|Home | About | Journals | Submit | Contact Us | Français|
Aberrant expression and hyperactivation of the human epidermal growth factor receptor 2 (ErbB2) constitute crucial molecular events underpinning gastric neoplastic transformation. Despite ErbB2 extracellular domain being a well-known target for glycosylation, its glycosylation profile and the molecular mechanisms through which it actively tunes tumorigenesis in gastric cancer (GC) cells remain elusive. We aimed at disclosing relevant ErbB2 glycan signatures and their functional impact on receptor’s biology in GC cells. The transcriptomic profile of cancer-relevant glycosylation enzymes, and the expression and activation of the ErbB receptors were characterized in four GC cell lines. Cellular- and receptor-specific glycan profiling of ErbB2-overexpressing NCI-N87 cells unveiled a heterogeneous glycosylation pattern harboring the tumor-associated sialyl Lewis a (SLea) antigen. The expression of SLea and key enzymes integrating its biosynthetic pathway were strongly upregulated in this GC cell line. An association between the expression of ERBB2 and FUT3, a central gene in SLea biosynthesis, was disclosed in GC patients, further highlighting the crosstalk between ErbB2 and SLea expression. Moreover, cellular deglycosylation and CA 19.9 antibody-mediated blocking of SLea drastically altered ErbB2 expression and activation in NCI-N87 cells. Altogether, NCI-N87 cell line constitutes an appealing in vitro model to address glycan-mediated regulation of ErbB2 in GC.
Gastric cancer (GC) represents the fifth most incident human malignancy and the third leading cause of cancer-related deaths worldwide . Being frequently diagnosed at advanced stages, at which surgical resection and conventional chemotherapeutic approaches fall short in improving patients’ dismal prognosis, GC remains a globally heavy health burden .
Overexpression of the human epidermal growth factor receptor 2 (ErbB2), a direct consequence of the ERBB2 gene amplification, constitutes a well-established molecular hallmark of multiple human solid tumors, and has been consistently reported as a key molecular driver of gastric carcinogenesis, by triggering the aberrant activation of complex downstream signaling pathways that ultimately dictate malignant cell behavior (reviewed in ). Indeed, ErbB2 overexpression is observed in 7% to 34% of GC patients and constitutes a predictive factor of poor disease prognosis . The 185-kDa ErbB2 transmembrane protein belongs to the ErbB epidermal growth factor receptor family of receptor tyrosine kinases (RTKs). Three additional members are included in this family (ErbB1/epidermal growth factor receptor (EGFR), ErbB3 and ErbB4), with which ErbB2 shares structural homology. The ErbB receptors are composed of an extracellular ligand-binding domain, a short hydrophobic transmembrane domain and cytosolic tail with intrinsic tyrosine kinase activity (only absent in ErbB3) . By exhibiting a ligand-independent mode of activation and constituting the preferred heterodimerization partner of the remaining ErbB receptors, ErbB2 holds the most pronounced ability for potent and prolonged signal transduction. ErbB2 therefore carries an exponentially aggravated oncogenic and transforming potential and represents an excellent molecular candidate for selective anti-cancer targeted therapy [6,7]. In fact, the trastuzumab monoclonal antibody (MAb), targeting the subdomain IV of the ErbB2 extracellular region, has become the first personalized standard of care for patients harboring advanced ErbB2-positive GC, in combination with conventional chemotherapy . However, the emergence of acquired tumor resistance, through still elusive molecular mechanisms, has limited the expected benefits from this therapeutic strategy [9,10].
Protein glycosylation is the enzymatic addition of sugar moieties to specific amino acid residues, and consists of a complex and finely regulated cellular process, representing the most abundant type of protein posttranslational modifications [11,12]. Glycosylation results from the well-orchestrated action of a wide range of enzymes and organelles and is acknowledged to be crucial in several cellular, physiological and pathological processes, including the carcinogenesis of multiple organs, such as the stomach [12,13]. The abnormal expression and activity of glycosyltransferase and glycosidase enzymes, as well as alterations in the structure and expression levels of cancer-associated glycan epitopes, have been consistently linked to the dismal prognosis of cancer patients (reviewed in ). Despite the remarkable advances made in glycan-based analytical techniques, many of the molecular mechanisms through which aberrant glycosylation actively tunes malignant cell behavior within the intricate context of tumor biology are still unknown .
RTK maturation is tightly regulated by protein N-linked glycosylation . Furthermore, the glycosylation status of these transmembrane proteins has been shown to play a pivotal role in several aspects of receptor’s biology in the context of distinct human developmental and cancer models [16,17,18,19,20]. In particular, the extracellular domain of the ErbB2 receptor harbors seven putative sites for the anchorage of N-linked glycan structures (www.cbs.dtu.dk/NetNGlyc), five of which had their glycosylated status demonstrated in human mammary carcinoma cell lines [21,22]. However, protein glycosylation is highly context-dependent, which severely limits the extrapolation of relevant findings from one pathological setting to another . In this sense, the detailed glycosylation profile of this oncogenic RTK within the context of GC requires further elucidation.
