Galectin-1 From Cancer-Associated Fibroblasts Promotes the Invasion and Metastasis of Gastric Cancer Through TGF-β1-Induced Epithelial-Mesenchymal Transition

Background The gastric cancer (GC) microenvironment has important effects on biological behaviors, such as tumor cell invasion and metastasis. However, the mechanism by which the GC microenvironment promotes GC cell invasion and metastasis is unknown. The present study aimed to clarify the effects and mechanism of galectin-1 (GAL-1, encoded by LGALS1) on GC invasion and metastasis in the GC microenvironment. Methods The expression of GAL-1/ LGALS1 was determined using western blotting, immunohistochemistry, and quantitative real-time reverse transcription PCR in GC tissues. Besides, methods including stable transfection, Matrigel invasion and migration assays, and wound-healing assays in vitro; and metastasis assays in vivo, were also conducted. Results GAL-1 from cancer-associated broblasts (CAFs) induced the epithelial ‐ mesenchymal transition (EMT) of GC cells though the transforming growth factor beta (TGF-β1)/ Sma- and mad-related protein (Smad) pathway, and affected the prognosis of patients with GC. The level of GAL-1 was high in CAFs, and treating MGC-803 and SGC -7901 cell line with the conditioned medium from CAFs promoted their invasion and metastasis abilities. Overexpression of LGALS1 promoted the expression of TGF-β1 and induced EMT of GC cell lines. A TGF-β1 antagonist inhibited the invasion and migration of GC cells. In vivo, overexpression of LGALS1 promoted GC growth and metastasis, and the TGF-β1 antagonist dramatically reversed these events. Conclusions These ndings suggested that high expression of GAL-1 in the GC microenvironment predicts a poor prognosis in patients with GC by promoting the migration and invasion of GC cells via EMT through the TGF-β1/Smad signaling pathway. The results might provide new therapeutic targets to treat GC. GAL-1/ GC, GAL-1, TGF-β1, phosphorylated (p)-Smad2/3 in GCT and NGCT IHC.


