Nucleotide sugar transporter SLC35A2 is involved in promoting hepatocellular carcinoma metastasis by regulating cellular glycosylation

Abstract

Purpose

Recently, aberrant glycosylation has been recognized to be relate to malignant behaviors of cancer and outcomes of patients with various cancers. SLC35A2 plays an indispensable role on glycosylation as a nucleotide sugar transporter. However, effects of SLC35A2 on malignant behaviors of cancer cells and alteration of cancer cells surface glycosylation profiles are still not fully understood, particularly in hepatocellular carcinoma (HCC). Hence, from a glycosylation perspective, we investigated the effects of SLC35A2 on metastatic behaviors of HCC cells.

Methods

SLC35A2 expression in clinical samples and HCC cells was examined by immunohistochemical staining or Western blot/quantitative PCR and was regulated by RNA interference or vectors-mediated transfection. Effects of SLC35A2 expression alteration on metastatic behaviors and membrane glycan profile of HCC cells were observed by using respectively invasion, migration, cell adhesion assay, in vivo lung metastatic nude mouse model and lectins microarray. Co-location among proteins in HCC cells was observed by fluorescence microscope and detected by an in vitro co-immunoprecipitation assay.

Results

SLC35A2 was upregulated in HCC tissues, and is associated with poor prognosis of HCC patients. SLC35A2 expression alteration significantly affected the invasion, adhesion, metastasis and membrane glycan profile and led to the dysregulated expressions or glycosylation of cell adhesion-related molecules in HCC cells. Mechanistically, the maintenance of SLC35A2 activity is critical for the recruitment of the key galactosyltransferase B4GalT1, which is responsible for complex glycoconjugate and lactose biosynthesis, to Golgi apparatus in HCC cells.

Conclusion

SLC35A2 plays important roles in promoting HCC metastasis by regulating cellular glycosylation modification and inducing the cell adhesive ability of HCC cells.

Supplementary information

The online version contains supplementary material available at 10.1007/s13402-022-00749-7.

Introduction

As an integral feature of almost all biomolecules including nearly all cell surfaces and over 70% of secretary proteins, glycans contribute the most abundant and diverse post-translation modification [1]. At present, glycosylation has gained recognition for its pivotal role in virtually all aspects of the body, from embryogenesis to pathogenesis [2]. Thus, it is not surprising that glycosylation changes are also a universal feature of most major diseases, which are directly/indirectly associated with the alteration of at least one central structure of glycan-binding proteins[3, 4].

Until now, a clear correlation between aberrant glycosylation status of cancers and outcomes of patients has been demonstrated in numerous cancers [5, 6]. However, whether aberrant glycosylation as a result or cause for carcinogenesis is still a long-standing debate. In these years, many studies suggest that glycosylation of proteins with important functions, such as E-cadherin and immunoglobulin family receptors (e.g., CD44), altered the strength of cell-cell interaction in cancer, which led to cancer cells invasion and metastasis [7, 8]; moreover, aberrant glycosylation of proteins can also be defined as an essential indictor in defining stage and fate of cancer progression [9, 10]. However, the structural and functional complexity of protein glycosylation make it more difficult to define cancer cell phenotypes, compared to certain oncoproteins/oncogenes [6, 11, 12].

As we know, the glycosylation of glycoconjugates and the biosynthesis of polysaccharides require activated nucleotide-sugars, which serve as the substrates for glycosyltransferases [13]. In these biochemical processes, it is nucleotide sugar transporters (NSTs) that mediate nucleotide sugars transportation from the cytosol into the ER/Golgi lumen [14]. NSTs, classified as the drug metabolite transporter family (solute carrier family 35: SLC35), have been recognized as antiporters of intracellular nucleotide-sugar/nucleotide-monophosphate and nucleotide-sugar/nucleotide-sugar [15–17]. Furthermore, the activity of endogenous NSTs has been implicated in organogenesis, development, mammalian cellular immunity and pathogenicity of human diseases [15, 18]. For instance, a defect in a human NST have been described to be directly associated with congenital disorders of glycosylation (CDG) [19].

However, far fewer attentions are paid to NSTs, as one of the most important functional proteins in glycol chains synthesis, than to various glycosyltransferases in current disease-associated glycobiological studies, mainly because most researchers assume that expression and functions of glycosyltransferases are the major contributor for glycosylation process; in contrast, NSTs are only considered as the subordinated factor, for they merely provide substrates for glycosyltransferases [14]. Nevertheless, NSTs can modulate intracellular concentration of specific nucleotide-sugars through changing their activity under certain conditions. As an important member of NSTs family, SLC35A2 has been found to be closely associated with a variety of cancers [20, 21]. However, effects of SLC35A2 on cancer cells malignant behaviors of cancer cells and the alteration of cancer cells-surface glycosylation profiles are still not well-known, particularly in HCC.

In the present study, we examined SLC35A2 expression in clinical HCC samples and in various HCC cell lines with different metastatic potentials and investigated the effects of SLC35A2 on the metastatic behaviors of HCC cells and on the glycosylation pattern on HCC cell surface. Furthermore, we explored the role of SLC35A2 in recruiting the key galactosyltransferase B4GalT1 to Golgi apparatus in HCC cells. Altogether, our results suggest that SLC35A2 plays significant roles in promoting HCC metastasis, which could rely on altering the glycosylation modification on HCC cell surface and inducing cell adhesive ability.

Materials and methods

Clinical samples

Clinical samples were obtained based on an established protocol approved by the Ethics Committee of Zhongshan Hospital after provision of informed consent from patients with HCC. HCC samples were randomly collected from the patients undergoing curative resection at the First Affiliated Hospital of Guangxi Medical University by using a random number table. These samples were collected immediately after resection and immediately transported in liquid nitrogen and stored at -80℃ until use. The study (No.2022-E381-01) was conducted in accordance with the Declaration of Helsinki and the study protocol was approved by the ethics committee of the First Affiliated Hospital of Guangxi Medical University. Written informed consent was obtained from each participating patient.