In the present study, we have performed a molecular characterization of four GC cell lines (NCI-N87, AGS, MKN45 and MKN74), including the transcription analysis of cancer-relevant glycosylation enzymes, as well as the expression and activation status of the ErbB family of receptors. Additional cellular and receptor-specific characterization of NCI-N87, an in vitro model of ErbB2 overexpression and hyperactivation, unveiled a highly heterogeneous and complex glycosylation pattern encompassing well-established tumor-associated glycan determinants, such as the endothelial selectin-ligand sialyl Lewis a (SLea). The consistent expression of this glycan epitope and key enzymes of its biosynthetic pathway observed in ErbB2-overexpressing GC cells was further corroborated by in silico analysis. Moreover, an in silico association between the ERBB2 and FUT3 genes was disclosed in a cohort of GC patients. Finally, in vitro deglycosylation and antibody-mediated glycan-blocking of SLea drastically altered receptor’s expression and activation in ErbB2-overexpressing NCI-N87 GC cells. Altogether, these results portray the NCI-N87 cell line as a useful in vitro tool to study the regulatory role of glycans in ErbB2-driven gastric carcinogenesis.
To characterize the expression pattern of the four RTKs integrating the ErbB family (ErbB1/EGFR, ErbB2, ErbB3 and ErbB4), Western blot (WB) analysis was performed on whole cell lysates from four GC cell lines (NCI-N87, AGS, MKN45 and MKN74), representative for the heterogeneous nature of gastric cancer regarding ErbB2 status (Figure 1). Furthermore, the phosphorylation status of the tyrosine-kinase domains of ErbB2 and EGFR receptors was additionally determined. In general, the four GC cell lines exhibited heterogeneous expression and activation of the analyzed RTKs. The ERBB2-amplified and well-differentiated NCI-N87 cell line displayed the highest levels of total ErbB2 protein, and was the only cell line with strong constitutive activation of this receptor (Figure 1a). ErbB2 expression was also detected in the remaining three GC cell lines, although in considerably lower levels. Despite similar levels of total EGFR expression being observed for NCI-N87 and MKN74, the former cell line displayed substantially higher receptor phosphorylation (Figure 1b). Although in substantially lower levels, endogenous EGFR activation was also detected in AGS, MKN45 and MKN74 cell lines. As for ErbB3, all cell lines displayed comparable protein levels of the receptor, with the exception of AGS where ErbB3 expression was lower (Figure 1c). Expression of ErbB4 was particularly low across the four screened GC cell lines.
The presence of tumor-associated glycan epitopes at the surface of malignant cells is regulated by the coordinated expression and activity of specific glycosyltransferases. In this sense, prior to an ErbB2-targeted glycosylation analysis, we first evaluated the general glycophenotype of the four GC cell lines. The transcript levels of eight key genes encoding for glycosyltransferases previously reported to be implicated in human neoplastic transformation were assessed by real-time quantitative PCR (RT-qPCR) (Figure 2). The FUT3 and FUT8 genes encode for enzymes with fucosyltransferase activity responsible for the terminal (α1,3-linked) or core (α1,6-linked) fucosylation of their glycoconjugate targets, respectively. The MGAT3 and MGAT5 genes encode for enzymes that catalyze the addition of bisecting and branching N-acetylglucosamine (GlcNAc) during the biosynthesis of complex N-glycans, respectively. The ST3GAL3 and ST3GAL4 genes encode for sialyltransferases capable of adding α2,3-linked sialic acids to the terminal non-reducing end of glycan chains. The ST6GAL1 gene encodes for an enzyme capable of transferring an α2,6-linked sialic acid monosaccharide to N-glycan substrates. Finally, The B3GALT5 gene encodes for an enzyme involved in the biosynthesis of type 1 chains of oligosaccharides, which is prerequisite for the formation of Lewis a (Lea), Lewis b (Leb) and SLea antigens.
The obtained RT-qPCR uncovered a distinct glycosyltransferase expression signature for each of the screened GC cell lines. In particular, the B3GALT5 gene showed an expression range of multiple orders of magnitude across the four screened cell lines. Indeed, the diffuse type MKN45 cell line exhibits remarkably higher levels of this enzyme’s mRNA when compared to the other cell lines, especially in the case of intestinal type MKN74 cells where no mRNA was detected. A similar expression pattern was observed for ST6GAL1. However, regarding ST6GAL1, a generally higher transcript expression was consistently observed across the four screened GC cell lines, with the exception of MKN74, in comparison to the remaining screened glyco-genes. Illustrative of the cell line-specific glycosyltransferase expression signature, the ErbB2-overexpressing NCI-N87 cell line displayed the highest expression of FUT3, MGAT5 and ST3GAL4 genes, but also the lowest detected transcript levels of both FUT8 and ST3GAL3. The obtained results are in accordance with previous data published by our group, in respect to the GC cell lines and glycosyltransferase-coding genes commonly analyzed in both studies .
