β3 Integrin Promotes TGF-β1/H2O2/HOCl-Mediated Induction of Metastatic Phenotype of Hepatocellular Carcinoma Cells by Enhancing TGF-β1 Signaling

In addition to being an important mediator of migration and invasion of tumor cells, β3 integrin can also enhance TGF-β1 signaling. However, it is not known whether β3 might influence the induction of metastatic phenotype of tumor cells, especially non-metastatic tumor cells which express low level of β3. Here we report that H2O2 and HOCl, the reactive oxygen species produced by neutrophils, could cooperate with TGF-β1 to induce metastatic phenotype of non-metastatic hepatocellular carcinoma (HCC) cells. TGF-β1/H2O2/HOCl, but not TGF-β1 or H2O2/HOCl, induced β3 expression by triggering the enhanced activation of p38 MAPK. Intriguingly, β3 in turn promoted TGF-β1/H2O2/HOCl-mediated induction of metastatic phenotype of HCC cells by enhancing TGF-β1 signaling. β3 promoted TGF-β1/H2O2/HOCl-induced expression of itself via positive feed-back effect on p38 MAPK activation, and also promoted TGF-β1/H2O2/HOCl-induced expression of α3 and SNAI2 by enhancing the activation of ERK pathway, thus resulting in higher invasive capacity of HCC cells. By enhancing MAPK activation, β3 enabled TGF-β1 to augment the promoting effect of H2O2/HOCl on anoikis-resistance of HCC cells. TGF-β1/H2O2/HOCl-induced metastatic phenotype was sufficient for HCC cells to extravasate from circulation and form metastatic foci in an experimental metastasis model in nude mice. Inhibiting the function of β3 could suppress or abrogate the promoting effects of TGF-β1/H2O2/HOCl on invasive capacity, anoikis-resistance, and extravasation of HCC cells. These results suggest that β3 could function as a modulator to promote TGF-β1/H2O2/HOCl-mediated induction of metastatic phenotype of non-metastatic tumor cells, and that targeting β3 might be a potential approach in preventing the induction of metastatic phenotype of non-metastatic tumor cells.


Introduction
Integrin expression is crucial for the migratory and invasive capability of tumor cells. Hepatocellular carcinoma (HCC) cells express several integrins which have been identified as the mediators of their migration and invasion, including a1b1, a2b1, a3b1, a6b1, avb1, avb3, and avb5 [1][2][3][4][5][6]. Most of these a and b integrin subunits are moderately expressed in nonmetastatic HCC cells [3,6,7], whereas the expressions of a3 and b3 in these cells are very low or even negligible [4][5][6][7][8][9]. a3 and b3 are expressed in metastatic HCC cells [3][4][5], indicating that the up-regulation of a3 and b3 might be crucial for non-metastatic HCC cells to acquire metastatic phenotype. Moreover, b3 has also been found to modulate transforming growth factor b1 (TGF-b1) signaling in some types of cells [10,11]. However, it is not known whether b3 might be involved in the induction of metastatic phenotype of tumor cells by functioning as modulatory factor.
Previous studies showed that TGF-b1 can induce a3 expression in non-metastatic HCC cells [1,7], and suggested the idea that in hepatocellular carcinoma patients TGF-b1 triggers invasiveness of HCC cells by stimulating the expression of a3 integrin [1]. However, a3 expression is required but not sufficient for the invasiveness of HCC cells, since TGF-b1-treated non-metastatic HCC cells showed higher invasiveness only in the presence of exogenous matrix metalloproteinase (MMP) [1]. Given that avb3 could increase the invasive capacity of HCC cells [5], simultaneous up-regulation of both a3 and b3 might be required for higher invasiveness of HCC cells. Current knowledge of expression and function of b3 in non-metastatic HCC cells is very limited. TGF-b1 has been found to up-regulate b3 expression in other types of cells by activating p38 MAPK pathway, whilst b3 positively controls TGF-b1-induced p38 MAPK activation by promoting Src-mediated tyrosine phosphorylation of TbRII [10,11]. However, TGF-b1 was inefficient in up-regulating b3 expression in non-metastatic HCC cells [9], implying that TGF-b1 might be less efficient in inducing p38 MAPK activation in these cells. In this context, other factors which could promote the activation of p38 MAPK might cooperate with TGF-b1 to upregulate b3 expression in non-metastatic HCC cells.
Therefore, H 2 O 2 and HOCl might be potential candidates for cooperating with TGF-b1 to induce the expression of b3 in HCC cells. In this study, we investigated whether H 2 O 2 and HOCl could cooperate with TGF-b1 to induce the metastatic phenotype of non-metastatic HCC cells, and whether b3 expression is required for the induction. Our data showed that TGF-b1 could up-regulate the expression of b3 in presence of H 2 O 2 /HOCl. Intriguingly, b3 promoted TGF-b1/H 2 O 2 /HOCl-induced expression of a3 and SNAI2, and also enabled TGF-b1 to augment the promoting effect of H 2 O 2 /HOCl on anoikis-resistance, thus promoting TGF-b1/H 2 O 2 /HOCl-mediated induction of metastatic phenotype of HCC cells.

