Human trophoblast requires galectin-3 for cell migration and invasion

Invasive extravillous cytotrophoblast of the human placenta expresses galectins-1, -3, and -8 in vivo and in vitro. This study aimed to investigate the potential role of galectin-3 in cell migration and invasion, using recombinant human galectin-3 (rhgalectin-3), small molecule galectin inhibitor I47, and galectin-3 silencing. HTR-8/SVneo cell migration was stimulated by rhgalectin-3 and reduced by I47, which could be neutralised by rhgalectin-3. Inhibitor specificity and selectivity for the galectins expressed in extravillous trophoblast were validated in solid phase assays using recombinant galectin-1, -3, -8, confirming selectivity for galectin-3. HTR-8/SVneo cell migration and invasion, and invasion by isolated trophoblast cells in primary culture were significantly reduced in the presence of I47, which could be restored by rhgalectin-3. Upon HTR-8/SVneo cell treatment with galectin-3 siRNA both LGALS3 and galectin-3 protein were dramatically decreased. Silencing of galectin-3 induced significant reduction in cell migration and invasion, which was restored by rhgalectin-3. The influence on known mediators of cell invasion, MMP2 and -9, and integrins α1, α5, and β1 was followed in silenced cells, showing lower levels of MMPs and a large reduction in integrin subunit β1. These results show that galectin-3 acts as a pro-invasive autocrine/paracrine factor in trophoblast in vitro.


Galectin-3 detection, localisation and (sub)cellular distribution in HTR-8/SVneo cells.
Expression of galectin-3 has been previously documented for villous cytotrophoblast, cell columns, isolated cytotrophoblast and trophoblast derived cell lines 10,11,23 . Here, the expression pattern and subcellular distribution of galectin-3 in HTR-8/SVneo cells was further examined using polyclonal anti-galectin-3 antibodies. Galectin-3 was present at the plasma membrane and in cytoplasm, as evidenced by fluorescence cytochemistry in Fig. 1a. Flow cytometric analysis showed that ~9% of non-permeabilized (Fig. 1b) and ~97% of permeabilised ( Fig. 1c) HTR-8/SVneo cells were galectin-3 positive. Subcellular distribution of galectin-3 was investigated by immunoblot analysis of the fractions obtained (Fig. 1d). Galectin-3 appeared as a band of ~30 kDa in membrane, cytoplasmic, nuclear soluble and nuclear chromatin fractions (Fig. 1d), which is in line with the previously recorded presence of galectin-3 in the nucleus, cytoplasm and at the cell surface of other cell types 16 . Data from the Western blot (WB) regarding relative galectin-3 content showed that 64% of this lectin was found in the membrane fraction (comprised of solubilised plasma membrane and intracellular membranes), 19.5% in the cytoplasm, 12% in the nuclear soluble and 4.5% in the nuclear chromatin fraction. Purity of the subcellular fractions was demontrated using antibodies against marker proteins MEK1/2, α 5 integrin and POU5F1 (Fig. 1d).

Selective inhibition of galectin binding.
We investigated the possibility that galectin-3 participates in processes relevant for trophoblast function in vitro using two approaches: (1) by inhibition of galectin-3 lectin function with I 47 , a thiogalactoside inhibitor of galectin-3 carbohydrate binding site and (2) by transient galectin-3 knockdown using siRNA.
The selectivity of I 47 and its effect on HTR-8/SVneo cell viability were tested in preliminary experiments. At 1,000 ng/ml, I 47 (Fig. 2a) was found to significantly reduce binding of rhgalectin-3 to immobilised Matrigel glycoconjugates in solid phase assay (Fig. 2b) at the tested concentrations of rhgalectin-3 (100, 500, and 1,000 ng/ml). The I 47 , present in large excess and with high affinity for galectin-3, was able to prevent further binding of rhgalectin-3 at increasing concentrations to a complex mixture of ECM components contained in Matrigel coating. Little change from the baseline absorbance (A 450 0.2) with 0 ng/ml of rhgalectin-3 was detected with higher concentrations. Previously, some of the galectin-3 inhibitors were found to also bind one or more of the members of the galectin family, thus binding to other galectins expressed by the invasive trophoblast was tested here. To that end galectin-1, in form known as CS-galectin-1 mutant form, previously documented to maintain lectin acitivity, sugar binding specificity and affinity 26 , and rhgalectin-8 were tested for binding with or without the inhibitor I 47 . Binding to Matrigel glycoconjugates, incubated at the galectin concentrations of 100 and 1,000 ng/ml was not reduced in the presence of I 47 (1,000 ng/ml; Fig. 2c), and in case of galectin-8, a currently poorly understood increase in binding of galectin-8 at 1,000 ng/ml only was observed. This inhibitor had no effect on HTR-8/SVneo cell viability (Fig. 2d), when the MTT test was performed with I 47 concentrations of 10, 100 and 1,000 ng/ml. Taken together, these results demonstrate that I 47 is a selective galectin-3 inhibitor, with no effect on HTR-8/ SVneo cell viability, which makes it suitable at all studied concentrations for the functional tests shown below.