Since the NCI-N87 GC cell line exhibited the highest levels of both ErbB2 expression and activation, it was therefore selected for further glycan epitope characterization. Due to the subcellular localization of ErbB2, specific glycan-recognizing monoclonal antibodies and lectins were used for the immunofluorescent detection of the glycan structures expressed at the cell membrane of NCI-N87 cells (Figure 3). The obtained results unveiled a complex glycosylation signature, encompassing a range of structural and functionally diverse glycan structures, many of them previously described as key molecular players in the onset and progression of human cancer .
The cell membrane of NCI-N87 cells stained positive for all neutral (Lea, Leb and Lewis y (Ley)) and sialylated (SLea and sialyl Lewis x (SLex)) forms of the Lewis antigens, with the exception of Lewis x (Lex), for which no signal was detected. The expression of the tumor-associated sialyl Tn (STn) truncated O-glycan antigen was also undetected. This observation suggests that the detected terminal α2,6 sialic acid moieties by the lectin SNA are present on N-glycan chains. The presence at the cellular membrane of four additional cancer-relevant glycan structures (terminal and core fucose, high mannose, and bisecting and branching GlcNAc) was detected through their specific recognition by a panel of carbohydrate-binding lectins (Aleuria aurantia (AAL), Concanavalin A (ConA), Phaseolus vulgaris erythroagglutinin (PHA-E) and Phaseolus vulgaris leucoagglutinin (PHA-L), respectively). Regarding the enzymes responsible for the biosynthesis of distinct glycan determinants, these results are in agreement with the previously characterized transcription profile of this particular cell line (Figure 2). For example, the presence of SLex, terminal α2,6 sialic acid, bisecting and branched complex N-glycan determinants, consistently match the detected expression of the ST3GAL4, ST6GAL1, MGAT3 and MGAT5 genes, respectively.
With the aim of elucidating the ErbB2 glycosylation pattern in NCI-N87 cells, the receptor was immunoprecipitated from NCI-N87 total cell lysates. The yield and purity of receptor immunoprecipitation was evaluated by silver staining, and ErbB2 was validated by the subsequent peptide mass fingerprint analysis of the major 185-kDa band by matrix-assisted laser desorption ionization time-of-flight (MALDI/TOF-TOF) mass spectrometry (Figure 4a). The visible bands at 50 kDa correspond to the IgG heavy chains of the ErbB2-specific MAb used for receptor immunoprecipitation. As a control, a parallel immunoprecipitation reaction using normal rabbit IgGs was performed. As no signal in ErbB2 WB was detected, the specificity of receptor’s immunoprecipitation was validated (Figure 4b).
For the profiling of the glycan structures specifically carried by ErbB2 in NCI-N87 cells, the immunoprecipitated receptor was further characterized by both WB and lectin blot analysis with the previously used panel of glycan-recognizing antibodies and lectins found positive by immunofluorescence in NCI-N87 cells (Figure 4c). The 185-kDa band identified as the ErbB2 receptor displayed a specific glycosylation profile resembling the one of NCI-N87 cells. The only exception was SLex, for which no signal was observed by WB. Indeed, when SLex WB was performed on NCI-N87 whole cell lysates, a single band was visible with a corresponding molecular weight higher than 185-kDa, indicating that the SLex epitope is carried by a protein other than ErbB2 in this particular GC cell line (data not shown). To our knowledge, this is the first study identifying ErbB2 as a carrier of the tumor-associated SLea, a well-established molecular player in the malignant behavior of tumor cells.
Having identified ErbB2 as a protein carrier of the SLea antigen in ErbB2-overexpressing NCI-N87 cells, we aimed at dissecting the biosynthetic pathway of this cancer-relevant glycan epitope in this particular cell line, both in vitro and in silico. Lewis antigens constitute terminal structures in both N- and O-linked oligosaccharide chains . To elucidate which type of SLea-capped glycans are carried by ErbB2 in NCI-N87 cells, the immunoprecipitated receptor was digested with the PNGase F enzyme, a glycosidase capable of removing all N-linked glycans through the specific cleavage between the sugar’s innermost GlcNAc and the protein’s asparagine residue (Figure 5a). As expected, following PNGase F digestion, a clear shift in the molecular weight of ErbB2 was observed. This shift was accompanied by the disappearance of the 185-kDa SLea band, indicating that this terminal carbohydrate epitope is present on N-linked glycan chains.
