Manuscript bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. TheClick copyright here holder to view for this linked preprint References (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 2 3 Cingulin unfolds ZO-1 and organizes myosin-2B and g-actin to mechanoregulate apical and 4 tight junction membranes 5 6 7 8 9 10 11 Ekaterina Vasileva1&, Florian Rouaud1&, Domenica Spadaro1, Wenmao Huang2, Adai Colom3, 12 Arielle Flinois1, Jimit Shah1, Vera Dugina4, Christine Chaponnier5, Sophie Sluysmans1, Isabelle 13 Méan1, Lionel Jond1, Aurélien Roux3, Jie Yan2,6, and Sandra Citi*1 14 15 16 17 Departments of Cell Biology1 and Biochemistry3, Faculty of Sciences, Department of Pathology and 18 Immunology5, Faculty of Medicine, University of Geneva, 1205 Switzerland; Department of 19 Physics2 and Mechanobiology Institute6, National University of Singapore, 119074 Singapore; 20 Belozersky Institute of Physico-Chemical Biology4, Lomonosov Moscow State University, Moscow, 21 119192 Russia. 22 23 24 25 26 27 28 &Equal contribution 29 30 1*Corresponding author/Lead Contact: 31 Prof. Sandra Citi, 32 Department of Cell Biology 33 University of Geneva 34 30, Quai E. Ansermet 35 1205 Geneva, Switzerland 36 Tel. +41223796182 37 email Sandra.Citi@ unige.ch 38 39 Running title: Cingulin mechano-regulates the apical membrane 40 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

41 SUMMARY 42 43 How junctional proteins regulate the mechanics of the plasma membrane and how actin and 44 myosin isoforms are selectively localized at epithelial cell-cell junctions is poorly understood. Here 45 we show by atomic force indentation microscopy, immunofluorescence analysis and FLIM 46 membrane tension imaging that the tight junction (TJ) protein cingulin maintains apical surface 47 stiffness and TJ membrane tortuosity and down-regulates apico-lateral membrane tension in 48 MDCK cells. KO of cingulin in MDCK, mCCD and Eph4 cells results in a decrease in the juxta- 49 membrane accumulation of labeling for cytoplasmic myosin-2B (NM2B), g-actin, phalloidin and 50 ARHGEF18, but no detectable effect on myosin-2A (NM2A) and b-actin. Loss of paracingulin leads 51 to weaker mechanical phenotypes in MDCK cells, correlating with no detectable effect on the 52 junctional accumulation of myosins and actins. Cingulin and paracingulin form biomolecular 53 condensates, bind to the ZU5 domain of ZO-1, and are recruited as clients into ZO-1 condensates 54 in a ZU5-dependent manner. Cingulin binding to ZO-1 promotes the unfolding of ZO-1, as 55 determined by interaction with DbpA in cells lacking ZO-2 and in vitro. Cingulin promotes the 56 accumulation of a pool of ZO-1 at the TJ and is required in a ZU5-dependent manner for the 57 recruitment of phalloidin-labelled actin filaments into ZO-1 condensates, suggesting that ZU5- 58 cingulin interaction promotes ZO-1 interaction with actin filaments. Our results indicate that cingulin 59 tethers the juxta-membrane and apical branched g-actin-NM2B network to TJ to modulate ZO-1 60 conformation and the TJ assembly of a pool of ZO-1 and fine-tune the distribution of forces to 61 apical and TJ membranes. 62 63 Keywords: Cingulin, paracingulin, ZO-1, DbpA, ARHGEF18, membrane tension, mechanobiology, 64 actin, nonmuscle myosin. 65 66 67 INTRODUCTION 68 69 The cytoskeleton orchestrates cell shape, motility, internal architecture and mechanical properties, 70 and is involved in most physiological and pathological cellular processes. In epithelial and 71 endothelial tissues, the cytoskeleton is also crucial for the organization and physiology of cell-cell 72 junctions, including tight junctions (TJ), which provide a semipermeable seal for solute diffusion 73 across the paracellular pathway [1, 2], and adherens junctions (AJ), which establish and maintain 74 cell-cell adhesion [3]. Both TJ and AJ are associated with actin and myosin filaments and 75 microtubules, which regulate their dynamic assembly, disassembly and functions [4-7]. Force 76 generated by extracellular and intracellular cues is transduced by junctional mechano-sensing 77 proteins, which modify their conformations and interactions to regulate adhesion strength and cell 78 behavior [8, 9]. TJ and AJ proteins contribute to the organization and function of the cytoskeleton, 79 both through direct interactions with cytoskeletal proteins and through the recruitment of specific 80 activators and inactivators of Rho family GTPases [2, 10].

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

81 82 Most of the mechanical properties of epithelial cells depend on the actin- and myosin-containing 83 cell cortex: myosin-2 ATPase activity and unbranched F-actin polymerization increase cortex 84 tension, while branched actin networks decrease it [11]. Cytoplasmic isoforms of myosin have 85 different roles and localizations at epithelial junctions. Nonmuscle myosin-2B (NM2B) associates 86 with the branched actin meshwork proximal to the plasma membrane, whereas nonmuscle myosin- 87 2A (NM2A), which provides mechanical tugging force, sits on distant peri-junctional actin bundles 88 parallel to the junction [12]. Down-regulation of cytoplasmic b-actin, which is detected laterally and 89 junctionally in epithelial cells, inhibits AJ biogenesis, whereas down-regulation of g-actin, which is 90 localized apically and junctionally, impairs TJ, but not AJ [13, 14]. However, nothing is known about 91 the mechanisms that direct the selective spatial organization of cytoplasmic actin and myosin 92 isoforms at apical junctions. 93 94 The TJ cytoplasmic scaffolding proteins ZO-1 and ZO-2 are critically important not only to link the 95 actomyosin cytoskeleton to TJ transmembrane proteins, but also to control its apical/junctional 96 organization [15-19]. Depletion of ZO-1 results in increased junctional contractility coupled to 97 decreased NM2B integration into junctions [20-22], loss of tortuosity of the TJ membrane [20, 23], 98 increased apical stiffness [24], and altered organization of apical actin filaments [18, 19]. However, 99 the mechanisms through which ZO proteins organize the actomyosin cytoskeleton and mediate its 100 effects on TJ and apical membranes are not well understood. For example, the TJ barrier 101 phenotype in cells lacking ZO-1 can be rescued by constructs lacking the actin-binding region 102 (ABR) within the C-terminal half of ZO-1 [20], suggesting that indirect interactions with actin- and 103 myosin-binding proteins mediate ZO-1 cross-talk with actomyosin. Actomyosin tension and 104 dimerization control the conformation of ZO-1, which can be either stretched (unfolded) or folded 105 (auto-inhibited) [25]. The folded conformation was proposed to result from a mechanosensitive 106 intramolecular interaction between the C-terminal ZU5 (Cter) domain of ZO-1 and the PDZ3-SH3- 107 GUK (PSG) region [25]. Unfolding and multimerization of ZO proteins are required for their liquid- 108 liquid phase separation to drive TJ formation, regulated by phosphorylation and multivalent 109 interactions [26]. Although the N-terminal half of ZO-1 can by itself undergo multimerization and 110 phase separation [26], the C-terminal half of ZO-1 is required to confer mechano-sensitivity to 111 junctions [27], to provide fluidity to ZO-1 condensates [26], and to allow MLCK-dependent 112 regulation of the dynamic behavior of ZO-1 [28]. Together, these observations suggest that ZO-1 113 interactions mediated by its C-terminal region are critical for ZO-1 mechano-chemical signaling. 114 115 Among several actomyosin-associated proteins that bind to ZO-1 [5], cingulin (CGN) [29] and 116 paracingulin (JACOP/CGNL1) [30, 31] interact with actin and myosin [32-35], and with GEFs and 117 GAPs for Rho family GTPases, such as GEF-H1, ARHGEF18 and MgcRacGAP [22, 36-39]. 118 Cingulin is recruited to TJ by ZO-1 [40, 41], whereas paracingulin is recruited to TJ and the zonula 119 adhaerens (ZA) through multiple interactions, including with ZO-1 and PLEKHA7 [30, 42]. Cingulin 120 and paracingulin interact with ZO-1 through a conserved ZO-1 Interaction Motif (ZIM) at their N-

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

121 terminus [32, 40, 42]. Although the sequences of ZO-1 that interact with either cingulin or 122 paracingulin are not known, yeast-2-hybrid screen [43] and BioID experiments [44] indicate that 123 they are within the C-terminal region of ZO-1. Nothing is known about the role of cingulin and 124 paracingulin in the regulation of the mechanical properties of the apical and junctional membranes, 125 the organization of apical actomyosin filaments, and the conformation and junctional assembly of 126 ZO-1. Here we address these questions, by using CRISPR-KO epithelial cell lines and biophysical, 127 biochemical and cellular approaches. We show that cingulin regulates apical surface stiffness and 128 apicolateral membrane tension, junctional accumulation of NM2B and g-actin, and ZO-1 129 conformation, TJ assembly and interaction with actin. 130 131 RESULTS 132 133 Knock-out of cingulin decreases apical surface stiffness and increases junctional 134 membrane tension 135 136 To ask if cingulin and paracingulin regulate the mechanical properties of epithelial cells, we 137 generated clonal lines of epithelial cells KO for either cingulin, paracingulin or both (Figure S1). 138 Immunoblotting, immunofluorescence and genomic sequence analysis validated the KO lines 139 (Figure S1). In MDCK cells, the single and double CGN/CGNL1 KO clonal lines showed 140 proliferation curves and expression levels of other junctional proteins similar to wild-type (WT), 141 except for an increase in paracingulin levels in CGN-KO cells (Figure S1). 142 143 Atomic force indentation microscopy was used to measure the stiffness of the apical surface in 144 MDCK clonal lines (Figure 1A). Force-indentation curves of the MDCK cells were fitted by Hertz 145 model [45] (Figure 1B) to obtain the Young’s modulus (Figure 1C). The Young’s modulus of CGN- 146 KO and CGN/CGNL1 double-KO MDCK cells was less than half the value of WT cells (e.g. 0.0015 147 MPa and 0.0014 MPa, compared to 0.0037 MPa) (Figure 1C and Table S1), indicating a significant 148 loss of stiffness of CGN-KO and double-KO cells. The decrease in Young’s modulus for CGNL1- 149 KO MDCK cells was smaller (0.0021 MPa), indicating that paracingulin is less important in 150 regulating apical stiffness, and that cingulin is epistatic to CGNL1. 151 152 The shape of the apical surface of CGN-KO cells was examined by scanning electron microscopy 153 of confluent MDCK monolayers. WT cells displayed a slightly extruding, dome-shaped apical 154 surface (Figure 1D), whereas the apical surface of CGN-KO cells appeared flatter (Figure 1E). 155 156 Next, we used the FliptR membrane probe [46] to measure membrane tension along the apico- 157 lateral region of the plasma membrane of MDCK cells. Probe lifetime, which correlates with 158 membrane tension [46], was strongly increased in CGN-KO and double-KO MDCK cells, and 159 slightly, but still significantly increased in CGNL1-KO cells (Figure 1F). By analyzing lifetime 160 emission of the FliptR probe as a function of the distance from the apical surface, WT cells showed

