIR) cells, and the stable transfectant, UMUC3 (shXIAP/⌬BIR/sh-c-Myc), and its vector transfectant, UMUC3 (shXIAP/⌬BIR/nonsense), were established and identified as shown in Fig. 7A. The results obtained from the anchorage-independent growth assay revealed that effective knockdown of c-Myc by sh-c-Myc-2 and sh-c-Myc-4 could abolish RING domain-mediated BC cell anchorage-independent growth (Fig. 7, B and C), migration, and invasion (Fig. 7, D-F). Collectively, our results strongly indicate that XIAP RING-mediated up-regulation of C-Myc is crucial for RING domain promotion of anchorage-independent growth and invasion in human BC cells.

Discussion
XIAP possesses three BIR domains in the N terminus and one RING domain in the C terminus. XIAP BIR domains mainly interact with caspases to regulate cell apoptosis, whereas the RING domain of XIAP can function as an E3 ligase to bind to caspase-3 and mitochondrial XIAP inhibitor SMAC/ Diablo, as well as mediate the proteasomal degradation of itself (23,24). Recently, the E3 ubiquitin ligase function of the XIAP RING domain has been further studied. XIAP has been found to be a ubiquitin E3 ligase for Mdm2 to promote its degradation (25). It is reported that the XIAP RING domain controls the protein stability of Cdc42 through directly conjugating polyubiquitin chains to the lysine 166 of Cdc42 (26). XIAP could also ubiquitinate a highly conserved Lys residue in adenylyl cyclase isoforms and thereby accelerate their endocytosis and degradation (27). In this study, we found that the XIAP RING domain stabilized c-Myc protein without direct interaction. Thus, the regulation of c-Myc protein by the XIAP RING domain might be through an indirect regulatory mechanism.
The c-Myc oncogene is a "master regulator" in development of many tumors. Amplified copies of c-Myc may play a key role in the progression of highly metastatic prostate cancer and cutaneous melanoma (28). In contrast, highly metastatic melanoma and bladder carcinoma cells did not show amplified c-myc gene, although it is overexpressed in these cancers (12). Therefore, other mechanisms may be involved in up-regulation of c-myc in BC. Previous studies have shown that the dual c-myc P1/P2 promoters are important for its transcription, which involves multiple positively and negatively acting transcription factors and signaling pathways (29,30). A recent study also shows that the translation of c-Myc is promoted by its mRNA N 6 -methyladenosine nucleotide modification in human myeloid leukemia cells (31). Two phosphorylation sites of c-Myc

Figure 7. Knockdown of c-Myc exhibited an inhibitory effect on XIAP-RING domain-promoted anchorage-independent growth and invasion abilities of bladder cancer cells.
A, the shRNA was used to knock down c-Myc expression in UMUC3 (shXIAP/⌬BIR) cells, and the stable transfectants were then subjected to Western blotting for determination of the indicated c-Myc expression. GAPDH was used as the protein-loading control. B and C, the indicated transfectants were subjected to an anchorage-independent growth assay. D-F, the indicated transfectants were subjected to migration and invasion assays. *, significant difference as compared with vector transfectant (p Ͻ 0.05). #, significant difference as compared with nonsense cells (p Ͻ 0.05). Error bars, S.D.

Figure 6. Restoration of c-Myc expression reversed the down-regulation of anchorage-independent growth and invasion abilities in XIAP-deficient cells.
A and B, c-Myc was stably overexpressed in XIAP knockdown UMUC3 (A) and T24T (B) cells, and the stable transfectants were then subjected to Western blotting for identification of c-Myc expression. GAPDH was used as the protein-loading control. C-F, the indicated UMUC3 (C and D) and T24T (E and F) transfectants were subjected to an anchorage-independent growth assay. G-I, the indicated transfectants were subjected to a migration and invasion assay. *, significant difference as compared with nonsense cells (p Ͻ 0.05). #, significant difference as compared with vector transfectant (p Ͻ 0.05). Error bars, S.D.

XIAP RING domain stabilization of c-Myc
protein, Thr-58 and Ser-62, exhibit opposing roles in the control of c-Myc protein stability (32). It has been showed that GSK3-induced c-Myc phosphorylation at Thr-58 is critical for ensuring its transient and timely protein degradation (33). In the present study, we discovered that inhibition of Thr-58 phosphorylation, which was mediated by increased phosphorylation of GSK-3␤ at Ser-9, was critical for stabilization of c-Myc protein by the XIAP RING domain. Our study provides novel evidence that protein degradation plays an important role in the regulation of c-Myc expression in BCs.
GSK3␤ is a multifunctional serine/threonine kinase that participates in a diverse array of cell functions (34). Phosphorylation of GSK3␤ at Ser-9 leads to its inactivation by proteasomal degradation and has been connected to many pathological conditions, including cancer (35). Several kinase-driven pathways phosphorylate GSK3␤ at Ser-9, including protein kinase A, protein kinase B/Akt, p90 ribosomal S6 kinase/mitogen-activated protein kinase-activating protein (p90RSK/MAPKAP), and p70 ribosomal S6 kinase (p70S6K) (19,36). Our previous study indicated that ERK1/2 activation could be promoted by XIAP in T24T cells (9). Our current studies indicate that the XIAP RING domain was crucial for ERK1/2 activation and GSK phosphorylation at Ser-9 in human BC cells, which further mediated the c-Myc protein phosphorylation at Thr-58 and BC cell anchorage-independent growth and invasion. Although the kinase-driven pathway that is responsible for XIAP RING domain-induced ERK1/2 activation still needs to be addressed, the discovery of the XIAP RING domain in promotion of GSK-3␤ phosphorylation at Ser-9 provides new insight into the function of XIAP in BC tumor biology.
In summary, we found that the XIAP RING domain, but not the BIR domain, is crucial for XIAP-mediated up-regulation of c-Myc protein expression. Mechanistically, the XIAP RING domain could stabilize c-Myc protein through inhibition of its phosphorylation at Thr-58, which is mediated by increasing phosphorylation of GSK-3␤ at Ser-9. The results from functional studies reveal that c-Myc is a crucial effector responsible forXIAPRINGdomain-mediatedpromotionofanchorage-independent growth and invasion in BC cells. Our study uncovers a novel function of the RING domain of XIAP in regulating the c-Myc stability, further offering new theoretical support for using the XIAP RING domain and c-Myc as targets for BC cancer therapy.