To establish a valid causal association between the performed transcriptomic analysis of glycosyltransferase-coding genes and the expression of SLea in the NCI-N87 GC cell line (Figure 2 and Figure 3, respectively), in silico analysis, using the transcriptomic Barretina CellLine data deposited in the OncomineTM platform (www.oncomine.org), was performed. For this purpose, the either exceptionally high or low expression of each gene of interest, as compared to a total of 35 GC cell lines, was assessed in the four GC cell lines (Figure 5b, right panel). Amongst the four analyzed GC cell lines, NCI-N87 clearly exhibited the most robust expression profile of genes relevant for SLea biosynthesis and thus emerged as the most promising candidate to consistently express this glycan epitope. Besides the pronounced overexpression of SLea-relevant genes (FUT3, TSTA3, SLC35C1, and SLC35A1), no downregulation of enzyme-coding genes directly implicated in SLea biosynthesis is observed, contrary to what occurs for the remaining three GC cell lines. The generated in silico results are in robust alignment with the ones obtained in the performed transcriptomic analysis (Figure 2). Illustrating this observation is the overexpression of FUT3 by the ERBB2-amplified NCI-N87, and of B3GALT5 by MKN45. It is also worth noting that, although unrelated to SLea biosynthesis, ST3GAL4 is downregulated in AGS cells, and ST6GAL1 is downregulated in MKN45, NCI-N87 and AGS when compared to MKN74 cell line. To validate the robustness of the in silico data, the four GC cell lines were screened for the expression of SLea by WB (Figure 5c). Remarkably, and in agreement with the in silico results, the only GC cell line exhibiting a strong SLea immunodetection was NCI-N87. Our data therefore suggest that the marked upregulation of SLea in NCI-N87 cells may be driven by high FUT3 expression levels.
Finally, we sought to investigate whether the apparent ErbB2-SLea crosstalk could also be observed in patients harboring GC. For this purpose, an in silico analysis based on the transcriptomic data of the Ooi dataset extracted from the OncomineTM platform was carried out. Among the 200 primary gastric carcinomas of the referred dataset, a statistically highly significant positive association (Spearman’s rank correlation p < 0.0001) between the expression of ERBB2 and FUT3 genes was observed (Figure 5d). The latter gene encodes for the FucT-III glycosyltransferase, also known as the Lewis enzyme, which is solely responsible for the biosynthesis of SLea . In conclusion, a higher ERBB2 expression was associated to FUT3 upregulation in the GC patients, and vice-versa.
Protein N-linked glycosylation actively regulates RTK maturation, translocation to the cell membrane and signaling activity . Having identified the major types of glycans carried by ErbB2 in NCI-N87 cells, we aimed at assessing the functional impact of glycosylation-targeting agents on receptor’s expression and activation status in this particular cell line. NCI-N87 cells were subjected to increasing concentrations of two distinct treatment regimens: tunicamycin, a natural inhibitor of the enzyme catalyzing the first step of N-glycan biosynthesis, and the CA 19.9 MAb targeting the SLea epitope. Following 24 h of treatment, WB analysis of both total and phosphorylated forms of ErbB2 was performed on freshly collected protein extracts (Figure 6). Only the highest concentration of the tunicamycin compound (1000 ng/mL) exerted a pronounced effect on the levels of both total and activated protein (Figure 6a). In regard to total receptor’s expression, a second band with lower molecular weight strongly indicates the presence of a deglycosylated version of ErbB2, in addition to the heavier 185-kDa band corresponding to the fully glycosylated receptor. Moreover, the same concentration of tunicamycin led to a significant reduction of receptor’s phosphorylation. Since ErbB2 phosphorylation seems to be only detected at 185-kDa, the observed reduction can be due to a general decrease in the abundance of the fully matured transmembrane receptor, which is the one targeted by phosphorylation. As for the treatment of NCI-N87 cells with the CA 19.9 MAb, drastic effects on both forms of the receptor could also be clearly observed. While the lowest CA 19.9 concentration (0.2 µg/mL) produced no visible effects on ErbB2 total expression, and a mild, yet detectable, reduction on the ErbB2 phosphorylation, the highest MAb concentration (2 µg/mL) led to a severe reduction of total receptor’s levels and a complete abrogation of receptor’s phosphorylation.
The extracellular region of the ErbB2 mitogenic receptor is a well-documented target for extensive glycosylation, which has been shown to actively promote human carcinogenesis through the regulation of receptor’s total expression, subcellular localization and signaling potential [15,26,27]. However, the ErbB2 glycosylation signature in GC cells and the molecular mechanisms through which it may tune receptor biology towards malignancy remain poorly understood. The present study aimed at disclosing glycan determinants carried by ErbB2 that may act as functional regulators of the receptor in ErbB2-driven gastric carcinogenesis. The NCI-N87 cell line was therefore selected, amongst the four screened cell lines, as the most appropriate in vitro model of ErbB2-positive GC.
Despite the similar high levels of total EGFR expression observed in NCI-N87 and MKN74, the former GC cell line exhibited the most pronounced activation of this receptor, and also of ErbB2, as previously described . This observation may reflect the homo- or hetero-oligomerization capacity of the ErbB transmembrane receptors, which constitutes a well-established pre-requisite for intracellular activation . The ErbB2 protein is overexpressed in NCI-N87 cells as a direct consequence of gene amplification and constitutes the preferred heterodimerization partner of EGFR [30,31]. Therefore, in the NCI-N87 GC cell line, the high levels of expression of the ErbB2 RTK at the cell membrane lead not only to its own hyperactivation, but also of the remaining members of the family. This is in agreement with the significantly lower levels of both EGFR endogenous activation ErbB2 total expression observed in the three other GC cell lines. However, whether ErbB2-specific glycosylation plays an active part on dimer formation, although reported for other ErbB receptors, remains a question to be addressed [16,32,33,34,35].
