Fig 1.
Glycomic profiling of native versus cultured, myofibroblastic RPE by lectin blot analysis reveals alterations in the glycan expression pattern upon EMT in vitro.
Equal amounts of whole cellular protein lysates (7.5 μg) derived from native (N) and cultured, myofibroblastic (M) human RPE cells of passage 3 were separated by SDS page. Blots were incubated with the biotin-coupled plant lectins as indicated followed by peroxidase-coupled streptavidin (large boxes, above). Normalization to GAPDH (small boxes, below) shows equal total protein amounts of native RPE cells (N) and passaged RPE cells (M) were loaded in both slots. Molecular weight in kDa is indicated on the left (MW). The experiment was repeated at least three times with protein lysates from different donors of different age groups and different primary RPE cell lines of passage 3–7. A representative blot is shown.
Fig 2.
Complex-type N-glycans but not O-glycans are required for Gal-3 binding to the RPE cell surface.
(A-C) Cultured human RPE cells were treated with swainsonine to block N-glycan elongation followed by incubation with Gal-3 or plant lectins for 30 minutes at 4°C as described in the Materials and Methods Section. (A) Effectiveness of swainsonine treatment is shown by decreased staining with the plant lectin PHA-L (dotted line) compared with control treated cells (thick line). (B) Reduction of complex-type N-glycans on RPE cells results in decreased Gal-3 binding (dotted line). (C) Binding of the control lectin WGA is not modified. (D-F) Treatment of cultured RPE cells with deoxymannojirimycin (DMNJ), another inhibitor of N-glycan branching, reduces binding of the plant lectin PHA-L (dotted line) (D) and Gal-3 (dotted line) (E) compared with control treated cells (thick line) whereas binding of the control lectin WGA (dotted line) is not affected (F). (G-I) RPE cells were treated with BenzylGalNAc, which inhibits elongation of O-glycans and can compete with sialyltransferases resulting in decreased O-glycan sialylation, allowing increased branching of O-glycans. Decreased O-glycan sialylation is shown by increased staining with the plant lectin PNA (dotted line) compared with untreated controls (thick line) (G). (H) Increased accessibility of branched O-glycans does not alter binding of Gal-3 to RPE cells (dotted line), showing that branched O-glycans are not required for Gal-3 binding. (I) Binding of PHA-E as a control lectin is not altered by benzylGalNAc treatment. (J-L) Treatment of RPE cells with neuraminidase increases Gal-3 binding to the surface of cultured human RPE cells. Cultured, myofibroblastic RPE cells were incubated with or without Vibrio Cholera neuraminidase for 30 minutes at 37°C. (J) Binding of MAL-2 (specific for α2,3 sialic acid residues) (thick line) and (L) SNA (specific for α2,6 sialic acid residues) (thick line) confirms the presence of sialic acid residues on glycans on the RPE cell surface. Both, MAL-2 and SNA binding, is reduced after treatment with neuraminidase (dotted line). Binding of Gal-3 (K) to cultured human RPE is increased by removal of sialic acids (dotted line). Results are representative for three independent experiments.
Fig 3.
Effect of glycan synthesis inhibitors on Gal-3 mediated inhibition of RPE cell attachment and spreading.
Cultured, myofibroblastic human RPE cells were pretreated with or without the N-glycan synthesis inhibitor DMNJ or the O-glycan synthesis inhibitor benzylGalNAc in the medium for 72 hours. Suspended RPE cells were then incubated with Gal-3 and allowed to attach on fibronectin as described in the Materials and Methods Section (A). The N-glycan synthesis inhibitor DMNJ partially reverses the Gal-3 mediated effect, whereas the O-glycan synthesis inhibitor did not influence the inhibition of RPE attachment by Gal-3. Values indicate means ±SD of four experiments performed in triplicate and are expressed as percentage of controls without galectin-3 in the medium. Co, untreated control. (B-H) For spreading assays suspended RPE cells were pretreated with glycan synthesis inhibitors as described above and then incubated for 35 min with 125 μg/mL galectin-3 (Gal-3) allowed to spread on fibronectin as described in the Materials and Methods Section. Cells were then fixed, stained with giemsa and observed by light microscopy (magnification 20×). Representative light microscopic fields are shown. Exposure of RPE cells to Gal-3 reduces RPE spreading on fibronectin (C), whereas untreated controls (B) and cells pretreated with DMNJ (D) or BenzylGalNAc (F) alone establish broad cytoplasmic halos. Pretreatment with DMNJ partially abolishes the Gal-3 mediated effect (E), whereas benzylGalNAc failed to reduce the Gal-3 mediated inhibition of RPE cell spreading (G). (H) Quantification of cell spreading shown in A-F. Results were obtained from evaluation of five separate fields by examining at least 100 cells per field. Spreading cells were defined as cells with cytoplasmic protrusions and perinuclear halo formation, and non-spreading cells as rounded cells with small extent of cytoplasmic spreading. Values indicate means±SD of three experiments performed in duplicate and are expressed as percentage of total cells present on the respective microscopic field. Statistical analysis was performed using Mann–Whitney U test and a p-value <0.05 was considered as statistically significant (*).