Background
Gastric cancer (GC) is one of most malignant tumors of digestive tract, with a high invasive ability.
Worldwide, more than 1 million new cases of GC occur every year, and the incidence and mortality of GC are the fth and third among all malignant tumors [1]. Thus, GC seriously threatens human health, especially in East Asian countries, such as China, South Korea, and Japan [2]. Although there has been considerable progress in the diagnosis and treatment of GC, the 5-year survival rate is still less than 30% in most countries because of the di culty of early diagnosis and the high recurrence and metastasis of GC [3]. Therefore, it is important to explore the molecular mechanism of GC invasion and metastasis and seek effective molecular markers for early diagnosis and prognosis, to improve the diagnosis and treatment of GC and improve the survival rate of patients with GC.
For a long time, tumor biologists only focused on the biological characteristics of tumor cells, ignored the decisive role of the microenvironmental components of non-tumor cells in tumorigenesis and development, which stalled the development of anti-tumor therapies. Recently, researchers found that the tumor and its microenvironment form an integrated structure, and began to study of the effect of the microenvironment non-tumor cells on tumorigenesis and development. The tumor microenvironment is a complex and functional environment, which includes the extracellular matrix and multiple types of stromal cells, such as mesenchymal stem cells, macrophages, in ammatory cells, and cancer-associated broblasts (CAFs) [4]. CAFs comprise a group of highly homologous activated broblasts from different parts of the tumor that have unique phenotypes [5]. Recent studies have shown that CAFs are highly activated in GC tissues, and are closely associated with the malignant potential of GC, such as tumor size, tumor invasion, metastasis, metabolism, and remodeling [6]. Further studies con rmed that CAFs promote the tumorigenesis, development, invasion, metastasis, and other malignant potentials of GC cells via the secretion of various cytokines that act on GC cells [5,7].
Galectin-1 (GAL-1), a member of the lectin family, is encoded by the LGALS1 gene, and is characterized by its a nity for glycans containing β galactosides [8]. In the tumor microenvironment, GAL-1 is a multivalent carbohydrate binding protein that mediates malignant cellular activities by cross-linking glycoproteins [9]. For example, GAL-1 can cluster cell surface glycoproteins, form lattices and larger aggregates, and cross-link receptors thought to be involved in various mechanisms [9,10]. GAL-1 can be activated by paracrine or autocrine sugar-dependent interactions with β-galactoside-containing glycoconjugates in the extracellular environment, and participates in tumor cell adhesion, migration, invasion, tumor-induced angiogenesis, and apoptosis via multiple interactions [11,12]. Previous studies have shown that GAL-1 is overexpressed in a variety of malignant tumors, including GC [13]. Generally, in malignant tumor tissues with high expression of GAL-1, tumor progression is regulated by interactions with sugar complexes in the tumor microenvironment. Speci cally, GAL-1 serves as a scaffold for vascular growth and vascular network formation by establishing physical connections between vascular endothelial cells and the extracellular matrix in the tumor microenvironment, which provides the necessary physical support for new vasculature [14][15][16] . However, how GAL-1 regulates GC invasion and metastasis in the GC microenvironment remains elusive.
The invasion and metastasis of GC is a complex multifactorial and multistep process. Many studies have found that the epithelial-mesenchymal transition (EMT) leads to downregulation of epithelial-associated markers, such as E-cadherin, and upregulation of mesenchymal markers, such as Vimentin, which plays an important role in the invasion and metastasis of GC [17]. During EMT, epithelial cells lose their polarity, reduce their contact with peripheral cells and stromal cells, and reduce intercellular interactions, which enhance cell migration and motility. Numerous signaling pathways are involved in EMT through cooperation and antagonism, such as transforming growth factor beta (TGF-β), Wnt/beta-catenin, and Ras-mitogen activated protein kinase (MAPKl). [18][19][20]. TGF-β is considered the most important factor that induces EMT in developmental processes and other pathological states. In some cultured epithelial cell lines, simple TGF-β stimulation can induce EMT. TGF-β signaling-mediated EMT can be achieved by classical the Sma-and Mad-related protein (Smad) pathway or non-Smad pathway. In the classical Smad pathway, TGF-β activates Smad2 and Smad3 and combines with Smad4. Then, Smad complexes will be transferred to the nucleus to mediate the inhibition or activation of target genes together with transcription factors. Meanwhile, Smad complexes can also induce the expression of microRNA in the nucleus, which inhibits the signature protein expression of epithelial cells and promotes the expression of proteins that confer mesenchymal cell properties, which facilitating EMT [21]. TGF-β signaling-mediated non-Smad signaling pathway, can activate phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)-protein kinase B (AKT)-mechanistic target of rapamycin (mTOR) signals for transcriptional regulation [22], and activated AKT can also trigger EMT by inhibiting ribonucleoprotein transcriptional regulation [23].
Our previous study showed that GAL-1 promoted the invasion, metastasis, and vasculogenic mimicry of GC via EMT [24]. However, the role of TGF-β signaling pathways in GAL-1-mediated promotion of GC EMT remains unclear. Clarifying the effects of GAL-1 on GC invasion and metastasis in GC microenvironment, and its molecular mechanism, will provide a new perspective and therapeutic targets to treat GC.

Materials And Methods
Tumor tissue samples Patients with primary gastric adenocarcinoma (n = 127) were enrolled in the present study. None of the patients had undergone preoperative neoadjuvant radiotherapy or chemotherapy, and all patients received radical gastrectomy with D2 lymphadenectomy at the Department of Gastrointestinal Surgery, Taizhou Clinical Medical School of Nanjing Medical University (Taizhou People's Hospital) of Jiangsu province. GC tissues and matching adjacent gastric mucosa tissues for IHC were formalin-xed and para n-embedded. GC tissues and matching adjacent mucosal tissues for molecular analysis were collected from 15 patients, which were stored in liquid nitrogen. Our study was approved by the Clinical Research Ethics Committee of Taizhou People's Hospital (TZRY-EC-12-068). Provided all patients with details regarding the assessment procedure, and all patients gave informed written consent.

