Regulation of claudin/zonula occludens-1 complexes by hetero-claudin interactions

Abstract

Claudins are tetraspan transmembrane tight-junction proteins that regulate epithelial barriers. In the distal airspaces of the lung, alveolar epithelial tight junctions are crucial to regulate airspace fluid. Chronic alcohol abuse weakens alveolar tight junctions, priming the lung for acute respiratory distress syndrome, a frequently lethal condition caused by airspace flooding. Here we demonstrate that in response to alcohol, increased claudin-5 paradoxically accompanies an increase in paracellular leak and rearrangement of alveolar tight junctions. Claudin-5 is necessary and sufficient to diminish alveolar epithelial barrier function by impairing the ability of claudin-18 to interact with a scaffold protein, zonula occludens 1 (ZO-1), demonstrating that one claudin affects the ability of another claudin to interact with the tight-junction scaffold. Critically, a claudin-5 peptide mimetic reverses the deleterious effects of alcohol on alveolar barrier function. Thus, claudin controlled claudin-scaffold protein interactions are a novel target to regulate tight-junction permeability.

Figure 1. Alcohol dependent upregulation of claudin-5 is necessary and sufficient to impair alveolar barrier function.

AECs from alcohol fed rats and controls were cultured on Transwell permeable supports and then transepithelial resistance (TER) (a) and dye flux with calcein (b) and Texas Red Dextran (c) were measured. Alcohol-exposed AECs showed a significantly lower TER (n=6, *P<0.001, unpaired two-tailed t-test) as well as significantly higher calcein (n=3, *P<0.001, two way ANOVA with Bonferroni multiple comparisons test) and Texas Red Dextran permeability (n=3, *P<0.001, two way ANOVA) versus cells from control fed rats. (d,e) By immunoblot, alcohol exposure significantly decreased claudin-4 expression (n=3, *P=0.002, t-test) and significantly increased claudin-5 expression by AECs (n=3, ^#P=0.005, t-test). (f–h) Control AECs were transduced with adenovector YFP-claudin-5 at MOI of 2.5 or 5 or with EGFP adenovector at MOI of 5 as a control. The EGFP/EYFP doublet has been seen by others and has no bearing on our results since untagged claudin-5 has a comparable effect on AECs (Supplementary Fig. 5). (f,g) YFP-claudin-5 at MOI of 5 significantly increased claudin-5 expression (n=3, *P=0.022, one way ANOVA with Tukey multiple comparisons test) and (h) decreased TER (n=3, ^#P=0.0005 versus EGFP transduced control AECs; ^†P=0.028 versus EGFP transduced alcohol exposed cells, one way ANOVA). In (h), TER of alcohol exposed cells was significantly lower than comparable control cells (n=3, *P=0.036, one way ANOVA). (i–k) Claudin-5 protein expression in alcohol-exposed AECs was depleted using a lentiviral system delivering shRNA targeting claudin-5 or control scrambled shRNAs. (i,j) Claudin-5 was significantly depleted by specific shRNAs versus scrambled shRNA treated cells (n=4, *P=0.006, ^#P=0.036, one way ANOVA). (k) decreased claudin-5 expression in alcohol-exposed cells significantly increased TER as compared with cells transduced with scrambled shRNAs (n=4, ^#P<0.001, ^†P<0.001, one way ANOVA). TER of cells from alcohol exposed cells treated with shRNA was significantly lower than comparable control cells (n=4, *P<0.001, one way ANOVA). All quantitative data represents average ±s.e.m.

Figure 2. Increased claudin-5 expression enhances the formation of tight junction spikes.