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

161 a sharp gradient of tension along 4 µm, from the basal to the apical surface (slope 0.182, Figure 162 1G), whereas the gradient for cingulin-KO cells was less sharp (slope 0.067, Figure 1G), indicating 163 increased tension throughout this region. The slopes for CGNL1-KO and double KO cells were 164 0.098 and 0.081, respectively (Figure 1G) suggesting a less important role of paracingulin in the 165 maintenance of a sharp membrane tension gradient. 166 167 Together, these observations indicate that cingulin, and to a lesser extent paracingulin, is required 168 to maintain the stiffness of the apical surface membrane and down-regulate tension at the apico- 169 lateral, junctional membrane. 170 171 Cingulin maintains the tortuosity of the TJ membrane and is required for the correct 172 organization of nonmuscle myosin-2B (NM2B) and g-actin at junctions 173 174 To understand how cingulin and paracingulin regulate the mechanical properties of the apicolateral 175 plasma membrane, we examined the shape of the TJ membrane in MDCK cells and the spatial 176 organization of cytoplasmic myosin and actin isoforms in different cell types KO for one of both 177 proteins. In polarized WT MDCK cells labeling for cingulin and occludin was wavy, whereas 178 b-catenin labeling, corresponding to the zonula adhaerens (ZA) and the AJ, was straight (Figure 179 2A, and Figure 2B-WT), suggesting that the membranes of TJ and ZA/AJ are subjected to different 180 tensile forces. To quantify TJ membrane tortuosity, we used the zigzag index [23]. KO of either 181 cingulin or both cingulin and paracingulin resulted in straight TJ membranes (Figure 2B) and a 182 large drop of the zig-zag index (Figure 2C), whereas KO of paracingulin resulted in a considerably 183 smaller reduction in the zig-zag index (Figure 2B-C). The decreased TJ membrane tortuosity of 184 CGN-KO MDCK cells was rescued by re-expression of either canine, mouse or human cingulin 185 (Figure 2D-E, Figure S2A, S2C), but not by overexpression of ZO-1 (Figure S2A, S2C). 186 Furthermore, tortuosity of the TJ membrane in the background of WT cells was increased by 187 overexpression either of cingulin or ZO-1 (Figure S2A-B), the latter being associated with increased 188 junctional cingulin (Figure S2D). 189 190 Next, we examined the distribution of myosins and actins in mixed cultures of WT and KO cells, 191 using antibodies against cytoplasmic nonmuscle myosin-2A (NM2A), myosin-2B (NM2B), g-actin 192 and b-actin isoforms, and fluorescently labeled phalloidin. In WT MDCK cells NM2A and NM2B 193 showed a junctional localization (arrows in Figure 2F-G, insets), and a diffuse apical distribution 194 (circles in Figure 2F-G), this latter likely corresponding to the terminal web and apical cortex. In 195 CGN-KO MDCK cells, labeling for NM2A was similar to WT (arrow, Figure 2F), whereas in CGN- 196 KO cells there was a decrease in both junctional and apical NM2B labeling (arrowhead and dotted 197 circle, Figure 2G), as well as a decrease in junctional labeling for phalloidin (arrowhead, Figure 198 2H). A similar decrease in NM2B and phalloidin junctional labeling and no detectable change in 199 NM2A labeling was observed using CGN-KO+WT mixed cultures of Eph4 cells (Figure S2G) and 200 mCCD cells (Figure S2K). However, the TJ membrane loss-of-tortuosity phenotype could not be

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

201 detected in Eph4 and mCCD cells, since both WT and CGN-KO cells had straight TJ membranes 202 (Figure S2G-L). The junctional accumulation of ARHGEF18 (p114RhoGEF), which regulates Rho 203 signaling and actomyosin organization at junctions [39], was also decreased in CGN-KO MDCK 204 cells (arrowhead, Figure 2I). NM2B, phalloidin and ARHGEF18 labeling were rescued in CGN-KO 205 cells by re-expression of GFP-tagged cingulin (arrows, Figure 2J-L), whereas GFP alone had no 206 effect (arrowheads, Figure S2F). In contrast, in mixed cultures of WT and CGNL1-KO WT cells 207 there were no clearly detectable differences in the accumulation of NM2A, NM2B, phalloidin and 208 ARHGEF18 at junctions (arrows, Figure S2E). Next, we investigated the distribution of cytoplasmic 209 b-actin and g-actin isoforms, using specific antibodies [13]. Since the labeling of MDCK cells with 210 these antibodies was less clear, we used mCCD and Eph4 cells. Strikingly, junctional labeling of g- 211 actin, which was intense in the juxtamembrane region of WT cells, was reduced in intensity and 212 shifted farther away from the membrane in CGN-KO mCCD cells (arrowhead, Figure 2M inset). In 213 contrast, the distribution of labeling for b-actin, which was farther away from the membrane than g- 214 actin, was similar in WT and CGN-KO mCCD cells (arrows, Figure 2N). Quantification of line 215 scans of labeling across the junction and colocalization with PLEKHA7 confirmed that KO of 216 cingulin resulted in a shift of g-actin labeling farther from the membrane, with respect to WT cells 217 (Figure 2O-Q), and no detectable change for b-actin (Figure 2P-R). Similar results were obtained in 218 Eph4 cells, where KO of cingulin resulted in decreased junctional labeling for NM2B, phalloidin and 219 g-actin, but no detectable change in NM2A and b-actin labeling (arrows and arrowheads in Figure 220 S2G). Similar to MDCK cells, KO of paracingulin in Eph4 and mCCD cells did not result in clearly 221 detectable differences in labeling for myosins, phalloidin and actins (Figure S2H, S2L). The 222 decreased junctional localization of NM2B and F-actin in CGN-KO Eph4 cells was rescued by re- 223 expression of cingulin (Figure S2I-J), and of chimeric constructs comprising the head domain of 224 CGNL1 and the rod+tail domain of CGN (Figure S2J) [30, 32], indicating that the rod+tail region of 225 cingulin is mechanistically involved. In summary, cingulin plays a key role in maintaining membrane 226 tortuosity and organizing NM2B and g-actin at TJ of epithelial cells, whereas paracingulin has a 227 minor role in maintaining membrane tortuosity in MDCK cells and does not induce clearly 228 detectable changes in the organization of actin and myosin filaments in this experimental model. 229 230 Cingulin binding to the ZU5 domain of ZO-1 unfolds ZO-1 231 232 To investigate the functional consequences of the cingulin-dependent organization of NM2B and g- 233 actin at TJ, we focused on ZO-1, the major protein implicated in TJ-actin interactions, and the 234 protein that recruits cingulin to TJ. Specifically, since ZO-1 monomers undergo tension-dependent 235 stretching (unfolding) and folding [25], we hypothesized that by promoting the accumulation of 236 NM2B near TJ, cingulin might facilitate the transduction of force onto ZO-1 monomers, and thus 237 promote their unfolding. Alternatively, cingulin may affect ZO-1 conformation by binding to a region 238 involved in intramolecular interactions, such as the C-terminal ZU5 domain (Cter, [25]). 239

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

240 To test these hypotheses, we first mapped the region of ZO-1 that binds to cingulin. A N-terminal 241 fragment of cingulin that contains the ZO-1-Interaction-Motif (ZIM) (CGN(1-70)) [40] was used as a 242 bait to pull down GFP-tagged fragments of the C-terminus of ZO-1 (residues 888-1748)(Figure 243 S3A). Only ZO-1 preys that contained the ZU5 domain (ZU5=Cter, residues 1619-1748 [25]) 244 interacted with GST-CGN(1-70) (Figure 3A), and the binding of the ZIM to the isolated ZU5 245 fragment (1619-1748) was >4-fold stronger than to other fragments (Figure S3B). Deleting the ZIM- 246 containing first 70 residues of cingulin (CGN-D1-70) essentially abolished the interaction between 247 bacterially expressed ZU5 and full length cingulin (Figure 3B). Fragmenting the ZU5 domain 248 strongly decreased the interaction with the CGN-(1-70) bait (Figure S3 C-F), indicating that an 249 intact ZU5 domain is required for high affinity binding of ZO-1 to cingulin. Full-length paracingulin 250 also interacted, albeit weaker than cingulin, with the ZU5 bait, and deleting its first 80 residues 251 abolished the interaction (Figure 3C). Larger ZIM-containing fragments of paracingulin also showed 252 weaker binding to ZO-1, with respect to cingulin (Figure S3G). Junctional cingulin labeling was 253 undetectable in ZO-1-KO cells (Figure S3H) (see also [41]), whereas paracingulin labeling was only 254 reduced, but not abolished in ZO-1-KO Eph4 cells (Figure S3I-J). The loss of cingulin labeling in 255 ZO-1-KO cells was rescued by expressing full-length ZO-1, but not ZO-1 lacking the ZU5 domain 256 (ZO-1-DZU5) (Figure 3D). As an alternative approach to test ZO-1 binding to cingulin and 257 paracingulin, we examined their recruitment in ZO-1 condensates obtained by exogenous 258 overexpression of ZO-1 [26]. Both cingulin and paracingulin were recruited into ZO-1 condensates 259 in a ZU5-dependent manner (Figure 3E-F). Together, these results demonstrate that the ZU5 260 domain of ZO-1 binds to cingulin and paracingulin and indicate that only a fraction of paracingulin is 261 recruited to TJ by ZO-1, consistent with previous observations [30, 42]. 262 263 Second, we asked whether cingulin binding to the ZU5 domain promotes ZO-1 unfolding. We 264 proposed that when ZO-1 is in the folded conformation, the ZU5 domain binds to the PDZ3-SH3- 265 GUK region of ZO-1 (ZPSG1) [25]. Thus, we wondered whether cingulin binding induces ZO-1 266 unfolding by competing with ZU5-dependent intramolecular interactions. To test this, we 267 overexpressed the (1-70) fragment of cingulin, which contains the ZO-1-binding ZIM motif, in Eph4 268 cells, and used the junctional localization of DbpA as a readout for the ZO-1 stretched conformation 269 [25]. ZO-1 was sufficient to retain DbpA at junctions in cells depleted of ZO-2 (arrow Figure 3G, 270 upper panels), but DbpA junctional localization was reduced or undetectable in cells depleted of 271 ZO-2 and treated with the myosin inhibitor blebbistatin (arrowhead, Figure 3H, upper panel), as 272 shown previously [25]. GFP-CGN(1-70) was recruited to junctions (Figure 3G, bottom panel), and 273 its overexpression in ZO-2-depleted cells treated with blebbistatin rescued the junctional 274 localization of DbpA (arrow in Figure 3H, bottom panel), indicating ZO-1 unfolding [25]. To confirm 275 the role of cingulin in unfolding ZO-1, we used CGN-KO Eph4 cells. Depletion of ZO-2 in these 276 cells resulted in loss of DbpA at junctions (arrowhead, Figure 3I, bottom panel), indicating that loss 277 of cingulin phenocopies treatment with blebbistatin of ZO-2-depleted cells [25]. Furthermore, 278 although GST-DbpA failed to interact with full-length ZO-1 in vitro [25, 47], addition of increasing 279 amounts of GFP-CGN(1-70) resulted in detectable ZO-1 in GST-DbpA pulldowns (Figure 3J,

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

280 normalization of preys in Figure 3K), indicating ZO-1 unfolding. This interaction, and the rescue of 281 junctional DbpA by CGN(1-70) (Figure 3H) was not an artefact due to binding of CGN(1-70) to 282 DbpA, since this construct did not interact with DbpA, which could still bind to its known interactor

283 GEF-H1 [48], used as a positive control (Figure 3L). The dissociation equilibrium constant (Kd) for 284 the interaction of the CGN(1-70) with ZU5 was 40.4 nM (Figure S3K-M), lower that the calculated

