Zonula occludens-1 and -2 regulate apical cell structure and the zonula adherens cytoskeleton in polarized epithelia

Abstract

The structure and function of both adherens (AJ) and tight (TJ) junctions are dependent on the cortical actin cytoskeleton. The zonula occludens (ZO)-1 and -2 proteins have context-dependent interactions with both junction types and bind directly to F-actin and other cytoskeletal proteins, suggesting ZO-1 and -2 might regulate cytoskeletal activity at cell junctions. To address this hypothesis, we generated stable Madin-Darby canine kidney cell lines depleted of both ZO-1 and -2. Both paracellular permeability and the localization of TJ proteins are disrupted in ZO-1/-2-depleted cells. In addition, immunocytochemistry and electron microscopy revealed a significant expansion of the perijunctional actomyosin ring associated with the AJ. These structural changes are accompanied by a recruitment of 1-phosphomyosin light chain and Rho kinase 1, contraction of the actomyosin ring, and expansion of the apical domain. Despite these changes in the apical cytoskeleton, there are no detectable changes in cell polarity, localization of AJ proteins, or the organization of the basal and lateral actin cytoskeleton. We conclude that ZO proteins are required not only for TJ assembly but also for regulating the organization and functional activity of the apical cytoskeleton, particularly the perijunctional actomyosin ring, and we speculate that these activities are relevant both to cellular organization and epithelial morphogenesis.

FIGURE 1:

ZO-1 and -2 are effectively depleted in stable MDCK II Tet-Off cell lines. (A) Western blot of ZO-1, -2, and -3 polypeptides in control, dKD (lines 1–3), and in dKD cells expressing a Tet-inducible full-length ZO-1 rescue transgene (ZO1R). ZO-1 expression in dKD lines is ∼2–3% of that observed in control cells, while ZO-2 expression is almost undetectable. Note that ZO-3, while not directly targeted in dKD cells, is also reduced by up to 50%. In ZO1R cell lines, induction of ZO1R (I) restores ZO-1 and ZO-3 expression to levels equivalent to those in control cells, while ZO-2 expression remains suppressed. U, uninduced; I, induced. (B) Immunocytochemistry of ZO-1, -2, and -3 in the control and dKD-3 cell line. There is a dramatic reduction in the staining of all three peptides at the AJC, and the outline of the AJC is much more trapezoidal in dKD cells relative to control cells. All images are 1-μm-thick, maximum-density projections of the AJC. Scale bar: 10 μm.

FIGURE 2:

The localization of some, but not all, TJ proteins to the AJC is attenuated in ZO dKD cell lines. (A–C) A 1-μm-thick, maximum-density projection of the apical domain, en face, acquired at the AJC (see Materials and Methods). (A) Occludin and cingulin localization at the AJC is reduced, whereas JAM-A and tricellulin appear relatively normal in dKD cells relative to control cells. Claudin-4 staining (green) is included with tricellulin staining (red) to highlight tricellular junctions. (B) Claudin-1 and -2 staining at the AJC is reduced, whereas claudin-3 and -4 appear relatively normal in dKD cells. Note again the dramatic change in cell shape at the AJC in dKD cells. Scale bar: 10 μm.

FIGURE 3:

The paracellular barrier in dKD cells is selectively perturbed. (A) The TER is maintained in dKD cells, and is unaffected by expression of the full-length ZO-1 transgene. (B) The dilution potential of the dKD cell lines is significantly reduced, but is reversed by expression of the full-length ZO-1 rescue transgene. (C) Size selectivity of the paracellular barrier is measured by the paracellular movement of a graded series of PEG oligomers from 2.6–7.0 Å radius. The size selectivity of these uncharged solutes is lost in dKD cell. (A–C) n = 3 clones. (D) The flux of 3-kDa FITC-dextran is dramatically increased in dKD and uninduced (U) ZO1R epithelia relative to control or ZO-1 or -2 single-KD cell lines. The increased flux is partially reversed by expression (I) of ZO1R.

FIGURE 4:

There is a dramatic reorganization of the cytoskeleton within the AJC. (A) Control and dKD cells were fixed and stained with TRITC-phalloidin (F-actin) and antibodies against myosin IIB, and the acquired images are presented as a 1-μm-thick, maximum-density projection of the AJC. In dKD cells, the cell–cell contacts at the AJC are more linear, cell shape is more trapezoidal, and F-actin and myosin IIB are reorganized into a thick array of fibrils with a 400–600 Å repeat of myosin IIB. There are also distinct foci of F-actin and myosin IIB within this apical section of dKD cells that are not apparent in control cells. (B) The organization of F-actin and myosin IIB in a 1-μm-thick, maximum-density projection of the basal (substrate-associated) domain appears relatively normal. Scale bar: 10 μm.