Trophoblast cell migration is influenced by availability of galectin-3. Several lines of evidence have
pointed to the relevance of galectin-3 in cell adhesion and motility 27,28 . This possibility was tested in HTR-8/SVneo cells using cell adhesion and migration assays (Fig. 3). The involvement of galectin-3 in trophoblast cell migration was evaluated by wound healing assay. When HTR-8/SVneo cell migration was determined in conditions of increased availability of rhgalectin-3 (10, 100, 200, 500 and 1,000 ng/ml), a bell-shaped curve was observed (Fig. 3b). Significant stimulatory action of galectin-3 on cell migration was obtained at 100 and 200 ng/ml, to 134% and 130% of control values respectively. To rule out a possibility that the observed effect resulted from increased cell number/viability after treatment with rhgalectin-3, a MTT assay was performed in parallel. Exogenously added rhgalectin-3 decreased viable cell number significantly at 500 ng/ml to 77% of the control value (Fig. 3a). Thus, the cell migration increase with galectin-3 was not a result of increased cell numbers or metabolic activity. In addition, the cell migration test used is based on measuring the distance crossed by viable cells during the studied interval, and the obtained result suggests that individual cells were stimulated by galectin-3 to migrate further than in control. In order to assess lectin-type interactions in HTR-8/SVneo cell migration, cells were incubated with I 47 alone, in combination with rhgalectin-3 (Fig. 3c), or upon galectin-3 silencing (Fig. 5a).  As opposed to the effect of rhgalectin-3, the presence of I 47 (1,000 ng/ml) significantly decreased HTR-8/SVneo cell migration to 80% of the control level. When rhgalectin-3 (200 ng/ml) was added together with I 47 , inhibition of cell migration was abolished (Fig. 3c). Taking into consideration the concentrations of both I 47 , present in large excess, and rhgalectin-3 added at nanomolar concentration, it can be proposed that added galectin-3 was able to restore migration by acting jointly with HTR-8/SVneo cell released galectin-3, preferentially binding cell membrane ligands that contribute to cell migration. The inhibitor, however, may have a more restricted access to cell membrane/ECM binding sites in other cell based methods. This finding also shows that galectin-3 contributes to cell migration of HTR-8/SVneo cells through lectin-type interactions, being inhibitable by the inhibitor of the carbohydrate recognition domain of galectin-3, I 47 . In galectin-3 silenced HTR-8/SVneo cells cell migration was also reduced to 74% (Fig. 5a), which was restored by adding rhgalectin-3 (200 ng/ml) to 91% of control. Adhesion of HTR-8/SVneo cells was studied using plastic or ECM coated surfaces (Fig. 3d). HTR-8/SVneo cells were pretreated and incubated for 1 h in the absence (control) or presence of rhgalectin-3 (200 ng/ml) and I 47 (1,000 ng/ml) separately or combined. Regardless of the treatment or surface no difference in cell adhesion was observed.
