Galectin-1 promotes vasculogenic mimicry in gastric adenocarcinoma via the Hedgehog/GLI signaling pathway

GLI1 is overexpressed in GC tissues

In order to understand the role of Hh signaling in VM in GC, we detected the expression of GLI1 in 127 GC tissues and matched non-tumor tissues. For the GC tissues, the median GLI1 IHC score was 65.08 (6.73-164.78), whereas that of the matched non-tumor tissues was 35.21 (3.16-98.12) (Figure 1A). A significant difference in GLI1 expression was observed between GC and non-tumor tissues (P < 0.01; Figure 1B). The receiver operating curve statistic was used to distinguish positive and negative expression of GLI1 in 127 GC samples in accordance with the cutoff point for IHC scores. Scores ≥ 78.38 were considered to indicate positive expression (Figure 1C). This indicated a rate of positive GLI1 expression in GC tissues of 39.4% (50/127), and 8.66% (11/127) in the matched non-tumor tissues, which is a significant difference (χ² = 32.81; P < 0.01).

Figure 1. GLI1 is overexpressed in GC tissues (A) Representative images of IHC for GLI1 protein expression in GC tissues and matched non-tumor tissues. (B) GLI1 IHC scoring compared between tumor and matched non-tumor tissues. (C) ROC statistics were employed to estimate the cut-off points of GLI1 IHC scoring in human GC tissue.

GAL-1 expression is correlated with GLI1, and GLI1 expression is correlated with VM

To determine whether GAL-1 can activate Hh signaling and induce VM in GC tissue, we examined GAL-1 in GC tissue using IHC and examined VM using CD34/PAS double-staining. The median IHC score for GAL-1was 78.29 (9.51-186.24) in GC tissues. The GAL-1 IHC scores in GLI1-positive GC tissues were significantly higher than the GAL-1 IHC scores in GLI1-negative GC tissues (P <0.01; Figure 2A, 2B). GAL-1 IHC scores positively correlate with GLI1 IHC scores in GC tissues (r = 0.958; P < 0.01; Figure 2C). As shown in Figure 2D, one GC specimen exhibited endogenous cell-dependent vessels and VM. Red blood cells are indicated by the red arrow in VM and endogenous cell-dependent vessels. CD34/PAS double staining indicated VM in 29/127 cases (22.8%). GC tissues with GLI1 high expression were more likely to exhibit VM. The GLI1 IHC scores of GC tissues with VM were significantly higher than those without VM (P < 0.01; Figure 2E). These results suggested that GAL-1 may contribute to VM formation in GC through activating Hh signaling and promoting GLI1 expression.

Figure 2. GAL-1 expression is correlated with GLI1, and GLI1 expression is correlated with VM. (A) Representative images of IHC indicating GLI1 and GAL-1 protein expression in GC tissues (×200 magnification). (B) The GAL-1 IHC scores in GLI1-positive GC tissues were significantly higher than the GAL-1 IHC scores in GLI1-negative GC tissues(P < 0.01) (C) GAL-1 IHC scores were positively correlated with GLI1 IHC scores in GC tissues (r = 0.958; P < 0.01). (D) GC tissues exhibiting high GLI1 expression were more likely to form VM (endogenous cell-dependent vessels are indicated by the red dotted line, VM is indicated by the green dotted line, and red blood cells are indicated by red arrows in VM and endogenous cell-dependent vessels; ×400 magnification). (E) The GLI1 IHC scores in GC tissues with VM were significantly higher than those without VM (P < 0.01).

GAL-1 promotes GLI1 expression both in vitro and in vivo

To confirm whether GAL-1 activation of Hh signaling promotes GLI1 expression in GC, LGALS1 was overexpressed and silencing in MGC-803 and AGS cells, confirmed by western blotting (Figure 3A–3D) and qRT-PCR (Figure 3E–3H). According to these results, shGal1#3 and LV-LGALS1-OE were selected for silencing and overexpression of GAL-1 in subsequent experiments, respectively. Western blotting indicated that LGALS1-overexpressing (OE-LGALS1) MGC-803 cells efficiently induced GLI1 protein expression compared with untransduced MGC-803 cells (wild-type control (WC) or negative control-transfected (OE-CON) cells) (Figure 4A). LGALS1- silencing (sh-LGALS1) MGC-803 cells efficiently inhibited GLI1 protein expression compared with WC or sh-CON MGC-803 cells (Figure 4B). We next performed qRT-PCR to evaluate LGALS1 and GLI1 mRNA expression in all cell lines, the results of which were consistent with those of western blotting (Figure 4E and 4F).

