Apoptotic signaling by TNFR1 is inhibited by the α2-6 sialylation, but not α2-3 sialylation, of the TNFR1 N-glycans

Abstract

The TNF-TNFR1 signaling pathway plays a pivotal role in regulating the balance between cell survival and cell death. Upon binding to TNF, plasma membrane-localized TNFR1 initiates survival signaling, whereas TNFR1 internalization promotes caspase-mediated apoptosis. We previously reported that the α2-6 sialylation of TNFR1 by the tumor-associated sialyltransferase ST6GAL1 diverts signaling toward survival by inhibiting TNFR1 internalization. In the current investigation, we interrogated the mechanisms underlying sialylation-dependent regulation of TNFR1 and uncovered a novel role for α2-6 sialylation, but not α2-3 sialylation, in mediating apoptosis-resistance. Our studies utilized HEK293 cells with deletion of sialyltransferases that modify N-glycans with either α2-3-linked sialic acids (ST3GAL3/4/6) or α2-6-linked sialic acids (ST6GAL1/2). Additionally, ST6GAL1 was re-expressed in cells with ST6GAL1/2 deletion to restore α2-6 sialylation. Using total internal reflection fluorescence (TIRF) microscopy and BS³ cross-linking, we determined that, under basal conditions, cells expressing TNFR1 devoid of α2-6 sialylation displayed enhanced TNFR1 oligomerization, an event that poises cells for activation by TNF. Moreover, upon stimulation with TNF, greater internalization of TNFR1 was observed via time-lapse TIRF and flow cytometry, and this correlated with increased caspase-dependent apoptosis. These effects were reversed by ST6GAL1 re-expression. Conversely, eliminating α2-3 sialylation did not significantly alter TNFR1 clustering, internalization or apoptosis. We also evaluated the Fas receptor, given its structural similarity to TNFR1. As with TNFR1, α2-6 sialylation had a selective effect in protecting cells against Fas-mediated apoptosis. These results collectively suggest that ST6GAL1 may serve a unique function in shielding cancer cells from apoptotic stimuli within the tumor microenvironment.

Keywords: Fas; ST6GAL1; TNFR1; apoptosis; cancer; death receptors; glycosylation; sialic acid.

Figure 1

HEK293 cell models with selective sialyltransferase deletion and restoration.A, HEK293 cell lines were engineered with CRISPR/Cas-9-mediated deletion of sialyltransferases involved in the α2-3 or α2-6 sialylation of N-glycans. ΔST3 cells (α2-3 deficient) were generated by deletion of ST3GAL3, ST3GAL4, and ST3GAL6. ΔST6 cells (α2-6 deficient) were produced by deleting ST6GAL1 and ST6GAL2. Additionally, we created a stable, polyclonal cell line in which the α2-6 sialylation of N-glycans was restored in ΔST6 cells by the re-expression of ST6GAL1 (ΔST6-R). Wild-type (WT) cells represent the unedited HEK293 cell line. B, ST6GAL1 expression was evaluated by immunoblotting. Longer exposure of the blot was required to visualize the endogenous ST6GAL1 in WT and ΔST3 cells. The graph shown on the right depicts densitometric analyses of three independent blots for ST6GAL1. Densitometric units (DUs) for ST6GAL1 were normalized to β-actin, and then the relative DU values for the WT line were set at 1.0., with the other cell lines normalized to this value. Results are presented as mean ± S.D., and statistics were calculated using one-way ANOVA with Tukey’s multiple comparison test. (∗p < 0.05, ∗∗∗∗p < 0.0001). C, flow cytometry using SNA lectin confirmed reduced α2-6 sialylation in ΔST6 cells, as well as restoration of α2-6 sialylation in the ΔST6-R line. D and E, cell surface α2-3 sialylation was assessed by flow cytometry using the MAA lectin (D) or SiaFind α2-3 (E). F, Unsialylated N-glycans on the cell surface were evaluated by flow cytometry using ECL lectin.

