Cell culture and reagents

The human GBC cell lines NOZ, EH-GB-1, EH-GB-2, SGC-996 and GBC-SD were purchased from the Cell Bank of Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The OCUG-1 cell line was obtained from the Health Science Research Resources Bank (Osaka, Japan). The NOZ cell line was maintained in William’s medium (Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Gibco). The GBC-SD cells were maintained in DMEM medium (Gibco). The EH-GB-1, EH-GB-2, SGC-996 and OCUG-1 cells were cultured in RPMI 1640 (Gibco). DHA was purchased from Selleck Chemicals and dissolved in DMSO. Primary antibodies against TCTP, vinculin, paxillin and β-actin were purchased from Cell Signaling Technology (Beverly, MA, USA).

Patients and clinicopathological data

Tumor tissue specimens were obtained from 73 GBC patients without any prior radiotherapy or chemotherapy who had undergone radical cholecystectomy between 2008 and 2014 at the Department of General Surgery, Xinhua Hospital, School of Medicine, Shanghai Jiaotong University, China. Additionally, 103 patients with cholelithiasis who underwent simple cholecystectomy were included as controls. The patients provided consent for their tumor tissues to be used for clinical research, and the Shanghai Jiaotong University Xinhua Hospital Ethical Committee approved the research protocol. All diagnoses of GBC, cholelithiasis, and lymph node metastasis were confirmed via histopathological examination, and all tissue samples were fixed in 4% formalin immediately after removal and then embedded in paraffin for immunohistochemical staining.

Immunohistochemical analysis of GBC tissues

After the tissues were sections, deparaffinized and quenched to block endogenous peroxidase activity, they were incubated in 1% bovine serum albumin (BSA) in PBS. The slides were subsequently incubated first with rabbit anti-human-TCTP antibodies (1:500, CST) and then with goat anti-rabbit IgG antibodies. The slides were counterstained with ChemMate Hematoxylin (DakoCytomation) and mounted and observed under a microscope (Olympus). The TCTP expression via IHC was blinded read by two pathologists independently. The sections were semi-quantitatively scored according to the amount and level of immunoreactivity, as follows: 0 (less than 5% positive cells), 1 (5–25% positive cells), 2 (25–50% positive cells), and 3 (greater than 50% positive cells); and 0 (no coloration); 1 (pale yellow); 2 (yellow); and 3 (reddish-brown). The two scores were then multiplied to achieve the following composite scores: 0, negative (−); 1–3, weak (+); 4–6, moderate (++); and 7–9, strong (+++).

Real-time quantitative PCR (qRT-PCR)

Total RNA was extracted using Trizol LS reagent (Invitrogen). The mRNA was reverse-transcribed using a SuperScript First-Strand Synthesis System (Invitrogen). PCR amplification was performed using the following primers: TCTP, 5′- TATTGGACTACCGTGAGG -3′ (sense) and 5′- CTCGGTGGAAGGACAAACTC -3′ (antisense); and β-actin, 5′-TTAGTTGCGTTACACCCTTTC-3′ (sense) and 5′-ACCTTCACCGTTCCAGTTT3′ (antisense). Real-time quantitative RT-PCR was performed following a standard SYBR Green PCR protocol using SYBR® Premix Ex Taq™ (Takara, Dalian, China). The relative level of TCTP mRNA was normalized to the housekeeping gene β-actin.

Wound healing assay

GBC cells were seeded into 6-well plates (Becton Dickinson). After the cells had grown to 100% confluence, a wound was made by scraping a conventional pipette tip across the monolayer. The cells were washed in PBS and placed in medium supplemented with or without 10% FBS. After the cells were incubated for 18 h at 37 °C, they were fixed and photographed.

Cell migration and invasion assays

Cell transwell migration and invasion assay were performed using Boyden chambers (BD Biosciences). The cells (NOZ, GBC-SD, and OCUG-1 cells at 2 × 10⁴ cells/well and EH-GB-2 and SGC-996 cells at 3 × 10⁴ cells/well) were seeded in the upper chamber in DMEM without FBS. In the lower chamber, 600 μl of DMEM containing 10% FBS was added. After the cells were incubated for 24 h and then fixed and stained with crystal violet. The cells in the upper chamber were removed, and three randomly selected fields of cells migrating through the membrane were photographed and counted using ImageJ software (NIH).