The elaborate shaping of the cellular glycan landscape depends on the highly controlled and dynamic crosstalk between both glycosyltransferases and glycosidases . Interestingly, the observed expression levels of the selected glycogenes agree with the known competing enzymatic role that these glycosyltransferases play in the cell, in response to both external and internal stimuli. For example, the MGAT3 gene displays a generally lower expression across all four GC cell lines compared to MGAT5. The same can be observed for ST3GAL3 and ST3GAL4, respectively.
Although the expression of specific glycan epitopes depends on multiple factors, such as enzyme activity, substrate recognition and availability of both donor and acceptor molecule, in NCI-N87 cells we have observed a remarkable match between the transcriptomic profile of cancer-relevant glycosyltransferase-coding genes and the expression of their respective glycan products. This is illustrated by the correspondence between the observed expression of the MGAT3, MGAT5, ST3GAL3 and ST3GAL4, and ST6GAL1 genes, and the immunodetection of bisected and branched N-glycans, sialylated versions of the Lewis antigens, and the terminally α2,6-sialylated N-glycan chains, respectively, both at the cellular and receptor levels. Some of the disclosed glycan signatures have been consistently reported as active promoters of human gastric carcinogenesis. The expression of the Leb antigen at the surface of the stomach’s epithelial lining is crucial for Helicobacter pylori (H. pylori) pathogenic colonization of the host’s gastric mucosa . The α2,3 sialylated versions of the Lewis antigens, SLex and SLea, constitute additional motifs for H. pylori adhesion. Moreover, they are known ligands of selectins expressed at the surface of endothelial cells, and therefore function as molecular docking sites during the metastatic dissemination of circulating tumor cells (reviewed in ).
The highest mRNA expression of MGAT5 was detected in ErbB2-overexpressing NCI-N87 cells. Interestingly, a previous study reported that the transcriptional activation of the MGAT5 gene and the activity of its translated product, the GnT-V enzyme, were directly regulated by the ErbB2 oncogenic receptor and key transducers of its downstream signaling pathway [39,40]. Furthermore, in another study, the overexpression of the MGAT5 gene in mammary cells potentiated ErbB2-mediated malignancy in mouse models . These observations suggest that branched N-glycans, the direct products of the MGAT5 gene, and here disclosed as part of ErbB2 glycosylation signature, may function as active regulators of ErbB2-mediated signaling in NCI-N87 cells.
This study identified ErbB2 as a previously unreported protein carrier of the tumor-associated SLea epitope. In addition, both in silico and in vitro analysis of genes involved in SLea biosynthesis corroborate this finding, as illustrated by FUT3 overexpression in NCI-N87 cells. The FUT3 gene encodes for the only enzyme capable of the in vivo biosynthesis of SLea . In this sense, the disclosed statistically significant positive correlation between ERBB2 and FUT3 expression in GC patients additionally strengthens this interplay. Moreover, the dramatic impact observed on both total and phosphorylated ErbB2, following treatment of the NCI-N87 cells with an anti-SLea-specific MAb, further supports the association between ErbB2 status and SLea expression in this particular cell line. These results are in agreement with a previous study, where treatment of ErbB2-overexpressing tumor cells with anti-Ley MAb produced a similar impact in the receptor’s expression . The decreased detection of both total and activated ErbB2 following treatment with CA 19.9 MAb may reflect a rapid receptor internalization and recycling from the cell membrane following antibody-binding . The tunicamycin treatment, on the other hand, by completely abrogating the N-glycosylation pathway, prevents the “naked” and misfolded receptor from accumulating at the cell membrane and, therefore, of becoming activated . Our findings have uncovered an interesting regulatory interplay that should be further elucidated. Additional studies using the administration of other glycosylation inhibitors, glycan specific antibodies or glycosidases are warranted and may further unravel the molecular basis of the underlying mechanism.