285 Kd for the ZPSG-ZU5 interaction (66 nM, [25]). Collectively, these experiments indicate that high 286 affinity binding of the ZIM-containing region of cingulin to the ZU5 domain of ZO-1 is required for 287 cingulin recruitment to junctions and promotes the unfolding of ZO-1. 288 289 Cingulin promotes the accumulation of ZO-1 at tight junctions 290 291 Since ZO-1 unfolding promotes phase separation and TJ assembly of ZO-1 [26], we hypothesized 292 that cingulin, which is recruited to TJ by ZO-1, could in turn enhance the assembly of ZO-1 at TJ, in 293 a positive feedback loop. To test this, we compared ZO-1 immunofluorescent labeling in confluent 294 monolayers of WT cells, and cells KO for cingulin. Using either the ZA protein PLEKHA7 or the TJ 295 protein occludin as a reference junctional marker, ZO-1 labeling at apical junctions was reduced in 296 CGN-KO cells, when compared to WT, in mixed cultures of WT+CGN-KO Eph4 cells (Figure 4A, 297 Figure S4A-B) and MDCK cells (Figure 4B). In contrast, ZO-1 labeling was similar in mixed cultures 298 of WT and CGNL1-KO Eph4 (Figure 4C) and MDCK cells (Figure 4D). The same results were 299 obtained when examining the effect of KO of either CGN or CGNL1 in mCCD cells (Figure S4C-D, 300 G-H), and when labeling ZO-1 with different antibodies (Figure S4A-H), indicating that this 301 phenotype is not affected by cell type or epitope availability differences. Next, to confirm that the 302 reduction in ZO-1 labeling in CGN-KO cells was specifically due to the loss of cingulin, and not to 303 clone-dependent variations, we rescued the localization of ZO-1 through expression of exogenous 304 cingulin constructs. Full-length, myc-tagged cingulin rescued ZO-1 junctional accumulation (arrows 305 in Figure 4E, CGN FL-myc, and quantification in Figure 4G), whereas a mutant of cingulin lacking 306 the first 70 amino acid residues of cingulin, which contain the ZO-1-Interaction-Motif (ZIM) [40] did 307 not localize at junctions, and did not rescue ZO-1 junctional labeling (arrowheads, Figure 4E, CGN- 308 D1-70-myc, quantification in Figure 4G). Conversely, highly overexpressed full-length cingulin, but 309 not cingulin lacking the first 70 residues, enhanced the junctional accumulation of ZO-1 in 310 otherwise WT cells (Figure 4F), even though this increase was not statistically significant (Figure 311 4G), suggesting that in WT cells ZO-1 levels are already near saturation. By immunoblotting 312 analysis, the KO of either CGN or CGNL1 did not affect the total levels of ZO-1 protein and its a(+) 313 versus a(-) isoforms (Figure S1R, S4I-J), indicating that the reduced junctional labeling of ZO-1 is 314 not due protein degradation or to a loss of a specific isoform. 315 316 We next asked whether other ZO proteins are affected by the loss of cingulin. The junctional 317 labeling for ZO-2 was similar to WT in cells KO for either CGN or CGNL1 (Figure S4K-L), whereas 318 ZO-3 labeling was reduced in CGN-KO cells (Figure S4O-P), but not in CGNL1-KO cells (Figure 319 S4Q-R). ZO-3 labeling was rescued by re-expression of cingulin, but not overexpression of ZO-1

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

320 (Figure S4U-V). The protein expression levels of ZO-2 and ZO-3 were not affected by KO of CGN 321 either in MDCK cells (Figure S1R, S4M-N, S4S-T), or mCCD and Eph4 cells (S4M-N, S4S-T). 322 Together, these data indicate that cingulin controls the accumulation of a pool of ZO-1 and ZO-3, 323 but not ZO-2, at TJ. 324 325 ZO-1 condensates recruit NM2B and ARHGEF18 independently of cingulin, and cingulin 326 promotes interaction of ZO-1 with F-actin by interacting with the ZU5 domain 327 328 To gain more insight into the mechanism through which cingulin promotes the junctional 329 accumulation of ZO-1, we overexpressed GFP-tagged forms of cingulin, paracingulin and ZO-1 in 330 MDCK cells and analyzed the labeling of clients/interactors into phase-separated ZO-1 331 condensates (Figure 5 and Figure S5) [26]. ZO-1 undergoes liquid-liquid phase separation [26], 332 and previous observations indicate that either cingulin or paracingulin overexpression in MDCK 333 cells leads to the formation of coalescing dots, with dynamics that suggest phase separation [49]. 334 GFP-tagged cingulin and paracingulin, when overexpressed in cells, induced the formation of 335 cytoplasmic brightly labelled structures, which underwent dynamic fission and fusion, (Figure 5A-A’’ 336 and Figure S5A-A’), very similar to what observed for ZO-1 (Figure 5B-B’’), suggesting the 337 formation of phase-separated condensates. We found that while ZO-1 condensates recruited both 338 cingulin and paracingulin (Figure 3E-F) (see also [26]), cingulin and paracingulin condensates did 339 not recruit ZO-1 as a client (Figure S5B). Moreover, NM2B labeling was increased and detected in 340 bright cytoplasmic condensates of either CGN or ZO-1 (arrows, Figure 5C-D). In contrast, in the 341 case of phalloidin and ARHGEF18, labeling was redistributed in the cortical submembrane area of 342 cells that overexpressed CGN (double arrowheads, Figure 5C) and detected in strongly labelled 343 cytoplasmic condensates of ZO-1 (arrows, Figure 5D). Paracingulin condensates also showed 344 redistribution of ARHGEF18 in the submembrane cortex but not in the cytoplasm, similarly to 345 cingulin, but did not show increased or redistributed labeling for either NM2B or phalloidin (Figure 346 S5C). NM2B and ARHGEF18 were brightly labelled in cytoplasmic ZO-1 condensates even when 347 ZO-1 was overexpressed in the background of CGN-KO MDCK cells (Figure 5E), indicating that 348 the association of ZO-1 with NM2B and ARHGEF18 occurs independently of cingulin, at least in 349 the context of condensates. Importantly, the recruitment of phalloidin-labelled F-actin into ZO-1 350 condensates was significantly decreased in CGN-KO cells, as shown by the lack of brightly labelled 351 cytoplasmic dots (arrowhead, Figure 5E, middle panels), suggesting that cingulin promotes ZO-1 352 interaction with actin filaments. To test this hypothesis, we compared phalloidin staining in 353 condensates of either full length ZO-1, or of the ZO-1 mutant lacking the ZU5 domain (ZO-1-DZU5). 354 Strikingly, while in WT cells condensates made by overexpressing either full-length ZO-1 or ZO-1- 355 DZU5 recruited brightly phalloidin-labelled F-actin (arrows, Figure 5F, quantification in Figure 5H), 356 in the background of CGN-KO cells only the ZO-1-DZU5 construct, but not full-length ZO-1 could 357 recruit bright phalloidin-labelled F-actin (arrowhead and arrow, Figure 5G, quantification in Figure 358 5H). This suggests that in full-length ZO-1 the ZU5 domain inhibits ZO-1-F-actin interaction, and 359 that cingulin binding to the ZU5 antagonizes this inhibition.

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

360 361 DISCUSSION 362 363 The mechanical properties of cells are crucial to respond to environmental cues during 364 development, in adult tissue homeostasis, and in disease, and depend on actin and myosin, and 365 their mode of polymerization [11]. Here we provide evidence that cingulin regulates the mechanical 366 properties of the apical membrane, the junctional organization of g-actin and NM2B, and the 367 conformation, TJ assembly and F-actin interaction of ZO-1 (Figure 6, Figure S6). 368 369 The observation that straight TJ and reduced junctional accumulation of NM2B [20, 21, 23] occur in 370 CGN-KO cells in the presence of ZO-1 suggests that ZO-1 controls TJ membrane tortuosity and 371 NM2B organization by recruiting cingulin to TJ. A branched network of actin filaments is associated 372 with NM2B in the juxta-membrane region of apical junctions [12], and we propose that this 373 branched network comprises g-actin, and its tethering to TJ requires cingulin (Figure 6, Figure S6). 374 Cytoplasmic g-actin is not critical for junction assembly and organism viability [50, 51], but is 375 specifically associated with the apical surface and TJ in epithelia [13, 14]. Labeling for g-actin was 376 reduced and shifted farther away from the membrane in cingulin-KO cells, and phalloidin labeling 377 was decreased despite normal labeling of junctional b-actin. Since actin assembly at junctions is 378 regulated by tension [52], our results suggest that when cingulin tethers to TJ the NM2B- g-actin 379 lace-like web, this latter is maintained under tension, and phalloidin binds very well to it. The KO of 380 CGN uncouples the network from TJ and releases the tension, inducing the disassembly, and 381 possibly the partial collapse of the network onto the peri-junctional ring, and decreased phalloidin 382 labeling. Since the accumulation of NM2B is not abolished, but only altered/reduced in cells 383 depleted of either ZO-1 [20, 21, 23] or cingulin, redundant mechanisms must exist to recruit NM2B 384 to TJ and ZA. The mechanism through which cingulin tethers the g-actin-NM2B network to TJ could 385 be either through direct or indirect binding to myosin [32] or through recruitment of ARHGEF18 386 [39], or through modulation of ZO-1 conformation (see below), or a combination of these. Future 387 studies should test these hypotheses and determine whether and how g-actin is connected to the 388 ZA. Although we did not observe any effect of cingulin KO on the localization of either NM2A or b- 389 actin in our models, depletion of cingulin in human corneal epithelial (HCE) increases labeling for 390 NM2A [39]. Possible reasons for this apparent discrepancy include the difference in cytoskeletal 391 organization in stratified versus columnar/cuboidal epithelial cells, and the presence of residual 392 cingulin in depleted cells, which could result in altered signaling through regulators of Rho family 393 GTPases. 394 395 Actin and myosin control cellular mechanics, and here we studied how KO of CGN, CGNL1 or both 396 affected apical surface stiffness and apicolateral membrane tension. In our atomic force 397 microscopy experiments, the elastic modulus of WT MDCK cell is consistent with previously 398 reported MDCK stiffness [53]. The loss of cingulin, and to a lesser extent of paracingulin, reduces 399 the cell layer elastic stiffness, while monolayer integrity and junctions are maintained. This

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

400 phenotype can be explained by a role of the branched g-actin-NM2B network in providing 401 mechanical support to the apical surface, consistent with g-actin localization [13], and with the 402 observed collapse of the apical mesh-like adhesion structure in cells lining the spinal canal, upon 403 KO of NM2B in mice [54, 55]. Nonmuscle myosin isoforms have different enzymatic properties and 404 cellular roles [56], and the longer actin-attachment lifetimes and greater strain dependence of 405 NM2B allows it to maintain force more effectively than NM2A [57]. The loss of apical stiffness in 406 CGN-KO cells correlates with a flatter shape of the apical membrane, in contrast to the phenotype 407 of ZO-1-depleted cells, which show increased contractility and apical surface stiffness, and convex, 408 extruding apical membranes [18, 19, 24]. Since ZO-1 recruits cingulin to TJ, we conclude that the 409 complex cytoskeletal reorganization that occurs in ZO-1-depleted cells overrides the effects 410 generated by the secondary loss of cingulin. This is not surprising, considering ZO-1’s multivalent 411 interactions with proteins that bind to and regulate actomyosin filaments [58-60]. Using the FliptR 412 membrane tension probe, we found that KO of cingulin, and to a lesser extent of paracingulin, 413 increases overall apicolateral membrane tension, by reducing the sharp tension gradient existing in 414 WT cells. This suggests that frictional forces between the membrane and the dense TJ-associated 415 branched γ-actin network dampen the tension generated by the NM2A peri-junctional bundle, in 416 agreement with previous observations [11, 12, 19, 61, 62]. Concerning the weaker mechanical 417 phenotypes of paracingulin-KO cells, we speculate that similar mechanisms are involved, despite 418 the apparently normal localizations of myosins, actins and ARHGEF18 in CGNL1-KO cells. The 419 phenotypes may be weaker and harder to detect in CGNL1-KO cells because of the lower levels of 420 paracingulin versus cingulin in epithelial cells [63], the small fraction of paracingulin associated with 421 TJ versus ZA [30, 42] and the lower affinity of interaction of paracingulin with ZO-1. However, our 422 experiments in different cell types, and using chimeric molecules to rescue NM2B localization 423 indicate that cingulin and not paracingulin is specifically involved in the junctional assembly of 424 NM2B. In contrast, both proteins were reported to regulate ARHGEF18, although in distinct cellular 425 contexts [22, 39]. Thus, future studies should address the role of ARHGEF18 in membrane 426 mechanics and actin and myosin isoform localization. 427 428 We show that the ZU5 domain, a 110-residue domain positioned at the extreme C-terminus of ZO- 429 1 [5], binds to cingulin and paracingulin. Known interactors of the ZU5 domain of ZO-1 are the 430 Cdc42 effector kinase MRCKb, which is associated with actin at the leading edge of migrating cells, 431 but is not localized at TJ [64, 65], and the Rho GTP Exchange factor ARHGEF11, which is 432 recruited to TJ by ZO-1, but is not required for ZO-1 TJ targeting [58]. The ZU5-cingulin interaction 433 not only recruits cingulin to TJ, but also promotes the unfolding of ZO-1, as operationally defined by 434 the rescue of junctional DbpA labeling in ZO-2-depleted cells [25]. This suggests that loss of 435 cingulin phenocopies the effect of blebbistatin, and ZO-1 folding could thus be due to reduced 436 action of NM2B with ZO-1. However, our experiments in vitro using the CGN(1-70) fragment, which 437 is unlikely to recruit actomyosin, show that its binding to the ZU5 is sufficient to induce ZO-1-DbpA 438 interaction, suggesting that the CGN(1-70) fragment unfolds ZO-1 by antagonizing ZU5-dependent 439 intramolecular interactions. Thus, although ZO-1 obviously unfolds and forms condensates in the