Cell line and cell culture
The human gastric adenocarcinoma cell lines, MGC-803 and SGC-7901 were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Roswell Park Memorial Institute (RPMI) medium (Thermo Fisher Scienti c, Waltham, MA, USA) was used to culture the cells, The culture medium was supplemented with 10% (V/V) fetal bovine serum (FBS; Thermo Fisher Scienti c), 100 U/mL penicillin and 100 mg/mL streptomycin (Gibco, Grand Island, NY, USA). The cells were maintained at 37 °C in a humidi ed atmosphere containing 5% (V/V) CO 2 [24] When con uence reached 80%, the cells were passaged via trypsinization.

Lentiviral transduction
Lentiviral transduction was performed as previously described [24] . Lentiviral vectors for LGALS1 overexpression and silencing were constructed by Genechem Co. Ltd (Shanghai, China). The short hairpin RNA (shRNA) sequences and the lentiviral vector were designed in a previous report [15]. MGC-803 or SGC-7901 cells were seeded in 6-well plates at a concentration of 5 ´ 10 4 cells per well before lentiviral transduction. Cells were transduced with lentiviral vector and 10 μg/mL polybrene (Sigma-Aldrich, St. Louis, MO, USA) according to a multiplicity of infection of 10. The medium was replaced 12 h after transfection. Puromycin (Sigma-Aldrich) was added at 2 µg/mL to select stably transduced cell lines. After 48 h, the stably transduced cells were cultured with puromycin at 0.5 μg/mL. The transduction e ciency was con rmed under a uorescent microscope (OLYMPUS-U-HGLGPS-IX73, Olympus, Tokyo, Japan) after 72 h, and further con rmed by qRT-PCR and WB.

RNA extraction and qRT-PCR
An RNeasy Mini Kit (Invitrogen, Waltham, MA, USA) was used to extract total RNA, and a Reverse transcription kit (Takara, Shiga, Japan) was used to synthesize cDNA from the RNA. A SYBR Green dye kit (Roche Diagnostics, Mannheim, Germany) was used to performed qPCR and an iQ5 Multicolor Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) was used to analyze the reaction products. The thermocycling conditions were as follows: 95°C for 30 s; 40 cycles of 95 °C for 5 s, 60 °C for 30 s, and 72°C for 30 s. GAPDH was used as the reference control gene. The following primers were used: LGALS1 (forward): GCTGAACCTGGGCAAAGACAG an GTGGCGGTTGGGGAACTT (producing a 247 bp amplicon); and GAPDH (forward) TGACTTCAACAGCGACACCCA) and (reverse) CACCCTGTTGCTGTAGCCAAA (producing a 121 bp amplicon).

Western blotting analysis
Total cell extracts and nuclear extracts were prepared using an extraction kit (Beyotime, Shanghai, China).
Cell lysates (20 μg) were separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then the separated proteins were transferred to a nitrocellulose membrane (GE Healthcare Life Sciences, Pittsburgh, PA, USA). The blots were probed with antibodies against GAL-1, Ecadherin, Vimentin, TGF-β1, p-Smad2/3, and GAPDH at a dilution of 1:2000. Horseradish peroxidase (HRP)-conjugated goat anti-rabbit immunoglobulin was used as the secondary antibody also at a dilution of 1:2000. The immunoreactive protein bands were visualized using the West Pico chemiluminescent Substrate (Pierce, Carlsbad, CA, USA), and densitometric image analysis software (Image Master VDS; Pharmacia Biotech, Little Chalfont, UK) was used to quantify the visualized protein bands. The level of GAPDH was used as an internal reference. All experiments were performed in triplicate.
Wound-healing assay GC cells were cultured to a con uent monolayer (80-90 %) in a 6-well plate, scored across the cell surface to create a wound using a sterile plastic tip. The plates were washed three times with phosphatebuffered saline (PBS) to remove cellular debris, and then incubated with serum-free medium containing 10 μg/mL mitomycin C to block proliferation. The wound was photographed at 0 and 48 h. All assays were performed in triplicate.

Cell viability assay
The viability of the cells was determined using the MTT assay. Cells were seeded into a 96-well at bottom plate at 5 × 10 3 cells/well. The cells were cultured overnight at 37 °C with 5% CO 2 in a humidi ed atmosphere and then treated with ITD1 at various concentrations (5, 10, 15, and 20 μM). After 24 h, 20 μL of MTT was added to each well and incubated for 4 h. The supernatant was then removed, and 150 μL of DMSO was added to dissolve the formazan crystals. The absorbance was measured at 490 nm using a microplate reader (Model 550; Bio-Rad, Hercules, CA, USA).