Fig 4.
Knockdown of Mgat5 expression in cultured human RPE cells attenuates binding of Gal-3 to RPE cells.
(A) Western blot analysis of Mgat5 expression in RPE cells transfected with siRNA containing the same nucleotides as Mgat5 siRNA in random order (sc siRNA), and the same cell line transfected with 50 pmol and 100 pmol of double stranded siRNA complementary to Mgat5 (Mgat5 siRNA), respectively. Lysates containing approximately equal amounts of protein were separated by SDS-PAGE and blotted for immunochemical detection of Mgat5 content. Experiments were repeated at least three times. MW; molecular weight. (B) Quantification of Mgat5 gene silencing. Values are normalized to expression of tubulin. (C) Fluorescence micrographs of Gal-3 binding to the RPE cell surface. Ninety-six hours after transfection cells were treated with 60 μg/mL biotinylated Gal-3. Cells were then fixed and stained with a fluorescent streptavidin conjugate. Nuclei were counterstained with DAPI. Localization of Gal-3 binding was visualized by fluorescence microscopy at a 40 fold magnification. Scale bars represent 100 μm. Untreated cells exposed to streptavidin conjugate alone served as negative controls and exhibited no fluorescence signal (data not shown). (D-E) The target sequence derived from the genomic sequence of Mgat5 was inserted into a CRISPR-Cas9 nuclease expressing lentiviral vector and ARPE19 cells were transfected using Lipfectamine 2000 Plus reagent. (D) Western blot analysis of Mgat5 expression in ARPE19 cells transduced with guide RNA leading to specific knockdown of the Mgat5 (gMgat5), or cells tranduced with a CRISPR-Cas9 lentiviral vector encoding for an none-coding filler RNA (LV), or wild-type ARPE-19 cells. Lysates containing approximately equal amounts of protein were separated by SDS-PAGE and blotted for immunochemical detection of Mgat5 content. (E) Flow cytometric analysis of Gal-3 binding in Mgat5-knockout cells. CRISPR-Cas9-mediated Mgat5 knockdown of cultured RPE cells reduces cells surface binding of Gal-3, when compared to cells transfected with a non-coding control vector alone (LV). Histograms represent the number of counted cells versus relative fluorescence intensity. Transduction experiments have been repeated three times.
Fig 5.
Upregulation of Mgat5 and β1,6GlcNAc-branched N-glycosylation in myofibroblastic RPE.
(A) Total cellular RNA from native (N) and cultured, myofibroblastic human RPE cells (M) was analysed for β1,6-N-acetylglucosaminyltransferase (Mgat5) expression levels by quantitative real-time RT-PCR as described in the Materials and Methods Section. Data are presented as relative Mgat5 mRNA expression levels as compared with native RPE cells and represent the mean ± SD of at least 3 experiments. (B) Equal amounts of whole cellular protein lysates (50 μg) derived from native (N) and cultured, myofibroblastic (M) human RPE cells of passage 3 were separated by 10% SDS PAGE and Mgat5 levels detected by Western blot. Equal loading was confirmed by GAPDH. Experiments were repeated at least three times
Fig 6.
Native RPE cells exhibit little binding of Gal-3.
(A) Flow cytometric analysis of Gal-3 binding to native and myofibroblastic RPE cells. Binding of Gal-3 (gray) to native RPE cells was only slightly above background, whereas Gal-3 binding to myofibroblastic RPE cells was evident. ConA binding was markedly above background in both cell populations. Histograms represent the number of cells versus relative fluorescence intensity. Experiments have been repeated two times. (B) Lectin histochemistry of human RPE cells in situ. Human cadaver eyes were incubated with biotinylated lectins as indicated on the left and binding was visualized by incubation with streptavidin-coupled peroxidase and Vector VIP substrate™ (purple color). In control sections with the VIP substrate alone the RPE can be easily discerned by the characteristic brownish pigment (third panel from the top). RPE cells and extracellular matrix reacted strongly with ConA (top panel), whereas PHA-L (second panel from the top) did not recognize native RPE cells and exhibited a staining pattern comparable to that of substrate alone. Biotinylated Gal-3 (purple color, fourth panel) did not bind to native RPE in situ and staining patterns resembled closely the situation in untreated negative control eyes. RPE, retinal pigment epithelium; BM, Bruch´s membrane; CH, choriocapillaris.