Endogenous and added recombinant galectin-3 affect trophoblast cell invasion. The relevance of galectin-3 for trophoblast cell invasion was studied using the Matrigel invasion assay by isolated cytotrophoblast, which most faithfully replicates extravillous trophoblast in vivo, and HTR-8/SVneo cells, either treated with I 47 or specific galectin-3 siRNA. After transfection with siRNA targeting galectin-3, silencing was confirmed by qPCR and WB (Fig. 4). The expression levels of LGAL3 and galectin-3 protein were significantly decreased to 6% and 7.6%, respectively, of control levels ( Fig. 4a,b), showing high efficiency of galectin-3 silencing. Expression levels of LGAL1 and LGAL8 were not affected by transfection targeting galectin-3 ( Fig. 4a), demonstrating the selectivity of silencing. In the Matrigel invasion assay, trophoblast HTR-8/SVneo cells with down-regulated galectin-3 expression exhibited significantly reduced invasive ability down to 33% of the control level (Fig. 5b). When galectin-3 (200 ng/ml) was added to the system, inhibition of invasion with silenced galectin-3 cells was reduced to 45% of the control. The presence of I 47 decreased invasion by HTR-8/SVneo cells to 60% and cytotrophoblast (CT) to 50% of the control value (Fig. 5c,d, respectively). The addition of rhgalectin-3 (200 ng/ml), in a model system where both cell released and ECM contained galectin-3 are present, resulted in reversion of the effect of I 47 , to 79% in HTR-8/SVneo (Fig. 5c) and 89% CT cells (Fig. 5d). Taken together, the data demonstrate involvement of galectin-3 in trophoblast cell invasion.
In all invasion assays performed Matrigel, the basement membrane product composed of laminin, fibronectin and proteoglycans, was used 29 . In contrast to previously published data 30 , dot blot and WB analysis detected galectin-3 and galectin-1 in all tested basement membrane preparations, as bands of ~30 kDa and 14.5 kDa, respectively (Fig. 5e,f ). In addition, the 30 kDa band consistent with galectin-1 dimer was also detected. Galectin-8 was only barely visible in one of the three tested Matrigel preparations. In wound healing experiments the distances crossed after 24 h period were measured using an electronic grid; data from three experiments in duplicate are presented as percentages of untreated controls, given as mean ± SEM. (c) HTR-8/SVneo cell migration in response to I 47 (1,000 ng/ml) and rhgalectin-3 (200 ng/ml) alone, or in combination (200 ng/ml rhgalectin-3 + 1,000 ng/ml I 47 ). Data from three experiments in triplicate are presented as percentages of untreated controls, given as mean ± SEM. (d) Adhesion was studied after 1 h using rhgalectin-3 (200 ng/ml), I 47 (1,000 ng/ml) alone, or in combination (200 ng/ml rhgalectin-3 + 1,000 ng/ ml I 47 ), using plastic, or collagen type I, and Matrigel coated surfaces, as indicated. Data from three experiments in triplicate are expressed as percentages of the untreated control. The data were analyzed using the oneway analysis of variance (ANOVA) with Tukey or Newman-Keuls post hoc-test (α = 0.05) after passing Kolmogorov-Smirnov normality test. Differences were significant at *p < 0.05, **p < 0.01, and at ***p < 0.001.

Discussion
The data presented here for the first time document the relevance of galectin-3 for invasive trophoblast cell function in vitro. Galectin-3 is present in human placenta, as well as in choriocarcinoma cell lines, isolated trophoblast and the trophoblast derived HTR-8/SVneo cell line 10,11,22 . Under the experimental conditions employed here the reported dominant intracellular localisation of galectin-3 was confirmed by immunocytochemistry and flow cytometry, together with membrane staining. This finding was extended further with respect to intracellular distribution of galectin-3 between different cell compartments, showing that galectin-3 is present in LGALS3 was significantly reduced in galectin-3 silenced cells, while it had no effect on LGALS1 (a middle) or LGALS8 expression (a right), as evidenced by qPCR. One-way analysis of variance (ANOVA) with Tukey post-hoc test (α = 0.05) was used when data passed Shapiro-Wilk normality test (LGALS1 and LGALS8). When data didn't pass normality test (LGALS3), non-parametric Kruskal-Wallis