Cell invasion and migration assays
The invasive and migratory ability of GC cells were measured using 24-well Transwell units with polycarbonate lters (pore size, 8.0 μm; Corning, Corning, NY, USA). The upper Transwell inserts were rst coated with or without 100 µl of Matrigel basement membrane (BD Biosciences, San Diego, CA, USA) and 100 µL of serum-free RPMI medium, then 1× 10 5 cells were seeded on the upper Transwell inserts. The lower chamber was loaded with Medium (600 μL) containing 10% FBS as a chemoattractant. Cells were allowed to migrate or invade at 37 °C for 24 h, then the non-invasive or non-migratory cells were removed. The lters were xed using 4% paraformaldehyde, and the cells were stained with a 0.05% crystal violet solution. Six elds were selected randomly to count the cells under a microscope at 100× magni cation for each sample. All assays were performed in triplicate.

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 construct subcutaneous GC implantation model and lung metastasis model of GC. These experiments were approved by the Ethics Committee of Yang Zhou University (YZU-EC-JS2742). A laminar ow cabinet under pathogen-free conditions was provided to breed the mice. MGC-803 cells (wild-type control), LGALS1-overexpressing (OE-LGALS1) MGC-803 cells, and OE-LGALS1 MGC-803 treated with ITD1 were separately inoculated into the right ank or the tail vein of the athymic mice (2 ´ 10 6 cells/mouse; n = 6). The mice in the subcutaneous model group were sacri ced on day 21 and the mice in the lung metastasis model group were sacri ced on day 50. The subcutaneous GC tumors or the lung were harvested for hematoxylin and eosin (H&E) staining, WB and immunohistochemical staining.
Statistical analysis SPSS 20.0 (IBM Corp, Armonk, NY, USA) was used to conduct the statistical analysis. Means ± standard error were used to express the continuous variables, one-way analysis of variance (ANOVA) and Dunnett's t test was used for comparisons between groups based on the normal distribution of the data. Pearson or Spearman's correlation coe cients were used to determine the relationship between two variables. Values of P < 0.05 were regarded as statistically signi cant.

Results
1. GAL-1/ LGALS1 is overexpressed in human GC tissues, which is derived from CAFs and promotes lymph node metastasis in GC To con rm that LGALS1 is associated with the malignant behavior of GC, we initially performed quantitative real-time reverse transcription PCR (qRT-PCR) to evaluate LGALS1 mRNA expression in 15 matched gastric cancer tissues (GCT) and non-gastric cancer tissues (NGCT). We observed a signi cant difference in LGALS1 mRNA expression between GCT and NGCT: GCT exhibited signi cantly higher levels of LGALS1 mRNA than that in NGCT (P < 0.01; Fig. 1A). We then examined GAL-1 protein levels in 15 pairs of GCT and NGCT using western blotting (WB), the results of which were consistent with the qRT-PCR results, demonstrating higher levels of GAL-1 in GCT than in NGCT (P < 0.01; Fig. 1B). Besides, the expression of GAL-1 was detected in GCT and in adjacent NGCT of 127 patients with GC using immunohistochemistry (IHC), Image Pro Plus (Media cybernetics, San Diego, CA, USA) was used to evaluated the digital images of IHC. The median IHC scores of GAL-1 in GCT and NGCT were 78.29 (9.51-186.24) and 31.09 (5.89-123.45), respectively. We observed a signi cant difference in GAL-1 expression between GCT and NGCT( Fig. C and Fig. D; P < 0.01), and we found that GAL-1 was mainly expressed by stroma tissue of GC (Fig. 1C) . To further con rm which cells produced GAL-1/ LGALS1, we used IHC to detect alpha smooth muscle actin (SMA-α) in GCT from 127 patients with GC. We found that GAL-1 was derived from cancer-associated broblasts (CAFs) with a SMA-α protein positive phenotype (Fig. 1C). Moreover, we found that GC cells expressed low or no GAL-1 (Fig. 1C). Surprisingly, we found that the GAL-1 was highly expressed in GC cells from metastatic lymph node tissues when detected using IHC (Fig. 1E). GAL-1 expression correlated positively with SMA-α expression in GCT (r = 0.963, P < 0.01; Fig. 1F). We classi ed the GAL-1 IHC scores as positive and negative using Receiver Operating Curve (ROC) statistics, and found that the lymph node metastasis rate of the GAL-1 positive group was signi cantly higher (64/86) than that in the GAL-1 negative group (7/41) (P < 0.01; Fig. 1G).