Figure 2

Deletion of α2-6 sialylation, but not α2-3 sialylation, on TNFR1 enhances TNF-induced apoptosis.A, cell surface levels of TNFR1 were analyzed by flow cytometry. B, total levels of TNFR1 were evaluated by immunoblotting. Graph on right depicts relative densitometric units (DUs) for TNFR1 normalized to β-actin. The relative DU for WT cells was set at 1.0, and the other lines were normalized to this value. Data were analyzed by one-way ANOVA with Tukey’s test. C, Lectin pull-down assays were employed to evaluate the sialylation status of TNFR1. Cell lysates were precipitated with agarose-conjugated SNA, MAA, or ECL lectins, followed by immunoblotting for TNFR1, to determine the amount of α2-6 sialylated, α2-3 sialylated, or unsialylated TNFR1, respectively. D, schematic diagram of the TNF-TNFR1 signaling network. Upon stimulation by TNF, surface-localized TNFR1 recruits Complex I, which promotes survival signaling via NFκB (green). However, internalized TNFR1 induces the formation of Complex II, which drives apoptotic signaling (red). E, cells were treated with 100 ng/ml TNF for 5 h and assessed for cell death by a caspase 3/7 luminescence assay. Data were analyzed by two-way ANOVA with Tukey’s test. F, cells were treated with TNF for 5 h and immunoblotted for cleaved caspase 8 (cl. casp8 p41/43, and p18) and caspase 3 (cl. casp3). Graphs on right depict relative DUs for cleaved caspases 8 and 3 normalized to β-actin. The relative DU for TNF-treated ΔST6 cells was set at 1.0 because there was no detectable signal in either untreated or TNF-treated WT cells. Values for the other cell lines were normalized to the TNF-treated ΔST6 line. Data were analyzed by two-way ANOVA with Tukey’s test. G, immunoblotting for cl. casp8 in TNF-treated cells incubated with or without a TNFR1 function-blocking antibody (H398). Graph depicts relative DUs (cl. casp8/β-actin), with values normalized to the TNF-treated ΔST6 line. Statistics were calculated using two-way ANOVA with Tukey’s test. H, immunoblotting for cl. casp8 in TNF-treated cells incubated with or without Dyngo-4A, a dynamin inhibitor that blocks TNFR1 internalization. The graph depicts relative DUs (cl. casp8/β-actin), with values normalized to the TNF-treated ΔST6 line. Data were analyzed by two-way ANOVA with Tukey’s test. I, cells were treated with or without TNF for 10 min and then lysates were immunoblotted for phosphorylated (p-, pSer536) and total (t-) NFκB p65. Relative DUs were calculated by first normalizing p-NFκB to t-NFκB, and then the p/t NFκB ratio was normalized to β-actin. The relative DU for TNF-treated WT cells was normalized to 1.0. Data were analyzed by two-way ANOVA with Tukey’s test. J, cells were treated with or without TNF for 3 h and immunoblotted for A20. The graph depicts relative DUs (A20/β-actin), with values normalized to the TNF-treated WT line. Data were analyzed by two-way ANOVA with Tukey’s test. For all graphs in Figure 2, results are shown as mean ± S.D. from three independent experiments. (nonsignificant (ns) p > 0.05, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).

Figure 3

Evaluation of TNFR1 clustering by TIRF microscopy.A, schematic diagram illustrating the difference between epifluorescence and TIRF microscopy. Epifluorescence imaging produces an out-of-focus image with a diffused TNFR1 protein signal. TIRF illumination allows selective acquisition of a higher signal-to-noise image focused within a ∼100 nm region at the plasma membrane-coverslip interface. Scale bar represents 20 μm. B, TIRF microscopy was performed on untreated cells or cells treated with 100 ng/ml TNF for 15 min. Cells were fixed and stained with an anti-TNFR1 antibody to visualize TNFR1 clusters. Each image depicts TNFR1 clusters within a single cell. Scale bars represent 20 μm for the field of view and 2 μm for the region of interest (ROI). C, RICM was used to define the cell boundary (yellow line), and this boundary was overlaid onto the TIRF image to measure TIRF signals relative to the cell area.

Figure 4

Cells lacking α2-6 sialylated TNFR1 exhibit increased TNFR1 oligomerization in the absence of ligand. Cells treated with or without TNF for 15 min were stained with an anti-TNFR1 antibody and then imaged using TIRF microscopy to detect TNFR1-positive punctae, representing TNFR1 clusters. A, the average intensity per image (arbitrary fluorescence units per μm², AFM/μm²). B, the percentage of the basal cell membrane occupied by TNFR1 clusters. C, the average diameter (μm) of TNFR1 clusters per image. D, the number of TNFR1 clusters per μm². Graphs in panels (A–D) depict mean ± S.D. from two independent experiments with 60 cells analyzed per group with statistical analysis by two-way ANOVA with Tukey’s test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001). E, cross-linking of receptors using BS³, followed by TNFR1 immunoblotting, was conducted on untreated cells to evaluate the formation of TNFR1 oligomers in the absence of TNF. F, TNFR1 oligomers (>250 kDa) and monomers were quantified by densitometry, and data were plotted as the ratio of oligomers to monomers. Values for the WT line were set at 1.0, and the other cell lines were normalized to this value. Graph depicts mean ± S.D. from three independent experiments. Data were analyzed using a one-way ANOVA with Tukey’s test (∗p < 0.05, ∗∗p < 0.01).

