Mechanism of IFN-gamma-induced endocytosis of tight junction proteins: myosin II-dependent vacuolarization of the apical plasma membrane

Abstract

Disruption of epithelial barrier by proinflammatory cytokines such as IFN-gamma represents a major pathophysiological consequence of intestinal inflammation. We have previously shown that IFN-gamma increases paracellular permeability in model T84 epithelial cells by inducing endocytosis of tight junction (TJ) proteins occludin, JAM-A, and claudin-1. The present study was designed to dissect mechanisms of IFN-gamma-induced endocytosis of epithelial TJ proteins. IFN-gamma treatment of T84 cells resulted in internalization of TJ proteins into large actin-coated vacuoles that originated from the apical plasma membrane and resembled the vacuolar apical compartment (VAC) previously observed in epithelial cells that lose cell polarity. The IFN-gamma dependent formation of VACs required ATPase activity of a myosin II motor but was not dependent on rapid turnover of F-actin. In addition, activated myosin II was observed to colocalize with VACs after IFN-gamma exposure. Pharmacological analyses revealed that formation of VACs and endocytosis of TJ proteins was mediated by Rho-associated kinase (ROCK) but not myosin light chain kinase (MLCK). Furthermore, IFN-gamma treatment resulted in activation of Rho GTPase and induced expressional up-regulation of ROCK. These results, for the first time, suggest that IFN-gamma induces endocytosis of epithelial TJ proteins via RhoA/ROCK-mediated, myosin II-dependent formation of VACs.

Figure 1.

IFN-γ induces internalization of occludin into F-actin-coated vacuoles in intestinal epithelial cells. T84 cells were incubated with IFN-γ for 48 h and then labeled for F-actin and either occludin or JAM-A. Reconstructed confocal images in the x-z and x-y plane revealed internalization of occludin (A; arrowheads), as well as JAM-A (B; arrowheads), into subapical F-actin-coated vacuoles (arrows). Scale bar, 10 μm.

Figure 2.

Time course of IFN-γ-induced internalization of occludin and development of subapical F-actin-coated vacuoles. Confluent T84 monolayers were incubated with 100 U/ml IFN-γ for 12-48 h. Internalization of occludin (arrowheads) into F-actin-coated vacuoles (arrows) was determined by fluorescence labeling and confocal microscopy. After 24 h of IFN-γ incubation, organization of occludin and F-actin was analogous to that observed in polarized control epithelial cells. The first evidence of occludin internalization into F-actin-coated vacuoles was observed after 38 h of IFN-γ incubation. This effect was augmented at 48 h of IFN-γ incubation. Scale bar, 10 μm.

Figure 3.

IFN-γ induced apical vacuoles are derived from the apical plasma membrane. Epithelial monolayers were incubated with 100 U/ml IFN-γ for 48 h. followed by fixation and double labeling for F-actin and either villin or syntaxin 3 and syntaxin 4. (A) In IFN-γ-treated T84 cells, reconstructed images in the x-z plane reveal colocalization of villin with apical F-actin-coated vacuoles (arrows) and accumulation of syntaxin 3 but not syntaxin 4 inside the vacuoles (arrows). (B) T84 cells were preincubated for 36 h with IFN-γ followed by selective biotinylation of either apical or basolateral plasma membranes. Cells were subsequent incubated in IFN-γ-containing medium for an additional 4 h. Note the presence of biotin label in the VACs (arrows) only if biotin was applied apically. (C) T84 cells were preincubated for 36 h with IFN-γ followed by either apical or basolateral application of 0.1% (wt/vol) tannic acid for additional 4 h. Note that selective inhibition of endocytosis from the apical plasma membrane prevents formation of F-actin-coated vacuoles (arrows), whereas inhibition of basolateral endocytosis is ineffective. Scale bar, 10 μm.

Figure 4.

Inhibition of F-actin depolymerization does not affect formation of VACs and endocytosis of TJ proteins. Confluent T84 monolayers were preincubated for 37 h with IFN-γ followed by incubation with either F-actin stabilizing drug jasplakinolide (1 μM) or vehicle for additional 1 h. Note that F-actin stabilization failed to inhibit either formation of VACs (arrows) or internalization of occludin (arrowheads). Scale bar, 10 μm.