Although a steady decline of its incidence and mortality rates has been observed over the past decades, GC still remains a lethal disease with worldwide coverage . Furthermore, the discovery of novel and reliable biomarkers for early disease diagnosis remains a challenging issue in GC research. On the other hand, the abnormal expression of specific glycoconjugates by malignant cells benefits from a degree of unprecedented specificity and constitutes, in this way, a foreseeable promising source of novel diagnostic, prognostic and therapeutic molecular tools. The SLea carbohydrate antigen is deeply involved in multiple molecular processes underlying neoplastic transformation [44,45,46,47,48]. Moreover, the frequent upregulation of circulating SLea-modified glycoconjugates in individuals bearing gastrointestinal tumors led to the clinical implementation of the CA 19.9 serological assay. The SLea-specific CA 19.9 assay has become widely used for monitoring disease progression and patient response to therapy [49,50,51]. Since the shedding of the ErbB2 extracellular domain by malignant cells into the bloodstream is commonly observed in patients harboring ErbB2-positive cancers, circulating ErbB2 has been previously proposed as a novel circulating biomarker and, curiously, had its performance found comparable to the one of CA 19.9 [52,53,54]. In this light, our findings open a new research line: if proven that ErbB2 is, in fact, an in vivo protein carrier of SLea in ErbB2-positive GC patients, assessing ErbB2 modified with SLea in the sera of patients might surpass the clinical performance of either circulating ErbB2 or SLea.
Regardless of the initial enthusiasm following drug development, the use of the trastuzumab MAb in the treatment of ErbB2-expressing advanced GC frequently fails to ameliorate the overall survival of the majority of patients . Nevertheless, due to its consistently observed overexpression in a significant fraction of patients, the ErbB2 transmembrane receptor remains one of the most promising molecular targets for the clinical management of GC. In the near future, it is likely that advances in the field of glycobiology will allow the improvement of classic protein targets in human malignancy, alongside the widespread usage of novel glycoconjugates as effective theranostic tools. In this study, the NCI-N87 cell line is reported as an interesting in vitro model system to study the regulatory role of glycans in ErbB2-positive GC.
Four human GC cell lines (NCI-N87, AGS, MKN45, and MKN74) were used in this study. The intestinal type NCI-N87, AGS and MKN74 cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA) and the diffuse type MKN45 cell line was obtained from the Japanese Cancer Research Bank (Tsukuba, Japan). Cell line identity was authenticated by standard short tandem-repeat based DNA profiling. All four cell lines were grown in a monolayer culture and maintained at 37 °C in an atmosphere of 5% CO2, in complete growth medium: RPMI 1640 GlutaMAX (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% heat-inactivated fetal bovine serum (Biowest, Riverside, MO, USA). Cultured cell lines were routinely tested for mycoplasma contamination.
Total protein lysates were collected by scrapping confluent cultures of each cell line using lysis buffer 17 (R&D Systems, Minneapolis, MN, USA) supplemented with 1 mM sodium orthovanadate (Sigma-Aldrich, St. Louis, MO, USA), 1 mM phenylmethanesulfonyl fluoride (PMSF) (Sigma-Aldrich) and cOmpleteTM protease inhibitor cocktail (Roche, Basel, Switzerland; Sigma-Aldrich). Protein concentration of whole cell lysates was determined using the DC protein assay (BioRad, Hercules, CA, USA). Distinct amounts of total protein lysates (15 µg for ErbB2, pErbB2, EGFR and pEGFR, 30 µg for ErbB3, 50 µg for ErbB4, and 50 µg for SLea analysis) were denaturated and negatively charged by boiling for 10 min in Laemmli buffer containing 5% β-mercaptoethanol at 100 °C, subjected to 8% SDS-PAGE, and blotted onto a nitrocellulose membrane (GE Healthcare Life Sciences, Chicago, IL, USA). Following 1 h blocking at room temperature with either 5% non-fat milk or 5% of bovine serum albumin (BSA) (Sigma-Aldrich) in 0.1% Tween® 20 (Sigma-Aldrich) in tris buffered saline (TBS), membranes were incubated overnight at 4 °C with primary antibodies. For biotinylated lectins, membranes were blocked with 2% polyvinylpyrrolidone (Sigma-Aldrich) in TBS overnight at 4 °C and, after washing, incubated for 1 h at room temperature with the corresponding lectin. Then, membranes were washed three times for 10 min with TBS-Tween® 20, and were incubated for 1 h at room temperature with peroxidase-conjugated secondary antibodies D1/2000 (Santa Cruz Biotechnology, Dallas, TX, USA, and Jackson ImmunoResearch, West Grove, PA, USA) or peroxidase-conjugated streptavidin D1/100,000 (GE Healthcare) for Western and lectin blot, respectively. Following three additional washes in TBS-Tween® 20, chemiluminescence signal was obtained using the ECL WB detection reagent and films (both from GE Healthcare). All WB experiments were performed in triplicate. The information regarding dilutions used for antibodies and lectins is provided in Table 1.