Plasmids and stable cell transfection
The specific shRNA targeting human XIAP and c-Myc was purchased from Open Biosystems (Lafayette, CO). The overexpression of ⌬RING and ⌬BIR plasmids (21) was described in our previous studies (8,37). The c-Myc expression plasmid (38) was a gift from Dr. Rosalie Sears (Oregon Health and Science University, Portland, OR). GSK-3␤ S9A expression plasmid and the plasmid encoding dominant negative ERK2, which contained a K52R mutation of rat ERK2, were created as described previously (39,40). Cell transfections were performed with PolyJet TM DNA in Vitro Transfection Reagent (SignaGen Laboratories, Rockville, MD) according to the manufacturer's instructions. For stable transfection, cell cultures were subjected to hygromycin B, G418, or puromycin selection according to the resistance of plasmids, and cells surviving were pooled as stable mass transfectants.

Western blot analysis
Whole-cell extracts were prepared with the cell lysis buffer (10 mM Tris-HCl, pH 7.4, 1% SDS, and 1 mM Na 3 VO 4 ) as described in our previous studies (41). 50 g of proteins were resolved by SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and probed with the indicated primary antibodies together with the alkaline phosphatase-conjugated secondary antibody. Signals were detected by the enhanced chemifluorescence Western blotting system as described in a previous report (22). The images were acquired, and the protein levels were quantified by using the Typhoon FLA 7000 imager (GE Healthcare).

RT-PCR and quantitative RT-PCR
Total RNA was extracted using the TRIzol reagent as described in the manufacturer's instructions (Invitrogen). 5 g of total RNA was used for first-strand cDNA synthesis with oligo(dT) primer by the SuperScript IV first-strand synthesis system (Invitrogen). Specific primers were used for PCR amplification. The primers used in this study were as follows: human c-myc, forward (5Ј-AAC ACA CAA CGT CTT GGA GC-3Ј) and reverse (5Ј-CCT TAC GCA CAA GAG TTC CG-3Ј); human gapdh, forward (5Ј-AGA AGG CTG GGG CTC ATT TG-3Ј) and reverse (5Ј-AGG GGC CAT CCA CAG TCT TC-3Ј); mouse c-myc, forward (5Ј-TCT CCA CTC ACC AGC ACA ACT ACG-3Ј) and reverse (5Ј-ATC TGC TTC AGG ACC CT-3Ј); and mouse gapdh, forward (5Ј-TGC AGT GGC AAA GTG GAG ATT-3Ј) and reverse (5Ј-TTT TGG CTC CAC CCT TCA AGT-3Ј). The quantitative RT-PCR analysis was carried out using the SYBR Green PCR kit (Qiagen, Santa Clarita, CA) and the 7900HT Fast Real-time PCR system (Applied Biosystems, Carlsbad, CA). The ⌬⌬CT value was used to calculate the relative expression of c-myc mRNA, using gapdh as an endogenous control.

Anchorage-independent growth assay
Anchorage-independent growth ability was evaluated in soft agar as described in our previous studies (42). Briefly, 3 ml of XIAP RING domain stabilization of c-Myc 0.5% agar in basal modified Eagle's medium supplemented with 10% FBS was layered onto each well of 6-well tissue culture plates. 1 ml of 0.35% agar medium with cells (1 ϫ 10 4 cells) was then layered on top of the 0.5% agar layer. Plates were incubated at 37°C in 5% CO 2 for 2-3 weeks, and the colonies with more than 32 cells were scored and are presented as colonies/10 4 cells.

Transwell migration and invasion assay
The migration and invasion kit was purchased from BD Falcon, and the assay was performed according to the manufacturer's instructions. 30,000 cells were seeded in the upper well with 0.1% serum medium and in the lower part with complete medium, and culture continued for 24 h. Then cells both on the inside and outside of the chamber were fixed with 3.7% formalin for 2 min, washed twice with PBS, transferred to 100% methanol for 20 min, washed twice again, and then finally stained by Giemsa (diluted 1:20 with PBS) at room temperature for 15 min in the dark. They were again washed twice, and then the noninvaded cells were scraped off with a cotton swab (PBS-wetted) four times. The photographs were taken with an Olympus DP71 camera, and the number of cells was calculated by ImageJ software.

Statistical analysis
Student's t test was used to determine the significance between different groups. p Ͻ 0.05 was considered as a significant difference between compared groups.