Transfection for knockdown and overexpression experiments

GBC cells were transfected to create TCTP-overexpressing or -knockdown cells, as previously described [17]. A short interfering RNA sequence that targeted Cdc42 (5′- TTCAGCAATGCAGACAATTAA-3′) was designed and synthesized by GenePharma (Shanghai Co., Ltd.). RNAi targeting human Cdc42 and a non-targeting control sequence were transfected using lip2000 (Invitrogen) according to the manufacturer’s instructions. A plasmid encoding myc-tagged active Cdc42 Q61L was purchased from Addgene (Addgene, Cambridge, USA).

Western blot analysis

GBC cell lines were treated with 40 μM DHA for 48 h, and whole-cell lysates were prepared for western blot analysis according to a previous report [22].

Immunofluorescence analysis

GBC cells were seeded onto glass slides and treated with DHA for 48 h. The cells were then washed once with PBS, fixed with 4% paraformaldehyde in PBS for 20 min, permeabilized with 0.1% Triton X-100 in PBS for 30 min at room temperature, and then blocked in 3% BSA in PBS for 1 h. The cells were then stained using an anti-Vinculin antibody (1:100 dilution, overnight at 4 °C). The cells were stained for F-actin using phalloidin (Beyotime Biotechnology). Cell nuclei were stained using DAPI, and the samples were mounted using DAPI Fluoromount-G media (Southern Biotech). The slides were viewed using an Olympus confocal microscope (Leica, Switzerland).

Cell adhesion and spreading

GBC cells were treated with DHA for 48 h, and 1 × 10⁴ cells were then resuspended in Opti-MEM and seeded on fibronectin-coated (25 mg/mL) coverslips for up to 6 h. The cells were fixed and processed for immunofluorescence staining and then examined using confocal microscopy. For cell spreading, the sizes of at least 50 cells were measured using Volocity software (PerkinElmer). To analyze cell adhesion, the number of cells that were attached to the surface was counted in three randomly selected fields under 40 × magnification.

In vivo metastasis assay

Female Balb/c SCID mice (6–8 weeks of age, weight 18–20 g) were purchased from the Shanghai Laboratory Animal Center of the Chinese Academy of Sciences (Shanghai, China). All mice were housed in specific pathogen-free (SPF) conditions according to the guidelines of the Ethics Committee of Xinhua Hospital, School of Medicine, Shanghai Jiaotong University. To study the effect of DHA on GBC cell metastasis in vivo, we used liver metastasis and lung metastasis mouse models. To induce liver metastasis, the spleens of the mice were injected with 2 × 10⁶ NOZ or EH-GB-2 cells. DHA was transfused via daily intraperitoneal injections, starting on the day after the surgical operation at a dose of 50 mg/kg. The control group was injected with PBS. After 2 months, the mice were euthanized, and the livers were removed for HE staining. In the lung metastasis model, NOZ and EH-GB-2 cells were infected with a lentiviral luciferase vector to generate NOZ-luciferase and EH-GB-2-luciferase cells. A total of 1 × 10⁶ cells of each of these lines were then injected into the mice via the lateral tail vein. DHA was transfused via daily intraperitoneal injections starting on the day after injection at a dose of 50 mg/kg. The trace of the fluorescent cells was monitored every 2 weeks using an IVIS Lumina II (Caliper Life Sciences, USA). The survival times were recorded for 90 days. The mice were sacrificed by CO₂ inhalation or pentobarbital injection when they had lost >10% of their body weight or when they became moribund (Fatal Plus). To observe tumor metastasis in lung, mice lungs were injected with 1 ml of 15% India ink (Yasutomo, San Francisco, CA) to visualize individual tumor nodules using a standard protocol.

Statistical analysis

Mann–Whitney U test was used to compare TCTP expression between the GBC patients. Kaplan-Meier plots were used for the survival analysis. All data are expressed as the mean values ± standard errors of at least three independent experiments. Statistical significance was calculated using the Mann–Whitney U test, and a p value less than 0.05 was considered significant in all tests. All analyses were performed using SPSS software version 19.0 (SPSS Inc., Chicago, IL, United States).

