Recombinant mutant forms of human galectin-9 that lack the linker peptides were expressed and purified as described (17). Alexa Fluor^® 647 Annexin V, propidium iodide (PI), PE anti-human Tim-3 conjugated antibody (clone: F38-2E2) and its isotype, Human TruStain FcX IgG and Annexin V binding buffer were obtained from Biolegend (San Diego, CA, USA). Anti-DYRK1A antibody (ab65220) was provided by Abcam (Cambridge, UK). Anti-β-actin monoclonal antibody (clone: AC-15) was supplied from Sigma-Aldrich (St. Louis, MO, USA).

The HCC cell line HLE was obtained from the Japanese Cancer Research Resources Bank (Osaka, Japan). Li-7 and Huh7 were supplied by Riken Cell Bank (Tsukuba, Japan). The two Chinese hamster ovary (CHO) cell lines, CHO-K1 (Tim-3 knocked out cells) and CHO-Tim-3 (Tim-3 overexpressed cells), were gifts from Dr Vijay Kuchroo (Harvard Medical School). HLE, CHO-K1 and CHO-Tim-3 cells were grown in DMEM, Li-7 in RPMI-1640, and Huh7 in MEM, supplemented with 10% fetal bovine serum (FBS) and 100 mg/l of penicillin-streptomycin in a humidified atmosphere with 5% of CO₂ at 37°C.

Cell proliferation assays were conducted in HLE, Li-7 and Huh7 cells with WST-8 (18). Each cell type (5×10³) was seeded into 96-well plates and cultured in 100 μl of culture medium for 24 h. Cells were treated with 0.01, 0.03, 0.1, 0.3 or 1 μM galectin-9. In addition, 30 mM of lactose was added to inhibit the binding of galectin-9, and sucrose was used as a control (19).

We conducted a flow cytometric analysis with Annexin V and PI to clarify whether galectin-9 has apoptotic and necrotic effects on HCC (18,20). Briefly, we treated HLE, Li-7 and Huh7 cells (1×10⁵ cells in 200 μl of culture medium) with 0.3 μM galectin-9 for 12 h at 37°C. The cells were analyzed with the Mofro Astrios flow cytometer (Beckman Coulter, Indianapolis, IN, USA). The results were analyzed by Kaluza software (Beckman Coulter).

The expression of Tim-3 on the cell surface was characterized by a single-color flow cytometric analysis (7). The CHO-K1 cell line was used as the negative control and the CHO-Tim-3 cells were used as the positive control. HLE, CHO-K1 and CHO-Tim-3 cells were dissociated from a monolayer culture with enzyme-free cell dissociation buffer. Li-7 and Huh7 cells were dissociated with 0.05% of trypsin and incubated for 12 h in a non-adhesive 96-well microplate. The flow cytometric analysis was conducted using the Cytomics FC 500 flow cytometer (Beckman Coulter), and the results were analyzed by Kaluza software.

We analyzed the effects of galectin-9 on the cell-cycle profile of HCC cells by using the Cell Cycle Phase Determination Kit (Cayman Chemical, Ann Arbor, MI, USA). The assays were conducted as described (21,22). The flow cytometric analysis was conducted using the Cytomics FC 500 flow cytometer, and the percentage of cells in different phases of the cell cycle was analyzed using the Kaluza software.

We evaluated the amounts of caspase-cleaved keratin18 (CCK18) by using the M30 Apoptosense ELISA kit (Peviva, Bromma, Sweden) (23). Each cell line (5×10³) was seeded into 96-well plates and cultured in 100 μl of culture medium for 24 h. The cells were then treated with 0.3 μM galectin-9. The rest of the assay procedures were performed according to the manufacturer’s instructions. The amounts of antigen in the controls and samples were calculated by interpolation into a standard curve.

Animal experiments were performed according to the guidelines of the Kagawa University Committee on Experimental Animals and the guidelines regarding the use of animals tissued by the UK National Cancer Research Institute (24).

We purchased female athymic mice (BALB/c-nu/nu; 8-week-old; 20–25 g) from Japan SLC (Shizuoka, Japan). The animals were maintained under specific pathogen-free conditions using a laminar air flow rack and had continuous free access to food sterilized by irradiation with γ-rays, which was purchased from CLEA Japan (Tokyo, Japan) and autoclaved water.

We used tissues instead of cell suspensions for the xenograft model of Li-7 cells because a 90-day period would be necessary for the tumors to grow to palpable size in mice injected with a cell suspension of Li-7 cells. Li-7 tissues were transplanted subcutaneously (s.c.) in the flank of nude mice for initial passage. The tumor tissues were harvested when the tumors grew to 15 mm in diameter, and the tumor tissue was cut into pieces 3 mm in diameter.