FIGURE 5:

F-actin forms large parallel arrays adjacent to the plasma membrane at the most apical aspect of the lateral domain. (A and B) TEM cross-sections of control (A) and dKD (B) MDCK cells at the apical domain. Note the increased electron density at the lateral membrane (brackets), almost immediately below the microvilli, in dKD cells relative to control cells. Note also that membrane contacts, or “kisses,” characteristic of TJs are still observed in dKD cells (arrows). (C and D) En face section of control (C) and dKD (D) epithelial monolayers. There are large parallel filaments in dKD (D) cells that are never observed in control cells. Scale bars: (A and B) 100 nm; (C and D) 200 nm.

FIGURE 6:

The morphology of the apical domain is distorted in ZO dKD cells. (A) TEM of a cross-section of epithelia formed by control and dKD cells. dKD cells are taller when measured from substrate to the AJC, irregularly packed, and often have a distortion of the apical plasma membrane. Scale bar: 5 μm. (B) SEM of control and dKD cells. The distortion of the apical plasma membrane is immediately apparent, and there are long membrane extensions that are distinct from the microspikes normally apparent on the apical domain of control cells. Scale bars: 5 μm (low mag.); 1 μm (high mag.).

FIGURE 7:

ZO-depleted cells have normal cell polarity and form cysts with a single lumen in three-dimensional collagen culture. (A) Confocal Z-section reconstruction of control and dKD cells fixed and stained with antibodies against apical (gp135, ezrin), lateral (α-, β-catenin), and cell polarity (PAR3) markers. The polarized distribution of all markers is maintained, although the apical distortion and altered cell packing are readily apparent. (B) Confocal section through cysts cultured from single cells plated in a collagen suspension. Cysts were fixed and stained with rhodamine-phalloidin and antibodies against the apical marker gp135 and ZO-1. Indentations (arrowheads) and distensions (arrow) of the apical plasma membrane are apparent in dKD cells. Scale bar: 20 μm.

FIGURE 8:

The localization of AJ proteins is unaltered in ZO-depleted cells. (B) Monolayers of control and dKD cells were fixed and stained with antibodies against the indicated AJ proteins; the images are displayed as 1-μm-thick, maximum-density projections of the AJC. Although the characteristic change in cell shape of dKD cells is apparent, the lateral distribution of the core components of the AJs appear similar to those of control cells. Scale bar: 10 μm.

FIGURE 9:

ROCK-1 and 1p-MLC are recruited to contractile actomyosin arrays in dKD cells. (A) ZO-depleted cells have a smaller apical diameter than control MDCK cells. dKD cells were transfected with GFP, diluted 1:20 into control MDCK cells, and examined 4 d after plating. The X,Y,Z views are maximum-density projections of the entire confocal volume. Arrows highlight the cell–cell contacts of the ZO-depleted cell with its neighbors. Scale bar: 10 μm. (B) Apical area of control and dKD cells calculated in ImageJ from perimeter traces. Data were compiled from seven different fields, similar to (A). Control, n = 103; Z2Z1 dKD n = 18. (C) Immunocytochemistry of ROCK-1 and 1p-MLC in control and dKD cells. Scale bar: 10 μm. (D) Western blot of 1p-MLC and total regulatory myosin light chain (RMLC) in control, dKD cells, or dKD cells uninduced (U) or induced (I) to express ZO1R. Note that the overall cellular levels of 1p-MLC are unchanged in dKD cells relative to control or ZO1R (U) cells. (E) RhoA-GTP pulldown assay; the cellular levels of RhoA-GTP in control, ZO-1 single-KD, and ZO dKD cells were analyzed using a standard pulldown assay with RBD. Western blots of total RhoA and RBD-bound fractions were stained with antibodies against RhoA.

FIGURE 10:

Blebbistatin, but not the ROCK inhibitor Y-27632, can reverse the accumulation of actin arrays at the AJC. (A) Monolayers of control and dKD cells were treated with vehicle or 30 μM of the ROCK inhibitor Y-27632 for 15 h and subsequently fixed and stained with TRITC-phalloidin (F-actin) or antibodies specific for myosin IIB. Images are presented as maximum-density projections at the AJC, as described in Materials and Methods. (B) As in (A), but monolayers were treated with vehicle or 100 μM blebbistatin. Although there are clear changes in the distribution of F-actin at the AJC, the F-actin/myosin IIB arrays characteristic of dKD cells are still present. In contrast, the actomyosin arrays are absent in the blebbistatin-treated cells. Scale bar: 10 μm.