CAF-derived GAL-1/LGALS1 promotes the invasion and metastasis ability of GC cell lines in vitro.
To further clarify the origin of GAL-1/LGALS1, we cultured CAFs from human GC tissue, and GC cell lines (MGC-803, BGC-823, AGS and SGC-7901), in vitro and investigated the GAL-1 levels in all cells using WB, which con rmed that GAL-1 was strongly expressed in CAFs, with lower levels in GC cells (Fig. 2A). Meanwhile, we performed qRT-PCR to evaluate the expression of LGALS1 mRNA in all cell lines, the results of which were consistent with those of WB (Fig. 2B). MGC-803 cells and SGC-7901 were then treated with conditioned medium (CM) from CAFs (CM-CAFs) for 72 h, and then the expression of GAL-1 and LGALS1 mRNA were monitored by WB and qRT-PCR. Compared with untreated MGC-803 and SGC-7901 cells, we found that treatment of MGC-803 and SGC-7901 cells with CM-CAFs signi cantly increased the GAL-1 protein and LGALS1 mRNA levels ( Fig. 2C and 2D). We observed that the proliferation of MGC-803 and SGC-7901 cells were increased when treated with CM-CAFs for 48 h, and the proliferation effect was abolished when the medium contained 10 μg/mL mitomycin C (Fig. 2E). As shown in Figure 2F, MGC-803 and SGC-7901 cells treated with CM-CAFs exhibited a signi cantly enhanced migration capacity compared with that of the wild-type (untreated) control; the fold changes in migration are shown in Fig 2G (P < 0.01). Transwell assays showed that MGC-803 and SGC-7901 cells treated with CM-CAFs had increased cell invasion and migration abilities (P < 0.01, Figure 2H and 2I).
GAL-1/LGALS1 promotes EMT in GC, and EMT promotes lymph node metastasis of GC To determine whether GAL-1/LGALS1 promotes the invasion and migration abilities of GC cells via EMT, we examined the EMT-related biomarkers Vimentin and E-Cadherin in GC tissue using IHC. We identi ed E-cadherin-positive staining in GC cells and Vimentin staining in tumor stroma cells in 85 cases, indicating that EMT had not occurred (Fig. 3A). In 42 cases, the expression of Vimentin was positive in both tumor stroma cells and GC cells, and E-cadherin was decreased in those samples, indicating that EMT had occurred (Fig. 3B). The EMT rate in the GAL-1 positive group was 41.86% (36/86), while that in the GAL-1 negative group was 14.63% (6/41), suggesting that GAL-1 promotes signi cantly EMT in GC (P < 0.05, Fig. 3C). In addition, we found that GC cases with EMT were more prone to lymph node metastasis: There were 30 cases with lymph node metastasis among 42 patients with GC with EMT (71.43%, 30/42), while there were 41 cases with lymph node metastasis among 85 cases of GC without EMT (48.24%, 41/85) (P < 0.05, Fig. 3D). GC cells with or without EMT in metastatic lymph nodes are shown in Fig. 3E and 3F.