H 2 O 2 /HOCl cooperates with TGF-b1 to induce higher invasive capacity of HCC cells
To investigate whether H 2 O 2 and HOCl could cooperate with TGF-b1 to induce the metastatic phenotype of non-metastatic HCC cells, we first analyzed the effect of TGF-b1, H 2 O 2 and HOCl on invasive capacity of HepG2 and Huh7 cells. The result showed that the invasive capacity of tumor cells was gradually increased after prolonged treatment ( Figure 1A). Much higher invasive capacity of tumor cells was induced by TGF-b1 in presence of both H 2 O 2 and HOCl, but not each of them alone ( Figure 1B). Consistently, TGF-b1/H 2 O 2 /HOCl was most efficient in promoting the polymerization of actin in tumor cells ( Figure 1C) and the production of active MMP-2 and MMP-9 by tumor cells ( Figure 1D) in response to ECM molecules (matrigel), which are important for migratory and invasive properties of tumor cells [18][19][20]. These results indicated that TGF-b1 could induce much higher invasive capacity of HCC cells in presence of H 2 O 2 /HOCl, whereas TGF-b1 alone was less efficient.

TGF-b1/H 2 O 2 /HOCl induces metastatic phenotype of HCC cells
We then tested the metastatic capability of HCC cells by using an experimental metastasis model in nude mice. Tumor cell arrest and extravasation in the lung of mice were assessed 5 h and 48 h, respectively, after i.v. injection of tumor cells. The pre-treatment with TGF-b1/H 2 O 2 /HOCl increased tumor cell arrest and resulted in the extravasation of tumor cells in the lung ( Figure  2A), whereas pre-treatment with TGF-b1 or H 2 O 2 /HOCl did not promote tumor cell extravasation (Figure 2A). After inoculation via tail vein, the metastatic foci were only observed in the lung tissues of the mice inoculated with the tumor cells pre-treated with TGF-b1/H 2 O 2 /HOCl ( Figure 2B, 2C). These results demonstrated that TGF-b1/H 2 O 2 /HOCl could induce the metastatic phenotype of HCC cells.
TGF-b1 was inefficient in inducing the sustained activation of p38 MAPK in HepG2 cells ( Figure 3C). Co-stimulation with TGF-b1/H 2 O 2 /HOCl enhanced the transient activation of p38 MAPK, and also gradually enhanced the sustained activation of p38 MAPK ( Figure 3C). To ascertain whether b3 was involved in the enhancement of the sustained activation of p38 MAPK, we used b3 shRNA to suppress the up-regulation of b3 expression ( Figure 3D). Intriguingly, b3 shRNA significantly reduced the phosphorylation level of p38 MAPK induced by TGF-b1/H 2 O 2 / HOCl ( Figure 3E), suggesting that b3 promoted TGF-b1-induced activation of p38 MAPK pathway. To confirm this, we stimulated HepG2 cells with TGF-b1/H 2 O 2 /HOCl in presence of SU6656 (Src inhibitor), since inhibiting Src activity could prevent the ability of b3 integrin to enhance TGF-b1 signaling [10,11]. The result showed that SU6656 significantly reduced the phosphorylation level of p38 MAPK induced by TGF-b1/H 2 O 2 /HOCl ( Figure 3F, S1), suggesting that b3 augmented p38 MAPK activation by enhancing TGF-b1 signaling.
