Toll-like receptor signalling via IRAK4 affects epithelial integrity and tightness through regulation of junctional tension

ABSTRACT Toll-like receptors (TLRs) in mammalian systems are well known for their role in innate immunity. In addition, TLRs also fulfil crucial functions outside immunity, including the dorsoventral patterning function of the original Toll receptor in Drosophila and neurogenesis in mice. Recent discoveries in flies suggested key roles for TLRs in epithelial cells in patterning of junctional cytoskeletal activity. Here, we address the function of TLRs and the downstream key signal transduction component IRAK4 in human epithelial cells. Using differentiated human Caco-2 cells as a model for the intestinal epithelium, we show that these cells exhibit baseline TLR signalling, as revealed by p-IRAK4, and that blocking IRAK4 function leads to a loss of epithelial tightness involving key changes at tight and adherens junctions, such as a loss of epithelial tension and changes in junctional actomyosin. Changes upon IRAK-4 inhibition are conserved in human bronchial epithelial cells. Knockdown of IRAK4 and certain TLRs phenocopies the inhibitor treatment. These data suggest a model whereby TLR receptors near epithelial junctions might be involved in a continuous sensing of the epithelial state to promote epithelial tightness and integrity.


Fig. S1 ,
Fig. S1, related to Figure 1.TLRs are constitutively present and localised in a polarised position in epithelial Caco-2 cells.Localisation of TLR4 (A-A'''), TLR6-YFP (B-B''') in epithelial Caco-2 cells.Z-projections of apical-most confocal sections are shown.Antibody labelling against TLR4 and overexpressed TLR6-YFP are in green, phalloidin to label cell boundaries is in magenta.A'', B'' show an apical-basal cross-section of the epithelium (nuclei stained with DAPI in blue), apical is up, individual channels for TLR4 and TLR6-YFP are shown below.A''', B''' show magnifications of cell boundary regions, the insets show the single channel for TLR4 or TLR6-YFP, respectively.C-F Constitutive TLR signaling in Caco-2 monolayers does not involve nuclear NFkB.3-week post-confluence Caco2 monolayers have little to no nuclear NFkB (C) whereas monolayers treated with 10µM Interleukin-1b, a canonical TLR pathway agonist, show strong nuclear NFkB labeling after 30 mins (D).Caco2 monolayers treated with IRAK4-inhibitor (see Fig. 2) like the DMSO control show no nuclear NFkB (E, F).Cell outlines are labelled with phalloidin revealing F-actin.

FigH
Fig. S2, related to Figure 2. IRAK4 inhibition affects tight junction integrity and barrier function.Treatment of 3-week post-confluent Caco-2 cells with 10µM of either PF06650888 (A) or AS 2444697 (B) IRAK4 inhibitor induces a dose-dependent reduction in TEER that is reversible upon inhibitor removal (recovery), indicating that the epithelial barrier at tight junctions is reenforced upon washout of the inhibitor.Shown are SEM of n=3 separate transwells per experimental condition (DMSO, AS, PF).Statistical significance for the final timepoint was determined by unpaired Student's t-test.C Transepithelial diffusion of a fluorescent dextran FD4 across mature Caco-2 monolayers.Diffusion into the basal compartment is higher following 10μM PF-06650833 pre-treatment, suggesting diminished barrier integrity.Each data point represents an independent transwell (N=3).Statistical significance determined by one way ANOVA with Dunnett's multiple comparison test (* = p<0.05).D-D' Effect of 10µM PF inhibitor-treatment on junctional components.Z-projections of confocal sections covering the apical-lateral junctional area are shown.The junctional intensity of Claudin-1 is decreased compared to control (DMSO)-treatment.Quantification is in Fig. 2H.E-G E-Cadherin at apical junctions appears to change its distribution within lateral spot adherens junctions (see schematic in G) upon 10µM IRAK4 inhibitor-treatment (PF).ZO-1 (E, F) is shown in comparison to E-Cadherin (E', F') and to mark the apical-most end of the lateral sides.H Treatment with IRAK4 inhibitor (PF) at different concentrations does not affect total IRAK4 levels.GAPDH is shown as loading control.I, J Analysis of protein levels for control (DMSO)-and 10µM PF-treated (I) or 10µM AS-treated (J) Caco-2 cell monolayers.Immunoblots of samples revealed with the indicated antibodies are shown.