Immunohistochemistry (IHC) analysis

The IHC assay was carried out according to our previous protocols [22]. Primary polyclonal antibody rabbit anti-SLC35A2 (cat no. 13657-1-AP) was purchased from Proteintech Group Inc. (IL, USA). The appropriate concentration of 1:100 was determined through titrating the antibody against normal controls before it was used on the clinical samples tissue microarray.

Cell culture

Five human HCC cell lines including Hep3B, HepG2, MHCC97L, MHCC97H and HCCLM6 with different metastasis potentials were obtained from the Liver Cancer Institute at Zhongshan Hospital, Fudan University and were routinely cultured in high-glucose DMEM medium (Gibco BRL, USA), supplemented with 10% fetal bovine serum (Hyclone, USA) at 37℃ in 5% CO₂. When the cells were grown to the indicated confluency, transfection was carried out and cells were harvested at the indicated time by treating with 0.25% trypsin and 0.02% EDTA. After being rinsed three times with PBS, cells were centrifuged for further experiments including RNA isolation and protein extraction.

Lentivirus-mediated SLC35A2 deficiency (Lv-shSLC35A2) and SLC35A2 overexpression

Three different sequences targeted to 3 different sites of SLC35A2 mRNA (NM_0005660) were designed without off-target effects and cloned into expression vector pGCsi-SLC35A2shRNA. After validation of inhibited efficiency in HCC cells, a most efficient target sequence (5’-CAGUAUGUUGCCAUCUCUA-3’) was identified, with its structure subsequently modified their structure to form a short hairpin RNA (shRNA-2): 5’-CCGGCAGTATGTTGCCA TCTCTATTCAAGAGATAGAGATGGCAACATACTGTTTTTG-3’. Viral vector generation was obtained by co-transfection of 293T cells by the calcium phosphate precipitation (CPP) method on 10-cm plates with 20 µg of PGC-LV, 15 µg of pHelper1.0EGFP, and 10 µg of pHelper2.0. SLC35A2 overexpression was constructed by GeneCopoeia Biotechnology Co., Ltd. (Guangzhou, China) through using OmicsLink™ expression system (pReceiver-Lv170). Infectious lentiviruses were harvested, centrifuged to eliminate cell debris, and then filtered through 0.22-µm cellulose acetate filters. Infectious titer was determined by fluorescence-activated cell sorting analysis of EGFP positive in 293T cells. A multiplicity of infection (MOI) of 20 for HCC cells in serum-free growth medium was used. Furthermore, transfected cells were selected by 1 µg/mL puromycin and used in subsequent assays. Meanwhile, a rescue assay was performed in SLC35A2-knockdown HCC cells. In detail, synonymous mutations were introduced into the target sequence of shSLC35A2 (5’- CAGUAAGUUGCCACCUAUA − 3’, mutations underlined) in SLC35A2 ORF (Open Read Frame) to generate a shSLC35A2-resistant overexpressed SLC35A2 vector (Re-SLC35A2).

Construction of pcDNA3.1-based wild-type and mutant vectors

The coding sequences of SLC35A2 and B4GalT1 were amplified by RT-PCR and cloned respectively into pcDNA3.1(+)-flag and pcDNA3.1(+)-HA eukaryotic expression vectors. Moreover, the de novo mutant in SLC35A2 gene (c.797G > T, p.G266V) was constructed by using site-directed mutagenesis system. After which, these cloned sequences were confirmed by carrying out a sequencing analysis. According to the manufacturer’s protocol, the transfection of these vectors into HCC cells was lipofectamine 3000-mediated, according to the manufacturer’s protocol. Co-location of SLC35A2 with B4GalT1 in HCC cells through co-transfecting pcDNA3.1-HA-SLC35A2 and pcDNA3.1-Flag-B4GalT1 was observed by fluorescence microscope (Leica, DM500).

Western blot

Equivalent protein amounts were separated on a denaturing SDS polyacrylamide gel and transferred onto polyvinylidene difluoride (PVDF) membrane. After being blocked with 5% nonfat dry milk in PBS containing 0.05% Tween-20, membranes were incubated with primary antibodies (Supplementary Table 1). For visualization, horseradish peroxidase linked anti-rabbit secondary antibodies (Cell Signaling) and ECL-Plus blotting substrate detection kit (Pierce) were used. Quantification of Western blots was analyzed densitometrically by using Quantity One software (Bio-Rad Laboratories).

Polymerase chain response

RT-PCR assay was carried out according to manufacturer’s instruction. In brief, 1–2µg RNA was incubated at 70℃ for 5 min and then placed on ice. After addition of oligo-(dT)₁₈ primer, RT reaction was performed at 42℃ for 60 min followed by 97℃ for 2 min. For PCR amplification, 1µL cDNA from RT was used in a final volume of 25µL, PCR program for SLC35A2 (F: 3’- GAATGCCTCCCTCATCCTCAG-5’; R: 3’- CCTTTGAGCACTTCCGCCAT-5’) was seen in Supplementary Table 1) or B4GalT1 (F:3’-TCTTTATTTGGCATTGGATATCC-5’; R: 3’- GTGTGTGCATGGAAGATGCC-5’) was performed at 95℃ for 5 min, followed by 25 cycles of 95℃ for 30s, 60℃ for 50s and 72℃ for 1 min and a final extension at 72℃ for 10 min. For quantitative RT-PCR (qRT-PCR), Platinum® SYBR® Green qPCR Super Mix kit (Invitrogen, Carlsbad, CA, USA) was used. Reactions were carried out using three independent technical replicates for each sample which was quantified in IQ5 real-time PCR system (Bio-rad, USA). GAPDH was selected as the internal reference.

Invasion and migration assay

In vitro invasion assay was performed by utilizing 24-well Transwell cell culture plates with polycarbonate membrane (8-µm pore size; Costar, USA). At first, 1 × 10⁵ cells were seeded onto upper chambers of the membrane coated by Matrigel (0.8 µg/µL); afterwards, the lower chambers were filled with 500 µL complete medium as a resource of chemo-attractants. After 24 h, HCC cells were fixed with 0.5% glutaraldehyde and stained by crystal violet. Subsequently, HCC cells of the upper membrane were wiped by cotton swabs, leaving the cells on the underside. For the migration assay, experimental procedures are the same as the in vitro invasion assay described above except that the filter was not coated with Matrigel. Finally, HCC cells on the lower membrane were observed and counted by microscope.