AECs isolated from alcohol or control fed rats were cultured for 5–7 days on transwell permeable supports and immunolabeled for claudin-18. (a) Cells from alcohol fed rats showed enhancement of tight junction spikes, that are claudin-18 projections perpendicular to the cell-cell interface (d; arrowhead). Square regions in the top panels correspond to magnified images in below (Bar, 10 μm). (b) Control AECs transduced with YFP-claudin-5 increased the appearance of tight junction spikes as determined by labeling for cldn-18 or cldn-5 (Bar, 10 μm). (c) Alcohol-exposed AECs transduced with claudin-5 shRNA had a decrease in tight junction spikes (Bar, 10 μm). (d–f) Quantification of the % of cells containing 3 or more tight junction spikes oriented towards the nucleus demonstrated that alcohol exposed and YFP-claudin-5 transduced AECs had significantly more spikes than comparable controls. (d) Control versus alcohol: n=11 fields, *P=0.035, unpaired two-tailed t-test. (e) EGFP versus YFP-claudin-5: n=11 fields, *P<0.001, unpaired two-tailed t-test. (f) Alcohol exposed AECs transduced with claudin-5 shRNA1 had significantly fewer spikes than cells treated with control shRNA (n=5 fields, *P=0.011, one way ANOVA with Tukey multiple comparisons test). Cells treated with shRNA2 showed a trend towards decreased spikes (n=5, ^#P=0.18, one way ANOVA with Tukey multiple comparisons test) (g). Control AECs were partially transfected with YFP-claudin-18 then fixed and immunolabeled for ZO-1. YFP-claudin-18 expressing cells adjacent to untransfected cells showed uptake of YFP-claudin-18 in intracellular vesicles (arrows, Bar, 10 μm). (h) Magnified images corresponding to the square region in (g) showing spike associated claudin-18 internalized into adjacent cells. Arrowheads show areas where claudin-18 does not co-localize with ZO-1. The arrow indicates a structure where YFP-claudin-18 and ZO-1 co-localize. All quantitative data represents average ±s.e.m.

Figure 3. Claudin-containing vesicles bud from and fuse with tight junction spikes.

(a–d) Live cell imaging was performed with alcohol exposed AECs transduced with Adenovirus encoding either YFP-claudin-5 (a,b) or YFP-claudin-18 (c,d). Shown are still images from videos acquired over a 20 min time period with a frame capture of 30 s intervals. Labeled vesicles containing YFP-claudin-5 or YFP-claudin-18 were found to both fuse to (a,c) and bud from (b,d) tight junction spikes, demonstrating that these are dynamic structures. The top left panel in each series is a lower magnification image, the square region represents the time series, which is time stamped in seconds. Bar, 5 μm. (e) Cells from alcohol fed or control fed rats were cultured for 7 days and then treated with either DMSO vehicle control or the dynamin inhibitor Dynasore at varying concentrations for 4 h at 37 °C in serum free media. The cells were then fixed and immunolabeled for claudin-18. Representative images show vehicle-treated and 160 μM Dynasore treated cells. Arrowheads show tight junction spikes. Bar, 10 μm. (f,g) Quantification of the % cells containing 3 or more tight junction spikes oriented towards the nucleus demonstrated that 160 μM Dynasore significantly decreased the number of cells from alcohol fed rats containing spikes (n=8-9 fields, *P=0.002, one way ANOVA with Tukey multiple comparisons test) (f). Dynasore did not have a significant effect on spike formation by control cells (g). All quantitative data represents average±s.e.m.

Figure 4. Claudin-5 induced by alcohol decreases ZO-1:claudin-18 co-localization as determined by super-resolution microscopy.

(a–c) AECs isolated from alcohol (alc) or control (con) fed rats were cultured, immunolabeled and imaged by STORM. Cells were double-labeled for claudin-18 and ZO-1 (a), claudin-5 and claudin-18 (b) or claudin-5 and ZO-1 (c). Images were analyzed for protein co-localization (d–f). Alcohol exposed AECs showed a reduction in the co-localization between claudin-18 and ZO-1 and an increase in co-localization between claudin-18 and claudin-5. Co-localization between claudin-5 and ZO-1 was comparable for both control and alcohol exposed cells. Arrowheads denote areas of co-localization. Bar, 1 μm. (d–f) Quantification of co-localization using STORM images demonstrated a significant change. In alcohol-exposed AECs there was a significant decrease in ZO-1:claudin-18 (n=4 fields (control), n=3 fields (alcohol exposed AECs),*P=0.014, unpaired two-tailed t-test) (d) which correlated with a significant increase in claudin-18:claudin-5 co-localization (n=3 fields, *P=0.039, unpaired two-tailed t-test) (e). ZO-1:claudin-5 co-localization was unchanged (n=4 fields, unpaired two-tailed t-test) (f). Data in (d–f) represent average±s.e.m.

Figure 5. Claudin-5 induced by alcohol decreases ZO-1:claudin-18 co-localization as determined by proximity ligation assay.