Mice were inoculated with a tumor fragment of Li-7 (25). Huh7 cells (2×10⁶ cells per animal) were also injected s.c. in nude mice. When the xenografts were palpable as a mass of >6 mm in diameter in all recipients, the mice were randomly assigned to the control group or a treated group. Treated mice (n=6) were injected s.c. with galectin-9 (90 μg) three times a week. Only phosphate-buffered saline (PBS) was administered to the control group (n=6). The tumor growth was monitored by the same investigators (K. Fujita and T. Masaki). The tumor volume was calculated as follows: tumor volume (mm³) = [tumor length (mm) × tumor width (mm²)]/2. The animals bearing an excessive tumor volume (>2,000 mm³) were euthanized for ethical reasons (24).

The tumor tissues were fixed in 4% buffered paraformaldehyde. We performed a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay to detect apoptotic cells using the TACS2 TdT-DAB In Situ Apoptosis Detection kit (Trevigen, Gaithersburg, MD, USA) according to the manufacturer’s instructions (18).

We determined the amount of caspase-4, -8 and -9 by using the caspase-4, FLICE/caspase-8 and caspase-9 colorimetric assay kits (BioVision, Milpitas, CA, USA), respectively (18). HLE and Li-7 cells were cultured with or without 0.3 μM galectin-9, and the subsequent procedures were performed according to the manufacturer’s instructions.

We extracted the total RNA of cultured cells or tumor tissues using the miRNeasy Mini kit (Qiagen, Venlo, The Netherlands) as described (21,22,26). Cells were cultured with or without galectin-9 for 24 h in pentaplicate. The samples were labeled using a miRCURY Hy3 Power Labeling kit (Exiqon, Vedbaek, Denmark) and hybridized on a human miRNA Oligo chip, version 14.0 (Toray, Tokyo, Japan). Scanning was conducted with the 3D-Gene Scanner 3000 (Toray). We used 3D-Gene extraction software (ver. 1.2, Toray) to read the raw intensity of the image. The raw data were analyzed with GeneSpringGX (ver. 10.0, Agilent Technologies, Santa Clara, CA, USA) and quantile normalized (27). We calculated the fold changes in miRNA expression level between the treated groups and control group. Hierarchical clustering was accomplished using the furthest neighbor method and Pearson’s product-moment correlation coefficient as a metric.

Li-7 cells were transfected with miR-1246 mimic (mirVana miRNA mimic, Ambion, Carlsbad, CA, USA), miR-6131 mimic, and Lipofectamine 2000 according to the manufacturer’s protocol. HCC cells were transfected with miR-1246 mimic, miR-6131 mimic, or negative control #1 (Ambion) under galectin-9. The WST-8 assay and ELISA of CCK18 were conducted at the indicated time-points.

The cell lysate was processed according to the methods described in our previous reports (21,22). All the steps were carried out at 4°C. Protein concentrations were measured with a spectrophotometer NanoDrop 2000 (Thermo Fisher Scientific, Wilmington, DE, USA). Samples were electrophoresed through 10% SDS-PAGE, and the proteins were transferred to nitrocellulose membranes. The membranes were incubated with primary antibodies after blocking and then incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies. Immunoreactive proteins were visualized with an enhanced chemiluminescence detection system (Perkin-Elmer, San Jose, CA, USA) on X-ray film.

Target sites for miR-1246 found in DYRK1A 3′UTR are aligned among 16 mammals by TargetScan.

Data were analyzed by paired t-test for tumor volumes in vivo, which were analyzed by two-way repeated-measures analysis of variance (ANOVA). P-values <0.05 were recognized as significant. All statistical analyses were performed with Prism 6 software (Graph Pad Software, La Jolla, CA, USA).

We screened the inhibitory effect of galectin-9 on HCC by performing a WST-8 assay, and we found that galectin-9 inhibited the cell proliferation of the HLE and Li-7 cell lines in a dose- and time-dependent manner. The inhibition of cell proliferation was more evident in the HLE cells than the Li-7 cells. However, no anti-proliferative effect of galectin-9 was detected in Huh7 cells (Fig. 1A and B). The effect of galectin-9 was antagonized by 30 mM lactose, suggesting that the β-galactoside binding nature of galectin-9 is essential for the activity (Fig. 1C).

We then performed flow cytometric analysis to assess the contribution of apoptosis and necrosis to the suppression of tumor growth. As shown in Fig. 2A and B, the antiproliferative effects of galectin-9 on the HLE and Li-7 cells was not due to necrosis, but rather to the induction of an early apoptotic change that was Annexin V-positive and PI-negative (18,20). The Huh7 cells did not exhibit early apoptotic or necrotic changes after treatment with galectin-9. These results resemble the data of WST-8 assay (Fig. 1A and B).