To further clarify the role of H 2 O 2 /HOCl, we removed H 2 O 2 / HOCl 48 h after stimulation with TGF-b1/H 2 O 2 /HOCl, and continuously stimulated HepG2 cells with TGF-b1. The result showed that both ITGB3 expression and the phosphorylation level of p38 MAPK were significantly reduced if H 2 O 2 and HOCl were removed ( Figure 3G), suggesting that the continuous existence of H 2 O 2 /HOCl was required for inducing higher activation level of p38 MAPK and higher expression of ITGB3 gene.
b3 augments the promoting effect of TGF-b1/H 2 O 2 /HOCl on invasive capacity To ascertain the role of b3 integrin in TGF-b1/H 2 O 2 /HOClinduced invasiveness, we further detected the invasive migration of TGF-b1/H 2 O 2 /HOCl-treated HepG2 cells in presence of a3 and avb3 blocking antibodies. Blocking a3 almost abolished the invasiveness of HepG2 cells. Blocking avb3 partially but significantly suppressed the invasive migration ( Figure 4A). Intriguingly, if up-regulation of b3 expression in HepG2 cells was suppressed by shRNA, TGF-b1/H 2 O 2 /HOCl induced much lower invasive capacity of the cells ( Figure 4B). Moreover, inhibiting Src activity with SU6656 significantly suppressed the promoting effect of TGF-b1/H 2 O 2 /HOCl on invasive capacity of HepG2 cells ( Figure 4C). On the other hand, HepG2 cells did not acquire higher invasive capacity when b3 was overexpressed in the cells only by transfection with b3 expression vector (data not shown), indicating that b3 alone could not increase the invasive capacity of non-metastatic HCC cells without the stimulation with TGF-b1/H 2 O 2 /HOCl. These results suggested that in addition to being a mediator of invasive migration, b3 integrin could function as a modulator to promote the effect of TGF-b1/H 2 O 2 /HOCl on invasiveness of HCC cells by enhancing TGF-b1 signaling.
b3 promotes TGF-b1/H 2 O 2 /HOCl-mediated up-regulation of a3 and SNAI2 expression Based on the above results, we next investigated whether b3 might influence the expression of a3. TGF-b1 induced a3 expression in HepG2 cells ( Figure 5A). Intriguingly, much higher expression level of a3 was induced by TGF-b1/H 2 O 2 /HOCl. We then analyzed the signaling pathway(s) involved in up-regulation of ITGA3 expression by stimulating HepG2 cells with TGF-b1/ H 2 O 2 /HOCl in presence of SB203580, PD98059, SP600125, wortmannin, QNZ, and SIS3. The result showed that p38 MAPK, ERK, and JNK pathways were involved in up-regulating ITGA3 expression ( Figure 5B). Among them, ERK pathway was the most efficient one. In line with this, the sustained activation of ERK pathway in HepG2 cells was gradually enhanced by stimulation with TGF-b1/H 2 O 2 /HOCl ( Figure 5C).
Both Smad and ERK pathways are involved in up-regulating the expression of SNAI2 [21] which positively controls a3b1-mediated   migration of tumor cells [22]. TGF-b1-induced activation of Smad pathway was also gradually enhanced in the presence of H 2 O 2 / HOCl ( Figure S2). Consistently, TGF-b1/H 2 O 2 /HOCl induced higher expression of SNAI2 in HepG2 cells ( Figure 5D). Inhibiting b3 expression with shRNA did not influence the activation of Smad pathway (data not shown), but suppressed TGF-b1/H 2 O 2 /HOClinduced activation of ERK, and also suppressed the up-regulation of ITGA3 and SNAI2 expression ( Figure 5E). Taken together, these results suggested that the up-regulation of b3 enhanced the sustained activation of ERK pathway, thus promoting TGF-b1/ H 2 O 2 /HOCl-induced expression of both a3 and SNAI2.