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

440 absence of cingulin, cingulin binding might regulate a pool of folded ZO-1, lower its critical 441 concentration for liquid phase separation, and/or increase the kinetic of formation and the size of 442 ZO-1 condensates. Moreover, our analysis of ZO-1 condensates in either WT or CGN-KO 443 backgrounds indicates that one function of the ZU5 domain could be to inhibit the binding of F-actin 444 to the ABR domain, and binding of cingulin could relieve this inhibition (Figure S6). The idea that 445 the ZU5 domain might interact with upstream, ABR-containing sequences of the C-terminal half of 446 ZO-1 is consistent with the observation that the binding of CGN(1-70) to the isolated ZU5 domain is 447 stronger than to constructs encompassing larger C-terminal fragments (Figure 3A, Figure S3D). 448 There is also evidence from FRAP studies that the ABR domain is involved in intramolecular ZO-1 449 interactions [28]. Moreover, the ABR lies within a disordered region of the C-terminal half of ZO-1 450 and shows low affinity of binding to actin [66], suggesting that conformational changes driven by 451 ligand-ZU5 interactions may promote the ABR interaction with actin filaments. These hypotheses 452 need to be tested by future studies. 453 454 The role of cingulin in promoting the efficient accumulation of ZO-1 at TJ provides a mechanistic 455 explanation for the observed ZU5-dependent stabilization of ZO-1 at junctions [67, 68]. As noted 456 above, this stabilization could be due to cingulin promoting the unfolding and phase separation of 457 an otherwise soluble pool of ZO-1, by relieving a ZU5-dependent intramolecular inhibition which 458 modulates the binding of actin filaments to the ABR. Finally, tension is required for TJ formation 459 [69, 70] and cingulin may promote tension-dependent accumulation of ZO-1 by recruiting to 460 junctions ARHGEF18, which activates junctional RhoA [39]. Interestingly, KO of cingulin but not 461 paracingulin reduced junctional ARHGEF18 in our epithelial cell models, although an interaction of 462 ARHGEF18 with paracingulin was reported in endothelial cells [22]. Analysis of condensates also 463 indicates that ZO-1 can associate with ARHGEF18 independently of cingulin. Since ZO-1 junctional 464 levels are only reduced, but not abolished by KO of CGN, other mechanisms which unfold/stretch 465 ZO-1 are sufficient to allow most of ZO-1 to assemble at junctions in the absence of cingulin, in 466 agreement with previous data [71, 72]. These other mechanisms include heterodimerization, 467 actomyosin tension applied indirectly, through ZO-1-interactors, and phosphorylation (Figure S6), 468 whereas the role of cingulin is to unfold the ZU5 domain, promote the complete TJ assembly of ZO- 469 1, and tether it to the NM2B-g-actin network. 470 471 We also provide evidence indicating that in addition to ZO-1 [26], cingulin and paracingulin undergo 472 phase separation, at least when expressed at high concentrations in cells. In a previous study, we 473 showed that overexpressed GFP-tagged cingulin and paracingulin form structures that undergo 474 dynamic fusions and relaxations, disappear upon mitosis, and rapidly associate with cell-cell 475 contacts immediately following cytokinesis [49]. Cingulin and paracingulin contain intrinsically 476 disordered N-terminal head and C-terminal tail domains, and a coiled-coil domain, all of which 477 could drive phase separation, by enabling multivalent homo- and heterotypic protein-protein 478 interactions [73, 74]. Furthermore, both cingulin and paracingulin are expected to form 479 condensates based on prediction algorhythms [75]. Although additional work is required to

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

480 characterize in detail the biophysics and biochemistry of cingulin and paracingulin condensates and 481 establish their physiological relevance, here we show that ZO-1 condensates recruit cingulin and 482 paracingulin, but cingulin and paracingulin condensates do not recruit ZO-1, in agreement with 483 previous data [22, 39-41, 71, 72]. 484 485 In summary, cingulin and paracingulin regulate the mechanical properties of the apical surface and 486 the apicolateral plasma membrane, and cingulin regulates the conformation and TJ assembly of 487 ZO-1 by organizing NM2B and g-actin at the TJ-apical complex. Importantly, both cingulin and 488 paracingulin associate with microtubules, and cingulin organizes the epithelial planar apical 489 network of microtubules [6, 35, 76, 77]. Thus, the present results suggest that cingulin and 490 paracingulin are major players in the cross-talk between the microtubule and actin cytoskeletons at 491 junctions and apical membranes, to fine-tune the mechano-regulation of morphogenesis, TJ barrier 492 function, and other physiological and pathological aspects of epithelial and endothelial cells. 493 494 ACKNOWLEDGEMENTS 495 496 This study was supported by the Swiss National Fund for Scientific Research (N. 31003A_152899 497 and N. 31003A_172809 to S.C., N. 31003A_130520, N.31003A_149975 and N.31003A_173087 to 498 A.R.), European Research Council Consolidator Grant N° 311536 (to AR). W.H. and J. Y. are 499 funded by the Singapore Ministry of Education Academic Research through the MOE Research 500 Scholarship Block (RSB) scheme (to W. H.) and the Singapore Ministry of Education under the 501 Research Centres of Excellence program (to J. Y.). We thank Jerôme Bosset and Christoph Bauer 502 (BioImaging Facility, Faculty of Sciences) for help with scanning electron microscopy, the cited 503 colleagues for gifts of reagents, and David Shore and Jean-Claude Martinou for comments on the 504 manuscript. 505 506 AUTHOR CONTRIBUTIONS 507 Conceptualization, S.C., E.V., F.R., D.S.; Methodology, E.V., F.R., D.S., W.H., A.C., A.F., J.S., and 508 S.C.; Validation, E.V., F.R., D.S., W.H., A.C., A.F., J.S., and S.C.; Formal Analysis, E.V., F.R., 509 D.S., W.H., and A.C.; Investigation, E.V., F.R., D.S., W.H., A.C., A.F., J.S., I.M., L.J., and S.C.; 510 Resources, E.V., F.R., D.S., V.D., C.C., S.S., I.M., L.J., and S.C.; Data Curation, E.V., F.R., D.S., 511 W.H., A.C., A.F., J.S. I.M., and L.J.; Writing – Original Draft, S.C., E.V., F.R., D.S., W.H., A.C., 512 A.R., J.Y; Writing – Review & Editing, S.C., E.V., F.R., W.H., A.C., A.F., J.S., V.D., C.C., S.S., A.R., 513 J.Y.; Visualization, S.C., E.V., F.R., D.S., W.H., A.C., A.F.; Supervision, S.C., J. Y. and A.R..; 514 Project Administration, S.C.; Funding Acquisition, S.C., J. Y. and A.R.. 515 516 DECLARATION OF INTERESTS 517 The Authors declare no conflicting interest. 518 519

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

520 FIGURES TITLES AND LEGENDS 521 522 Figure 1. Cingulin KO results in decreased apical membrane stiffness and increased 523 apicolateral membrane tension. 524 (A) Schematic diagram of experimental setup. 525 (B) Representative force-indentation curves of WT, CGN-KO, CGNL1-KO and CGN/CGNL1-double 526 KO MDCK cell lines fitted with Hertz (Spherical) model. 527 (C) Averaged stiffness (Young's modulus) of WT, CGN-KO, CGNL1-KO and CGN/CGNL1-double 528 KO MDCK cell lines. 529 (D-E) Representative SEM images of MDCK WT (D) and CGN-KO (E) cell lines. Scale bars, 10 530 μm. 531 (F) Average lifetimes of the FliptR probe for WT, CGN-KO, CGNL1-KO and CGN/CGNL1-double 532 KO MDCK cell lines. 533 (G) FliptR lifetime changes along the apicolateral junctional membrane of WT, CGN-KO, CGNL1- 534 KO and CGN/CGNL1-double KO MDCK cell lines. Slopes of fitted linear regression curves (dashed 535 lines) are indicated. 536 Bars (C, F, G) show mean±SD. 537 Related Figure S1 shows phenotypic characterization of MDCK KO cells. 538 539 Figure 2. Cingulin is required to maintain membrane tortuosity, and normal accumulation of 540 myosin-2B, actin and ARHGEF18 at TJ. 541 (A-B) Immunofluorescent localization of β-catenin and cingulin in WT cells (A) and occludin in WT 542 and KO lines (B) using MDCK cells grown on Transwell filters. 543 (C) Quantifications of membrane tortuosity (zigzag index) in WT, CGN-KO, CGNL1-KO and 544 CGN/CGNL1-double KO MDCK cell lines. 545 (D-E) Immunofluorescent localization of occludin and GFP-c/mCGN/GFP (D) and quantifications of 546 membrane tortuosity (zigzag index) (E) in CGN-KO MDCK cells rescued with either GFP- 547 canisCGN (cCGN), or GFP-mouseCGN (mCGN) or GFP. Arrows and arrowheads indicate 548 junctional localization and lack of junctional localization, respectively. White lines outline shape of 549 junctions. 550 (F-I) Immunofluorescent localization of NM2A (F), NM2B (G), F-actin (TRITC-phalloidin) (H) and 551 ARHGEF18 (I) in mixed WT+CGN-KO MDCK cultures. Arrows and arrowheads indicating normal 552 versus reduced/undetectable labeling are placed in magnified areas for the red channel and in 553 regular magnification areas in CGN and ZO-1 channels. Continuous and dotted circles indicate 554 normal and reduced, respectively, apical cortical labeling for NM2A and NM2B. 555 (J-L) Immunofluorescent localization of NM2B (J), F-actin (TRITC-phalloidin) (K) and ARHGEF18 556 (L) in CGN-KO cells rescued with full length canis cingulin. 557 (M-R) Immunofluorescent localization (M-N) and line-scans of immunofluorescence signal across 558 junctions (O-P) of γ-actin (M, O) and β-actin (N, P) in junctions from either WT or KO cells within 559 mixed WT-CGN-KO mCCD cultures. Arrows and arrowheads in magnified insets show normal and