GAL-1/LGALS1 promotes the migration and invasion of GC cells in vitro through TGF-β/Smad signaling pathways
To investigate whether GAL-1/LGALS1 promotes the migration and invasion of GC cells via activation of the TGF-β/Smad pathway, the TGF-β/Smad signaling pathway speci c antagonist, ITD1, was used to explore the relationship between GAL-1-induced migration and invasion of GC cells in GC and TGFβ/Smad pathway activation. The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay indicated that 10 μM ITD1 did not affect the numbers of MGC-803 and SGC-7901 cells at 48 h. OE-LGALS1 increased the number of cells; however, the proliferation effect was abolished using 10 μg/mL mitomycin C. OE-LGALS1 MGC-803 cells exhibited signi cantly enhanced migration compared with the wild-type control (WC) and negative control-transfected MGC-803 cells (OE-con) (P < 0.01, Fig. 6A). However, the migration capacity was abolished when the medium contained 10 μM ITD1( P < 0.01, Fig.  6A). The fold changes in migration are shown in Fig 6B. To con rm that GAL-1 promotes migration and invasion in vitro, we repeated these experiments in SGC-7901 cells. OE-LGALS1 SGC-7901 cells showed an enhanced migration capacity compared with WC and OE-con SGC-7901 and 10 μM ITD1 abolished the migration capacity ( Fig. 6C and 6D). As shown in Figure  6E, Transwell assays demonstrated increased invasiveness of MGC-803 and SGC-7901 cells transfected with LV-LGALS1-OE, and the addition of 10 μM ITD1 abolished this capacity for invasion.

GAL-1/LGALS1 promotes GC cell metastasis in vivo through the TGF-β/Smad signaling pathway
Through clinical analysis and in vitro experiments, we demonstrated that GAL-1/LGALS1 promotes GC migration, invasion, and metastasis by an EMT-mediated process via the TGF-β/Smad signaling pathway. Subcutaneous GC implantation and lung metastasis models in athymic BALB/c mice (n = 6/group) were generated to further elucidate whether GAL-1/LGALS1 promotes GC growth and metastasis through EMT induced by activating TGF-β/Smad signaling. At 21 days after subcutaneous GC implantation, tumors in the LGALS1 overexpression group (OE-LGALS1) were larger and heavier than those in the WC group, and the tumors formed by OE-LGALS1 MGC-803 cells treated with ITD1 were signi cantly smaller and lighter than those in the WC group ( Figure 7A, B and C; P < 0.01). From day 12 onward, the tumor volume in the OE-LGALS1 group was signi cantly higher compared with that in the WC group, and the volumes of the tumors in the OE-LGALS1 MGC-803 cells treated with ITD1 were signi cantly lower (P < 0.01 and P < 0.05; Fig. 7B). Western blotting showed that the levels of Vimentin in the OE-LGALS1 group were increased, and the levels E-cadherin were decreased, which suggested that LGALS1 overexpression promoted EMT. Compared with the WC group, TGF-β1 and p-Smad2/3 levels were elevated in the OE-LGALS1 group (P < 0.01, Fig. 7D). However, ITD1 treatment reduced the levels of p-Smad2/3 signi cantly (P < 0.01, Fig. 7D), and inhibited EMT in the implanted GC cells ( Figure 7D) .
Larger pulmonary metastases were found in all the mice in the OE-LGALS1 group after fty days in the lung metastasis models. Four cases of pulmonary metastasis were found in the WC group, while the ITD1-treated OE-LGALS1 group did not have any pulmonary metastases (P = 0.002; Figure 7E). Pulmonary metastases in the OE-LGALS1 group and the WC group were identi ed using hematoxylin and eosin (H&E) staining ( Figure 7F). IHC staining was used to examine the TGF-β/Smad signaling markers in the pulmonary metastases, which indicated that GAL-1, Vimentin, TGF-β1, and p-Smad 2/3 levels were increased signi cantly and the E-cadherin level was decreased in the pulmonary metastases tissues of the OE-LGALS1 group ( Figure 7G). These results suggested that GAL-1 plays an important role in GC metastasis and EMT, and that TGF-β/Smad signaling pathway potentially contributes to this process.