The above results suggested that the higher and sustained activation of p38 MAPK, ERK, and Smad pathways was necessary for TGF-b1/H 2 O 2 /HOCl to induce the invasive capacity of HCC cells. To further confirm this, we added SB203580, PD98059, and SIS3 to the cell culture 96 h after stimulation and thereafter. Each of these inhibitors significantly suppressed the promoting effect of TGF-b1/H 2 O 2 /HOCl on invasive migration and extravasation of HepG2 cells ( Figure S3), suggesting that the sustained activation of these pathways was indeed required for TGF-b1/H 2 O 2 /HOCl to induce higher invasive capacity of HCC cells.
b3 enables TGF-b1 to promote the anoikis-resistance of HCC cells TGF-b1 has the potential to induce apoptosis of tumor cells in a Smad-dependent manner [23]. We therefore further investigated whether TGF-b1/H 2 O 2 /HOCl might increase or decrease the apoptosis-resistance of HCC cells. TGF-b1 could induce transient activation of Smad pathway, but was inefficient in inducing the sustained activation of Smad pathway in HCC cells ( Figure S2). Consistently, the treatment with TGF-b1 alone promoted the apoptosis of HepG2 cells after 48-h culture, whereas the apoptosis was gradually reduced after prolonged stimulation ( Figure S4A). Importantly, the apoptosis of HepG2 cells was further reduced in presence of H 2 O 2 /HOCl ( Figure S4A).
The prolonged treatment with TGF-b1/H 2 O 2 /HOCl reduced the expression of pro-apoptotic genes (BAX, BIM, BID), and increased the expression of anti-apoptotic genes (MCL1, BCL2, c-FLIP) ( Figure S4B). These genes also influence mitochondrial pathway and extrinsic pathway involved in anoikis [24,25]. We therefore further investigated whether the treatment with TGF-b1/H 2 O 2 /HOCl might increase the anoikis-resistance of HepG2 cells. Pre-treatment with TGF-b1 alone slightly increased the apoptosis of tumor cells cultured under anchorage-independent condition (anoikis) ( Figure 6A). The anoikis of tumor cells was reduced by the pre-treatment with H 2 O 2 /HOCl. Intriguingly, TGF-b1 augmented the effect of H 2 O 2 /HOCl ( Figure 6A). However, if b3 expression was suppressed with shRNA, TGF-b1 could not augment the promoting effect of H 2 O 2 /HOCl on anoikis-resistance ( Figure 6B). We therefore further analyzed the effect of Smad and MAPK pathways on anoikis-resistance. The results showed that inhibiting Smad3 further reduced anoikis of HCC cells, whereas inhibiting MAPK pathways increased the anoikis of the cells ( Figure 6C). These results suggested that the upregulation of b3 enabled TGF-b1 to promote anoikis-resistance by enhancing the activation of MAPK pathways.