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

560 altered junctional labeling, respectively. (Q-R) show the ratio between the full width at half 561 maximum signal intensity (FWHM) for γ-actin (Q) and β-actin (R), ratioed to PLEKHA7 signal 562 (cyan) (N=30 junctions). 563 Data in (C, E, Q, R) are represented as mean±SD. Scale bars= 10 μm. 564 Related Figure S1 shows phenotypic characterization of MDCK and mCCD KO cells. 565 Related Figure S2 shows analysis of zigzag index, NM2A, NM2B, phalloidin, ARHGEF18, g-actin 566 and b-actin in different cell types and experimental conditions. 567 Related Figure S6D shows schematic diagram of interaction of cingulin with ZO-1, g-actin filaments 568 and NM2B. 569 570 Figure 3. Cingulin binding to the ZU5 domain unfolds ZO-1 571 (A-B) Interaction between the GST-CGN(1-70) bait (red) and GFP-tagged fragments of ZO-1 (input 572 preys, green) (A), and between the GST-ZU5 bait and GFP-tagged cingulin preys (B), as shown by 573 IB with anti-GFP. Numbers indicate migration of pre-stained size markers. PonceauS-stained baits 574 are shown below the IB. GST alone is used as a negative control in all GST pulldown experiments. 575 (C) Interaction between the GST-ZU5 bait and HA-tagged paracingulin full length and mutated 576 (CGNL1) preys. 577 (D) Immunofluorescent localization of endogenous CGN (CGN, green), myc-tagged ZO-1 (red) and 578 PLEKHA7 (reference junctional marker, cyan) in ZO-1-KO Eph4 cells rescued either with FL-ZO-1 579 (top) or with a C-terminal truncation of ZO-1 lacking the ZU5 domain (bottom). Arrows and 580 arrowheads indicate normal and reduced/undetected staining, respectively. 581 (E-F) Recruitment of CGN (E) and CGNL1 (F) in condensates of full-length myc-ZO-1-HA (top), but 582 not in condensates of ZO-1 lacking the ZU5 domain (ZO-1-DZU5) (bottom). 583 (G-I) CGN(1-70) promotes ZO-1 unfolding in cells. Junctional recruitment of DbpA in WT Eph4 cells 584 depleted of ZO-2 and overexpressing ether GFP or GFP-CGN(1-70) (G-H) either untreated (G) or 585 treated with blebbistatin (H), and in CGN-KO Eph4 cells treated with either si-control or si-ZO2 (I). 586 Asterisks indicate ZO-2-depleted cells. Arrows and arrowheads indicate normal and 587 reduced/undetected junctional staining, respectively. 588 (J-L). Cingulin (1-70) promotes ZO-1 unfolding in vitro. Immunoblot analysis (J) of full-length ZO-1 589 prey (purified from baculovirus-infected insect cells) in GST pulldowns using GST-DbpA as a bait 590 and increasing amounts of either GFP-CGN(1-70) or GFP as third protein. (K) shows normalization 591 of third protein. (L) Immunoblot analysis of HA-DbpA prey in GST pulldowns using GST-CGN(1-70) 592 and GST-GEF-H1 as baits [47]. 593 Scale bars, 10 μm. 594 Related Figure S1 shows phenotypic characterization of Eph4 KO cells. 595 Related Figure S3 shows binding of cingulin and paracingulin fragments to ZU5, localization of

596 CGN and CGNL1 in ZO-1-KO Eph4 cells, and measurement of the Kd of interaction between GST- 597 CGN(1-70) and ZU5. 598 Related Figure S6 shows schematic diagram of how cingulin binds to and unfolds the ZU5 domain 599 of ZO-1, thus connecting it to actin filaments.

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

600 601 Figure 4. Cingulin promotes the accumulation of a pool of ZO-1 at TJ 602 (A-D) KO of cingulin (A-B) but not paracingulin (C-D) decreases the junctional accumulation of ZO- 603 1 in Eph4 (A, C) and MDCK (B, D) cells. 604 (E-F) Immunofluorescent localization of ZO-1 (ZO-1), CGN rescue constructs full length (CGN-FL), 605 mutated (CGN-Δ1-70) and GFP myc-tagged constructs (myc) and PLEKHA7 (PLEKHA7) in either 606 CGN-KO (E) or WT (F) cultures of Eph4 cells. 607 Arrows and arrowheads indicate normal and reduced/undetected junctional labeling, respectively. 608 Scale bars, 10 μm. 609 (G) Quantification of ZO-1 junctional immunofluorescence signal in either WT or CGN-KO cells 610 after rescue with either GFP-myc or full-length or mutated CGN constructs (E-F). 611 Data are represented as mean±SD. 612 Related Figure S1 shows phenotypic characterization of KO cells. 613 Related Figure S4 shows IF analysis and quantifications of IF labeling for ZO proteins in additional 614 cell lines, and IB analyses. 615 616 Figure 5. Cingulin is required for recruitment of F-actin but not NM2B to ZO-1 condensates. 617 (A-B) CGN (A) and ZO-1 (B) both form round droplets that fuse into larger structures over time (A’- 618 B’), indicating the formation of condensates. 619 (C-E) Labeling for NM2B (upper panels), F-actin (TRITC-phalloidin) (middle panels) and 620 ARHGEF18 (bottom panels) in cell overexpressing either CGN in MDCK WT (C), or ZO-1 in WT 621 (D) or CGN-KO (E) MDCK cells. 622 (F-G) Recruitment of F-actin (TRITC-phalloidin) in condensates of full-length myc-ZO-1-HA (top) 623 and ZO-1 lacking the ZU5 domain (ZO-1-DZU5-HA) (bottom) in WT (F) or CGN-KO (G) MDCK 624 cells. 625 Arrows indicate phase-separated condensates, arrowheads indicate reduced/undetected labeling, 626 double arrowheads indicated redistributed subcortical labeling. 627 (A, B, C-G) Scale bars, 10 μm. (A’, B’) Scale bars, 0.05 μm. 628 (H) Quantification of relative fluorescent intensity of F-actin recruited to ZO-1 condensates in WT 629 and CGN-KO MDCKII (F-G). Data are represented as mean±SD. 630 Related Figure S1 shows phenotypic characterization of MDCK CGN-KO cells. 631 Related Figure S5 shows IF analysis of ZO-1 in CGN and CGNL1 condensates, and NM2B, 632 phalloidin and ARHGEF18 labeling in CGNL1 condensates. 633 Related Figure S6C-D shows schematic diagram of how ZU5 folds back on the ABR in the 634 absence of cingulin, how cingulin binding opens the ZU5 domain, thus allowing the ABR to connect 635 to actin filaments. 636 637 Figure 6. The effects of cingulin KO on membrane shape and cytoskeleton organization. 638 (A-B) Schematic hypothetical organization of cingulin, paracingulin, myosin and actin filaments in 639 the apicolateral junctional region of WT (A) and CGN-KO (B) cells. Selected transmembrane and

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

640 junctional proteins of tight junctions (TJ) and zonula adhaerens (ZA) are shown (see graphical 641 legend). In WT cells g-actin is tethered to the ZO-1-cingulin complex, and this link is lost in CGN- 642 KO cells, resulting in thinning of the branched g-actin network and uncoupling from the TJ 643 membrane, straightening of the TJ membrane, and reduced accumulation of ZO-1. 644 (C) Graphical legend. Note that many additional transmembrane and cytoplasmic junctional 645 proteins of TJ, such as ZO-2 and ZO-3, which form heterodimers with ZO-1 have been omitted 646 from the scheme, for the sake of clarity. ZO-1 is represented in “open” and “closed” forms, 647 depending on the conformation and intramolecular interactions of the ZU5 domain, regulated by 648 interaction with cingulin. In cells that contain ZO-2, ZO-1 is unfolded through dimerization. 649 Related Figure S6 shows an additional scheme detailing the folding/unfolding of soluble and 650 membrane-associated pools of ZO-1, its interactions with TJ transmembrane and other proteins, 651 and cingulin-mediated interaction with NM2B and g-actin.

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

652 STAR ★Methods

653 Resource Availability 654 655 Lead Contact 656 Further information and requests for resources and reagents should be directed to and will be 657 fulfilled by the Lead Contact, Sandra Citi ([email protected]). 658 659 Materials Availability 660 Reagents generated in this study will be made available on request, but we may require a 661 payment for shipping and a completed Materials Transfer Agreement. 662 663 Data and Code Availability 664 This study did not generate any unique datasets or code. 665

666 Experimental Model and Subject Details

667 Eph4 (mouse mammary epithelial cell line) WT and ZO-1-KO cells, MDCK (Madin-Darby 668 Canine Kidney – II), mCCD (mouse cortical collecting duct epithelial cell line) and HEK293T

669 cells were cultured at 37°C, 5% CO2 in DMEM containing 10% or 20% FBS (for mCCD). For 670 Eph4, MDCK and mCCD culture media were supplemented with 1% non-essential amino 671 acids, 100 units/ml penicillin and 100 μg/ml streptomycin [25, 63, 78]. The sex of Eph4 and 672 MDCK lines was female. For other lines, sex was not listed in the information available to us. 673 We trusted the providers of the cells for their authentication. 674 675 Cell lines KO for cingulin (CGN) and paracingulin (CGNL1) were generated using 676 CRISPR/Cas9 gene editing technology, designing guide-RNA (gRNA) using the Zhang Lab 677 CRISPR design tool, targeting exons which are present in all major transcripts of CGN and 678 CGNL1. For mouse CGN and CGNL1 (mCGN, mCGNL1) the target sequences for the 679 CRISPR/Cas9, to generate Eph4 and mCCD KO lines, were selected in exon 2 (Key 680 Resource Table). For canine CGN and CGNL1 (cCGN, cCGNL1) the target sequences for 681 the CRISPR/Cas9 for MDCK cells were selected in exon 1 (Key Resource Table). The 682 gRNAs were subcloned into the BbsI site of pSpCas9(BB)-2A-GFP (PX458) CRISPR 683 plasmid. Cells were transfected using Lipofectamine 2000. At 48h post-transfection, single 684 cells were sorted (using a Beckman Coulter MoFlo Astrios sorter, Flow Cytometry Service, 685 University of Geneva Medical School) into 96-well tissue culture plates. Single clones were 686 further amplified and screened for the KO using immunoblot and immunofluorescence 687 analyses, based on which two-three CGN-KO and CGNL1-KO clones (one CGNL1-KO clone 688 for Eph4) were selected and verified by sequencing. For sequencing, genomic DNA was 689 extracted using DNeasy Blood & Tissue kit, the genomic loci of CGN and CGNL1 (around 690 500 bp upstream and 600 downstream of target sequence) were amplified by PCR using

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

691 specific primers (Key Resource Table) and subcloned into pBluescript II KS(+) using 692 restriction enzymes SacII and XhoI for mCGN, XbaI and EcoRI for mCGNL1 and cCGN, 693 BamHI and SalI for cCGNL1. The T7 primer was used for sequencing inserts for genotyping.

694 To generate CGN/CGNL1 double-KO MDCK clones, CGN-KO (to generate clone 21D3) or 695 CGNL1-KO (to generate clone 11C9) clones were transfected with above-mentioned CGNL1- 696 KO or CGN-KO CRISPR/Cas9 constructs, respectively, sorted, amplified and screened as 697 described above for single KO.

698 To generate MDCK CGNL1-KO cells stably expressing YFP-myc, cells were transfected 699 using Lipofectamine 2000. At 48h post-transfection, single cells were sorted into 96-well 700 tissue culture plates and were grown in the presence of 200 µg/ml Hygromycin B prior to 701 clone amplification.

702

703 Method Details 704 705 Antibodies and immunofluorescence

706 Antibodies are described in Key Resources Table. 707 For immunofluorescence (IF) cells were cultured either on glass coverslips in 24-well plates 708 for 3 days seeded at a density of 1-2 x 105 cells/well (Figure 3; Figure 4; Figure S1; Figure S2 709 (B, C, D, G, H, I, J); Figure S3 (H, I); Figure S4) or on 24-mm Transwell filters for 5 days 710 seeded at a density of 5 x 105 cells (Figure 2; Figure S2 (E, F, K, L)).