Figure 3. Lentiviral vectors were used to overexpress or silencing LGALS1 using 3 shRNAs. Western blot analysis showing decreased GAL-1 protein expression in (A) MGC-803 cells, and (C) AGS cells transfected with LGALS1 shRNAs, compared with cells transfected with control shRNA (shCtl). Western blot showing GAL-1 protein expression following overexpression of LGALS1 (OE-LGALS1) compared to the empty vector (OE-Con) in (B) MGC-803 cells, and (D) AGS cells. Reduced expression levels of LGALS1 mRNA in (E) MGC-803 cells, and (G) AGS cells transfected with LGALS1 shRNAs. Increased expression levels of LGALS1 with OE-LGALS1 in (F) MGC-803 cells, and (H) AGS cells transfected with OE-LGALS1. GAPDH was used as a loading control. (All n = 3, ^*P < 0.05, ^**P < 0.01).

To confirm that GAL-1 promotes GLI1 expression in vitro, we repeated these experiments in AGS cells. OE-LGALS1 AGS cells efficiently induced GLI1 protein expression (Figure 4C) and sh-LGALS1 AGS cells efficiently inhibited GLI1 protein expression (Figure 4D). These results were confirmed at the mRNA level by qRT-PCR (Figure 4G and 4H).

Figure 4. LGALS1 promotes GLI1 expression in vitro. (A and C) Western blotting indicating that OE- LGALS1 efficiently induced GLI1 protein expression compared with wild-type control and negative control-transduced cells. (B and D) Transfection with shLGALS1#3 inhibited GLI1 protein expression compared with wild-type control and negative control-transduced cells. These results were confirmed at the mRNA level by qRT-PCR (E–H). GAPDH served as a loading control. (All n = 3, ^*P < 0.05, ^**P < 0.01).

IHC scores of GC tissues from the subcutaneous GC mice indicated that GLI1 expression was positively correlated with that of GAL-1 in all groups (P < 0.05; Figure 5A). Compared with the WC and OE-CON groups, GLI1 expression was elevated in the OE-LGALS1 group (P < 0.01; Figure 5B) and reduced in the sh-LGALS1 group (P < 0.01; Figure 5C). LGALS1 and GLI1 mRNA expression was consistent with the IHC results (Figure 5D and 5E).

Figure 5. LGALS1 enhances GLI1 expression in vivo. (A) GLI1 and GAL-1 IHC scores were positively correlated in subcutaneous GC tissues from all groups (P < 0.05; ×200 magnification). (B) IHC scores indicating that GLI1 expression was elevated in the OE-LGALS1 group (P < 0.01). (C) IHC scores indicating that GLI1 expression was abrogated in the shLGALS1#3 group (P < 0.01). LGALS1 and GLI1 mRNA expression levels in (D) OE LGALS1-transduced and untransduced cells, and (E) shLGALS1#3-transduced and untransduced cells.

GLI1 promotes VM in vivo and in vitro

To confirm whether GLI1 promotes VM in vivo and in vitro, lentiviral vectors were used to silencing or overexpress GLI1 in MGC-803 and AGS cells. GLI1 silencing and overexpression was confirmed by western blotting (Figure 6A–6D) and qRT-PCR (Figure 6E–6H). According to these results, shGLI1#3 (for GLI1 silencing) and LV-GLI1-OE (for GLI1 overexpression) were selected for subsequent experiments in vivo and in vitro.

Figure 6. Lentiviral vectors were used to overexpress or silencing GLI1. Western blot analysis of GLI1 protein expression in (A) MGC-803 cells, and (C) AGS cells transfected with control shRNA (shCtl) or different GLI1 targeting sequences (shGLI1#1, shGLI1#2, shGLI1#3). Western blot showing GLI1 protein expression following overexpression of GLI1 (OE-GLI1) or the empty vector (OE-CON) transfection in (B) MGC-803 cells, and (D) AGS cells. Expression levels of GLI1 mRNA in (E and G) shGLI1 transduced and untransduced cells, and (F and H) OE-GLI1 transduced and untransduced cells. GAPDH served as a loading control. (n = 3, ^*P < 0.05, ^**P < 0.01).