Discussion
Both extrahepatic metastasis and intrahepatic metastasis of HCC cells involve the step of extravasation from circulation [27,28], which requires higher invasive capacity and anoikisresistance of tumor cells. TGF-b1 could induce the invasive capacity of non-metastatic HCC cells to some extent as shown by our data and others [1,7]. Nevertheless, TGF-b1-treated HCC cells were unable to extravasate from circulation. Our data in present study showed that H 2 O 2 /HOCl could cooperate with TGF-b1 to induce higher invasive capacity and anoikis-resistance of non-metastatic HCC cells. Consistently, TGF-b1/H 2 O 2 / HOCl-induced metastatic phenotype was sufficient for HCC cells to extravasate from circulation and form metastatic foci in the secondary sites. H 2 O 2 /HOCl enhanced TGF-b1 signaling, which was crucial for inducing higher invasive capacity and anoikisresistance of non-metastatic HCC cells. Importantly, b3 played an indispensable role in enhancing TGF-b1 signaling, and therefore was required for TGF-b1/H 2 O 2 /HOCl-mediated induction of metastatic phenotype of non-metastatic HCC cells.
The prolonged treatment with TGF-b1/H 2 O 2 /HOCl was required for inducing the metastatic phenotype of non-metastatic HCC cells, since the expression of b3 was gradually increased. Our data showed that TGF-b1 was inefficient in inducing the expression of b3 in non-metastatic HCC cells, which is consistent with the result reported by Nejjari et al [9]. The activation of p38 MAPK pathway induced by either TGF-b1 or H 2 O 2 /HOCl was not sufficient for up-regulating b3 expression, suggesting that the sustained and higher activation of p38 MAPK pathway was required for inducing b3 expression in non-metastatic HCC cells. H 2 O 2 /HOCl cooperated with TGF-b1 to induce higher activation level of p38 MAPK, thus up-regulating the expression of b3. The requirement for the continuous existence of H 2 O 2 /HOCl implicated that the attenuation of PTP activity was required for the sustained activation of p38 MAPK pathway. On the other hand, the up-regulation of b3 in turn enhanced TGF-b1 signaling, resulting in the gradually enhanced activation of p38 MAPK pathway in non-metastatic HCC cells. If the up-regulation of b3 expression was suppressed, the sustained activation of p38 MAPK was maintained at much lower level. Moreover, if the function of b3 was suppressed, TGF-b1/H 2 O 2 /HOCl-induced activation of p38 MAPK was not sufficient for inducing higher expression of b3. Therefore, H 2 O 2 /HOCl cooperation with TGF-b1 actually augmented p38 MAPK-b3 feed-back regulation, thus resulting in the gradual increase of both b3 expression and p38 MAPK activation. Since the expression of b3 was very low in nonmetastatic HCC cells, the feed-back regulation was gradually enhanced, which might explain the requirement for the prolonged stimulation with TGF-b1/H 2 O 2 /HOCl.
In presence of H 2 O 2 /HOCl, TGF-b1-induced activation of Smad and MAPK pathways was gradually enhanced. H 2 O 2 / HOCl promoted the sustained activation of Smad pathway by down-regulating the expression of Nm23-H1 (our unpublished data), whereas the up-regulation of b3 expression was crucial for the enhanced activation of MAPK pathways. Inhibiting the expression and function of b3 did not influence the activation of Smad pathway in HCC cells, suggesting that b3 could not influence the activity of TbRI. It has been found that b3 regulates TGF-b signaling by interacting physically with TbRII and promoting Src-mediated tyrosine phosphorylation of TbRII, which is essential for the ability of TGF-b1 to activate MAPKs [10,11]. Our data showed that inhibiting either b3 or Src could significantly suppress the sustained activation of MAPK pathways after prolonged stimulation with TGF-b1/H 2 O 2 /HOCl, suggesting that b3-Src-mediated modulation of TbRII was crucial for the higher and sustained activation of MAPK pathways. Importantly, inducing higher and sustained activation of MAPK pathways was necessary for TGF-b1/H 2 O 2 /HOCl to induce higher invasive capacity and anoikis-resistance of non-metastatic HCC cells.