Cell adhesion assay

Cell adhesion assay was carried out by utilizing CytoSelectTM cell adhesion assay (Cell Biolabs, Inc.). In brief, the Fibronectin, Collagen IV and Laminin Adhesion Plate or E-selectin-coated flat-bottom culture plates were allowed to warm up at room temperature for 10 min under sterile conditions. A cell suspension containing 0.5 × 10⁶ cells/ml in serum free media was prepared. Each well was filled with 150 µL of the cell suspension (BSA-covered wells are supplied as a control). After 60 min of incubation, each well of the media was extracted and washed separately and 4–5 times with 250 µL PBS. PBS from each well was removed and 200 µL of cell stain solution was added and incubated for 10 min at room temperature. Afterwards, the cell stain solution was discarded from these wells and then these wells were washed 4–5 times with 500 µL deionized water. After the wells were dried, each well was added with 200 µL of extraction solution and then incubated 10 min on an orbital shaker. 150 µL solution from each extracted sample was transferred to a 96-well microtiter plate and the OD 560 nm was measured in a plate reader. For cell-cell adhesion assay, human endothelial cells (HEC) were seeded in 96-well plates and cultured for 48–72 h until 100% confluent. Then pretreated or untreated HCC cells with EGFP protein were added to each well as 2.0 × 10⁴/well and incubated at 37℃ for 60 min. After being washed three times, plates were subjected to fluorescence microscopy and remaining HCC cells were counted.

In vivo metastasis assay

An orthotopic liver xenograft model was established for tumor growth and metastasis analysis. In brief, 1 × 10⁷ MHCC97H cells with/without SLC35A2 knockdown or overexpression were resuspended in PBS buffer and injected subcutaneously into the capsule of the left hepatic lobe of male BALB/c nude mice. Tumor weight (g) and volume ((mm³) = [width² (mm²) × length(mm)]/2) were monitored. After 7 weeks, mice were sacrificed and the lungs were removed and fixed in formalin, then embedded in paraffin. Lung colonization was meticulously quantified by histological examination under microscope after consecutive tissue Sect. (5 μm/each). The Animal Experimentation Ethics Committees of our institute (Zhongshan hospital, Fudan university) approved the animal study.

Immunoprecipitation of E-cadherin

Total protein (750 µg) of MHCC97H cells was precleared with protein G-sepharose beads (50 ul) for 1.2 h. Following centrifugation, the supernatant was incubated overnight with E-cadherin mAb (5 µg). The cells were then incubated with protein G-sepharose for 2 h. Next, the beads were washed three times with an immunoprecipitation buffer. The immune complexes were released by boiling 5 min at 95℃ in sampling buffer and the immunoprecipitants were subjected to 7.5% SDS-PAGE and the separated proteins were transferred to a nitrocellulose membrane. The mobility shift was evaluated.

Cell surface glycosylation pattern analysis

Thirteen kinds of 1 mg/mL tumor-associated lectins (Vector laboratory, USA) were dissolved with chip-spotting buffer (CapitalBio, Beijing, China) and spotted on gel-substrate chip using a microarray printing robot Smart Arrayer-48 (CapitalBio, Beijing, China). Each lectin point has 6 repeats. Then the chip was incubated in a vacuum chamber with humidity greater than 80% at 25℃ overnight to immobilize the lectins. Cells were cultured with at least 80% confluence and then were harvested by cell-scraping followed by centrifugation at 1200 rpm for 5 min. After being washed 4–5 times with 1× PBS (PH 7.4), the cell pellet was resuspended in 0.5-1 mL cy3 in NaHCO3-NaH2CO3 (PH 9.3) at a concentration of 1 × 10⁶ cells/mL. After incubation for 30–60 min at room temperature, stained cells were washed three times in 1× PBS and then resuspended in 500 mL PBS; then cells were added into sub-array and incubated in the dark for 15–30 min in humidified incubator at 37℃. The binding fluorescence signals were obtained with fluorescence scanner LuxScan 3.0 (CapitalBio, Beijing, China). After the background value was subtracted, the net intensity value for each spot was calculated. The dye-bias-corrected ratios were calculated using median rectification. T-statistical analysis with Significance Analysis of Microarrays (SAM, version 2.1, Stanford University, CA, USA) software was used to determine q value using median adjusted data.

In vitro co-immunoprecipitation assay

The cell culture dishes were placed on ice and the cells were washed with ice cold PBS. After the PBS was drained, then the cells were treated with buffer (170 mM NaCl, 50 mM Tris-HCl, pH 8.0, 50 mM NaF, 0.5% NP-40) containing protease inhibitors. After clearing with protein-G-sepharose, supernatants were incubated with protein-G-sepharose conjugated with anti-GP73, anti-Flag or anti-HA monoclonal antibody at 4 °C overnight. Subsequently, protein-beads complexes were washed with PBS for three times and then re-suspended in 20 µL loading buffer. Lastly, the proteins were detected by western blotting.

Bioinformatics analysis

The expression data of patients with live hepatocellular carcinoma (LIHC, n = 371) and matched normal tissues (n = 50) was downloaded from TCGA database (https://portal.gdc.cancer.gov) by using the data transfer tool. The differential expression analysis of SLC35A2 among a variety of groups in LIHC was performed through UALCAN web server (http://ualcan.path.uab.edu). Overall survival probability and progress-free survival probability were plotted by using the Kaplan-Meier analysis, and their comparison was performed at the 0.05 level of significance. The gene list that positively correlated with SLC35A2 in human hepatocellular carcinoma was downloaded from the public cancer UALCAN databases (http://ualcan.path.uab.edu). Then Metascape database (http://www.metascape.org, v3.5.20220101) was used for pathway enrichment analysis, in which P value less than 0.05 was considered as significant.