test was used followed by Dunn's Multiple Range test (α nuclear chromatin, nuclear soluble, cytoplasmic and cell membrane fractions. The membrane fraction associated galectin-3 was by far the most abundant, which seems to contradict the finding for intact unpermeabilised cells evidenced by flow cytometry. This however likely stems from the different membranes analysed by the two methods. Namely, the outer aspect only of the plasma membrane of non-permeabilised cells is accessible for antibody binding in flow cytometry, as opposed to the complex mixture of disrupted plasma, mitochondrial and ER/Golgi membranes after cell fractionation. Considering plasma membrane, the two methods used can detect galectin-3 at distinct locations, since galectin-3 engaged in lipid rafts, trapped in extracellular lattices, or within the cytoplasmic aspect of the plasma membrane documented for other cell types 31,32 could be inaccessible for antibody binding in flow cytometry of trophoblast shown here. Our finding of galectin-3 in nuclear compartments is consistent with published data and galectin-3 function in early assembly of the splicing machinery for mRNA-processing and nuclear export 33 . Based on their binding characteristics galectins are expected to influence multiple processes in cells exposed to or expressing them. Galectin-1 and galectin-3 are differentially expressed by endometrium during the menstrual cycle and early pregnancy suggestive of specific roles in embryo implantation 21 . Along with the finding that galectin-3 is absent from non-decidualised endometrium and is not present in the uterus of non-pregnant females 20 , it has been proposed that it has a pregnancy related function in the uterine environment. The interstitial trophoblast invasion characterized by interaction of dispersed EVT with decidual stromal cells and the matrix they produce is one of the main interests of our group. Decidual stromal cells of first trimester pregnancy are known for abundant expression of both galectin-1 and galectin-3 10,22 , so the invasive trophoblast in vivo is exposed to both intracellular and extracellular galectin-1 and -3. We have previously shown that trophoblast cell invasion in vitro is dependent on both endogenous and exogenous galectin-1 11 , while data from the literature support the relevance of galectin-1 for maintenance of pregnancy in vivo 34,35 . Based on galectin binding specificity and potentially shared ligands, it was of interest to determine whether galectin-3 participates in processes affecting trophoblast cell migration and cell invasion. Recently, we have shown that trophoblast HTR-8/SVneo cells release galectin-1, -3 and -8 into culture medium 36 . Thus, two approaches were taken, the first one with either increased, or blocked extracellular galectin-3, and the other based on reduction of intracellular galectin-3 by targeting its mRNA. Several previous studies employing mouse animal models used 33DFTG (formerly known as TD139), a synthetic inhibitor of lectin-type interactions of galectin-3 [37][38][39][40] . In the present study, thiogalactoside inhibitor, I 47 that mimics N-acetyllactosamine was used, based on its reported specificity and high affinity for galectin-3 41 . This was supported by our solid phase assay data with immobilised Matrigel glycoconjugates and rhgalectin-3, CS-galectin-1, and rhgalectin-8, showing that I 47 decreases the binding of rhgalectin-3 only. Therefore, it was concluded that I 47 met the initial criteria for use in cell-based assays with trophoblast cells expressing galectin-1, -3, -8. In many cell types both endogenous and exogenous galectin-3 have been implicated in cell viability and survival 42,43 inducing either positive or negative effects on cell proliferation and apoptosis. Hence, its influence on HTR-8/SVneo cells was assessed, showing a small, concentration dependent, reduction in cell numbers and/ or metabolic activity in rhgalectin-3 treated cells. This was not accompanied with change in apoptosis rate (not shown). Neither treatment with inhibitor I 47 , nor galectin-3 silencing induced any alteration in MTT (not shown). Therefore, the influence of galectin-3 on trophoblast cell viability, survival and apoptosis was not studied further.
