Fig 2: Effects of HCELL/CD44 engagement on immunomodulatory properties of human MSCs.Human MSCs derived from adipose tissue (hAdMSCs) or bone marrow (hBMMSCs) were fucosylated (“Fuc”: FuchAdMSCs or FuchBMMSCs) or buffer-treated (unmodified (“U”): UhAdMSCs or UhBMMSCs) and cultured in presence of different concentrations of E-selectin (mE-Ig) or hyaluronic acid (HA) for 3 days at 37 °C. Thereafter, culture supernatants were harvested and analyzed by ELISA for a levels of anti-inflammatory molecules interleukin-10 (IL-10) and TGFβ, or b IDO and NO metabolites (e.g., NO2−/NO3−). Cells cultured in the absence of HA or in presence of control IgG served as negative controls. Also, as controls to assess specificity of E-selectin binding, FuchAdMSCs or FuchBMMSCs were treated with sialidase (“sialFuchAdMSCs” or “sialFuchBMMSCs”) to cleave terminal sialic acid from sLex (thereby abrogating binding to E-selectin). As shown in panels, levels of analyzed immunomodulatory molecules significantly increased following hMSC co-incubation with either E-selectin or HA (**p < 0.01 or ***p < 0.001); levels of immunomodulatory molecules did not rise in sialFuchMSCs co-incubated with E-selectin (compared to FuchMSCs, ΔΔp < 0.01 or ΔΔΔp < 0.001, respectively). Overarching bar lines reflect comparisons of immunomodulatory molecule levels between supernatants of hAdMSCs and of hBMMSCs following co-incubation with HA or E-selectin: notably, supernatant levels of all analyzed immunomodulatory molecules were significantly higher (differences of #p < 0.05, ##p < 0.01, or ###p < 0.001, as shown) in cultures of adipose-derived hMSCs compared to those of bone marrow-derived hMSCs, most conspicuously for IL-10 (###p < 0.001). Data are presented as the mean ± SD of n = 3 separate experiments and analyzed by one-way ANOVA with Tukey’s multiple-comparisons test.

Fig 3: Bioinformatics analysis of the 22 glycoproteins expressing SLeA and showing affinity to E-selectin that were present in all cell lines towards biomarker discovery. (A) Identification of glycoproteins showing elevated gene expression. The 22 glycoproteins set was comprehensively matched against existing transcriptomics data in the Oncomine database. Only five have previously been reported to be overexpressed in GC, occurring throughout intestinal, mixed and diffuse type lesions as highlighted in the right. Notably, KRT5 and HIST1H1D were significantly overexpressed in all types, PSMD2 was mainly associated with the intestinal and mixed phenotype, ANXA2 was linked to intestinal and diffuse type but not mixed lesions, whereas SPTAN1 was overexpressed solely in mixed type lesions. (B) Comparison between transcripts and protein levels for the identified glycoproteins. Transcriptomics findings (RNAseq; Oncomine) were matched against protein expression in GC (immunohistochemistry, Human Protein Atlas). KRT5 showed low/no expression at the protein level, thus not reflecting increased transcriptional activity. HIST1H1D, PSMD2, ANXA2 and SPTAN1 showed both increased gene expression and high protein levels in GC. (C) Target score for the identified glycoproteins in GC. Briefly, higher target scores are given to glycoproteins highly expressed at the cell membrane in cancer cells and showing low levels of expression in healthy tissues. Aspects related with poor prognosis significantly contribute to increase the scoring potential. Sub-cellular re-localization of proteins from intracellular compartments in healthy cells to the cancer cell membrane, an aspect that frequently occurs in cancer, is also highly score. Conversely, proteins showing a high expression in multiple healthy tissues are penalized in comparison to those showing a more cancer-specific expression pattern. According to this protocol, NCL (that was not found overexpressed in cancer tissues) ranked first as a potentially targetable biomarker due to its cancer specific nature. (D) Protein–protein networks highlighting the main biological functions played by the identified glycoproteins. The map demonstrates a significant functional correlation of identified glycoproteins with cell differentiation, cytoskeleton organization, cell–cell adhesion mediated by cadherins, as well as translation and mediation of responses to epidermal growth factor stimulus. Notably, NCL appears to play a key role in the latter two processes.

Fig 4: Expression of SLeA and affinity to E-selectin of different GC cell models. (A) SLeA expression in different cell models (AGS, MKN-74, MKN-45, KATO-III, N87 and OCUM-1). The FACS histogram in panel A highlights the levels of SLeA in AGS, MKN-74, MKN-45, KATO-III, N87 and OCUM-1 cells, ranked from the lowest to highest expression. Notably, AGS and MKN-74 were negative for the antigen, whereas N87 and OCUM-1 presented the highest expression levels. (B) Evaluation of the origin of SLeA expression based on N-deglycosylation with PnGase F. The histogram in panel B corresponds to control experiments involving PNGase F (for N-deglycosylation) and neuraminidase (NeuAse, sialidase) for cell line N87. This confirmed the specificity of the anti-SLeA antibody for sialylated structures and that the SLeA signals arises mostly from N-glycans. However, a still significant subpopulation of cells conserved SLeA antigen expression, suggesting also the presence of O-glycans and/or glycolipids carrying this alteration. (C) Summary of SLeA expression levels before and after N-deglycosylation and neuraminidase digestion for the six cell lines. Panel C summarizes SLeA evaluation for each cell line resulting from flow cytometry experiments, in agreement with the observations in panel A. Exposure to PNGase F had little effect on MKN-45, KATO-III and OCUM-1 cell lines, suggesting that the majority of the SLeA may arise from the O-glycosylation of proteins and glycolipids. In N87 the major source of SLeA expression appears to be N-glycans. Nevertheless, considerable amounts of SLeA remain, suggesting that SLeA is also carried by O-glycans. Moreover, these signals were significantly decreased when cells were treated with neuraminidase prior to analysis, reinforcing the specificity of antibody recognition. * p < 0.05; ** p < 0.01; *** p < 0.001; and **** p < 0.0001. (Student’s t-test for three independent replicates) (D) E-selectin affinity for SLeA negative gastric cancer cells lines (AGS and MKN-74) and cells expressing high amounts of the antigen (OCUM-1 and N87) by fluorescence microscopy. The panel shows that E-selectin binding (green) is dependent on SLeA expression (red). Nuclei were counterstained with DAPI (blue).

Fig 5: Specificity of the immunohistochemistry staining of colon adenocarcinoma tissue with E-Ig chimera. Images were taken with a 10× magnification, in sequences from the same tissue section of the same paraffin block of tumor tissue. Brown colour indicates positive reactivity and shows expression of E-selectin ligands (a). For control staining was performed in the absence of E-Ig (b), absence of anti-CD62E monoclonal antibody (c) and in presence of a calcium chelant - EDTA (d)