The up-regulation of b3 resulted in the higher expression of both a3 and SNAI2 by enhancing the activation of MAPK pathways. Previous study showed that TGF-b1 induced a3 expression in nonmetastatic HCC cells, but the cells did not secrete MMP [1]. The reason might be that TGF-b1 alone could not induce higher expression of SNAI2 in non-metastatic HCC cells as shown by our data. SNAI2 has a positive effect on a3b1-mediated invasiveness of tumor cells [22], since SNAI2 promotes the production of MMP-2 and MMP-9 [19,20]. Both Smad and ERK pathways are involved in up-regulating the expression of SNAI2 [21]. TGF-b1/H 2 O 2 / HOCl, but not TGF-b1 alone, induced much higher expression of SNAI2 by inducing higher and sustained activation of both Smad and ERK pathway. Although b3 did not influence the activation of Smad pathway, its enhancing effect on the activation of ERK pathway was indispensable for the up-regulation of SNAI2 expression. Inhibiting the expression or function of b3 could significantly suppress the expression of SNAI2. Our result is also supported by another report that inhibiting ERK signaling blocked TGF-b1-induced SNAI2 expression in oral squamous cell carcinoma cells [20]. Therefore, up-regulation of b3 was crucial for TGF-b1/H 2 O 2 /HOCl to induce higher expression of SNAI2 in nonmetastatic HCC cells. On the other hand, TGF-b1/H 2 O 2 /HOCl could induce much higher expression of a3 due to positive effect of b3 on the sustained activation of p38 MAPK and ERK pathways. In this context, up-regulation of b3 could promote both a3 expression and a3b1-mediated invasive migration of HCC cells.
TGF-b1 has the potential to induce apoptosis of tumor cells in a Smad-dependent manner [23]. The treatment with TGF-b1 alone within a relatively short period of time could promote the apoptosis in HCC cells as shown by our data and others [29], whereas the apoptosis was reduced after prolonged stimulation, possibly due to the inefficiency of TGF-b1 in inducing the sustained activation of Smad pathway and the proliferation of  surviving cells. TGF-b1 alone could not promote the anoikisresistance of HCC cells, which might be one of the reasons that TGF-b1-treated HCC cells were unable to extravasate from circulation. H 2 O 2 /HOCl promoted the anoikis-resistance of HCC cells, since H 2 O 2 and HOCl could activate NF-kB [30,31], which can activate the expression of a group of antiapoptotic genes [32,33]. Although the enhancement of TGF-b1-induced Smad activation by H 2 O 2 /HOCl might have negative effect on anoikisresistance, the up-regulation of b3 reduced the effect of Smad pathway by enhancing the activation of MAPK pathways. The enhanced activation of MAPK pathways could promote apoptosisresistance of HCC cells, and antagonize the negative effect of Smad pathway on apoptosis-resistance [23]. Therefore, the upregulation of b3 enabled TGF-b1 to augment the promoting effect of H 2 O 2 /HOCl on anoikis-resistance. In line with this, TGF-b1 augmented the effect of H 2 O 2 /HOCl if b3 expression was upregulated, but attenuated the effect of H 2 O 2 /HOCl if the upregulation of b3 expression was suppressed.
In summary, in this study we demonstrated that b3 expression in non-metastatic HCC cells was up-regulated by TGF-b1 in presence of H 2 O 2 /HOCl. Importantly, b3 could promote TGF-b1/H 2 O 2 /HOCl-mediated induction of metastatic phenotype of non-metastatic tumor cells by enhancing TGF-b1 signaling. Simply increasing b3 expression might not be sufficient for promoting the metastatic capability, since b3 could not influence the activation of Smad pathway. However, TGF-b1/H 2 O 2 / HOCl could not induce the metastatic phenotype of HCC cells without b3. Our findings in this study suggest that targeting b3 might be a potential approach in preventing the induction of metastatic phenotype of non-metastatic tumor cells.