Figure 5.

Inhibition of de novo actin polymerization does not disassemble VACs in IFN-γ-treated cells. Confluent T84 monolayers were preincubated for 40 h with IFN-γ followed by 15-min incubation with either G-actin sequestering drug latrunculin B (5 μM) or vehicle. Note that inhibition of de novo actin polymerization by sequestration of G-actin does not cause disassembly of VACs in IFN-γ-treated cells (arrows). Scale bar, 10 μm.

Figure 6.

Actin-polymerizing proteins are not enriched in VACs. T84 cells were treated with IFN-γ for 48 h, fixed, and double-labeled with F-actin and either vasodilator-stimulated phosphoprotein (VASP) or actin-related protein (Arp) 3. Note that neither VASP nor Arp 3 is enriched within VACs. Scale bar, 10 μm.

Figure 7.

Myosin II activity mediates formation of VACs and endocytosis of TJ proteins in IFN-γ-treated cells.T84 cells were preincubated with IFN-γ for 36 h followed by additional 4-h incubation with either the selective nonmuscle myosin II inhibitor blebbistatin (50 μM) or vehicle. Localization of occludin and F-actin was determined by immunofluorescence labeling/confocal microscopy. Note that inhibition of myosin II completely blocks formation of VACs and internalization of occludin in IFN-γ-treated cells. Scale bar, 10 μm.

Figure 8.

In IFN-γ-treated cells myosin II is hyperphosphorylated and is enriched at VACs. T84 cells were incubated for 48 h with IFN-γ, fixed, and double-labeled for F-actin and either MNMM IIA (A) or phosphorylated MLC (C). Note significant accumulation of both MNMMIIA and p-MLC with F-actin in VACs (arrows). Scale bar, 10 μm. (B) Confluent T84 monolayers were incubated with medium only or IFN-γ for 40 h and lysed, and the amount of total, mono- and diphosphorylated MLC was determined by Western blotting. Note that IFN-γ treatment causes significant increase in the level of mono- and diphosphorylated MLC. Representative Western blots and results of densitometric quantification are shown. Data are presented as mean ± SE (n = 3); ^* p < 0.05 compared with the nontreated control.

Figure 9.

Formation of VACs and endocytosis of TJ proteins depend on activity of Rho-associated kinase but not myosin light chain kinase. T84 cells were preincubated with IFN-γ for 36 h followed by additional 4-h incubation with either selective Rho-kinase inhibitor Y-27632 (20 μM), a specific membrane permeant myosin light chain kinase inhibitor PIK (250 μM), or vehicle. Organization of F-actin and distribution of occludin was determined by immunofluorescence labeling and confocal microscopy. Note that inhibition of Rho-associated kinase completely blocks formation of VACs and internalization of occludin, whereas inhibition of myosin light chain kinase is ineffective. Scale bar, 10 μm.

Figure 10.

IFN-γ up-regulates expression of Rho-associated kinase, activates Rho and does not affect phosphorylation of myosin phosphatase. (A) T84 monolayers were incubated without (controls) or with IFN-γ for 38 h, and protein expression of Rho-associated kinase (ROCK) and myosin light chain kinase (MLCK) was determined by Western blotting. To ensure equal protein loading, membranes were stripped and reprobed with anti-actin antibody. IFN-γ significantly upregulated expression of ROCK, whereas MLCK level was unaffected. Representative Western blot and results of densitometric quantification are shown. Data are presented as mean ± SE (n = 3); ^* p < 0.05 compared with the nontreated control. (B) Rho activation status in control and IFN-γ-treated (38 h) T84 cells was examined by Rhotekin pull-down assay as described in Materials and Methods. IFN-γ significantly increased amount of active RhoA without affecting a total level of this protein. Data are presented as mean ± SE (n = 3); ^* p < 0.05 compared with the nontreated control. (C) T84 cells were incubated with or without IFN-γ for 38 h, and amounts of total and phosphorylated myosin phosphatase target subunit (MYPT) was determined by Western blotting. IFN-γ did not influence expression of either total or phosphorylated MYPT.