Discussion
The growth and metastasis of a malignant tumor is a complex and multi-step process. Recent studies show that a variety of growth factors and cytokines released by tumor stromal cells create a microenvironment suitable for tumor growth, which promotes tumor proliferation, growth, invasion, and escape from immune surveillance, leading to distant metastasis [25]. GAL-1 is a multivalent carbohydrate-binding protein that regulates the activity of malignant tumor cells by cross-linking glycoproteins in the tumor microenvironment [26]. GAL-1 is synthesized on cytoplasmic ribosomes with a prototypical acetylated N-terminus but no signal peptide [27], then GAL-1 is transported from the nucleus to the medial side of the cell membrane and secreted into the extracellular tumor microenvironment [27,28] In extracellular environment, GAL-1 has high a nity with β-galactosides [27], and regulates homotypic aggregation of cancer cells by interacting with cell surface sugar complexes [8]. Furthermore, GAL-1 can mediate the adhesion of tumor cells to the extracellular matrix [12]. Laminin, bronectin, and other glycoproteins present in the basement membrane provide the necessary binding sites for GAL-1 to crosslink cells with the extracellular matrix, thus GAL-1 regulates cancer cell adhesion during metastasis through glycoproteins [12]. Conversely, GAL-1 inhibits adhesion between tumor cells and the extracellular matrix by competitive binding of matrix glycoproteins or cell surface sugar complexes [29]. This suggests another role of GAL-1, the regulation of tumor cell isolation, which allows tumor cells to detach from the primary site and migrate to secondary sites [29]. In the extracellular matrix, GAL-1 regulates the adhesion between tumor cells and the extracellular matrix, regulates the binding of the extracellular matrix to glycoproteins, and enhances the activities of proteolytic enzymes, thereby promoting the metastasis of tumor cells [12].
GAL-1 has been found in multiple tumor cells, including melanoma, lung cancer [30], pancreatic cancer [31], bladder cancer [32], thyroid cancer [33], cervical cancer [34], and colorectal cancer [35]. In the present study, we showed that the GAL-1 protein and LGALS1 mRNA levels in GC tissue were signi cantly higher than those in NGCT, suggesting that GAL-1/LGALS1 is associated with the malignant biological behavior of GC. We also observed that the GAL-1 in GC tissue is secreted by SMA-α-positive CAFs in the tumor microenvironment. WB and qRT-PCR also con rmed that the high expression of GAL-1 in CAFs. Intriguingly, GC cells showed low or no GAL-1 expression in GCT, but GC cells in metastatic lymph nodes showed high expression of GAL-1. Moreover, we treated GC cells with conditioned medium from CAFs cells, which increased GAL-1 expression in MGC803 and SGC-7901 cell lines, and promoted the invasion and metastasis of GC cell lines signi cantly. This indicated that GAL-1/LGALS1 promotes GC cells to acquire a metastatic phenotype and promotes GC cell metastasis. However, how GAL-1/ LGALS1 in the microenvironment regulates the invasion and metastasis of GC remains mostly unknown.
Many studies showed that EMT is a crucial step in the invasion and metastasis of malignant tumors [20]. Therefore, we hypothesized that GAL-1/LGALS1 promotes the invasion and metastasis of GC through EMT. To test this hypothesis, we examined the EMT-related biomarkers Vimentin and E-Cadherin in GC and metastatic lymph nodes with IHC, and found that the level of GAL-1 was related to EMT in GCT. In addition, we found that GC with EMT were more prone to lymph node metastasis.
The TGF-β family plays a vital role in EMT regulation [23]. The results of our IHC test also showed that the levels of TGF-β1 and p-Smad 2/3 in GC tissue were signi cantly higher than those in gastric mucosal tissue, and the level of TGF-β1 in GCT with GAL-1 positive expression was signi cantly higher than that in GAL-1 negative expression tissues. Moreover, the expression of GAL-1 in GCT correlated positively with TGF-β1 expression. These results suggested that GAL-1 in the GC microenvironment potential activates the TGF-β/Smad signaling pathway .