Matrigel invasion assay
Matrigel invasion assay was performed using Boyden chambers (Transwell, Corning, Inc., Corning, NY). The transwell filters were coated with matrigel (BD Biosciences). The lower chambers were filled with DMEM medium containing 10% FBS. 1610 5 tumor cells were placed in the upper compartment. After 24-h incubation at 37uC in a humidified incubator with 5% CO 2 , the non-invading cells were removed. The invasive cells attached to the lower surface of membrane insert were fixed, stained, and counted under a microscope from 5 randomly chosen fields in each membrane. The average number of the cells per field was calculated. When indicated, the cells were pre-incubated with 10 mg/ml of anti-a3 antibody (Santa Cruz Biotechnology) or anti-avb3 antibody (Chemicon) for 30 min. Matrigel invasion assay was then performed in the presence of antibody.

Analysis for actin polymerization
Tumor cells were incubated in matrigel-coated plate for 5 h. The cells were then fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and then stained with rhodaminephalloidin (Invitrogen) according to the manufacturer's protocol to visualize the cells with highly polymerized actin.

MMP assay by gelatin zymography
Tumor cells were cultured for 48 h in DMEM medium containing 1% FBS in presence of pre-coated matrigel. The assay of MMP-2 and MMP-9 in supernatants was performed as described previously [35].

Immunofluorescence and histology
Tumor cells were injected into mice via tail vein (2610 6 cells/ mouse). The lung tissues were harvested 4 weeks after inoculation. Frozen tissue sections were prepared and subjected to immunofluorescence analysis as previously described [36]. Anti-human HDGF (hepatoma-derived growth factor) antibody (Santa Cruz Biotechnology) was used as primary antibody. FITC-conjugated goat anti-rabbit IgG was used as secondary antibody. Images were obtained using a laser scanning confocal microscope (Olympus, FV500, Japan). For H&E staining, the lung tissues were embedded in paraffin according to standard histological procedures. Sections were stained with hematoxylin and eosin.

Assay of gene expression by real-time RT-PCR
Total RNA was extracted from cells with TRIzol reagent (Invitrogen). The relative quantity of mRNA was determined by real-time RT-PCR according to MIQE guidelines [37]. GAPDH, PPIA, and HPRT1 were chosen as reference genes. The relative expression of gene was calculated using GeNorm software. The primer sequences were as follows: ITGB3,

Flow cytometric analysis
Tumor cells were stained with FITC-conjugated mouse-antihuman b3 and a3 (Santa Cruz Biotechnology), or isotype control. Parameters were acquired on a FACS Calibur flow cytometer (BD Biosciences) and analyzed with CellQuest software (BD Biosciences). Percent staining was defined as the percentage of cells in the gate (M1) which was set to exclude ,99% of isotype control cells. The expression index was calculated by using the formula: mean fluorescence 6 percentage of positively stained cells [38].

Western blot assay
Western blot assay was done as described previously [39]. Abs were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and Cell Signaling Technology (Beverly, MA).

Cell transfection
To suppress b3 expression, tumor cells were transduced with b3 shRNA(h) lentiviral particles, or control shRNA lentiviral particles (Santa Cruz Biotech, Inc.) according to the manufacturer's protocol. After selection with puromycin, the cells were used for further experiments.

Assay of apoptosis and anoikis
For the assay of apoptosis, tumor cells were cultured under the indicated conditions for the indicated time. For the assay of anoikis, tumor cells were cultured (1610 6 /well) for 24 h in 6-well plates pre-coated with poly-HEMA (10 mg/ml, Sigma). The cells were then stained with Annexin V-FITC/Propidium Iodide (PI) apoptosis detection kit (BD Biosciences, San Diego, CA), and analyzed by flow cytometry.

Statistics
Data are pooled from three independent experiments with a total of six samples in each group. Results were expressed as mean value 6 SD and interpreted by one-way ANOVA. Differences were considered to be statistically significant when P , 0.05. Figure S1 The inhibitory effect of inhibitors on signaling pathways. HepG2 cells were untreated or treated for 7 days with T/H/H in absence or presence of SB203580 (10 mM), PD98059 (10 mM), SP600125 (10 mM), wortmannin (WT, 40 nM), QNZ (40 nM), SIS3 (2 mM), and SU6656 (10 mM). The phosphorylation of MK2 was detected to demonstrate the inhibition of p38 MAPK by SB203580. The phosphorylation of ERK was detected to demonstrate the inhibition of MEK by PD98059. The phosphorylation of c-Jun was detected to demonstrate the inhibition of JNK by SP600125. The phosphorylation of Akt was detected to demonstrate the inhibition of PI3K by wortmannin. The expression of iASPP was detected to demonstrate the inhibition of NF-kB by QNZ. The expression of PAI-1 was detected to demonstrate the inhibition of Smad3 by SIS3. The phosphorylation of p38 MAPK was detected to demonstrate the inhibition of Src by SU6566. (TIF)