Galectin-3 is also known to modulate cell-ECM interactions in cell adhesion, so plastic, collagen type I, and Matrigel coated surfaces were used in our model. Irrespective of the surface no change was observed in adhesion of cells treated with rhgalectin-3, I 47 , or their combination after 1 h. In other cell types galectin-3 has been shown to either promote or inhibit cell adhesion 27,44 , or produce no change, as found here. These conflicting findings indicate cell type specific effects of this lectin on cellular adhesion molecules. On the other hand, cell migration in the wound healing test was increased in the presence of recombinant rhgalectin-3 in a bell shaped manner (maximal at 100 and 200 ng/ml). The decrease when inhibitor I 47 was present could be reversed by rhgalectin-3. Similar concentration dependence and a maximally effective concentration were described for galectin-3 driven monocyte migration 28 . Invasive extravillous trophoblast in vivo would be exposed to extracellular galectin-3 from multiple sources, probably mostly released by decidual stromal cells or the trophoblast itself. Addressing the influence of extracellular galectin-3, our data showed that the small molecule inhibitor, I 47 , altered in vitro cell invasion. In the presence of inhibitor I 47 , both isolated cytotrophoblast and HTR-8/SVneo cell invasion was reduced, which seems to be reversible by the addition of rhgalectin-3. Inhibitory compounds that are based on disaccharides, which include I 47 , are considered relatively soluble, and not taken up well across membranes 17,45 (and personal communication with Hakon Leffler), thus it can be assumed that predominantly/exclusively extracellular galectin-3 was blocked. There are also several lines of evidence pointing to the relevance of galectin-3 for cancer cell migration and invasion 46 . Exogenously added galectin-3 contributed to the migration of many tumor cells mainly through interaction with ECM glycoprotein partners, such as fibronectin, laminin, collagen type IV 47 . Moreover, galectin-3 induced migration and invasion of mammary carcinoma cells involved integrin clustering and ligand induced integrin activation 48 . Although it is possible that the same can apply to normal human trophoblast migration and invasion and that our results indicate the relevance of lectin type interactions for this process, the molecules involved in I 47 inhibition remain to be elucidated. When galectin-3 was targeted
Scientific RepoRts | (2019) 9:2136 | https://doi.org/10.1038/s41598-018-38374-w by siRNA drastic reduction specific for this galectin was induced, resulting in decreased cell migration and invasion, which could be restored by addition of rhgalectin-3 (Fig. 5a,b). Analysis of galectin-3 in cell fractions after galectin-3 silencing revealed less galectin-3 in all compartments, leaving the possibility for reduced galectin-3 actions in these cells. In that context it is interesting to note that Matrigel, a commercial ECM widely used to study cell migration and invasion in a wide range of models, harbours certain galectins, and galectin-3 in particular (Fig. 5e,f), thus participating in base line cell invasion. The Matrigel galectin-3 contribution to trophoblast cell invasion shown here is however hard to fully appreciate, since HTR-8/SVneo cells also release galectin-3 into the medium 36 . On the other hand, the presence of galectins in Matrigel could be seen as a factor making this matrix a better model of decidual ECM. Nevertheless, the presence of galectin-3 specific inhibitor significantly reduced invasion of Matrigel by isolated cytotrophoblast and HTR-8/SVneo cells (Fig. 5c,d). The interaction of galectins with the ECM is however of great interest, since galectin-1 has been shown to influence T-cell survival differently depending on its presentation, whether soluble, or cell membrane and matrix bound 30 .
Based on the results of functional tests, it was further undertaken to determine whether any of the molecules known to participate in trophoblast cell migration/invasion, such as integrins and gelatinases (MMP2 and -9), were involved in reduction observed in galectin-3 silenced cells. Among the tested integrin subunits α 1 , α 5 , β 1 , forming receptors for fibronectin, collagen type IV and laminin, subunit β 1 was the only one that was significantly reduced at the RNA and protein levels with galectin-3 silencing. This finding is of particular interest since subunit β 1 is the binding partner of a large number of α subunits, thus forming ECM receptors that are expressed as trophoblast cells acquire physiological invasiveness in vivo. Less integrin subunit β 1 could thus lead to a decrease in cell migration/invasion. Our finding that silencing endogenous galectin-3 causes a decline in levels of integrin β 1 suggests considerable relevance of their interaction for the trophoblast. This concurs with the reports that galectin-3 promotes healing during re-epithelialization of corneal wounds and cell migration of a corneal HCLE cell line, by involving the β 1 integrin dimer α 3 β 1 as a binding partner 49,50 . Other factors modulating trophoblast invasion have also been associated with changes in β 1 integrins, such as triiodothyronine 51,52 . Previously, galectin-3 silencing was associated with another integrin subunit, β 3 , in an endometrial cell line RL95-2 53 . In cancer biology galectin-3 is overexpressed in migrating cells during wound healing, and during tumor invasion and metastasis 49,54 . Mechanisms of action by which galectin-3 can affect the intrinsic motility of cells include binding or upregulation of integrins, as a group of cell surface proteins that mediate cell adhesion to the extracellular matrix 50,55 .
