**2. Results**

#### *2.1. Establishment of Angulin-1 Knockout in MDCK C7 and HT-29/B6 Cells*

After transfection of MDCK C7 and HT-29/B6 cells with CRISPR/Cas9, HDR plasmids and sgRNA targeting angulin-1, puromycin-resistant cell clones were screened for angulin-1 knockout by Western blotting. In this study, two angulin-1 knockout clones and their controls were investigated in both cell lines (KO 18 and KO 36 and their respective control clones 14 and 18 in MDCK C7 cell line; and KO 12 and KO 32 and their respective controls 15 and 29 in HT-29/B6 cell line). The angulin-1 KO clones in both cell lines were monoallelic knockouts (Figure 1a,b, Supplementary Tables S1 and S2).

The localization of angulin-1 in the controls and its removal from the TJ in the knockouts was confirmed for all clones as dots at the tTJ by immunofluorescence confocal laser-scanning microscopy. ZO-1 [33] in HT-29/B6 cells and occludin [34] in MDCK C7 cells served as a TJ marker (Figure 1c,d).

**Figure 1.** *Cont*.

**Figure 1.** Expression and localization of angulin-1 in MDCK C7 and HT-29/B6 cells. Densitometric analysis of angulin-1 protein expression levels in control and angulin-1 knockout (**a**) MDCK C7 and (**b**) HT-29/B6 cells. sgRNA targeting angulin-1 led to a high decrease in angulin-1 expression in both cell lines (*n* = 9, MDCK C7: \*\*\* *p* ≤ 0.001 with regard to control 14 and ### *p* ≤ 0.001 with regard to control 18 and HT-29/B6: \*\*\* *p* ≤ 0.001 with regard to control 15 and ### *p* ≤ 0.001 with regard to control 29). Immunostaining of angulin-1 in the clones used throughout this study in (**c**) MDCK C7 and (**d**) HT-29/B6 cells. In the KO clones of both cell lines, angulin-1 disappeared from the tTJ in comparison with their controls. Angulin-1: green; Occludin or ZO-1: red; DAPI (nuclei): blue.

#### *2.2. Effects of Angulin-1 Knockout on Endogenous Proteins of MDCK C7 and HT-29/B6 Cells*

Knockout of one TJ protein may cause relevant variation in other proteins potentially involved in transepithelial water transport. Therefore, we examined levels of tricellulin, the three angulins, occludin, several claudins, aquaporin (AQP) water channels, AQP-1, -3, -4 and -7, SGLT1, and LI-cadherin (Figure 2a,b). As known from previous studies, claudin-2 is not expressed in MDCK C7 cells [12,35]. The densitometric analysis revealed some clonal variability in TJ protein expression between the knockout clones and the controls.

In detail, occludin, claudin-1, -3, -4, -5, -7, -8, and AQP-7 were reduced in the MDCK C7 angulin-1 KO 36 clone (Figure 2c). The MDCK C7 angulin-1 KO 18 clone showed a reduction in occludin, claudin-1, -5, and -7 and a slight increase in AQP-1 compared with the controls (Figure 2c). On the other hand, angulin-1 knockout resulted in increased levels of tricellulin, claudin-1, -5, -7 and -8 and LI-cadherin and decreased levels of claudin-2, AQP-4 and SGLT1 in HT-29/B6 angulin-1 KO 12 clone (Figure 2d). The HT-29/B6 angulin-1 KO 32 clone showed an upregulation of tricellulin, claudin-1, -2, -3, -5, -7, -8, and LI-cadherin (Figure 2d).

Clonal variability in protein expression was also observed between both control clones and both KO clones. None of the other two angulins were significantly altered in the knockout clones of both cell lines.

**Figure 2.** Angulin, tricellulin, occludin, claudin, and aquaporin (AQP) expression in control and angulin-1 knockout cells. Representative Western blots of angulin-1 knockout in (**a**) MDCK C7 and (**b**) HT-29/B6 cells. Densitometric analysis of protein expression levels in (**c**) MDCK C7 and (**d**) HT-29/B6 cells after angulin-1 knockout in comparison to the corresponding vector-transfected controls (*n* = 9, N = 3). β-Actin was used as an internal control for normalization to protein content. Statistical analysis was performed using one-way ANOVA test to compare between the four clones in both cell lines (MDCK C7: \* *p* ≤ 0.05, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001 with regard to control 14 and # *p* ≤ 0.05, ## *p* ≤ 0.01, ### *p* ≤ 0.001 with regard to control 18 and HT-29/B6: \* *p* ≤ 0.05, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001 with regard to control 15 and # *p* ≤ 0.05, ## *p* ≤ 0.01, ### *p* ≤ 0.001 with regard to control 29).