Statistical analysis

All the statistical and online analyses were implemented via utilizing R programming language (Version 4.1.3) and visualized by using GraphPad Prism 9.0 software. 95% confidence intervals (CIs), Hazard ratios (HRs) and median overall survival time were calculated in our Cox proportional-hazards model. Data were expressed as mean ± standard error (SE) and analyzed using analysis of variance (ANOVA). Student’s t-test was used for two-group comparisons. Pearson correlation was conducted to determine the association between the various factors. All statistical tests were conducted bilaterally. P < 0.05 was considered to be statistically significant.

Results

Association of SLC35A2 expression with HCC metastatic potential

To investigate the association of SLC35A2 expression with HCC metastatic potential, we first performed a bioinformatics analysis on SLC35A2 expression feature in HCC by exploring the public cancer databases including GEPIA, UALCAN and Oncomine databases. It was found that SLC35A2 expression was significantly increased in HCC, compared to normal tissues (Fig. 1A); the increased expression of SLC35A2 was also associated with tumor grades in HCC (Fig. 1B). Of note, it was further observed that SLC35A2 expression was significantly higher in HCC tissues with lymph node invasion or metastasis than those without it (Fig. 1C-D). Moreover, overall survival of HCC patients with SLC35A2 high-expression was significantly decreased, compared to those with SLC35A2 low-expression (45.6 vs. 70.1 months, P < 0.01) (Fig. 1E); meanwhile, there was also a significant difference of progress-free survival between patients with SLC35A2 high-expression and with SLC35A2 low-expression (13.1 vs. 29.7 months, P < 0.01) (Fig. 1F). Furthermore, by using constructed tissue chip including HCC tissues with and without metastasis lesions (n = 48, respectively), followed by immunohistochemical staining combined with IPP software analysis, we detected SLC35A2 expression and found that the IOD/ µm2 value of SLC35A2 staining in HCC with metastasis lesions was 0.74 ± 0.07, which was significantly higher than that of those without metastasis lesions (0.42 ± 0.04; P = 0.0026) (Fig. 1G-H). Moreover, mRNA and protein levels of SLC35A2 were also detected in 7 HCC thrombi and matched primary lesions. Notably, both mRNA and protein levels of SLC35A2 were dramatically higher in cancer thrombi than in primary lesions (Fig. 1I-K). In addition, SLC35A2 expression was detected in a variety of HCC cells (HepG2, Hep3B, MHCC97L, MHCC97H and HCCLM6) with a variety of different metastatic potentials. It was also shown that SLC35A2 expression in HCC cells (MHCC97L, MHCC97H and HCCLM6) with high metastatic potential was significantly higher than in HCC cells without nearly metastatic potentials (HepG2 and Hep3B) (Fig. 1L-N). Altogether, the results indicated that, SLC35A2 expression has a positive association with the metastatic potential of HCC, suggesting a necessity to explore the roles of SLC35A2 in promoting HCC metastasis.

Fig. 1.

SLC35A2 expression and prognostic value in HCC. Based on the liver hepatocellular carcinoma (LIHC) data from public cancer databases (TCGA) not including normal GTEx data, the difference of SLC35A2 expression between in normal tissues and in primary liver cancer (A), the association between SLC35A2 expression and tumor grade (B), lymph node invasion status (C) or metastatic status (D, no significant) were analyzed by independent t test through UALCAN web server. Data are presented as Median (upper quartile-lower quartile). Moreover, the overall survival probability (E) and progress-free survival probability (F) was plotted by using the Kaplan-Meier analysis. The comparison was performed at the 0.05 level of significance. Furthermore, a constructed tissue chip including HCC tissues with and without metastasis lesions (n = 48, respectively) was detected by SLC35A2 IHC staining combined with IPP software analysis. Representative images of IHC staining of SLC35A2 in HCCs with metastatic lesions (Case 38) and those without metastatic lesions (Case 22) was presented. Scale bars = 25 μm (G-H). The mRNA and protein levels of SLC35A2 were detected in 7 HCC thrombi and matched primary lesions by western blotting (I-J) and real-time PCR assays (K). In addition, the levels of SLC36A2 were detected by real-time PCR (L) and western blotting assays (M-N) in five HCC cell lines with different metastatic potentials. Experiments were carried out three times. Data are presented as Mean ± SE (independent t test). **P < 0.01;*P < 0.05