Total RNA from the four GC cell lines was isolated using TRI reagent (Sigma-Aldrich, St. Louis, MO, USA), according to manufacturer’s instructions. Following spectrophotometric assessment of RNA yield and quality, 5 µg of each cell line’s total RNA were reverse-transcribed into single-stranded cDNA using SuperscriptTM II Reverse Transcriptase kit (Invitrogen, Thermo Fisher Scientific), following manufacturer’s recommendations. Glycosyltransferase mRNA expression levels were quantified by RT-qPCR using TaqManTM Universal PCR Master Mix II, no UNG, and the specific TaqManTM gene expression assays, listed below, in a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA; Thermo Fisher Scientific). The mRNA expression levels of 18S Ribosomal 5 (RNA18S5) endogenous control were also measured for normalization of target gene abundance. RT-qPCR was carried out in triplicate for each biological sample and in duplicate for the negative controls. The following TaqManTM gene expression assays were used: FUT3 (Hs01868572_s1), FUT8 (Hs00189535_m1), MGAT3 (Hs02379589_s1), MGAT5 (Hs01073268_m1), ST3GAL3 (Hs00544035_m1), ST3GAL4 (Hs00920871_m1), ST6GAL1 (Hs00949382_m1), B3GALT5 (Hs00707757_s1), and RNA18S5 (Hs99999901_s1). Data were analyzed by ΔΔCt method .
Following trypsinization, NCI-N87 cells were washed twice with ice-cold phosphate buffered saline (PBS, pH 7.4) and air-dried overnight on 12-well microscope glass slides (Marienfeld Superior, Lauda-Königshofen, Germany). Prior to immunofluorescence, cells were fixed for 5 min at room temperature with 4% paraformaldehyde (Alfa Aesar, Haverhill, MA, USA; Thermo Fisher Scientific). Single immunofluorescence was performed as follows. Samples were rinsed three times in PBS for 5 min and permeabilized with 0.5% TritonTM X-100 (Sigma-Aldrich) in PBS for 10 min at 4 °C. Samples were rinsed three times with PBS and incubated for 30 min with goat non-immune serum (Dako, Agilent, Santa Clara, CA, USA) D1/5 in PBS 10% BSA or PBS 10% BSA, for MAb and lectin immunofluorescent labeling, respectively. Samples were rinsed in PBS and incubated overnight at 4 °C with MAbs and biotinylated lectins (Table 1) diluted in PBS containing 5% BSA (MAbs) or PBS (lectins). Samples were washed three times for 5 min in PBS and incubated for 45 min at room temperature with secondary antibodies labeled with either Alexa Fluor® 488 or 594 fluorochromes (Thermo Fisher Scientific) or FITC-conjugated streptavidin, for MAb- and lectin-mediated immunolabeling, respectively. Samples were washed three times for 5 min in PBS, incubated with 1 µg/mL DAPI for 5 min at room temperature for nuclear staining and mounted in VectaShield (Vector Laboratories, Burlingame, CA, USA). Images were acquired with a Zeiss Axio cam MRm and the AxioVision Rel. 4.8 software (both from Zeiss, Oberkochen, Germany). All experiments were performed in duplicate.
For ErbB2 immunoprecipitation, 400 µg of total protein from NCI-N87 whole cell lysates were pre-cleared with 30 µL of Protein G Sepharose Fast Flow beads (GE Healthcare Life Sciences) for 2 h at 4 °C. Separately, 60 µL of Protein G Sepharose Fast Flow beads were conjugated with 5 µL of the 29D8 ErbB2 MAb (Cell Signaling Technology, Danvers, MA, USA) for 2 h at 4 °C. Immunoprecipitation was performed by incubation of the pre-cleared lysate with the antibody-conjugated beads overnight at 4 °C. All steps were performed with gentle agitation. In parallel, the same immunoprecipitation was performed using normal rabbit IgGs (Merck Millipore, Burlington, MA, USA) as a negative control. After washing with ice-cold 0.1% Triton X-100 in PBS, the immune complexes were released and denaturated by boiling for 10 min in Laemmli buffer containing 5% β-mercaptoethanol at 100 °C. The immunoprecipitates were subjected to SDS-PAGE followed by either silver staining, using Pierce® Silver Stain for Mass Spectrometry kit (Thermo Fisher Scientific) according to the manufacturer’s instructions, or WB as described in Section 4.2. All experiments were performed in duplicate.
Following silver staining, the major visible band was excised from the gel, washed with 50% acetonitrile in 50 mM ammonium bicarbonate, digested for 3 h with 20 ng of trypsin at 37 °C, and analyzed on a MALDI mass spectrometer (4800 Plus MALDI TOF/TOF Analyzer, SCIEX) as described in . Proteins were identified by Peptide Mass Fingerprint using the Mascot software v2.5.1 (Matrix Science, London, UK). Protein searches were performed against the UniProt protein sequence database for the Homo sapiens taxonomic selection (2017_08, UP000005640 reviewed proteome, canonical proteins). The established search parameters were: up two missed cleavages allowed, cysteine carbamidomethylation as a fixed modification and methionine oxidation as a variable modification. The peptide tolerance was 20 ppm. Protein scores greater than 56 were considered to be significant (p < 0.05).
Following immunoprecipitation, bead-conjugated immune complexes were denaturated for 10 min at 100 °C with denaturating buffer, and digested with 375 units of PNGase F (both from New England Biolabs, Ipswich, MA, USA) overnight at 37 °C. The immunoprecipitated deglycosylated proteins were subjected to 8% SDS-PAGE together with the corresponding control sample (parallel reaction without PNGase F) and subsequent WB was performed as described in Section 4.2. All experiments were performed in duplicate.