#### *2.3. Effects of Angulin-1 Knockout on Tricellulin Localization in MDCK C7 and HT-29/B6 Cells*

It is known that angulin-1 knockout could have different side effects on the localization of other tight junction proteins, specifically tricellulin. Immunofluorescence studies in combination with confocal microscopy revealed a slight delocalization of tricellulin from the tTJs to the bTJ after angulin-1 knockout in both cell lines, whereas occludin was concentrated at the corners of tTJs (Figure 3). These data sugges<sup>t</sup> that angulin-1 has a role in the localization of tricellulin in both cell lines, nevertheless it does not play an essential role in the localization of other TJ proteins (data not shown).

**Figure 3.** Localization of tricellulin in control and angulin-1 knockout cells. (**a**) Immunostaining of tricellulin in MDCK C7 angulin-1 KO cells. After angulin-1 knockout, tricellulin was still located in the tTJ, but additionally in the bTJ. (**b**) Immunostaining of tricellulin in HT-29/B6 angulin-1 KO cells. After angulin-1 knockout, tricellulin was found in both tTJ and bTJ, similar to control cells.

#### *2.4. Effect of Angulin-1 Knockout on the TJ Ultrastructural Level in MDCK C7 and HT-29/B6 Cells*

To obtain insight of whether and to what extent angulin-1 influences the barrier properties of the TJ in MDCK C7 and HT-29/B6 cells, the ultrastructure of the TJs was analyzed by freeze-fracture electron microscopy (Figure 4). Although it would be interesting to investigate the tTJ structure in the angulin-1 KO cells, we could not find a sufficient number of tTJs in order to discuss the changes of the structure because the tTJs are quite rare and difficult to find in the freeze fracture replica samples.

Comparison of the bTJs (Figure 4a,b) of controls and angulin-1 KO clones showed no alteration in the ultrastructure and revealed a regular meshwork in both cell lines.

Regarding bTJs (Table 1), no alteration was found in the horizontally oriented filaments arranged perpendicular to the paracellular diffusion pathway between the controls and the angulin-1 knockouts in both cell lines. Neither the numbers of strands nor the meshwork depth was changed after angulin-1 KO (Table 1). The network density, which is calculated from the ratio of the number of strands to the network depth, did not differ between controls and angulin-1 knockout clones (Table 1). The number of breaks (>20 nm) per μm horizontal length of single-strands in bTJ was not significantly different between controls and the angulin-1 KO clones (Table 1). More importantly, analysis of strand appearance as being either of continuous- or particle- (pearl string) type revealed no changes that correlated with the observations reported above. Continuous strands appeared in all examined microscopy fields in control clones as well as in angulin-1 KO clones (Table 1). In addition, the TJs of the MDCK C7 controls and the angulin-1 KO clones were composed exclusively of linear strands; however, the HT-29/B6 control 29 and the angulin-1 KO 32 clones showed a slightly curved pattern (Table 1).

**Table 1.** Morphometric analysis of TJ ultrastructure of MDCK C7 and HT-29/B6 angulin-1 KO cells. No significant differences in TJ ultrastructure of control cells and angulin-1 KO cells could be detected using freeze-fracture electron microscopic analysis in MDCK C7 and HT-29/B6 cells.


The meshwork depth a is defined as the distance between the apical and the contra-apical strand. Breaks c are defined as strand discontinuities >20 nm within the compact TJ meshwork; their number is given per μm length of horizontally oriented strands. The density b of bTJ (bicellular tight junction) strands is the ratio of strand number and meshwork depth and given in number per pm meshwork depth.

**Figure 4.** Freeze-fracture electron microscopy of angulin-1 knockout in (**a**) MDCK C7 and (**b**) HT-29/B6 cells. Photos were taken at ×51,000; Bars: 200 nm. The bTJ strands of the control cells and angulin-1 KO clones revealed a regular meshwork, characterized by continuous-type areas. bTJs of angulin-1 KO cells showed no ultrastructural difference compared with control clones in both cell lines.

#### *2.5. Effects of Angulin-1 Knockout on Ion Permeability in MDCK C7 and HT-29/B6 Cells*

To examine whether the loss of angulin-1 affects the epithelial barrier development, the transepithelial resistance (TER), reflecting inverse overall ion permeability, was measured on controls and knockout clones.

TER was reduced in angulin-1 knockout clones in both cell lines (Figure 5a,b, Supplementary Tables S1 and S2). Angulin-1 KO caused a stronger decrease of TER in MDCK C7 cells than in HT-29/B6 cells (7 to 14 times in MDCK C7 cells and 2 to 7 times in HT-29/B6 cells).