FigEH
Fig. S3, related to Figure 4. IRAK4 inhibition leads to loss of epithelial tension at tight junctions.A-A'' In Caco-2 cells at 3-weeks-post confluence, NMIIA is organised into a prominent striated pattern at junctional contacts (A', arrowheads) and across the apical surface (A'').Upon PFinhibitor-treatment, NMIIA is less organised both at junctions (B') as well as within the apical surface (B'').Magnified areas are indicated by magenta boxes.Panels A and B are also shown as two-colour panels in Fig. 4 B and D. C-D''' Immunofluorescence of a18-a-Catenin compared to total a-Catenin (as quantified in Fig. 3K).Z-projections of confocal sections covering the apical-lateral junctional area are shown.E Total a-Catenin fluorescence intensity at junctions in control and PF-inhibitor treated Caco-2 cells.Values are extracted from a-Catenin/a18-a-Catenin ratios (Figure 4K) and statistical significance was determined using one-way ANOVA with Dunnett's multiple comparison test as either p<0.0001 (***) or p<0.00001 (****).F, G Cross-sections corresponding to panels A' and B' above illustrating a18-a-Catenin localisation at lateral junctions in control (DMSO) and 5µM PF-inhibitor treated Caco-2 cell monolayers.H Vinculin total protein levels do not change between control and PF-inhibitor-treated Caco-2 cells, as analysed by Western blotting.I-M In Caco-2 monolayers treated with siRNA(3) (see Figure 3) against IRAK4, both a-Catenin (I, J) as well as a18-a-Catenin (K, L) levels at junctions are reduced compared to control, with changes to a18-a-Catenin stronger than to total a-Catenin.Z-projections of confocal sections covering the apical-lateral junctional area are shown.M Quantification of a-Catenin/a18-a-Catenin ratio: n= 111 junctions were analysed for the scramble control, and n= 224 for the siRNA(3) against IRAK4.Statistical significance was determined using unpaired t-test as p<0.0001 (***).

Fig
Fig. S4, related to Figure 4. IRAK4 inhibition leads to changes at epithelial junctions one day post treatment.A-D' Changes to junctional components upon treatment of 3-week post-confluence Caco2 monolayers at 1 day post treatment, showing DMSO-treated controls (A, B, C, D) and 10 µM PF treated cells (A', B', C', D'), insets show individual junctions.E Quantification of changes shown in A-D'.For ZO-1, n=200 for the vehicle control, n=344 for 10 µM PF treatment; for E-cadherin, n=269 for the vehicle control, n=669 for 10 µM PF treatment; for phospho-a-Catenin to total a-Catenin ratio, n=244 for the vehicle control, n=266 for 10 µM PF treatment; for a18-a-Catenin to total a-Catenin ratio, n=419 for the vehicle control, n=504 for 10 µM PF treatment.Statistical significance was determined using unpaired Student's t-test as either non significant (n.s.), p<0.05(*) or p<0.0001 (****).