TCTP is associated with gallbladder cancer metastasis

To determine the role of TCTP in GBC progression, we used IHC to detect TCTP expression levels in 73 GBC specimens and 103 cholecystitis tissues (used as controls). More than 85% of the GBC specimens showed positive expression of TCTP (Additional file 1: Figure S1A). Despite the presence of inter-individual variation, TCTP protein levels were higher in GBC samples than in controls (Fig. 1a and statistical data, Fig. 1b). Furthermore, TCTP was expressed at higher levels in metastatic (including liver, lymph node and abdominal metastases) and invasive (including mircrovascular and neural invasion) GBC samples than in non-metastatic and non-invasive ones (Fig. 1c and d). We were particularly interested in evaluating the difference in TCTP expression levels between primary and metastatic tumors. We therefore obtained primary tumors with metastatic lymph nodes from 5 individual patients and sought to determine their TCTP mRNA expression levels using quantitative RT-PCR. In four out of five of these cases, the mRNA expression level of TCTP was noticeably higher in metastatic lymph nodes than in corresponding primary tumor tissues (Fig. 1e). To determine whether this increase in the expression of TCTP in tumors is potentially associated with reduced patient survival, we separated the GBC patients into the two following groups: 54 cases with high TCTP expression and 19 cases with low TCTP expression. As shown in Fig. 1f, the expression level of TCTP was negatively associated with patient survival. All of these data suggest that an increase in tumor expression of TCTP is associated with metastasis in patients with GBC.

TCTP promotes GBC cell migration and invasion

To further investigate the role of TCTP in GBC metastasis, we sought to determine the effect of depleting TCTP on GBC cell migration and invasion. We used shRNA transfection to knock down TCTP expression in the GBC cell lines NOZ and GBC-SD, which express high endogenous levels of TCTP (Fig. 2a). The shRNA efficiently knocked down TCTP expression by more than 90% (Fig. 2b) and was therefore used in subsequent functional studies. In the wound-healing assays, TCTP silencing blocked the migratory activity of NOZ and GBC-SD cells by 40–50% (Fig. 2c). Similarly, depleting TCTP in NOZ and GBC-SD cells resulted in decreases in cell migration and invasion by 80–90% and 65–70%, respectively, in comparison to scramble-treated control cells (Fig. 2d and e). These data suggest that TCTP is critical for migration and invasion in GBC cells.

DHA inhibits TCTP-dependent cell migration and invasion

Based on these findings, we next sought to determine whether inhibiting TCTP would reduce the migratory and invasive potential of GBC. DHA, a drug that has previously been used to treat malaria, has been shown to inhibit TCTP [20] (Additional file 1: Figure S1B, C). To test the hypothesis that DHA inhibits TCTP-mediated cell migration and invasion, we first assessed cell migration in multiple GBC lines, including the TCTP-positive cell lines NOZ, GBC-SD, OCUG-1, and EH-GB-1 and the TCTP-negative cell lines EH-GB-2 and SGC-996 (Fig. 2a). Treating NOZ, GBC-SD and OCUG-1 cells with DHA dramatically reduced migration (Fig. 3a and b). In contrast, DHA failed to inhibit migration in EH-GB-2 and SGC-996 cells (Fig. 3a and b). These findings indicate that DHA inhibits tumor cell migration in a TCTP-dependent manner. While these findings suggest an association between DHA’s anti-metastatic effects and TCTP status, tumor cell lines contain multiple mutations, each of which may also contribute to differential sensitivity to DHA. We engineered SGC-996 cells to over-express TCTP (Fig. 3c) and then treated these cells with DHA. SGC-996 cells that had acquired the expression of TCTP displayed 1.7-fold more cell migration than was observed in the cells expressing the control vector (Fig. 3d). However, in these cells, TCTP-dependent migration was completely blocked by treatment with DHA (Fig. 3d). In addition, while TCTP gene silencing reduced TCTP expression in NOZ cells (Fig. 3e), the cells in which TCTP was knocked down were not responsive to DHA and displayed no alterations in cell migration. In contrast, treating the control cells with DHA resulted in a 60% reduction in cell migration (Fig. 3f). Likewise, DHA abrogated invasion in TCTP-positive GBC-SD cells but did not block invasion in TCTP-negative EH-GB-1 cells (Additional file 1: Figure S1D). These data provide strong evidence supporting the hypothesis that DHA inhibits TCTP-mediated migration and invasion in GBC cells.

DHA affect cell attachment, spreading, and focal adhesion

It is well known that tumor metastasis is a multi-step cell process that involves cell attachment, spreading and adhesion and is not limited to migration and invasion. We used in vitro assays to estimate the effect of DHA on specific parameters associated with cell metastasis. As shown in Fig. 4a, GBC cells treated with DHA became oval-shaped and displayed inhibited cell pseudopodia formation. Moreover, treating cells with DHA resulted in an approximately 1.6-fold lower rate of attachment to a fibronectin-coated surface in both GBC-SD and OCUG-1 cells than was observed in untreated cells (Fig. 4b). Consistent with the results of cell attachment experiments, treatment with DHA also resulted in a marked reduction in spreading in both cell types (Fig. 4c). In addition, significant changes were observed in cell morphology following treatment with DHA. Whereas control cells exhibited an elongated, fibroblast-like morphology, DHA-treated cells had a flattened cell phenotype (Fig. 4d). Focal adhesions are composed of dynamic supramolecular structures that connect the actin cytoskeleton with the extracellular matrix. We examined focal adhesions by staining cells with antibodies against vinculin and E-cadherin, which are markers of focal adhesions. DHA induced focal adhesions in GBC-SD and OCUG-1 cells and increased the expression levels of vinculin and paxillin (Fig. 4d and e).