**Figure 5.** Functional analysis of angulin-1 knockout in MDCK C7 and HT-29/B6 cells. Effect of angulin-1 knockout on transepithelial resistance (TER) in (**a**) MDCK C7 (*n* = 50) and (**b**) HT-29/B6 cells (*n* = 43). Angulin-1 knockout strongly reduced TER in both cell lines. Statistical analysis was performed using one-way ANOVA test was used to compare between the four clones in both cell lines. (MDCK C7: \*\*\* *p* ≤ 0.001 with regard to control 14 and ### *p* ≤ 0.001 with regard to control 18 and HT-29/B6: \*\*\* *p* ≤ 0.001 with regard to control 15 and ### *p* ≤ 0.001 with regard to control 29).

#### *2.6. Effects of Angulin-1 Knockout on Macromolecule Permeability in MDCK C7 and HT-29/B6 Cells*

The next question was whether angulin-1 knockout also affected the permeability to macromolecules. Therefore, we examined the permeability of the different clones to FITC-dextran 4 kDa (FD4) in two different conditions: (1) under an osmotic gradient or (2) under isosmotic conditions. As a result, FD4 permeability exhibited a strong increase in the angulin-1 knockout in HT-29/B6 cells specifically under isosmotic conditions (Figure 6c,d, Supplementary Table S2). In contrast, angulin-1 knockout did not increase the permeability to FD4 in MDCK C7 cells under isosmotic conditions (Figure 6b, Supplementary Table S1). Interestingly, under an osmotic gradient, only the KO 18 clone in MDCK C7 cells showed a slightly increase in FD4 permeability (Figure 6a, Supplementary Table S1). In MDCK C7 cells, there was no difference between FD4 permeability measured under osmotic and isosmotic conditions, whereas in HT-29/B6 cells, the FD4 permeability measured under isosmotic conditions was higher than measured under osmotic conditions.

**Figure 6.** Functional analysis of angulin-1 knockout in MDCK C7 and HT-29/B6 cells. Effect of angulin-1 knockout on permeability for 4-kDa FITC-dextran (FD4) in MDCK C7 cells under an (**a**) osmotic gradient and (**b**) isosmotic condition. Permeability was increased only in the KO 18 clone under an osmotic gradient (*n* = 8–10). Effect of angulin-1 knockout on permeability for FD4 in HT-29/B6 cells under an (**c**) osmotic gradient and (**d**) isosmotic condition. Permeability was increased in both knockout clones under isosmotic conditions (*n* = 6–9), under osmotic gradient only the KO 32 clone showed an increase in permeability. Statistical analysis was performed using one-way ANOVA test was used to compare between the four clones in both cell lines (ns: not significant, MDCK C7: \* *p* ≤ 0.05 with regard to control 14 and HT-29/B6: \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001 with regard to control 15 and ### *p* ≤ 0.001 with regard to control 29).

#### *2.7. Effect of Angulin-1 Knockout on Transepithelial Water Transport in MDCK C7 and HT-29/B6 Cells*

To analyze the effect of angulin-1 knockout on water permeability of MDCK C7 and HT-29/B6 cells, water flux was measured after induction with an osmotic gradient produced by 100 mM mannitol either in the apical or the basolateral side of the cell layers (Figure 7, Supplementary Tables S1 and S2). This concentration produced measured osmolality gradients of 100 mOsm for 100 mM mannitol.

**Figure 7.** Water flux in control and angulin-1 knockout cells stimulated by an osmotic gradient on the apical or the basolateral side. Water flux induced by a gradient of 100 mM mannitol in (**<sup>a</sup>**,**b**) MDCK C7 and in (**<sup>c</sup>**,**d**) HT-29/B6 cells was unchanged. Statistical analysis was performed using one-way ANOVA test was used to compare between the four clones in both cell lines (ns: not significant).

Surprisingly, and in contrast to what was previously found after tricellulin knockdown in MDCK C7 cells, angulin-1 knockout did not significantly change the water flux under the osmotic gradient induced by 100 mOsm mannitol (Figure 7a,b) compared with their respective controls. In the same way, in HT-29/B6 angulin-1 KO cells, water flux did not significantly change under the osmotic gradient induced by mannitol (Figure 7c,d) in comparison with their controls.