Matrigel three-dimensional culture showed that OE-GLI1 transduction efficiently increased the capacity for tube-formation compared with WC or OE-CON MGC-803 and AGS cells. Meanwhile, sh-GLI1 transduction reduced the ability to form tube-like structures compared with WC or sh-CON MGC-803 and AGS cells (Figure 7A). Following subcutaneous GC implantation, the expression of GLI1 and the formation of VM structures in the OE-GLI1 and sh-GLI1 groups were detected by IHC and CD34/PAS double staining. CD34/PAS staining indicated that VM was significantly increased in the OE-GLI1 group and was not observed in the tumor tissues from the sh-GLI1 group (Figure 7B).

Figure 7. GLI1 promotes VM in vivo and in vitro. (A) Matrigel three-dimensional culture showing that GLI1 overexpression enhanced tube formation in MGC-803 and AGS cells, while silencing of GLI1 in MGC-803 and AGS cells inhibited their ability to form tube-like structures (×40 magnification; n = 3). (B) Expression of GLI1 and the formation of VM structures in the OE-GLI1 and sh-GLI1 subcutaneous GC mouse groups detected by IHC and CD34/PAS double staining. VM was significantly increased in the OE-GLI1 group, and was absent in the shGLI1#3 group (endogenous cell-dependent vessels are indicated by the red dotted line, VM is indicated by the green dotted line, and red blood cells are indicated by red arrows in VM and endogenous cell-dependent vessels; ×400 magnification).

GAL-1 promotes VM via the Hh signaling pathway

To investigate whether GAL-1 promotes VM in GC through activation of the Hh pathway, the GLI1-specific antagonist, GANT61, was used to explore the relationship between GAL-1-induced VM in GC and Hh pathway activation. An MTT assay indicated that 10 μM GANT61 did not affect cell number at 24 h. However, Matrigel three-dimensional culture demonstrated that OE-LGALS1 efficiently increased tube-formation capacity in MGC-803 cells and AGS cells, and that GANT61 can significantly inhibit this effect (Figure 8A). Meanwhile, sh-LGALS1 reduced formation of tube-like structures in MGC-803 cells and AGS cells. However, simultaneous silencing of LGALS1 and overexpression of GLI1 in MGC-803 cells and AGS cells rescued this reduction in tube-formation capacity (Figure 8B). These results suggest that GLI1 affects VM formation independent of GAL-1.

Figure 8. GAL-1 increases VM through the Hh signaling pathway. (A) Matrigel three-dimensional culture indicating that OE-LGALS1 efficiently increased tube-formation by MGC-803 cells and AGS cells, and the inhibition of this effect by GANT61. (B) sh-LGALS1 reduces the ability of MGC-803 cells and AGS cells to form tube-like structures. Simultaneous silencing of LGALS1 and overexpression of GLI1 rescued this reduction in tube-formation capacity. (×40 magnification; n=3).

To further investigate regulation of VM by GLI1, the expression of VM-related molecules was measured by qRT-PCR and western blotting in OE-GLI1 and sh-GLI1 MGC-803 and AGS cells, including MMP14, MMP2 and Ln5γ2 (LAMC2). We found that the mRNA expression of MMP2 (P < 0.05), MMP-14 (P < 0.05) and LAMC2 (P < 0.05) was significantly lower in sh-GLI1 cells than in the WC and negative control groups. The mRNA expression of MMP14 (P < 0.01), MMP2 (P < 0.01) and LAMC2 (P < 0.01) was significantly higher in OE-GLI1 cells than the WC and negative control groups (Figure 9A and 9B). Western blotting confirmed that protein expression was consistent with that of the mRNAs (Figure 9C and 9D).

Figure 9. GLI1 affects expression of VM markers - in MGC-803 cells and AGS cells. qRT-PCR analysis of MMP14, MMP2 and LAMC2 mRNA expression in (A) MGC-803 cells, and (B) AGS cells. Western blot analysis of MMP14, MMP2, and LAMC2 protein expression in (C) MGC-803 cells, and (D) AGS cells. (All n=3, ^*P < 0.05, ^**P < 0.01).

Tumor tissue samples

We enrolled 127 patients with gastric adenocarcinoma. No patients had received preoperative neoadjuvant chemotherapy or radiotherapy, but all patients underwent radical gastrectomy and D2 lymphadenectomy at the Gastrointestinal Surgery Department, Taizhou People’s Hospital of Jiangsu province. GC tissue and matching adjacent non-tumor tissue for immunohistochemistry (IHC) and histochemical staining were formalin-fixed and paraffin-embedded. GC tissue and matching adjacent non-tumor tissue was collected from 15 patients for molecular analysis. Our study was approved by the Clinical Research Ethics Committee of Taizhou People’s Hospital (TZRY-EC-12-068).