Cellular invasion requires proteolytic degradation of extracellular matrix molecules. The MMP system is a well-documented effector mechanism important for trophoblast invasion 2,56 . Matrix metalloproteinases, MMP2 and MMP9, have been shown to mediate the influence of various cytokines and growth factors on trophoblast cell invasion 52,57,58 . MMP9 is able to degrade type IV collagen of the basement membrane, fibronectin, laminin, elastin, entactin, gelatin, and proteoglycans 59,60 . The regulation of MMPs is believed to occur at the transcriptional level, by activation of latent proenzymes, and by inhibition of proteolytic activity 58 . In this study, a decrease in MMP2 and MMP9 mRNA and reduced level/gelatinolytic activity of MMP2 and MMP9 in the conditioned media of galectin-3 silenced cells were observed, pointing to the significant contribution of post-transcriptional regulation of MMP expression. This could be due to direct interaction of these molecule families, since all MMPs have potential glycosylation sites. Across the MMP family, conserved glycosylation sites appear mainly as N-glycosylations in the active site, which is compatible with the potential ligand role for MMP. Thus, decrease in intracellular galectin-3 could affect potential MMP binding and export to the cell membrane 61 . In addition, both molecules that function upstream (regulators) and those that function downstream (e.g. substrates) of MMPs are subject to glycosylation 61 . Alterations in the glycosylation status of these molecules, as well as the presence of binding lectins, might have direct or indirect implications on MMP function.
Others have reported that altered MMP levels may involve changes in mRNA half-life 62 and decreases in MMP stability 63 . Destabilization of mRNA is a possible post-transcriptional regulatory mechanism by which MMP2 and MMP9 activity is reduced in galectin-3 silenced trophoblast derived cells. However, the exact mechanism(s) for post-transcriptional MMP2 and MMP9 regulation by galectin-3 still remains to be elucidated. It is of interest to note that in other cell types galectin-3 has been associated with MMPs through EMMPRIN (extracellular matrix metalloproteinase inducer), also known as CD147 and basigin, a member of the immunoglobulin family that is present on the surface of tumor cells, which stimulates nearby fibroblasts to synthesize MMPs 64 . Thus, concomitant stimulation and colocalisation of galectin-3 with CD147 are associated with increased gelatinolytic activity in the actively ulcerating human cornea 65,66 . The highly glycosylated species of CD147 are responsible for the induction of matrix metalloproteinases 67 with β1, 6-branched polylactosamine residues constituting major components in this molecule 68 . Since these glycan residues are unique in that they mediate cell motility via modulation of carbohydrate dependent interactions between galectin-3 and many cell surface and extracellular molecules, such as fibronectin, laminin, and integrins, the involvement of CD147 and other molecules in invasive trophoblast needs to be investigated further. So far, the highly glycosylated 65 kDa EMMPRIN has been reported to participate in activation of MMPs during labor at term 69 .
Data regarding the clinical relevance of galectin-3 for normal pregnancy have only started to accumulate. Thus, it was shown that galectin-3 was significantly down-regulated in the extravillous trophoblast of IUGR placentas 70 . Should galectin-3 involvement in pregnancy be further supported by clinical data in various conditions pertinent to pregnancy outcome, the possibility to neutralize it by small molecule inhibitors, as shown here, may provide a useful experimental tool. At this point, it can be concluded that galectin-3 is relevant for invasive trophoblast in vitro, as we previously showed for galectin-1. Given the huge potential of these galectins to interact with diverse molecules, much remains to be learned about the roles galectins may play in the placental bed.