Impacts of SLC35A2 on HCC invasion, adhesion and metastasis

To investigate the impacts of SLC35A2 on the metastatic behaviors of HCC, we carried out a lentivirus-mediated expression alteration of SLC35A2 in HCC cells. The results indicated that, after transfection with Lv-shSLC35A2, the protein level of SLC35A2 was significantly decreased in HepG2, Hep3B and MHCC97H, respectively; when SLC35A2 rescued re-overexpression, the protein level of SLC35A2 was obviously increased, compared to SLC35A2 knockdown (Supplementary Fig. 1). Of note, stable integration and expression of lentivirus-mediated shSLC35A2 in HCC cells were monitored for more than 10 passages and did not alter their morphology, which remained epithelial in appearance. In recent years, carbohydrate determinants are known to be involved in mediating cancer cells metastasis and adhesion [22]; herein, impacts of SLC35A2 expression on HCC cells adhesion, invasion and metastasis were further investigated in this study. Before these investigations, effects of SLC35A2 expression on the cell viability of HCC cells were assessed. It was found that the cell viability of HCC cells was not significantly altered after SLC35A2 knockdown in the cells, nor was it significantly altered during the rescued re-overexpression of SLC35A2 (Supplementary Fig. 2), suggesting that SLC35A2 expression alteration had no significant effects on the cell viability of HCC cells. Subsequently, effects of SLC35A2 expression on the abilities of HCC cells to perform Matrigel invasion and migration were examined. As shown in Fig. 2A-C, cells across the membranes were significantly decreased after SLC35A2 knockdown in MHCC97H cells, compared to control cells; moreover, SLC35A2 re-expression clearly rescued the invasion and migration of MHCC97H cells. In contrast, these significant alterations were not observed in Hep3B and HepG2 without metastatic potentials. Furthermore, adhesive ability of HCC cells with SLC35A2 expression alteration to different matrix was tested. It was shown that, after SLC35A2 knockdown, the number of MHCC97H cells that adhered to Collagen IV, Fibronectin (FN), E-selectin and Laminin was significantly less than that of the control cells; while, the extent of decrease of cell adhesion to E-selectin was largest, followed by Laminin, then FN and Collagen IV. Moreover, the rescued expression of SLC35A2 enhanced the adhesion ability of MHCC97H cells; however, Hep3B or HepG2 with SLC35A2 expression alteration did not display significant alteration of cell adhesion to these matrixes except adhesion to laminin in Hep3B (Fig. 2D-F). Based on the finding above, adhesion of MHCC97H cells, but not Hep3B and HepG2 cells, with Lv-shSLC35A2 transfection followed by rescued expression to vascular endothelial cells was further observed. As expected, MHCC97H cells adhered to vascular endothelial cells were significantly decreased after SLC35A2 knockdown and rescue increased with SLC35A2 re-expression (Fig. 2G). It is well-known that the enhanced migration, invasion and adhesion of cancer cells are closely related to the increase of cancer metastasis [23][24]. Hence, we investigated the effect of SLC35A2 on tumor metastasis in vivo by utilizing the metastatic HCC cell line-MHCC97H cells. In the study, seven weeks after orthotopic liver xenograft of HCC cells with SLC35A2 knockdown (shSLC35A2) or SLC35A2 re-expression (Re-SLC35A2), tumor weight and volume were calculated and no difference was found between the control group, shSLC35A2 group and Re-SLC35A2 group (p > 0.05) (Fig. 2H-I); while the number of metastatic foci in lungs from the shSLC35A2 group was significantly less than those from the control group (p < 0.5), as well as significantly increased in lungs from the Re-SLC35A2group, compared to the shSLC35A2 group (Fig. 2J-K).

Fig. 2.

Effects of SLC35A2 on metastatic behaviors of HCC cells. The migration and invasion ability of HCC cells with SLC35A2 expression alteration were evaluated by a Transwell assay using a chamber. *P < 0.05 vs. MHCC97H; ^#P < 0.05 vs. shSLC35A2 (A-C). The adhesion of control, shSLC35A2 and Re-SLC35A2 transfected HCC cells (Hep3B, HepG2, MHCC97H) to one of a variety of extracellular matrix proteins including Fibronectin, Collagen IV, Laminin and E-selectin was quantified by using colorimetric detection. *P < 0.05 vs. MHCC97H or Hep3B; ^#P < 0.05 vs. shSLC35A2 (D-F). Moreover, the adhesion of metastatic MHCC97H cells with shSLC35A2 transfection followed by rescued expression to vascular endothelial cells was measured by a cell adhesion assay. **P < 0.01 vs. MHCC97H or Hep3B; ^#P < 0.05 vs. shSLC35A2(G). Data above are the mean ± SE of three independent experiments. Experiments in vitro were independently carried out three times. In addition, in vivo experiment, 7 weeks after orthotopic liver xenograft of MHCC97H cells with SLC35A2 expression alteration (n = 5), tumor weight (H) and volume (I) were displayed with no difference (P > 0.05); the number of metastatic foci in lung was meticulously quantified by histological examination under microscope after consecutive tissue sections. Scale bars = 25 μm; *P < 0.05 vs. MHCC97H or Hep3B; ^#P < 0.05 vs. shSLC35A2 (J-K)

Effect of SLC35A2 on membrane glycan profile and carbohydrate determinants of HCC cells

To elucidate the membrane glycan profile alteration of HCC cells after SLC35A2 expression alteration, lectins array experiments were performed through interaction of labeled cells with immobilized lectins on array. Compared to control, HCC cells (including Hep3B, HepG2 and MHCC97H) with SLC35A2 expression alteration present significantly different binding abilities to six classes of lectins, mainly including AAL, DSA, LCA, MAL-II and WGA, SNA (MHCC97H cells in Fig. 3A-C; Hep3B and HepG2 cells in Supplementary Fig. 3). Moreover, to investigate further whether a change exists in the levels of these specific carbohydrate determinants on the cell surface, such as Thomsen-Friedenreich (TF) antigen, sialyl Lewis A (SLe^a) and sialyl Lewis X (SLe^x), after SLC35A2 expression alteration, while taking the lower baseline levels of the three carbohydrate determinants in HCC cells into consideration, HCC cells with SLC35A2 overexpression were used for cytometry analysis. The result showed that SLC35A2 up-regulated expression induced TF antigen, SLe^a and SLe^x expression in HCC cells, few increases in HepG2, a moderate increase in Hep3B, and a distinct and significant increase in MHCC97H cells (Fig. 3D). Furthermore, some important cell adhesion-related molecules and glycosyltransferases were further investigated after SLC35A2 expression alteration in HCC cells. It was found that the expressions of cell adhesion-related molecules such as ICAM-1, integrin β1, integrinα2 and E-cadherin were significantly down-regulated, while the expressions of glycosyltransferases including Gnt3, Fut8 and B4GalT1 were not significantly altered in metastatic MHCC97H cells with SLC35A2 knockdown, and that the expressions of ICAM-1, integrin β1, integrinα2 and E-cadherin were partly rescued with SLC35A2 re-expression (Fig. 4A-B); in contrast, the expressions of these molecules and glycosyltransferases were not significantly changed in Hep3B and HepG2 without metastatic potentials (Fig. 4C-F). In addition, to further clarify whether the glycosylation of cell-adhesion molecules can be affected by SLC35A2 in metastatic MHCC97H cells, glycosylation of E-cadherin was also investigated after SLC35A2 knockdown in MHCC97H cells. The result from the western blotting assay showed that E-cadherin protein from HCC cells with SLC35A2 knockdown exhibited the mobility shifts at some extent (Fig. 4G).

Fig. 3.