711 IF for cells grown on coverslips was carried out by washing cells 2x with cold PBS, fixing in 712 methanol at -20° for 10 min, washing 3x5 min with PBS, incubating with primary antibody 713 (either at RT for 1 h, or for 16h at 4°C), followed by washing 3x with PBS, incubating with 714 secondary antibody (30 min at 37°C in a humidified chamber), washing 3x with PBS, and 715 mounting either with Vectashield with DAPI or Fluoromount-G. 716 For IF of actins (F-actin-phalloidin, b-actin, g-actin) in Eph4 cells, cells were fixed in room 717 temperature 1% PFA for 5 min, followed by rinsing 2x with PBS and incubating with methanol 718 at -20°C for 5 min, followed by gradual rehydration in PBS (3x replacing 50% of volume with 719 PBS), 2x washes in PBS [13, 14], and mounting with Fluoromount-G. 720 For IF of DbpA, Eph4 cells were permeabilized, fixed and labelled as described in [25]. 721 IF analysis of CGN, β-catenin and occludin in MDCK cells grown on Transwell, cells were 722 washed 2x with cold PBS, fixed in methanol for 16h at −20°C), followed by 1-min treatment 723 with acetone (−20°C). Filters were excised manually using a razor blade and hydrated in IMF

724 buffer (0.1% Triton X-100, 0.15 M NaCl, 5 mM EDTA, 20 mM HEPES, pH 7.5, 0.02% NaN3 as 725 preservative). 726 For IF analysis of NM2A, NM2B, F-actin-phalloidin and ARHGEF18 in MDCK cells grown on 727 Transwells, cells were washed with PBS containing Ca2+ and Mg2+ (PBS++), fixed in 1% PFA 728 pre-warmed at 30°C for 7 min, followed by rinsing 2x with PBS++ and incubating with

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

729 methanol at -20°C for 5 min, followed by gradual rehydration in PBS++ (3x replacing 50% of 730 volume with PBS), and 2x washes in PBS++ [13, 14]. Cells were permeabilized with 0.2% of 731 Tx100 in PBS++ 15min at RT and saturated 30 min with 2% of BSA in PBS++. 732 For mCCD cells grown on Transwells, fixation for IF of NM2A, NM2B and actins was carried 733 out using the same protocol above, except that final permeabilization and blocking steps were 734 carried out together, using PBS++ containing 0.2% Tx100 and 2% BSA (30 min at RT). 735 Incubation with primary antibodies was carried out for 16h at 4°C (in a humidified chamber) 736 and with secondary antibodies for 1-2h at RT. The filters were placed on glass slides with the 737 cells facing up and were mounted with Fluoromount-G. 738 Slides were imaged on a Zeiss LSM800 confocal microscope using a Plan-Apochromat 739 63x/1.40 oil objective at a resolution of 1024x1024 px or on an upright microscope Leica 740 DM4B (Figure S1 (E-G, K-L); Figure S2 (I)) using 63x oil objective at a resolution of 741 2048x2048 px; the pixel size=0.10 μm. Single confocal plane images are shown in Figure 2D; 742 Figure 3I; Figure 4; Figure S1 (M, Q); Figure S2 (B-D), Figure S4. Maximum intensity 743 projections of z-stack images are shown in Figure 2A (6 confocal planes over 8 µm, step 744 size=1.6 µm); Figure 2B (5 confocal planes over 1 µm, step size=0.25 µm); Figure 2 (F-L) and 745 Figure S2 (E-F) (4 confocal planes over 1.2 µm, step size=0.4 µm); Figures 2(M-N) and 746 Figure S2 (K-L) (5-6 confocal planes over 1-1.3 µm, step size=0.25 µm); Figure 3D (7 747 confocal planes over 6 µm, step size=1 µm; Figure 3 (G-H) (4 confocal planes over 3 µm, 748 step size=1 µm); Figure S2 (G-H, J) (5 confocal planes over 1.6 µm, step size=0.35 µm) 749 Figure S3 (H, I) (4-5 confocal planes over 1.1-1.4 µm, step size=0.4 µm. Images were 750 extracted from .lif, .lsm or .czi files using ImageJ, adjusted and cropped using Adobe 751 Photoshop, and assembled in Adobe Illustrator figures. 752 753 Measurement of cell proliferation 754 755 To measure cell proliferation, cells were plated in 12-well plate (75 000 cells/well) in duplicate, 756 and cells in one well were trypsinized and counted in a hemocytometer each day, for 6 days.

757 758 Plasmids

759 Full length (FL) mCGN (S2407) and mutant lacking ZIM domain (D1-70) (S2408) were 760 generated by PCR using appropriate oligonucleotides containing Kozak sequence and 761 subcloning into XhoI-KpnI sites of pcDNA3.1(-)/MycHis C. GFP-myc (S1166) construct was 762 generated by subcloning of GFP-myc into pcDNA3.1(-)/MycHis A. The FL mCGN (S2363) 763 and mCGNL1 (S2386) constructs were generated by PCR and subcloning into NotI-ClaI sites 764 of pTRE2Hyg containing GFP and myc tags, downstream of GFP and upstream of myc. 765 GFP-tagged human (h) CGN FL (S2508) (G-CGN-FL) and the mutant lacking ZIM domain (G- 766 CGN-D1-70) (S2509) were generated by PCR and subcloning into NotI-Acc65I sites of GFP- 767 myc (S1166) downstream of GFP.

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

768 FL cCGN (S1136) and FL cCGNL1 (S2432) HA-tagged in C-terminal were generated by PCR 769 using appropriate oligonucleotides containing Kozak sequence and subcloning into BamHI- 770 KpnI sites of pcDNA3.1(-)/MycHis A and BamHI-NotI sites of pcDNA3.1/Zeo(+), respectively. 771 Chimeras of cCGN and cCGNL1 HA-tagged in C-terminal (S2433: aa 1-579 from CGNL1 and 772 339-1190 from CGN; S2434: aa 1-338 from CGN and 580-1295 from CGNL1) were obtained 773 by triple PCR using appropriate oligonucleotides containing Kozak sequence and overlapping 774 region for fusion between the head and rod-tail sequences, before insertion into BamHI-NotI 775 sites of pcDNA3.1/Zeo(+). 776 FL cCGN (S2294=S1052) was described previously [79]. YFP-tagged cCGNL1 (S1137) and 777 YFP-myc (S1152) were described previously [34]. GST (S1851) and GFP (S1821) tagged 778 CGN (1-70aa) were constructed by PCR on cCGN and subcloned either into pGEX4T1 779 (EcoRI-NotI) or into GFP-containing pTRE2-hyg (S1210) [80] downstream of GFP (BglII- 780 NotI), respectively. The FL myc-hZO-1-HA (S1947) construct; GFP-tagged constructs of 781 fragments within the C-terminal region of hZO-1 (residues 888-1619 (S1807); 1150-1619 782 (S1808); 888-1748 (S1809); 1150-1748 (S1810); 1619-1748 (S1811)); and FL hZO-1 for 783 expression in insect cells were described previously [25]. GFP-tagged constructs of 784 fragments within the C-terminal region of hZO-1 (residues 1698-1748 (S1874)) and (residues 785 1550-1650 (S1875)) were generated by PCR and subcloned downstream of GFP into NotI- 786 KpnI sites of pCDNA3.1(+). The construct of myc-hZO-1-HA (S2161) lacking the ZU5 domain 787 (ZO1-DZU5, residues 1-1619) and GFP tagged mZO-1 (S2474) were described previously 788 [78]. GST-DbpA, CFP-HA and HA-DbpA were previously described [47]. The DbpA-binding 789 construct of GST fused to the C1 domain of GEF-H1 was a kind gift from K. Matter [48]. FL 790 hCGN (S2411), FL hCGNL1 (S2442) (CGNL1-FL-HA) and mutant lacking ZIM domain 791 (S2510) (CGNL1-D1-80-HA) were generated by PCR using appropriate oligonucleotides 792 containing Kozak sequence and subcloning upstream of HA into modified HA containing 793 pCDNA3.1 (BamHI-XbaI sites for S2411, BamHI-NotI sites for S2442 and S2510). 794 GST-tagged hZU5 (S1789), ZIM domain – containing fragments of hCGN (1-121aa) (S96) 795 and (1-353aa) (S97) and hCGNL1 (1-119aa) (S1024) were generated by PCR and 796 subcloning into pGEX4T1 (EcoRI-XhoI sites for S1789, S96, S97; BamHI-EcoRI sites for 797 S1024). 798 799 Recombinant Protein Expression, Glutathione S-Transferase (GST) Pulldown, and 800 Supernatant Depletion Assays

801 GST fusion proteins were expressed in BL21 bacteria and purified by affinity chromatography 802 on Glutathione Sepharose [47] or magnetic beads. Pulldowns were carried out using lysates 803 either from insect cells expressing FL ZO-1, or from lysates of HEK293T cells expressing 804 either GFP-tagged ZO-1 fragments; GFP-tagged CGN constructs, HA-tagged DbpA and 805 CGNL1 constructs (prey) as previously described [78, 81]. Lysates containing prey proteins 806 were normalized, and equivalent amounts of prey proteins were added to 5 µg of baits (GST

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

807 tagged DbpA, CGN fragments (residues: 1-70, 1-121, 1-353), CGNL1 (1-119), ZU-5 domain 808 of ZO-1, C1 domain of GEF-H1). 809

810 To determine the Kd of interaction between the ZU5 domain of ZO-1 (1619-1748=Cter) and 811 the CGN1-70 fusion protein, we used a quantitative GST-pulldown Supernatant Depletion 812 Assay [25, 47]. Increasing volumes (1, 2, 5, 10, 12.5, 15, 17.5, and 20 μl) of Glutathione 813 Sepharose beads pre-loaded with GST-CGN(1-70) bait (0.666 μM), were added to prey 814 protein (GFP-ZO-1-C-ter), in a total volume of 0.5 ml CO-IP buffer. As a negative control, 815 beads pre-incubated with CO-IP buffer (1, 2, 5, 10 μl) were added to the prey. After 16h 816 incubation at 4°C, beads were pelleted at 16,000 x g, and the prey proteins remaining in the 817 supernatant were analyzed by SDS-PAGE and immunoblotting. GST pulldown experiments to 818 assess the effect of CGN(1-70) on the interaction between DbpA and full-length ZO-1 were 819 carried out by incubating GST-DbpA (5 μg) with insect cell lysate containing full-length ZO-1, 820 in the presence of increasing volumes of normalized HEK293T cells lysates containing either 821 GFP-CGN(1-70) or GFP, this latter as a negative control. Although it is unlikely that sufficient 822 concentrations of additional interacting partners of baits and preys are present in the GST 823 pulldown assays and might influence some of the results, this possibility cannot be formally 824 excluded. Concentration of recombinant proteins was determined by densitometric analysis of 825 Coomassie-stained SDS gels, compared to a BSA standard curve. 826

827 Transfection, siRNA-mediated ZO-2 depletion and exogenous expression of proteins

828 For transfections for rescue experiments, cells were plated on glass round 12 mm coverslips 829 in 24-well plate (100 000 cells/well), transfected next day using Lipofectamine 2000 or 830 jetOPTIMUS DNA transfection reagent and fixed for IF 72 h post-transfection.

831 For transfections of cells grown on Transwells cells were transfected 24h after plating using 832 jetOPTIMUS according to the manufacturer’s protocol and fixed for IF 4 days after 833 transfection.