Following trypsinization of confluent NCI-N87 cultures, 5 × 105 cells were seeded in 6-well plates and allowed to attach for 24 h. Cells were then incubated for another 24 h with 0.01, 0.1 and 1 µg/mL of tunicamycin (Sigma-Aldrich) and 0.2 and 2 µg/mL of CA 19.9 MAb (Santa Cruz Biotechnology), in RPMI 1640 GlutaMAX supplemented with 10% heat-inactivated fetal bovine serum and simple RPMI 1640 GlutaMAX, respectively. DMSO (0.05%) and 2 µg/mL of mouse IgG1 isotype control (R&D Systems) were used as controls for tunicamycin and CA 19.9 treatment, respectively. Regarding CA 19.9 treatment, an additional control of complete growth medium was included. Finally, total protein extracts were collected for each experimental condition and WB was performed as described in Section 4.2. All experiments were performed in duplicate.
The gene expression profile analysis of the cell lines NCI-N87, AGS, MKN45 and MKN74 was performed using the Barretina CellLine dataset . In this dataset, the gene expression values of 913 human cancer cell lines were analyzed by Affymetrix GeneChip Human Genome U133 Plus 2.0 Array. After extracting the gene expression data, we normalized the log2 median centered intensity levels (i) of each gene probe among all 913 human cancer cell lines of the dataset into values ranging from 0 to 100 by using the formula:
The median for each normalized gene probe was determined among the 35 human gastric cancer cell lines. We defined the normal expression range for each gene transcript as the median ±10.
When more than one probe for a gene of interest was present, we chose the best probe according to following criteria: (1) probe accession _at over _a_at over _x_at over _s_at; (2) highest probe grading as annotated in www.affymetrix.com; and (3) highest sequence identity between probe and target sequence. If more than one probe remained after applying these criteria, the average among the remaining probes was used.
The association analysis was performed in GraphPad Prism 7 using the Ooi Gastric dataset . The Ooi Gastric dataset comprises expression data of 200 malignant primary gastric tumors obtained by Affymetrix GeneChip Human Genome U133 Plus 2.0 Array. Both datasets, Barretina and Ooi, used the same criteria to select the probes for ERBB2 and FUT3 as described above. The ERBB2 and FUT3 data were not normally distributed (determined by Shapiro–Wilk normality test) and association was therefore analyzed using Spearman’s correlation analysis.
This work was funded by the project NORTE-01-0145-FEDER-000029, supported by Norte Portugal Regional Programme (NORTE 2020), under the PORTUGAL 2020 Partnership Agreement, through the European Regional Development Fund (ERDF). This work was financed by FEDER—Fundo Europeu de Desenvolvimento Regional funds through COMPETE 2020—Operational Programme for Competitiveness and Internationalisation (POCI), Portugal 2020, and by Portuguese funds through FCT (Fundação para a Ciência e a Tecnologia)/Ministério da Ciência, Tecnologia e Inovação in the framework of the project “Institute for Research and Inovation in Health Sciences” (POCI-01-0145-FEDER-007274), the project POCI-01-0145-FEDER-016585 (PTDC/BBB EBI/0567/2014), and EU 7th framework programme ITN 316929. Mass spectrometry was performed at the Proteomics i3S Scientific Platform. Microscopy images were obtained at the Advance Light Microscopy i3S Scientific Platform. Henrique O. Duarte was supported by FCT through the FCT PhD Programmes and by Programa Operacional Potencial Humano (POPH), specifically by the BiotechHealth Programme (Doctoral Programme on Cellular and Molecular Biotechnology Applied to Health Sciences), with the reference PD/0016/2012 funded by FCT. Henrique O. Duarte received an individual grant (PD/BI/113948/2015) and a PhD studentship (PD/BD/128407/2017). Meritxell Balmaña has received a funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 748880.
|BSA||Bovine serum albumin|
|CA 19.9||Cancer antigen 19.9|
|EGFR||Epidermal growth factor receptor|
|ErbB2||Human epidermal growth factor receptor 2|
|PBS||Phosphate buffered saline|
|RT-qPCR||Real-time quantitative PCR|
|RTKs||Receptor tyrosine kinases|
|SLea||Sialyl Lewis a|
|SLex||Sialyl Lewis x|
|TBS||Tris buffered saline|
Celso A. Reis and Joana Gomes were responsible for conception and design of the study; Henrique O. Duarte, Meritxell Balmaña, Stefan Mereiter, Joana Gomes and Celso A. Reis designed the experiments; Henrique O. Duarte, Meritxell Balmaña and Joana Gomes performed all of the in vitro experiments; Stefan Mereiter performed the in silico and statistical analysis; and Hugo Osório performed the mass spectrometry analysis. All authors were involved in the preparation and critical revision of the manuscript prior to its submission.
The authors declare no conflict of interest.