In a parallel series of experiments, the water flux in the MDCK C7 angulin-1-depleted clones was measured after induction with an osmotic gradient produced by 37 mM 4-kDa dextran on the apical side because as we pointed out before, mannitol could go through the tTJ and interact with the pore or with the water molecules blocking their movement in the other direction. The water flux increased in both KO clones if 37 mM 4-kDa dextran (100 mOsm/Apical side) was used (Figure 8). This indicates that removal of angulin-1 from the tTJ altered the flux of water in a tight epithelium and that it is dependent on the chemical nature of the osmotic gradient.

 to

#### **Figure 8.** Water flux in angulin-1 knockout MDCK C7 cells induced by a gradient of 37 mM 4-kDa dextran on the apical side (*n* = 10–11). The transepithelial water transport increased in both angulin-1knockout clones. (\* *p* ≤ 0.05, \*\*\* *p* ≤ 0.001 with regard to control 14 and # *p* ≤ 0.05, ### *p* ≤ 0.001 with regardcontrol18).

As shown in Figure 9, there were differences in FD4 permeability under isosmotic and osmotic gradient conditions between both angulin-1 KO cell types, but there was no difference in FD4 permeability in MDCK C7 control and angulin-1 KO cells under both osmotic conditions. In contrast, in HT-29/B6 cells, no difference in control cells was found, except for an increase in FD4 permeability without any osmotic gradient (Figure 9). A possible explanation for this puzzling finding would be an interaction between the movement of water in the opposite direction to the movement of dextran under osmotic conditions in the tTJ of angulin-1 KO HT-29/B6 cells. This interaction appears not to exist in MDCK C7 angulin-1 KO cells.

**Figure 9.** Comparison of angulin-1 knockout effect on permeability for 4-kDa FITC-dextran (FD4) under isosmotic and osmotic gradient conditions in (**a**) MDCK C7 and (**b**) HT-29/B6 cells. The FD4 permeability was higher under an isosmotic condition than under an osmotic gradient in angulin-1 KO HT-29/B6 cells, whereas no differences were observed in angulin-1 KO MDCK C7 cells. (ns: not significant, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001).

#### *2.8. Effect of Tricellulin KD on Transepithelial Water Transport in HT-29/B6 Cells*

Previously, we described that in MDCK C7 cells tricellulin knockdown increased the transepithelial water flux compared to controls under different osmotic gradients [21]. In order to find out whether or not tricellulin has a similar effect in HT-29/B6 cells, tricellulin KD clones were generated and characterized in a similar way as we did for the MDCK C7 tricellulin KD clones (Table 2, Supplementary Figures S1–S4). Tricellulin KD in HT-29/B6 cells reduced the TER and increased the macromolecule permeability (Supplementary Figure S2). Interestingly, in tricellulin KD clones, the cation and water channel claudin-2 was upregulated (Supplementary Figure S3); however, it was not exclusively localized in the TJ, but also subjunctionally and intracellularly (Supplementary Figure S4a), and therefore this change did not affect ion and water permeability. In concordance with this, the ratio PNa/PCl did not change (Supplementary Figure S4b,c), which means that tricellulin increases the movement of ions without charge selectivity.

**Table 2.** Characteristics of HT-29/B6 tricellulin knockdown clones and the corresponding control. Two tricellulin knockdown clones and its corresponding control were analyzed in this study (Control 12, KD 11 and KD 17). Data of tricellulin expression have been obtained by densitometric analysis of Western blots using β-actin for normalization. Paracellular permeability measurements for FD4 were carried out in the Ussing chamber. Data of PNa/PCl permeability and absolute permeabilities for Na+ and Cl− (PNa, PCl) were obtained from dilution potential measurements in the Ussing chamber. Water flux measurements were performed in a modified Ussing chamber with water flux induced by different osmotic gradients.


Significances refer to the control. *n* number of experiments, \* *p* ≤ 0.05, \*\* *p* ≤ 0.01, \*\*\* *p* ≤ 0.001.

> Regarding tricellulin, its knockdown did not regulate the transepithelial water transport through the tTJ in HT-29/B6 cells (Figure 10). In HT-29/B6 cells claudin-2 is genuinely expressed and a large component of water transport may have travelled through the bicellular TJ so that a possible contribution of the tTJ may not have reached significant levels.

**Figure 10.** Water flux in control and tricellulin knockdown HT-29/B6 cells stimulated by an osmotic gradient. (**a**) Water flux induced by a gradient of 100 mM mannitol on the apical side. (**b**) Water flux induced by a gradient of 100 mM 4-kDa dextran on the apical side. (**c**) Water flux induced by a gradient of 37 mM 4-kDa dextran gradient on the apical side. (**d**) Water flux induced by a gradient of 37 mM 4-kDa dextran gradient on the basolateral side of the cell layer. (*n* = 8–10, ns: not significant, \* *p* ≤ 0.05).