We determined the levels of CCK18 by ELISA to establish whether apoptosis is involved in galectin-9-induced cell death, and we found that galectin-9 increased the levels of CCK18 in HLE and Li-7 cells as illustrated in Fig. 2C, suggesting that the apoptotic process following phosphatidylserine exposure proceeds to cut intermediate filaments of cells (23). A galectin-9-induced increase in CCK18 was also inhibited by one of the antagonists of galectin-9, 30 mM lactose, compared to 30 mM sucrose (Fig. 2D).

We further performed a flow cytometric analysis of the cell cycle to evaluate the contribution of cell cycle arrest to the suppression of HCC cell lines by galectin-9. HLE, Li7, and Huh7 cells were treated with 0.3 μM galectin-9, and we observed that the proportion of cells in S phase was very slightly increased in the HLE and Li-7 cells with the treatment by galectin-9 (Fig. 3A and B). These results suggested that galectin-9 inhibits tumor growth predominently through cancer cell apoptosis.

The cell surface expression of Tim-3, a specific ligand of galectin-9, was assessed by flow cytometric analysis because we speculated that a Tim-3/galectin-9 pathway is also involved in apoptosis of HCC cells, similarly to T lymphocytes (7). Intriguingly, we were not able to detect evident Tim-3 on HLE or Li-7 cells, suggesting that galectin-9 induces apoptosis of HCC cells through binding with an unknown ligand other than Tim-3 (Fig. 3C).

To determine whether galectin-9 affects tumor growth in vivo, we performed experiments with a xenograft animal model. As shown in Fig. 4A, galectin-9 significantly inhibited the tumor growth of Li-7 cells on day 7. On the contrary, the tumor growth of Huh7 cells was not suppressed in vivo with the treatment of galectin-9 (Fig. 4A). The antitumor effect in vivo resembled that observed in our in vitro experiments. The groups of mice did not show any significant differences in body weight or water consumption.

Apoptotic cells were detected by TUNEL assay in the tissues of Li-7 treated with galectin-9 as presented in Fig. 4B-b. However, apoptotic cells were not detected in control tissues (Fig. 4B-a). The results of the xenograft model analysis matched those observed in vitro.

We assessed the levels of caspase-4, -8 and -9 to clarify whether endoplasmic reticulum stress (ER stress), death receptor or mitochondrial pathway contributes to apoptosis induced by galectin-9 (28). The results of our in vitro and in vivo experiments (Fig. 4C) showed that the levels of caspase-4 and -9 were upregulated with the galectin-9 treatment, whereas caspase-8 was not changed by the galectin-9 treatment, suggesting that ER stress and mitochondrial pathway were involved in galectin-9-induced apoptosis.

To investigate the intracellular pathway of apoptosis, we screened the expression levels of miRNAs in Li-7 cells and xenografts, and we compared the miRNA profiles obtained with or without galectin-9. We used cells treated with galectin-9 for 24 h as in vitro samples and xenografts as the in vivo samples. The unsupervised hierarchical clustering analysis showed that the treated group clustered together and separately from the controls both in vitro and in vivo (Fig. 5).

We identified 14 miRNAs that were differently expressed (12 upregulated and two downregulated miRNAs) in culture (Table I) and 30 miRNAs differently expressed (14 upregulated and 16 downregulated) in xenograft tumor tissues (Table II). We found that two miRNAs, miR-1246 and miR-6131, were upregulated in both cultured cells and tumor tissues treated with galectin-9.

The microarray data obtained in this study are registered at the NCBI Gene expression Omnibus (GEO). The accession numbers are GSE55665 for the in vitro data and GSE55666 for the in vivo data.

We performed a WST-8 assay to determine whether the miR-1246 upregulated with galectin-9 treatment contributes to galectin-9-induced apoptosis. As shown in Fig. 6A, the tumor growth was inhibited in the transfection of miR-1246 in Li-7 cells under 0.1 μM galectin-9 treatment, but the suppression of tumor growth was not detected in the transfection of miR-1246 in Li-7 cells without galectin-9 treatment. miR-1246 also induced apoptosis with galectin-9 treatment in Li-7 cells (Fig. 6B). These data suggest that miR-1246 acted as the inhibitor molecule of the cell proliferation through apoptosis in Li-7 cells treated with galectin-9. The miR-6131, showed no additional inhibitory effect on the cell growth of Li-7 (Fig. 6C).

We speculated that galectin-9 suppresses the expression level of oncogene DYRK1A through the upregulation of miR-1246 in Li-7 cells and tissues (29). We used cells treated with galectin-9 for 24 h for in vitro samples. The western blot analysis showed that DYRK1A was downregulated by galectin-9 both in vitro and in vivo (Fig. 6D), suggesting that the miR-1246-DYRK1A-caspase-9 axis was involved in the galectin-9-evoked apoptosis. Target sites for miR-1246 found in DYRK1A 3′UTR are conserved among 16 mammals (Fig. 7A).