Fig. S5 ,
Fig. S5, related to Figure 4. IRAK4 inhibition leads to loss of epithelial integrity and changes at epithelial junctions in primary bronchial respiratory cells.A-B'' Primary bronchial epithelial cells 2 weeks post-confluence show well-established tight and adherens junctions as well as an apical actomyosin web surrounding apical cilia.A-A'' E-Cadherin (green), ZO-1 (magenta), F-actin (turquoise) and DAPI (blue) are shown, with A'' showing a cross section.B-B'' NMIIA (green), acetylated a-tubulin (turquoise) and DAPI (blue) are shown, with B'' showing a cross section.C Treatment of 2 week post-confluent primary bronchial cell monolayers with PF06650888 IRAK4 inhibitor induces a dose-dependent reduction in TEER that is reversible with 4 days of wash-out using inhibitor-free medium.Shown are SEM of n= 3 (DMSO), n=3 (PF) separate transwells.Statistical significance at day 4 and day 8 time point was determined by one-way ANOVA with Dunnett's multiple comparison test, with *<0.05, ** <0.005, ***<0.0005and n.s. as non significant.D-K Effect of PF inhibitor-treatment on junctional components.D-J' Z-projections of confocal sections covering the apical-lateral junctional area are shown for cells treated with PFinhibitor, with quantification shown in K. ZO-1 (D, D'), Claudin-3 (E, E') and E-Cadherin (F, F'), Vinculin (G, G'), NMIIA (H, H'), phospho-a-Catenin (I, I') and a18-a-Catenin (J, J') localisation at cell-cell junctions is reduced upon IRAK4 inhibition with 10µM PF. phospho-a-Catenin and a18-a-Catenin are shown and quantified in comparison to total a-Catenin (insets in I-J').Insets in D-H show individual junctions.K Quantification of fluorescence intensity changes at cell-cell junctions in control (DMSO), and 10µM PF treated primary bronchial cell layers.Total junctions analysed from 3 separate samples, one representative image per sample; for ZO1, n=535 for vehicle control, n=850 for 10µM PF; for Claudin-3, n=456 for vehicle control, n=255 for 10µM PF; for E-Cadherin, n=459 for vehicle control, n=650 for 10µM PF; for vinculin, n=580 for vehicle control, n=517 for 10µM PF; for NMIIA expressed as a ratio of junctional to central fluorescence, n=587 for vehicle control, n=473 for 10µM PF; for phospho-a-Catenin to total a-Catenin ratio, n=471 for the vehicle control, n=432 for 10 µM PF treatment; for a18-a-Catenin to total a-Catenin ratio, n=432 for the vehicle control, n=436 for 10 µM PF.Box-and-whisker plots in this and all subsequent figures show mean, 25 th and 75 th percentile, with extreme data points indicated by whiskers.Statistical significance was determined by unpaired t-test with Welch's correction or one-way ANOVA with Dunnett's multiple comparison test when comparing two or more drug treatments (**** = p<0.00001;*** = p<0.0001;* = p<0.01).

Fig. S6 ,
Fig. S6, related to Figure 5. Reduction of TLR levels leads to loss of epithelial tightness and changes at junctions.In siRNA-treated 1-week post-confluence Caco-2 monolayers with reduced TLR1 or TLR2 levels, localisation of junctional components to cell borders is affected.Z-projections of confocal sections covering the apical-lateral junctional area are shown.A-E Knock-down of TLR1, TLR2 and IRAK4 leads to MNIIA reduction at junctions.A-D show overviews of the apical and junctional area of a field of Caco-2 cells, with A'-D' showing higher magnifications at the level of junctions, illustrating the reduction in junctional striations upon siRNA treatment.Magenta boxes indicate the position of the magnified panels.
Development: doi:10.1242/dev.201893:Supplementary information Development • Supplementary information E Quantification of effects on NMIIA: n= 182 junctions were analysed for the scramble control and n= 142 for siTLR1 and n=187 for siTLR2 and n= 151 for siIRAK4.Statistical significance was determined by one way ANOVA with Dunnett's multiple comparison test as **** = p<0.00001.F-I' Knock-down of TLR1, TLR2 and IRAK4 leads to changes to E-Cadherin distribution at lateral junctions, similar to IRAK4-inhibitor treatment.F'-I' show higher magnifications of individual tilted lateral sides of Caco-2 cells, 'a' indicates the apical and 'b' the basal end of the lateral membranes shown.J Treatment of Caco-2 cells with 10µM of the TLR1/2 agonist CU-T12-9 leads to an increase in TEER after 5 days in culture (upon seeding at confluency).Error bars indicate SD of n= 3 (DMSO, CU-T12-9) separate transwells.Statistical significance was determined by unpaired Student's t-test as *** = p<0.0001.Development: doi:10.1242/dev.201893:Supplementary information Development • Supplementary information

Table S1 . Antibodies
Note, for anti TLR antibodies, the antibodies used in our study were the best performing in IF and WB of many that are available that we assessed.