834 For siRNA-mediated ZO-2 depletion (target sequence in Key Resource Table) cells were 835 transfected with Lipofectamine RNAiMAX 24h after plating, using siRNA negative control as 836 negative control. 837 838 For siRNA and DNA co-transfections, Eph4 cells were transfected next day after plating with 839 the mix of siRNA and DNA using Lipofectamine 2000, 48h post-transfection cells were treated 840 with 50 μM blebbistatin for 16 h and then fixed [25]. 841 842 HEK cells were plated in 10 cm dish (2x106 cells/dish), transfected next day using 843 Lipofectamine 2000 and lysed 48h post-transfection. 844 845 Overexpression and condensate analysis, live microscopy

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

846 847 To study biomolecular condensates of CGN, ZO-1 and CGNL1 (Figure 5 (A, B) and Figure 848 S5A respectively), 5x105 MDCKII cells were seeded onto 35-mm glass bottom culture dishes. 849 24h after plating cells were transfected with 8 µg of DNA per dish using JetOptimus®. GFP- 850 CGN or GFP-CGNL1 or GFP-ZO1 condensates were analyzed 48h post transfection. Live 851 imaging was performed on a Zeiss LSM780 confocal microscope using argon laser with the 852 488 nm excitation and PlanApo N60x oil /1.4 objective. Single plane images were recorded 853 every 2s for 100s at a resolution of 512x512 px. 854 To study recruitment of client proteins by condensates, 1x105 MDCKII cells were seeded onto 855 glass coverslips placed in 24 well plates. 24h after plating cells were transfected with 1,5 µg 856 of DNA per dish using JetOptimus®. 48h post transfection cells were fixed using protocol for 857 NM2A for MDCKII on Transwells, described in immunofluorescence section. Images were 858 taken using Leica DM4B (Figures 3 (E-F); 5 (C-E); S5 (B-D)) or Zeiss LSM800 for phalloidin 859 (Figures 5 (C-G); S5 (C-D)). Maximum intensity projections of four confocal planes 860 representing 1.2 µm final depth are shown for phalloidin. Imaging settings and treatment were 861 as described in immunofluorescence section. 862 863 Scanning Electron Microscopy

864 Confluent monolayers were rinsed with PBS and fixed with 2% glutaraldehyde for 20 min at 865 RT, after washing with 0,1M sodium cacodylate buffer cells were post-fixed with 1% osmium 866 tetraoxide in 0,1M sodium cacodylate buffer for 20 min at RT. Cells were rinsed with sodium 867 cacodylate buffer and water and then dehydrated with a series of ethanol washes (30, 50, 70, 868 95 and 100%), followed by acetone washes. Samples were dried using critical point dryer 869 (Leica EM CPD030), coated with gold and observed under the JEOL-6510LV low vacuum 870 scanning electron microscope.

871 Immunoblotting

872 Cell lysates were obtained using RIPA buffer containing 1X cOmplete or Pierce protease 873 inhibitor and immonoblotting was performed as previously described [25, Vasileva, 2017 874 #4538]. β-tubulin was used for protein loading normalization in immunoblots. Numbers on the 875 left of immunoblots indicate migration of molecular size markers (kDa). 876

877 Atomic force microscopy indentation measurements

878 The AFM-based indentation measurements were carried out using a commercial AFM 879 (Dimension FastScan, Icon Scanner, Bruker). A polystyrene bead (5 µm radius, Invitrogen) 880 was stuck on the tipless silicon nitride cantilever (MLCT-O10-E, Bruker) by epoxy fix. The 881 spring constant of the home-made cantilevers, calibrated each time before measurement by 882 thermal fluctuation method, were in the range of 0.10~0.15 N m-1.

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

883 All AFM indentation measurements were realized in cell culture medium at room temperature. 884 The cells were cultured in 60 mm petri dish for 36 h in incubator until forming monolayers 885 (confluency > 80%). In a typical experiment, the cantilever was brought to the cell layer with 886 the constant speed of 1 µm s-1 until reaching the maximum contact force of 5 nN, where the 887 maximum indentation distance of cells was in the range of 0.5-1.5 µm. Then the cantilever 888 was retracted and moved to another spot for the next cycle. A box pattern containing 100 889 spots in 40 µm × 40 µm region was set and typically 5-10 such regions were randomly 890 selected in each measurement to obtain the averaged stiffness of the cell.

891 The force-indentation traces were analyzed to obtain the Young’s modulus of the cells by 892 using the NanoScope Analysis program. After baseline correction and contact point 893 estimation, the approaching force-indentation curve was fitted with the Hertz (Spherical) 894 model (eq. 1) in the contact force range from 0.5 nN to 4.5 nN. Constant parameters and data 895 range were chosen to minimize the bias for different cell types.

896 897 !(#) = & ( .#'/0 [45] ' ()*+,) √

898 where F is the force of the cantilever, x is the indentation distance of the cell pressed by the 899 cantilever, E is the Young’s modulus of the cell layer, r is the radius of the spherical indenter, 900 and υ is the Poisson ratio. The Poisson ratio of cell is normally in the range of 0.3-0.5. We 901 chose υ=0.5 in all calculations.

902 Lifetime measurements

903 Lifetime imaging was carried out on MDCK cells seeded on 35-mm glass bottom dishes at a 904 density of 3x105 cells/dish and grown for 4 days. 1,5 μl of FliptR solution was added to each 905 dish prior imaging. FLIM imaging was performed as described [46]. 906 FLIM imaging was performed using a Nikon Eclipse Ti A1R microscope equipped with a Time 907 Correlated Single-Photon Counting module from PicoQuant [82]. Excitation was performed 908 using a pulsed 485nm laser (PicoQuant, LDH-D-C-485) operating at 20 MHz, and emission 909 signal was collected through a bandpass 600/50nm filter using a gated PMA hybrid 40 910 detector and a TimeHarp 260 PICO board (PicoQuant). 911

912 Quantification and Statistical Analysis

913 Data processing and analysis were performed in GraphPad Prism 8. All experiments were 914 carried out at least in duplicate, and data are shown either as dot-plots, as histograms or as 915 line-graph (with mean and standard deviation indicated). Statistical significance of quantitative 916 data was determined by unpaired two-tailed Student’s t-test (when comparing two sets of 917 data), ordinary one- or two-way ANOVA with Tukey’s post hoc test (for multiple comparisons), 918 (ns=not significant difference)=p>0.05, *p≤0.5, **p≤0.01, ***p≤0.001,**** p≤0.0001). 919

24 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

920 Analysis of immunofluorescence data 921 For the quantification of junctional immunofluorescent signal (Figure 4; Figure S3; Figure S4), 922 pixel intensity for each channel was measured in the selected junctional area using the 923 polyhedral tool of ImageJ, and the averaged background signal of the image was subtracted. 924 Relative intensity signal was expressed as a ratio between the signal of protein of interest and 925 an internal junctional reference (either PLEKHA7 or occludin, or PLEKHA6). 9-45 junctional 926 segments were analyzed. 927 For the measurement of the zig-zag index (L(TJ)/L(St)) (Figure 2 (C, E); Figure S2A): ratio 928 between actual length of bicellular junction and the distance between two vertexes), we used 929 the method described in [23], and measured the length of the TJ (L(TJ)) by using the 930 freehand line trace in ImageJ, and the straight length of junction (L(St)) by using a straight 931 line between vertexes. 45-240 junctions were analyzed. 932 To analyze intensity and distribution of b- and g-actins the line across junction was drawn 933 using Straight line tool in ImageJ and line scan graphs were generated by plotting profile. Full 934 width at half-maximum (FWHM) pixel intensity was determined for actins and PLEKHA7, and 935 the actin/PLEKHA7 ratio was calculated. 30 junctions were analyzed. 936 Each dot of dot-plot graphs represents one measurement, and data are shown in arbitrary 937 units (a.u.). 938 939 Analysis of condensates 940 To analyze recruitment of F-actin (TRITC-Phalloidin) into ZO-1 condensates (Figure 5) ZO-1 941 condensates were encircled using Polygon selection tool in ImageJ, relative fluorescent 942 intensities of ZO-1 and F-actin were measured, and F-actin/ZO-1 ratio was calculated after 943 background signal subtraction. 944 Each dot of dot-plot graphs represents one measurement, and relative intensities are shown 945 in arbitrary units (a.u.). 30 droplets were analyzed. 946 947 Analysis of immunoblotting data 948 For the quantification of immunoblot signals (Figure S3), densitometric analysis was carried 949 out on band signals from at least 3 separate experiments. Each time the signal was 950 normalized to tubulin, used as reference for total protein loading, and expressed as 951 percentage taking the strongest signal as 100%. 952 953 Analysis of atomic force microscopy indentation measurements 954 Each dot of dot-plot graphs represents one Young's modulus value from one indentation 955 curve. Note that the indentation is site-by-site, not cell-by-cell, representing the local stiffness 956 of the cells. 957 958 Analysis of lifetime measurements

25 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

959 SymPhoTime 64 software (PicoQuant) was used to fit fluorescence decay data (from full 960 images or regions of interest) to a dual exponential model after deconvolution for the 961 instrument response function (measured using the backscattered emission light of a 1 µM 962 fluorescein solution with 4M KI). Data was expressed as means ± standard deviation of the 963 mean. Slopes were determined based on simple linear regression model. 964

965

26 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

966 REFERENCES 967 968 1. Anderson, J.M., and Van Itallie, C.M. (2009). Physiology and function of the tight 969 junction. Cold Spring Harb. Perspect. Biol. 1:a002584, 1-16. 970 2. Zihni, C., Mills, C., Matter, K., and Balda, M.S. (2016). Tight junctions: from simple 971 barriers to multifunctional molecular gates. Nat Rev Mol Cell Biol 17, 564-580. 972 3. Takeichi, M. (2014). Dynamic contacts: rearranging adherens junctions to drive 973 epithelial remodelling. Nat Rev Mol Cell Biol 15, 397-410. 974 4. Meng, W., and Takeichi, M. (2009). Adherens junction: molecular architecture and 975 regulation. Cold Spring Harbor Perspectives in Biology, 976 doi:10.1101/cshperspect.a002899. 977 5. Fanning, A.S., and Anderson, J.M. (2009). Zonula occludens-1 and -2 are cytosolic 978 scaffolds that regulate the assembly of cellular junctions. Ann N Y Acad Sci 1165, 979 113-120. 980 6. Vasileva, E., and Citi, S. (2018). The role of microtubules in the regulation of 981 epithelial junctions. Tissue barriers, 1-20. 982 7. Buckley, A., and Turner, J.R. (2018). Cell Biology of Tight Junction Barrier Regulation 983 and Mucosal Disease. Cold Spring Harb Perspect Biol 10, doi: 984 10.1101/cshperspect.a029314. 985 8. Ladoux, B., Nelson, W.J., Yan, J., and Mege, R.M. (2015). The mechanotransduction 986 machinery at work at adherens junctions. Integr Biol (Camb) 7, 1109-1119. 987 9. Citi, S. (2019). The mechanobiology of tight junctions. Biophys Rev 11, 783-793. 988 10. Citi, S., Guerrera, D., Spadaro, D., and Shah, J. (2014). Epithelial junctions and Rho 989 family GTPases: the zonular signalosome. Small GTPases 5, 1-15. 990 11. Bruckner, B.R., and Janshoff, A. (2015). Elastic properties of epithelial cells probed 991 by atomic force microscopy. Biochim Biophys Acta 1853, 3075-3082. 992 12. Heuze, M.L., Sankara Narayana, G.H.N., D'Alessandro, J., Cellerin, V., Dang, T., 993 Williams, D.S., Van Hest, J.C., Marcq, P., Mege, R.M., and Ladoux, B. (2019). 994 Myosin II isoforms play distinct roles in adherens junction biogenesis. Elife 8, e46599. 995 13. Dugina, V., Zwaenepoel, I., Gabbiani, G., Clement, S., and Chaponnier, C. (2009). 996 Beta and gamma-cytoplasmic actins display distinct distribution and functional 997 diversity. J Cell Sci 122, 2980-2988. 998 14. Baranwal, S., Naydenov, N.G., Harris, G., Dugina, V., Morgan, K.G., Chaponnier, C., 999 and Ivanov, A.I. (2012). Nonredundant roles of cytoplasmic beta- and gamma-actin 1000 isoforms in regulation of epithelial apical junctions. Mol Biol Cell 23, 3542-3553. 1001 15. Fanning, A.S., Jameson, B.J., Jesaitis, L.A., and Anderson, J.M. (1998). The Tight 1002 Junction Protein ZO-1 Establishes a Link between the Transmembrane Protein 1003 Occludin and the Actin Cytoskeleton. J Biol Chem 273, 29745-29753. 1004 16. Van Itallie, C.M., Tietgens, A.J., and Anderson, J.M. (2017). Visualizing the dynamic 1005 coupling of claudin strands to the actin cytoskeleton through ZO-1. Mol Biol Cell 28, 1006 524-534. 1007 17. Fanning, A.S., Van Itallie, C.M., and Anderson, J.M. (2012). Zonula occludens-1 and 1008 -2 regulate apical cell structure and the zonula adherens cytoskeleton in polarized 1009 epithelia. Mol Biol Cell 23, 577-590. 1010 18. Odenwald, M.A., Choi, W., Kuo, W.T., Singh, G., Sailer, A., Wang, Y., Shen, L., 1011 Fanning, A.S., and Turner, J.R. (2018). The scaffolding protein ZO-1 coordinates 1012 actomyosin and epithelial apical specializations in vitro and in vivo. J Biol Chem 1013 293(45), 17317-17335. 1014 19. Choi, W., Acharya, B.R., Peyret, G., Fardin, M.A., Mege, R.M., Ladoux, B., Yap, A.S., 1015 Fanning, A.S., and Peifer, M. (2016). Remodeling the zonula adherens in response to 1016 tension and the role of afadin in this response. J Cell Biol 213, 243-260. 1017 20. Van Itallie, C.M., Fanning, A.S., Bridges, A., and Anderson, J.M. (2009). ZO-1 1018 stabilizes the tight junction solute barrier through coupling to the perijunctional 1019 cytoskeleton. Mol Biol Cell 20, 3930-3940. 1020 21. Yamazaki, Y., Umeda, K., Wada, M., Nada, S., Okada, M., Tsukita, S., and Tsukita, 1021 S. (2008). ZO-1- and ZO-2-dependent integration of myosin-2 to epithelial zonula 1022 adherens. Mol Biol Cell 19, 3801-3811. 1023 22. Tornavaca, O., Chia, M., Dufton, N., Almagro, L.O., Conway, D.E., Randi, A.M., 1024 Schwartz, M.A., Matter, K., and Balda, M.S. (2015). ZO-1 controls endothelial