All other reagents were of the best commercial grade available.  71 . Cytotrophoblast cells were isolated from the first trimester of pregnancy placentas from legal abortions (6-12 weeks) undertaken for non-medical reasons, at Obstetrics and Gynecology Clinic Narodni Front, Belgrade, Serbia, using sequential trypsin digestion, Percoll gradient centrifugation, and assessment, as previosuly described 11 . Briefly, after isolation, cells were identified as trophoblast by immunocytochemical staining for cytokeratin-7 (CK-7), and preparations with 95% cytokeratin positive cell were used. In addition, a small proportion of CD45-positive cells were removed using immunomagnetic beads prior to invasion assay. Cytotrophoblast was cultured in DMEM/F12 supplemented with 10% FCS and with antibiotic/antimycotic solution. This study was conducted according to local ethical standards (document GSP/05) and approved by Institutional Ethical Committee of the Institute for the Application of Nuclear Energy, INEP, University of Belgrade. (R&D) or goat anti-galectin-8 (R&D) were added and incubated for 2 h at room temperature (RT), with shaking. Wells were washed with PBS and incubated with biotinylated goat anti-rabbit IgG (for galectin-1) and horse anti-goat IgG (for galectin-3 and galectin-8). After 30 min, wells were washed, incubated with ABC for another 30 min and further with substrate and chromogen. The reaction was stopped with 0.2 M H 2 SO 4 . Non-specific binding was estimated by omitting the specific antibody. Subcellular fractionation. Subcellular fractionation of HTR-8/SVneo cells was carried out using the Subcellular Protein Fractionation Kit for Cultured Cells (ThermoScientific, Rockford, USA), according to the manufacturer's instructions. The procedure yields a cytosolic fraction (C), a membrane fraction (M), a nuclear soluble fraction (N) and a nuclear chromatin bound fraction (Nc). Protein concentration was then determined in each subcellular fraction using a BCA Protein Assay kit (ThermoScientific, Rockford, USA) and equal amounts of each fraction were loaded for SDS-PAGE electrophoresis. This analysis was performed on subcellular fractions obtained from cells cultured in complete RPMI 1640 medium and from cells after the transfection protocol. After transfer, membranes were incubated with goat anti-galectin-3. Relative galectin-3 content was calculated using densitometric evaluation of band intensities, since equal amounts of protein were used from each fraction. The obtained intensities were summed and that of each galectin-3 band was expressed as a percentage of total galectin-3. The following antibodies were used as controls for fraction purity: anti-MEK1, anti-integrin α5 and POU class 5 homeobox 1 (POU5F1, also known as OCT3/4) antibodies.  Cell viability assay (MTT). The viability of HTR-8/SVneo cells was assessed as described previously 73 .

Dot blot analysis.
HTR-8/SVneo cells were incubated for 24 h in treatment media containing different concentrations of rhgalectin-3 (1-1,000 ng/ml) or I 47 (10, 100 and 1,000 ng/ml) in 200 µl culture medium. In experiments with I 47 , DMSO supplemented medium was used as the control. The experiment was repeated three times with six replicates for each treatment group.
Cell adhesion. Adhesion assays were performed on plastic/uncoated wells or wells precoated with Collagen type I or Matrigel, as previously described 11 . HTR-8/SVneo cells were preincubated with rhgalectin-3 (200 ng/ml) or I 47 (1,000 ng/ml). Experiments were repeated three times in triplicate per treatment group.
Cell wounding and migration assay. The effect of various concentrations of rhgalectin-3 (10, 100, 200, 500 or 1,000 ng/ml), I 47 (1,000 ng/ml), rhgalectin-3 with I 47 and sigalectin-3 transfection on HTR-8/SVneo cell migration was investigated using the cell wounding assay as previously described 74 . HTR-8/SVneo cells were incubated until confluent, and scraped off using a sterile pipette tip. After rinsing, RPMI 1640 (containing 0.1% BSA) without or with rhgalectin-3 or I 47 was added. The pre-selected fields were photographed at zero point and after 24 h. The distances crossed after 24 h period were measured using an electronic grid and the distances Scientific RepoRts | (2019) 9:2136 | https://doi.org/10.1038/s41598-018-38374-w crossed by the cells were determined. The mean value for the controls was set to 100% and the data are expressed as percentages of the control value. The experiment was repeated three times in duplicate.
Cell invasion assay. Transwell Matrigel invasion assays were conducted as described previously 11 .
The invasion assay was run for 24 h using isolated CT and HTR-8/SVneo cells without or treated with I 47 , or rhgalectin-3 + I 47 . Invasion assay was also performed using HTR-8/SVneo cells transfected with sigalectin-3, as well as a negative control (S), and reagent-only (L). The treatments were prepared in DMEM/F12 for CT or RPMI 1640 for HTR-8/SVneo, both containing 0.1% BSA. Invaded CT cells were stained using anti-CK-7 antibody, and by Giemsa for HTR-8/SVneo cells. Cells on the underside of the filters and the occupied pores were counted in non-overlapping fields of the entire insert. Data presented are expressed as percentages of the control values. The experiment was repeated three times in duplicate.
Statistical analysis. The data were analysed using GraphPad Prism Demo Software (GraphPad Software, Inc., La Jolla California USA) and evaluation detailed in figure legends.

Data Availability
The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.