Animal models

Five-week-old male athymic mice were purchased from the Comparative Medicine Centre of Yangzhou University (Yang Zhou, JiangSu, China), and used to establish subcutaneous GC implantation. These experiments were approved by the Ethics Committee of Yang Zhou University (YZU-EC-JS2352). The mice were bred in a laminar flow cabinet under pathogen-free conditions. LGALS1-overexpressing (OE-LGALS1) MGC-803 cells, LGALS1- silencing (sh-LGALS1) MGC-803 cells, GLI1-overexpressing (OE-GLI1) MGC-803 cells, silencing GLI1 MGC-803 cells (sh-GLI1), wild-type control MGC-803 cells or vector control MGC-803 cells were inoculated into the right flank (2x10⁶ cells/mouse; n = 6). Mice were sacrificed on day 21 and the subcutaneous GC tumors were harvested. Subcutaneous GC tumors were examined by hematoxylin and eosin (H&E) staining and CD34-PAS dual staining.

Reagents and antibodies

GANT61 was purchased from Sigma Biotechnology (St. Louis, MO, USA). The anti- GAL-1 antibody was purchased from Cell Signaling Technology (13888S and 12936S, Danvers, MA, USA), the anti-GLI1 was purchased from Abcam (ab217326, Cambridge, UK) and Bioss (bs-1206R, Beijing, China), anti-MMP14 (ab78738), anti-MMP2(ab92536), anti-LAMC2 (ab210959 and ab274376) and anti-CD34(ab81289) antibodies were purchased from Abcam (Cambridge, UK), and the anti-GAPDH antibody(sc-47724), horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG(sc-516102) and goat anti-rabbit IgG(sc-2357) were provided by Santa Cruz Biotechnology (Santa Cruz, CA, USA). The PAS staining kit was provided by Leagene Biotechnology Co, Ltd (Beijing, People’s Republic of China). The MTT assay kit and dimethyl sulfoxide (DMSO) were provided by Sigma Biotechnology (St. Louis, MO, USA).

Cell line and cell culture

The human gastric adenocarcinoma cell line, MGC-803, was provided by The Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in RPMI (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% (V/V) fetal bovine serum (FBS; Thermo Fisher Scientific), 100 U/ml penicillin and 100 mg/ml streptomycin (Gibco, Grand Island, Waltham, MA, USA). Cells were maintained at 37°C in a humidified atmosphere containing 5% (V/V) CO₂ [15], and passaged via trypsinization when 80% confluence was reached.

Lentiviral transduction

Lentiviral transduction was performed as previously described [14]. Lentiviral vectors for LGALS1 and GLI1 overexpression and silencing were constructed by Genechem Co. Ltd (Shanghai, China). Three different short hair RNAs (shRNAs) were designed against both LGALS1 and GLI1 and cloned into the GV248(hU6-MCS-Ubiquitin-EGFP-IRES-puromycin) vector. The shRNA sequences of for LGALS1 and GLI1 are shown in Table 1. The GV358 (Ubi-MCS-3FLAG-SV40-EGFP-IRES-puromycin) lentiviral vector was constructed to upregulate the expression of LGALS1 and GLI1. Before lentiviral transduction, MGC-803 or AGS cells were seeded in 6-well plates at a concentration of 5x10⁴ cells per well. According to a multiplicity of infection of 10, cells were transduced with lentiviral vector and 10 μg/ml polybrene (Sigma-Aldrich, St. Louis, MO, USA). Additionally, a non-targeting negative control lentiviral vector was transduced by the same approach. Medium was replaced 12 h after transduction, and 2 μg/ml puromycin (Sigma-Aldrich) was added after a further 48 h for selection of stably transduced cell lines. The stably transduced cells were cultured at 37°C in the presence of 0.5 μg/mL puromycin. After 72 h, transduction efficiency was evaluated using a fluorescent microscope (OLYMPUS-U-HGLGPS-IX73), and further confirmed by quantitative reverse transcription-PCR (qRT-PCR) or western blotting.