TGF-β1 is the prototypic member of a large family of functionally and structurally related proteins, which includes its close relatives TGF-β2 and TGF-β3, and growth and differentiation factors (GDF), Mullerianinhibiting substance (MIS), and bone morphogenetic proteins (BMPs) [36]. In the 1980s, TGF-β1 was discovered as a secreted factor that induced the growth of normal rat kidney cells in soft agar. TGF-β1 transmits signals from cell surface receptors to the nucleus, and the Smads family of proteins play key roles in this process. The Smads family are divided into receptor-regulated (R-) Smads (R-SMAD1, -2, -3, -5 and -8), the common (Co-) Smads (Smad4), and inhibitory I-Smads (Smad -6 and -7). R-Smads interact with, and become phosphorylated, by activated TGF-β type I receptor kinases. Co-Smads form heteromeric complexes with activated R-SMADs, while I-Smads antagonize canonical Smad signaling [37]. In the case of the canonical Smad pathway, TGF-β1 exerts its cellular effects through cell surface TGF-β type I and type II receptors. TGF-β1 binds to type II receptors, which recruit and phosphorylate type I receptors, and the type I receptors rephosphorylate R-Smad proteins (Smad2 and Smad3), which bind to coSmad (Smad4). The R-Smad/coSmad complex aggregates in the nucleus as a transcription factor and cooperates with other transcription regulators to modulate target gene expression; therefore, Smad2/3 act as intracellular transcriptional effectors of TGF-β family receptor signaling.
In the present study, IHC examination of GC tissue showed that the GAL-1 levels in the GC microenvironment correlated positively correlated with TGF-β1 in GCT, and GCT with high galectin expression had signi cantly higher levels of p-Smad2/3. The levels of TGF-β1 and p-Smad2/3 increased in GC cell lines overexpressing LGALS1, which underwent EMT and showed enhanced invasion and migration abilities. The overexpression of LGALS1 promoted subcutaneous tumor growth and lung metastasis in nude mice. This subcutaneous growth and lung metastasis was inhibited in nude mice when the cell lines were treated with ITD1, a speci c inhibitor of the TGF-β signaling pathway. This further con rmed that GAL-1/LGALS1 promotes the invasion and metastasis of GC through the TGF-β/Smad signaling pathway.
TGF-β receptors are widely expressed in human cells and mediate various biological phenomena, such as embryonic development, organogenesis, immune surveillance, and tissue repair. Alterations in TGF-β signaling lead to many diseases, including cancer. At the early stage of a malignant tumor, TGF-β signaling acts to inhibit tumorigenesis by inducing apoptosis of premalignant cells. Meanwhile, when cancer cells acquire oncogenic mutations, they become resistant to TGF-β-induced apoptosis. TGF-β can induce tumor cells to undergo EMT, leading to metastasis and chemotherapy resistance [38].
EMT-induced changes in tumor cells and the microenvironment are the main factors that help cancer stem cells to escape from the primary site [39]. Previous studies have found that TGF-β, chemokine 4/12, interleukin-6, and tumor necrosis factor-α in the tumor microenvironment can enhance EMT. EMT results in more epithelial growth factors being secreted by tumor cells and leads to an acidic, hypoxic, and high interstitial uid pressure state in the microenvironment, which activates CAFs [40]. In the present study, we found that activated CAFs in GC secreted GAL-1, which can activate the TGF-β/Smad signaling pathway and promote EMT in GC. Therefore, this represents a positive feedback loop that ultimately promotes the invasion and metastasis of GC.
The clinical specimen analysis and the in vitro and in vivo experiments all con rmed that GAL-1 can promote the invasion and metastasis of GC through the TGF-β/Smad signaling pathway. However, the mechanism by which GAL-1/LGALS1 in GC microenvironment activates the TGF-β/Smad signaling pathway in GC cells remains elusive. Further experiments are required to reveal its speci c molecular mechanism to provide new targets for targeted therapy of GC.

Conclusion
Taken together, the results of the present study suggest that GAL-1, derived from CAFs in the GC microenvironment, promotes GC cells to acquire a metastatic phenotype, leading to lymph node metastasis. TGF-β/Smad signaling pathway components are activated in GC by GAL-1, resulting in EMT, which controls the initiation of invasion and metastasis in GC. These ndings provided evidence that the microenvironment and the GAL-1/TGF-β/Smad pathway plays important roles in invasion and metastasis in GC. This provides novel insights into the mechanisms underlying invasion and metastasis in GC, and might contribute towards the identi cation of a therapeutic target for GC.
However, further study is required to explore the mechanism of GAL-1/LGALS1-mediated activation of TGF-β/Smad signaling. Taking the microenvironment of GC as the breakthrough point for the treatment of GC, inhibition of this signaling pathway might reverse the EMT of GC and improve therapeutic outcomes for patients with refractory GC.    (E) Representative images of metastasis (yellow arrows) in the lungs at 50 days after inoculation, and