Figure 5

Live cell TIRF reveals that TNFR1 lacking α2-6 sialylation is more rapidly internalized upon TNF stimulation. Time-lapse TIRF imaging of live cells transfected with TNFR1-GFP was performed to evaluate TNFR1 dynamics. Images were acquired every 2 min beginning immediately before TNF treatment (0 min) and extending for 30 min afterward. A–D, representative images of WT (A), ΔST3 (B), ΔST6 (C), and ΔST6-R (D), cells before and 10, 20, and 30 min after TNF treatment. Each individual image depicts TNFR1 clusters within a single cell. Scale bar for the field of view = 5 μm, region of interest (ROI) = 1 μm. E, the average fluorescence intensity of TIRF images was quantified and normalized to the average fluorescence intensity of the time 0/untreated cells (dashed red line) to enable a direct comparison of TNF-induced changes in intensity. F, graph depicts the fold differences in relative intensity for cells treated with TNF for 10 min as compared with untreated cells (dashed red line). Values represent mean ± S.D. from two independent experiments using 12 cells analyzed per group. Statistical analyses were conducted using one-way ANOVA with Tukey’s test (∗∗∗p < 0.001, ∗∗∗∗p < 0.0001). G, the proportion of the plasma membrane occupied with TNFR1-GFP clusters was measured. Values for TNF-treated cells were normalized to time 0/untreated cells (dashed red line). H, graph depicts the fold differences in the proportion of membrane occupied by TNFR1-GFP clusters in cells treated with TNF for 10 min as compared with untreated cells (dashed red line). Values represent mean ± S.D. from two independent experiments using 10 cells analyzed per group. Statistical analyses were performed using one-way ANOVA with Tukey’s test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

Figure 6

The loss of α2-6 sialylation on TNFR1 promotes enhanced TNF-induced TNFR1 internalization as measured by flow cytometry.A, cells were treated with Flag-tagged TNF on ice for 60 min, which allows the binding of TNF to TNFR1 but does not permit the internalization of TNF-TNFR1 complexes. The amount of TNF bound to the cell surface at the end of the 60-min binding interval was measured by flow cytometry using an antibody against the Flag tag. A representative experiment is shown. B, TNF binding to the cell surface was quantified by measuring MFI, and MFI values were then normalized to the WT cell line. The graph depicts mean ± S.D. from three independent experiments with statistical analysis by one-way ANOVA with Tukey’s test (ns: p > 0.05). C, following the 60-min TNF binding interval, cells were switched to 37 °C for 30 min to allow the internalization of TNF-TNFR1 complexes. The TNF-TNFR1 complexes remaining on the cell surface after the temperature switch were measured by staining cells with the anti-Flag antibody. Representative histograms depict the level of surface TNF-TNFR1 complexes before (4 °C) and after (37 °C) the internalization step. D, the MFI values for surface TNF-TNFR1 complexes before and after the internalization step were compared to obtain a measurement of the degree of TNFR1 internalization. Values for internalization were normalized to the WT cell line. The graph depicts mean ± S.D. from three independent experiments with statistical analysis by one-way ANOVA with Tukey’s test (∗p < 0.05, ∗∗p < 0.01).

Figure 7

The loss of α2-6 sialylation on the Fas death receptor enhances Fas-mediated apoptosis.A, schematic diagrams of the human TNFR1 and Fas structures. Potential sites for the addition of sialylated N-glycans are indicated by pink diamonds. CRD: Cysteine rich domain; TM: transmembrane helix. B, molecular models from the crystal structures of the extracellular domains are shown for: the TNF-TNFR1 complex (PDB: 7KP7, note that this structure lacks CRD4); and the FasL-Fas complex (homology model based on PDB 7KP7, 5L19, 3TJE). The individual monomers within the ligand trimers (TNF and FasL) are indicated by varying shades of grey, and the monomers within the receptor trimers (TNFR1 and Fas) are shown in shades of blue and green. Sites for N-glycosylation are indicated in pink. C, flow cytometry was performed to assess the levels of Fas on the cell surface. D, total Fas expression was evaluated by immunoblotting. The graph depicts relative DUs (Fas/β-actin), with values for the WT line normalized to 1.0. Data were analyzed by one-way ANOVA with Tukey’s test. E, cell lysates were precipitated with agarose-conjugated SNA, MAA, or ECL lectin and immunoblotted for Fas to determine the amount of α2-6, α2-3 sialylated, or unsialylated Fas, respectively. F, cells were treated for 5 h in the presence or absence of the Fas agonistic antibody, CH11. Fas-mediated apoptosis was assessed by a caspase 3/7 luminescence assay. Data were analyzed using two-way ANOVA with Tukey’s test. G, results from the caspase 3/7 activity assays in panel F were normalized to the level of Fas expression, as measured by densitometric analyses of Fas immunoblots in panel D. Data were analyzed using one-way ANOVA with Tukey’s test. H, lysates from cells treated with or without the agonistic CH11 antibody were immunoblotted for cl. casp8 (p41/43, and p18) and cl. casp3. Graphs depict relative DUs for cl. casp8 or cl. casp3 normalized to β-actin. Relative DU for the CH11-treated ΔST6 line was normalized to 1.0 because there was no detectable signal in either untreated or CH11-treated WT cells. Data were analyzed using two-way ANOVA with Tukey’s test. For all graphs in Figure 7, data are presented as mean ± S.D. from three independent experiments. (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001).