RNA extraction and qRT-PCR

Total RNA was extracted using an RNeasy Mini Kit (Invitrogen, Waltham, MA, USA), and cDNA was synthesized using a Reverse transcription kit (Takara, Shiga, Japan), according to manufacturer’s instructions. qRT-PCR was performed with SYBR Green dye (Roche Diagnostics, Mannheim, Germany) and analyzed using an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The thermocycling conditions were as follows: 95°C for 30 s, then by 40 cycles of 95°C for 5 s, 60°C for 30 s and 72°C for 30 s. Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as the reference control gene. The primers sequences used are listed in Table 2.

Western blotting

Extraction kit (Beyotime, Shanghai, China) was used to isolate total-cell extracts and nuclear extracts. For all samples, 20 μg cell lysate was separated by 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and the separated proteins were transferred into a nitrocellulose membrane (GE Healthcare Life Sciences, Pittsburgh, PA, USA). Blots were probed with anti-GLI1, anti-GAL-1, anti-MMP14, anti-MMP2, anti-LAMC2 or anti-GAPDH primary antibody at a dilution of 1:2,000. The secondary HRP antibody was also used at a dilution of 1:2,000. Protein bands were visualized using West Picochemiluminescent Substrate (Pierce, Carlsbad, CA, USA) and quantified using densitometric image analysis software (Image Master VDS; Pharmacia Biotech). GAPDH was used as a loading control. All western blotting analysis assays were performed in triplicate.

Cell viability assay

The 3-(4-5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazoliumbromide dye reduction assay (MTT assay) was used to determine cell viability. Cells were seeded into a 96-well flat bottom plate at 5×10³ cells/well. The cells were cultured at 37°C with 5% CO₂ in a humidified atmosphere. After an overnight incubation, the cells were treated with various concentrations of GANT61 (5μM, 10 μM, 15 μM and 20 μM). After GANT61 treatment for 24 h, 20 μL MTT was added to each well and incubated for 4 h. Following the subsequent removal of the supernatant, 150 μL dimethyl sulfoxide (DMSO) was added to dissolve the formazan crystals. Absorbance was measured at 490 nm with an Enzyme-linked Immunosorbent Detector (Model 550; Bio-Rad, Hercules, CA, USA).

IHC

IHC was performed for human GC tissues, matching adjacent non-tumor tissue and murine subcutaneous GC tumor tissue. GAL-1 (dilution, 1:200), GLI1 (dilution, 1:200) or CD34 (dilution, 1:100) primary antibodies were incubated with the slides overnight at 4°C, the following steps were performed as per our previous report [13]. GAL-1 and GLI1 staining was imaged digitally using the same light exposure and evaluated using Image Pro Plus (IPP), a digitalized IHC scoring program (Media Cybernetics, San Diego, CA). The immunostaining results are expressed as the percentage of positively stained cells multiplied by staining intensity score (0, 1, 2 or 3) to yield scores of 0-300.

Three-dimensional culture

Coating of 24-well plates with 200 μl growth factor-reduced Matrigel (BD Biosciences, San Diego, CA, USA) was carried out, and incubated at 37°C for 1 h for polymerization. A total of 1x10⁵ cells in 600 μl of medium were plated onto the polymerized gel and incubated cells at 37°C for 24 h. The cells were imaged under an inverted microscope (OLYMPUS-U-HGLGPS-IX73). All experiments were performed in triplicate.

CD34-PAS dual staining and VM evaluation

All specimens were formalin-fixed and paraffin-embedded. After CD34 IHC described above, PAS staining was performed, according to the kit manufacturer’s protocol. After the DAB reaction, sections were treated with 0.5% periodic acid solution for 10 min. Prior to staining with Schiff solution for 20 min at room temperature, sections were washed with distilled water 3 times. Sections were then counterstained with hematoxylin, dehydrated, cleared and mounted. VM was analyzed at x200 magnification in 10 randomly selected fields of view. Channel-like structures stained positively for PAS and negatively for CD34, and containing red cells were considered to have a VM-positive status.

Statistical analysis

Statistical analysis was conducted using SPSS software (version 16; Chicago, IL, USA). Categorical variables were compared using the Chi-squared test. Continuous variables are presented as the mean ± standard error of the mean, Student’s t-test was used to compare between groups based on the normal distribution of the data. Pearson or Spearman’s correlation coefficient were used to determine the relationship between two variables. P < 0.05 was considered to indicate a statistically significant difference.

Ethical approval

This clinical study was approved by the Clinical Research Ethics Committee of Taizhou People’s Hospital (TZRY-EC-12-068), all patients consented to participate in our study. The animal experiments were approved by the Ethics Committee of Yang Zhou University (YZU-EC-JS2352).