(a,b) AECs isolated from alcohol (alc) or control (con) fed rats were cultured, immunolabeled and analyzed using the proximity ligation assay (PLA). Cells were PLA-labeled for claudin-18 and ZO-1, claudin-5 and claudin-18 or claudin-5 and ZO-1. Images in (b) are magnifications of regions in (a) as denoted by the squares. Bar, 20 μm. Negative controls are shown in Supplementary Fig. 10. Alcohol exposed AECs showed a reduction in the co-localization between claudin-18 and ZO-1 and an increase in co-localization between claudin-18 and claudin-5. Co-localization between claudin-5 and ZO-1 was comparable for both control and alcohol exposed cells. (c–e) Quantification of co-localization using PLA demonstrated a significant change. In alcohol-exposed AECs there was a significant decrease in ZO-1:claudin-18 (n=6 fields,*P=0.018, unpaired two-tailed t-test) (c) which correlated with a significant increase in claudin-18:claudin-5 co-localization (n=10 fields, *P=0.026, unpaired two-tailed t-test) (d). ZO-1:claudin-5 co-localization was unchanged (n=6 fields, unpaired two-tailed t-test) (e). Data in (c–e) represent average±s.e.m.

Figure 6. Claudin-5 expression is sufficient to decrease ZO-1:claudin-18 co-localization and increase claudin-18 solubilization.

(a) Control or YFP-claudin-5 transduced AECs were cultured for 6 days, immunolabeled and then imaged by STORM for claudin-18 and ZO-1. Increased claudin-5 expression decreased the extent of ZO-1:claudin-18 co-localization (see text). Arrowheads show sites of co-localization. Bar, 1 μm. (b,c) Biochemical analysis of protein insolubility was assessed by a Triton X-100 solubilization assay comparing control AECs to YFP-claudin-5 transduced cells. At 6 days in culture, AECs were harvested and extracted using 0.1% Triton X-100, an aliquot of total protein (T) was set aside and the remainder was centrifuged to separate Triton X-100 soluble (S) and insoluble (I) fractions that were measured by immunoblot for claudin-18, claudin-5 and ZO-1. Quantification of the soluble fraction revealed that YFP-claudin-5 expression significantly increased claudin-18 solubility from 35.2±1.8 to 42.1±0.6 (n=3, *P=0.003, unpaired two-tailed t-test) while claudin-5 and ZO-1 solubility did not significantly change. All quantitative data represents average±s.e.m.

Figure 7. A claudin-5 extracellular domain mimetic increases barrier function of alcohol-exposed AECs.

(a–f) AECs isolated from control (a,c,e) or alcohol fed rats (b, d, f) were cultured on Transwell permeable supports for 5 days and then either untreated (un), or incubated with 10 μM control peptide (con; Ac-LYQY-NH₂) or a claudin-5 extracellular domain mimetic peptide (C5; Ac-EFYDP-NH₂) for 16 h. The cells were examined for barrier function by transepithelial resistance (TER) (a,b) and paracellular flux of calcein (c,d) and 10 kDa Texas Red dextran (e,f). The C5 peptide had little effect on barrier function of control AECs (a,c,e) however, it significantly increased TER (*P=0.014 versus untreated; ^#P=0.042 versus control; n=6, one way ANOVA with Tukey multiple comparisons test) (b), and decreased paracellular flux of calcein (*P=0.007 versus untreated; ^#P=0.054 versus control; n=3, one way ANOVA with Tukey multiple comparisons test) (d), and Texas Red Dextran (*P=0.009 versus untreated; ^#P=0.040 versus control; n=3, one way ANOVA with Tukey multiple comparisons test) (f). (g) AECs as treated above were processed and examined by immunofluorescence for claudin-18 localization. Bar, 20 μm. Cells from alcohol fed rats showed a decrease in tight junction spikes, that was significantly less than that of untreated controls and alcoholic AECs that were either untreated or treated with a control peptide (*P<0.001 versus untreated; ^#P<0.001 versus control peptide; ^†P=0.041 versus untreated control AECs, n=9–11 fields from two independent experiments, one way ANOVA with Tukey multiple comparisons test) (h). Claudin-5 immunofluorescence is shown in Supplementary Fig. 11. (i–l) AECs as treated above were processed and examined by immunoblot for claudin-5, claudin-18 and ZO-1. Cells from alcohol-fed rats that were treated with the C5 peptide showed a significant and specific decrease in claudin-5 (*P=0.042 versus untreated; ^#P=0.016 versus control; n=9, one way ANOVA with Tukey multiple comparisons test) (l). All quantitative data represents average±s.e.m.