Effect of SLC35A2 on membrane glycan profile and carbohydrate determinants of HCC cells. A lectins array experiment was performed in control, shSLC35A2 and Re-SLC35A2 transfected MHCC97H cells (data in Hep3B and HepG2 cells can be found in sFig3) through the interaction of labeled cells with immobilized lectins on array; combination of visual inspection of scanned images and fluorescence signal over background was considered as a positive interaction (A). Different binding abilities to a variety of lectins among these MHCC97H cells were calculated; data were presented as mean ± SE, *P < 0.05 vs. MHCC97H; ^#P < 0.05 vs. shSLC35A2 (*p < 0.005) Experiments were independently carried out three times. (B) Schematic diagrams of sugar chain structure binding to six differential lectins including AAL, DSA, LCA, MAL-II and WGA, SNA were further presented (C). The expression of three specific carbohydrate determinants, such as Thomsen-Friedenreich (TF) antigen, sialyl Lewis A (SLe^a) and sialyl Lewis X (SLe^x) was analyzed in HCC cells with SLC35A2 overexpression by using a flow cytometer (D)

Fig. 4.

Effect of SLC35A2 on cell adhesion-related molecules and glycosyltransferases of HCC cells. The expressions of cell adhesion-related molecules such as ICAM-1, integrin β1, integrinα2, E-cadherin and glycosyltransferases including Gnt3, Fut8 and B4GalT1 in control, shSLC35A2 and Re-SLC35A2 transfected cells (HepG2, Hep3B and MHCC97H) were detected by using a western blotting assay (A, C, E); data were presented as mean ± SE, *P < 0.05 vs. MHCC97H or HepG2; ^#P < 0.05 vs. shSLC35A2; Experiments were independently carried out three times (B, D, F). Moreover, glycosylation of E-cadherin was investigated after SLC35A2 knockdown by using a western blotting assay and the mobility shift was observed (E). In addition, a KEGG pathway enrichment analysis was performed on genes, which were positively correlated with SLC35A2 in HCC by exploring the public cancer UALCAN databases (http://ualcan.path.uab.edu) (F)

SLC35A2 was necessary in recruiting B4GalT1 to golgi apparatus

To further explore the mechanism of how SLC35A2 affects the intracellular glycosylation, we performed a KEGG pathway enrichment analysis on genes positively correlated with SLC35A2 in HCC by exploring the public cancer UALCAN databases and found that these genes were enriched in histone modification, RNA splicing and Golgi vesicle transport, and nucleocytoplasmic/nuclear transport (Fig. 4H). As mentioned above, SLC35A2 can transport UDP-galactose, the substrates for galactosyltransferase, from the cytosol into Golgi vesicles, an important organelle for glycosylation modification of proteins. Meanwhile it is well known that among galactosyltransferase, β-1,4-galactosyltransferase (B4GalT1) is unique because it encodes an enzyme that participates both in glycoconjugate and lactose biosynthesis [25]. Herein, the levels of B4GalT1 were detected in a variety of HCC cell lines. From these findings, we discovered that the levels of B4GalT1 were higher in MHCC97L, MHCC97H and HCCLM6 cells with high metastatic potential, compared to Hep3B and HepG2 without metastatic potentials (Fig. 5A-B). Of note, in HCC cells, the overexpression of B4GalT1 did not significantly affect the levels of SLC35A2 protein in HCC cells (Supplementary Fig. 4). Furthermore, it was also found that SLC35A2 was significantly co-located with B4GalT1 in HCC cells through co-transfecting pcDNA3.1-HA-SLC35A2 and pcDNA3.1-Flag-B4GalT1 (Fig. 5C). It is also well-known that Golgi apparatus provides the location for glycosylation catalyzed by various glycosyltransferase in cells. Hence, Golgi marker GP73 was further detected in SLC35A2-B4GalT1 complex. The result showed that in metastatic HCC cells, Golgi marker GP73 was obviously detectable in SLC35A2-B4GalT1 coimmunoprecipitated complex; in contrast, in HCC cells without metastatic potential, GP73 was barely detected (Fig. 5D-E). To investigate the role of SLC35A2 in B4GalT1 translocation to Golgi apparatus, we detected the binding of GP73 and B4GalT1 in HCC cells with SLC35A2 knockdown and found that SLC35A2-knockdown induced the dislocation of B4GalT1 with Golgi marker GP73 (Fig. 5 F). In addition, a sequencing analysis revealed that a de novo mutation (c.797G > T, p.G266V) in SLC35A2 gene can lead to the loss of normal SLC35A2 activity [26]. Here, it was further found that compared to wild-type SLC35A2, SLC35A2 (G266V) impeded the binding of B4GalT1 to GP73 (Fig. 5G).

Fig. 5.

Effect of SLC35A2 on recruiting B4GalT1 to Golgi apparatus. The levels of B4GalT1were detected by using real-time PCR (A) and western blotting assays (B) in five HCC cell lines with different metastatic potentials. Experiments were carried out three times. Data were presented as mean ± SE (independent t test). The co-location of SLC35A2 with B4GalT1 in HCC cells through co-transfecting pcDNA3.1-HA-SLC35A2 and pcDNA3.1-Flag-B4GalT1 was further observed by using a fluorescence microscope (Leica, DM500); Scale bars = 10 μm (C). Moreover, the coimmunoprecipitation of SLC35A2 with B4GalT1 in HCC cells was detected by performing an in vitro co-immunoprecipitation assay with anti-flag and anti-HA, respectively (D-E). In addition, the binding of B4GalT1 to GP73 in HCC cells (MHCC97H) with SLC35A2 knockdown (F) and WT or mutant SLC35A2 (G) was evaluated by performing an in vitro co-immunoprecipitation assay with anti-GP73 (G)

Discussion

In recent years, accumulating information regarding molecular aspects of nucleotide sugar transport (NSTs) have been obtained [27, 28]; however, we still know little about the relationship of their aberrant expression and function, as well as their association with cancer cell malignant phenotypes. As an important NST, SLC35A2 holds specific UDP-galactose transporter activity and carbohydrate-proton symporter activity, which are responsible for transporting mainly UDP-galactose from the cytosol into Golgi vesicles where glycosyl-transferases function and where glycans are generated [20]. Mutations in SLC35A2 can cause some diseases associated with congenital disorder of glycosylation, such as Congenital Disorder Of Glycosylation (CDG) and Isolated Focal Cortical Dysplasia (IFCD) [29]. However, until now, there are few studies that explore the biological roles of SLC35A2 in cell phenotypes, particularly in cancer cells. To our knowledge, this is the first study to elucidate the effects of SLC35A2 on the malignant metastatic behaviors of HCC cells and investigate its clinical significance.