DHA reduces Cdc42 activation in TCTP-expressing GBC cells

Our data showed that DHA inhibits GBC cell migration in a TCTP-dependent manner, and recent studies have demonstrated that TCTP activates Cdc42 [16]. We therefore sought to determine whether DHA inhibits Cdc42 activity in GBC cells. GBC cells were treated with DHA for 48 h, and Cdc42 activation was then analyzed using immunoblotting against Cdc42-GTP. Treatment with DHA resulted in an approximately 60–70% reduction in Cdc42 activation in NOZ and GBC-SD cells (Fig. 5a). However, DHA did not reduce Cdc42 activation in SGC-996 cells, which do not express TCTP (Fig. 5b). These data suggest that DHA inhibits the TCTP-mediated activation of Cdc42. We next performed experiments to explore whether Cdc42 is required for TCTP-mediated migration. First, Cdc42 was depleted in NOZ cells using gene knockdown (Fig. 5c). Similar to treatment with DHA (Fig. 3), knocking down Cdc42 reduced cell migration (Fig. 5d). However, DHA had no further inhibitory effect on Cdc42-depleted cells, suggesting that Cdc42 is required for DHA-mediated inhibitory effects. We next tested whether the inhibitory effects of DHA could be overcome by constitutive activation of Cdc42. NOZ cells were transfected with an empty vector or active Cdc42Q61L (Fig. 5e) and then treated with DHA for 48 h. They were then seeded in a migration transwell chamber. DHA-inhibited cell migration was completely reversed by the expression of active Cdc42Q61L (Fig. 5f). Taken together, these data suggest that in GBC cells, DHA inhibits cell migration by inactivating Cdc42.

DHA reduces TCTP-dependent metastasis in vivo

These earlier in vitro promising results suggest that DHA serves as an effective inhibitor to block TCTP-dependent GBC metastasis. To test this hypothesis, we established in vivo models of GBC metastasis. As liver metastasis accounts for the primary metastasis in GBC, liver metastasis models were induced using TCTP-positive NOZ and TCTP-negative EH-GB-2 cell lines. The GBC cells were injected into the spleen to establish a spleen-to-liver metastasis model in immunodeficient mice, which were subsequently treated with DHA or vehicle control (PBS) via an IP injection every day for 8 weeks. After the mice were euthanized, their livers were collected, and the number of macroscopic metastatic lesions was quantified. Metastases could be easily detected in the liver in the vehicle-treated control group (Fig. 6a). Significantly fewer metastases (50% fewer) were observed in the DHA-treated mice than in the vehicle-treated controls (Fig. 6b). Similarly, 4 out of 10 mice developed liver metastases in response to treatment with DHA, whereas 7 of 10 vehicle control mice developed liver metastases (Fig. 6d). In contrast, DHA had no effect on metastases in mice injected with the TCTP-negative cell line EH-GB-2 (Fig. 6c and e). To further evaluate the effect of DHA on GBC cell-distant metastases, we employed an animal model of experimental pulmonary metastasis. First, GBC cells were engineered to express firefly luciferase by infecting the cells with a lentivirus containing the corresponding recombinant plasmid. We then inoculated immunodeficient mice with the GBC cells via tail intravenous injections. The mice were subsequently treated with DHA or vehicle control (PBS) via IP injection every day. The resulting fluorescent signals were monitored every week. As shown in Fig. 6f, an increasing amount of proton flux was observed in the vehicle-treated mice, indicating that more tumor cells had migrated to the lung in this group than in the DHA-treated group. Upon necropsy, we found that treatment with DHA resulted in significantly fewer metastases than were observed in the vehicle-treated control mice (Fig. 6g). Consistent with this reduction in metastases, treatment with DHA increased survival times to 3–4 weeks long than was observed in the control mice (Fig. 6h). However, the DHA efficacy was not observed in tumor metastases developed from the TCTP-negative line EH-GB-2 (Additional file 1: Figure S1E-G). Taken together, these data support the hypothesis that DHA inhibits TCTP-mediated GBC metastases and improves mouse survival.