In this study we found a significant increase of SLC35A2 expression in metastatic foci compared to primary lesions as well as in metastatic HCC cells compared to non-metastatic HCC cells. It was also observed that this increase was more frequent in HCC patients with lymph node or distant metastasis than in patients without metastasis; meanwhile high-expressed SLC35A2 indicated the poor outcomes in HCC. These results suggested that up-regulated expression of SLC35A2 was in accordance with the enhancement of metastatic capability of HCC cells and poor outcomes of HCC patients and might be recognized as a potential intervention target for treating advanced HCC in the future.

To determine the functional consequences of SLC35A2 aberrant expression, we effectively knocked down intercellular SLC35A2 expression HCC cell lines with different metastatic potentials, followed by a rescue assay to re-confirm the effects of SLC35A2 on the metastatic behaviors of HCC cells. Based on observations to HCC cells biological behavioral changes after SLC35A2 deficiency, we found that the knockdown of SLC35A2 didn’t result in alteration of HCC cell viability, in spite of some reports in which some NSTs are stated to be responsible for a retarded growth in human [17], but instead significantly decreased their invasion, migration and adhesion, which can be rescued with re-expression of SLC35A2 in HCC cells. Notably, in adhesion assay, we coated the flat-bottom plate with different extracellular matrix, taking into account of substrate specificity of each nucleotide-sugar transporter, which is responsible for specific glycol-structure [30], and found that after intercellular SLC35A2 deficiency, abilities of metastatic MHCC97H cells adhesion to collagen IV, FN, E-selectin and laminin was decreased, but the extent of decrease of HCC cells adhesion to E-selectin was largest, followed by laminin, then FN and collagen IV; such inhibition was partly or significantly reversed with SLC35A2 re-expression. Relatively speaking, in non-metastatic HCC cells-Hep3B and HepG2, the adhesion assay presented no significant change during SLC35A2 expression alteration. These results suggested the relatively unique effect of SLC35A2 on metastatic HCC cells and not non-metastatic HCC cells. Mounting evidence suggests that carbohydrates expressed in tumor cells are directly involved in cell-extracellular matrix adhesion [31]. More importantly, the glycosylation changes associated with cancer cell adhesion include the under- and/or overexpression of naturally occurring glycans [32]. Although extra molecular mechanism of SLC35A2 knockdown-induced down-regulated adhesion to extracellular matrix, particularly E-selectin, was not investigated in this study, it was partly explained that down-regulated levels of SLC35A2 could decrease the expression of galactose-containing markers of HCC cells surface such as the TF antigen, SLe^a and SLe^x. These carbohydrate determinants are believed to be linked with the metastatic nature of these cells and the cell adhesion qualities of cancerous cells are altered [33]. They are recognized by selectins or other carbohydrate-binding proteins or by complementary carbohydrates (through carbohydrate-carbohydrate interaction) [34]. Furthermore, in our study, we also found that up-regulated expression of SLC35A2 through exogenous SLC35A2 transfection induced TF antigen, SLe^a and SLe^x expression in metastatic HCC cells rather than non-metastatic HCC cells. This result indirectly supported the above explanation regarding different adhesion ability of HCC cells with SLC35A2 expression alteration to various extracellular matrixes, which may contribute to their metastatic potentials.

Recently, it has also been suggested that changes in glycosylation profiles occur during the development of HCC [35, 36]. However, little is known concerning the effects of some specific glycosyl epitopes on malignant phenotypes of cancer cells, particularly on liver cancer, because the analysis of protein glycosylation processing seems especially relevant to liver pathology, due to this organ’s major influence on the homeostasis of glycoproteins [37]. From the NST’s point of view, the issue mentioned above might be addressed. To elucidate the effect of SLC35A2 on the glycosylation pattern of HCC cells surface, we performed a lectins microarray analysis on a gel substrate lectins microarray, which has been established in our lab [38], through comparing the surface glycosylation profile of SCL35A2-knockdown/re-expression HCC cells with parental HCC cells. Significant differential binding ability to six classes of lectins, including AAL, DSA, LCA, MAL-II and WGA, SNA, was observed, suggesting that intercellular SCL35A2 knockdown affected glycol-structure, such as core fucose (α1–6 linked), α2-3sialic acid, (GlcNAc)n and terminal galactose. These cell surface glycosylation changes have been reported to be associated with tumor development and progress [39]. For example, an increase in sialylation and polysialic acid synthesis is the most frequent tumor-associated aberrant glycosylation, as the glycosyl epitopes promotes cancer invasion and metastasis [40].

In addition, because the mechanisms through which specific glycosyl epitopes induce invasive and metastatic phenotypes of tumor cells are extremely complicated [41]; at the same time, in present, there is also lack of efficient tool to explore specific glycosyl structure [42]. Despite such a dilemma, it has been accepted that an alternative strategy to the analysis of glycosylation is the analysis of specific protein-associated glycans [43], because glycosylation of functionally important membrane proteins may alter tumor cell adhesion or motility in a direction that either promotes or inhibits invasion and metastasis [44, 45]. As a cell surface protein, E-cadherin is glycosylated; thus, disturbances in E-cadherin-based adhesion can contribute to tumor progression [46], including HCC [47]. Characterizing the glycosylation profile of E-cadherin may evaluate the importance of glycosylation modifications in tumor malignant phenotype, such as that involved in the intrahepatic metastasis of HCC [48]. Therefore, we analyzed the glycosylation extent alteration of purified E-cadherin from SCL35A2-knockdown HCC cells. Down-regulated glycosylation of E-cadherin suggested that SLC35A2 was involved in being of carbohydrate determinant of E-cadherin in HCC cells. This is in accordance with some reports in which E-cadherin displayed high mannose type glycans as well as β1,6-branched oligosaccharides with poly-N-acetyl lactosamine structures and α2,3-linked sialic acid residues [43]. However, the molecular mechanism needs to be confirmed in further experiments.

As we have known, the glycan portion of glycoproteins, proteoglycans and polysaccharides is synthesized and modified by glycosyltransferases located in the lumen of the ER and Golgi apparatus [49]. During the process, a variety of nucleotide sugars need to be transported by specific transporters from cytosol into the lumen of the ER and Golgi compartment [15]. Notably, as a unique case among the beta4GalT family, B4GalT1 has an exclusive specificity for the donor substrate UDP-galactose and is involved in glycoconjugate and lactose biosynthesis, suggesting a close association between SLC35A2 and B4GalT1 [25]. In our study, we found that the level of B4GalT1 in metastatic HCC cells was higher than that of non-metastatic HCC cells and was not affected by SLC35A2 expression alteration and vice versa. Furthermore, interaction between SLC35A2 and B4GalT1 in HCC cells, in accordance with some reports [50], may collaboratively achieve a specific glycosylation structure. In metastatic HCC cells, Golgi marker GP73 was detectable in the SLC35A2-B4GalT1 complex; however, in HCC cells without metastatic potentials, GP73 was barely detected. The results implied that the amount of SLC35A2-B4GalT1 complex resident in Golgi apparatus may be associated with metastatic behaviors of HCC cells and that the translocation of SLC35A2-B4GalT1 complex into Golgi apparatus is necessary for malignant glycosylation features, which are associated with metastatic phenotype of HCC. Furthermore, it was also demonstrated that the delivery of B4GalT1 to Golgi apparatus requires the involvement of SLC35A2 in HCC cells and that G266 is a critical site in maintaining the function of B4GalT1. Altogether, the results indicated that the translocation of SLC35A2-B4GalT1 complex into Golgi apparatus is necessary for malignant glycosylation features and is associated with metastatic feature in HCC.

However, there are some limitations in the present study, partly due to technical drawback. For example, questions regarding whether SCL35A2 can generate unique glycol-structure phenotype in HCC cells? What is the extra molecular mechanism of SLC35A2 involved in HCC cells invasion, adhesion and metastasis remain yet to be answered. Nevertheless, the current data are sufficient to demonstrate that SLC35A2 may function as a regulator of HCC metastasis process, particularly in adhesion, through modulating cell surface carbohydrate determinants and inducing cell adhesive ability. In general, cell social function is predominantly affected by glycosylation [51]. At present, various reagents that block carbohydrate-mediated tumor cell adhesion or block glycosylation processing have been shown to inhibit tumor cell metastasis [52]. This study provides an experimental basis for the further development of “anti-adhesion target therapy” through blocking glycosylation signaling.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

NSTs

nucleotide sugar transporters

ER

endoplasmic reticulum

SLC35

solute carrier family 35

CDG

congenital disorders of glycosylation

SLC35A2

solute carrier family 35 member A2

HCC

hepatocellular carcinoma

MEM

minimum essential medium

DMEM

dulbecco’s modified eagle medium

EDTA

ethylene diamine tetraacetic acid

PBS

phosphate buffered saline

RNA

ribonucleic acid

mRNA

messenger ribonucleic acid

shRNA

short hairpin ribonucleic acid

CPP

calcium phosphate precipitation

MOI

multiplicity of infection

B4GalT1

beta-1,4-galactosyltransferase 1

RT-PCR

reverse transcription-polymerase chain reaction

RT

reverse transcription

SDS

sodium dodecyl sulfate

PVDF

polyvinylidene difluoride

PCR

polymerase chain reaction

cDNA

complementary deoxyribonucleic acid

qRT-PCR

quantitative reverse transcription-polymerase chain reaction

GAPDH

lyceraldehyde-3-phosphate dehydrogenase

ACTB

actin beta

BSA

bovine serum albumin

OD

optical density

HEC

human endothelial cells

SDS–PAGE

sodium dodecyl sulphate-polyacrylamide gel electrophoresis

GP73

golgi protein 73

LIHC

liver hepatocellular carcinoma

FN

fibronectin

AAL

aleuria aurantia lectin

DSA

datura stramonium

LCA

lens culinaris agglutinin

MAL-II

maackia amurensis lectin II

WGA

wheat germ agglutinin

SNA

sambucus nigra

TF

thomsen-friedenreich

SLe^a

sialyl Lewis A

SLe^x

sialyl Lewis X

Gnt3

n-acetylglucosaminyltransferase-3

Fut8

fucosyltransferase8

(GlcNAc)n

n-acetylglucosamine

OS

overall survival

mAb

monoclonal antibody

SAM

Significance Analysis of Microarrays

ANOVA

analysis of variance

SE

standard error

CDG

Congenital Disorder of Glycosylation

IFCD

Isolated Focal Cortical Dysplasia

Author contribution

GK and ZJB designed the experiments; CHX, WSK, GDM and YKK, CHP, HYL, LMM generated the data including supplementary data; CHP collected and provided the clinical samples; WSK and YKK performed bioinformatics and statistics analysis; CHX, WSK, ZJB and GK drafted the manuscript; all authors read and revised the manuscript; GK and ZJB guided the project.

Hongxia Cheng, Sikai Wang and Dongmei Gao share equal contribution to the work.

Funding

This work was financially supported by National Natural Science Foundation of China (81872492), Shanghai Pujiang Program (15PJD007) and Shanghai Natural Science Fund (15ZR1406300).

Data availability

All available material and data are presented in the manuscript.

Code availability

Not applicable.

Declarations

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not applicable.

Conflicts of interest/competing interests

The authors declare no conflicts of interest.