27 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1025 adherens junctions, cell-cell tension, angiogenesis, and barrier formation. J Cell Biol 1026 208, 821-838. 1027 23. Tokuda, S., Higashi, T., and Furuse, M. (2014). ZO-1 Knockout by TALEN-Mediated 1028 Gene Targeting in MDCK Cells: Involvement of ZO-1 in the Regulation of 1029 Cytoskeleton and Cell Shape. PLoS One 9, e104994. 1030 24. Cartagena-Rivera, A.X., Van Itallie, C.M., Anderson, J.M., and Chadwick, R.S. 1031 (2017). Apical surface supracellular mechanical properties in polarized epithelium 1032 using noninvasive acoustic force spectroscopy. Nature communications 8, 1030. 1033 25. Spadaro, D., Le, S., Laroche, T., Mean, I., Jond, L., Yan, J., and Citi, S. (2017). 1034 Tension-Dependent Stretching Activates ZO-1 to Control the Junctional Localization 1035 of Its Interactors. Curr Biol 27, 3783-3795 e3788. 1036 26. Beutel, O., Maraspini, R., Pombo-Garcia, K., Martin-Lemaitre, C., and Honigmann, A. 1037 (2019). Phase Separation of Zonula Occludens Proteins Drives Formation of Tight 1038 Junctions. Cell 179, 923-936 e911. 1039 27. Schwayer, C., Shamipour, S., Pranjic-Ferscha, K., Schauer, A., Balda, M., Tada, M., 1040 Matter, K., and Heisenberg, C.P. (2019). Mechanosensation of Tight Junctions 1041 Depends on ZO-1 Phase Separation and Flow. Cell 179, 937-952 e918. 1042 28. Yu, D., Marchiando, A., Weber , C., Raleigh , D., Wang, Y., Shen, L., and Turner, J. 1043 (2010). MLCK-dependent exchange and actin binding region-dependent anchoring of 1044 ZO-1 regulate tight junction barrier function. Proc. natl. Acad. Sci. USA 107, 8237- 1045 8241. 1046 29. Citi, S., Sabanay, H., Jakes, R., Geiger, B., and Kendrick-Jones, J. (1988). Cingulin, 1047 a new peripheral component of tight junctions. Nature 333, 272-276. 1048 30. Ohnishi, H., Nakahara, T., Furuse, K., Sasaki, H., Tsukita, S., and Furuse, M. (2004). 1049 JACOP, a novel plaque protein localizing at the apical junctional complex with 1050 sequence similarity to cingulin. J Biol Chem 279, 46014-46022. 1051 31. Guillemot, L., and Citi, S. (2006). Cingulin, a cytoskeleton-associated protein of the 1052 tight junction. Tight Junctions, 54-63. 1053 32. Cordenonsi, M., D'Atri, F., Hammar, E., Parry, D.A., Kendrick-Jones, J., Shore, D., 1054 and Citi, S. (1999). Cingulin Contains Globular and Coiled-coil Domains and Interacts 1055 with ZO-1, ZO-2, ZO-3, and Myosin. J Cell Biol 147, 1569-1582. 1056 33. D'Atri, F., and Citi, S. (2001). Cingulin interacts with F-actin in vitro. FEBS Lett 507, 1057 21-24. 1058 34. Paschoud, S., Guillemot, L., and Citi, S. (2012). Distinct domains of paracingulin are 1059 involved in its targeting to the actin cytoskeleton and regulation of apical junction 1060 assembly. Journal of Biological Chemistry 287, 13159-13169. 1061 35. Yano, T., Torisawa, T., Oiwa, K., and Tsukita, S. (2018). AMPK-dependent 1062 phosphorylation of cingulin reversibly regulates its binding to actin filaments and 1063 microtubules. Sci Rep 8, 15550. 1064 36. Aijaz, S., D'Atri, F., Citi, S., Balda, M.S., and Matter, K. (2005). Binding of GEF-H1 to 1065 the tight junction-associated adaptor cingulin results in inhibition of Rho signaling and 1066 G1/S phase transition. Dev Cell 8, 777-786. 1067 37. Guillemot, L., Guerrera, D., Spadaro, D., Tapia, R., Jond, L., and Citi, S. (2014). 1068 MgcRacGAP interacts with cingulin and paracingulin to regulate Rac1 activation and 1069 development of the tight junction barrier during epithelial junction assembly. Mol Biol 1070 Cell 25, 1995-2005. 1071 38. Guillemot, L., Paschoud, S., Jond, L., Foglia, A., and Citi, S. (2008). Paracingulin 1072 regulates the activity of Rac1 and RhoA GTPases by recruiting Tiam1 and GEF-H1 to 1073 epithelial junctions. Molecular biology of the cell 19, 4442-4453. 1074 39. Terry, S.J., Zihni, C., Elbediwy, A., Vitiello, E., Leefa Chong San, I.V., Balda, M.S., 1075 and Matter, K. (2011). Spatially restricted activation of RhoA signalling at epithelial 1076 junctions by p114RhoGEF drives junction formation and morphogenesis. Nat Cell 1077 Biol 13, 159-166. 1078 40. D'Atri, F., Nadalutti, F., and Citi, S. (2002). Evidence for a functional interaction 1079 between cingulin and ZO-1 in cultured cells. Journal of Biological Chemistry 277, 1080 27757-27764. 1081 41. Umeda, K., Matsui, T., Nakayama, M., Furuse, K., Sasaki, H., Furuse, M., and 1082 Tsukita, S. (2004). Establishment and characterization of cultured epithelial cells 1083 lacking expression of ZO-1. J Biol Chem 279, 44785-44794.

28 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1084 42. Pulimeno, P., Paschoud, S., and Citi, S. (2011). A role for ZO-1 and PLEKHA7 in 1085 recruiting paracingulin to tight and adherens junctions of epithelial cells. Journal of 1086 Biological Chemistry 286, 16743-16750. 1087 43. Citi, S., Pulimeno, P., and Paschoud, S. (2012). Cingulin, paracingulin, and 1088 PLEKHA7: signaling and cytoskeletal adaptors at the apical junctional complex. 1089 Annals of the New York Academy of Sciences 1257, 125-132. 1090 44. Van Itallie, C.M., Aponte, A., Tietgens, A.J., Gucek, M., Fredriksson, K., and 1091 Anderson, J.M. (2013). The N and C termini of ZO-1 are surrounded by distinct 1092 proteins and functional protein networks. J Biol Chem 288, 13775-13788. 1093 45. Harris, A.R., and Charras, G.T. (2011). Experimental validation of atomic force 1094 microscopy-based cell elasticity measurements. Nanotechnology 22, 345102. 1095 46. Colom, A., Derivery, E., Soleimanpour, S., Tomba, C., Molin, M.D., Sakai, N., 1096 Gonzalez-Gaitan, M., Matile, S., and Roux, A. (2018). A fluorescent membrane 1097 tension probe. Nat Chem 10, 1118-1125. 1098 47. Spadaro, D., Tapia, R., Jond, L., Sudol, M., Fanning, A.S., and Citi, S. (2014). ZO 1099 Proteins Redundantly Regulate the Transcription Factor DbpA/ZONAB. J Biol Chem 1100 289, 22500-22511. 1101 48. Nie, M., Aijaz, S., Leefa Chong San, I.V., Balda, M.S., and Matter, K. (2009). The Y- 1102 box factor ZONAB/DbpA associates with GEF-H1/Lfc and mediates Rho-stimulated 1103 transcription. EMBO Rep 10, 1125-1131. 1104 49. Paschoud, S., Yu, D., Pulimeno, P., Jond, L., Turner, J.R., and Citi, S. (2011). 1105 Cingulin and paracingulin show similar dynamic behaviour, but are recruited 1106 independently to junctions. Molecular membrane biology 28, 123-135. 1107 50. Belyantseva, I.A., Perrin, B.J., Sonnemann, K.J., Zhu, M., Stepanyan, R., McGee, J., 1108 Frolenkov, G.I., Walsh, E.J., Friderici, K.H., Friedman, T.B., et al. (2009). Gamma- 1109 actin is required for cytoskeletal maintenance but not development. Proc Natl Acad 1110 Sci U S A 106, 9703-9708. 1111 51. Vedula, P., and Kashina, A. (2018). The makings of the 'actin code': regulation of 1112 actin's biological function at the amino acid and nucleotide level. J Cell Sci 131. 1113 52. Leerberg, J.M., Gomez, G.A., Verma, S., Moussa, E.J., Wu, S.K., Priya, R., Hoffman, 1114 B.D., Grashoff, C., Schwartz, M.A., and Yap, A.S. (2014). Tension-sensitive actin 1115 assembly supports contractility at the epithelial zonula adherens. Curr Biol 24, 1689- 1116 1699. 1117 53. Nehls, S., Noding, H., Karsch, S., Ries, F., and Janshoff, A. (2019). Stiffness of 1118 MDCK II Cells Depends on Confluency and Cell Size. Biophys J 116, 2204-2211. 1119 54. Ma, X., Bao, J., and Adelstein, R.S. (2007). Loss of cell adhesion causes 1120 hydrocephalus in nonmuscle myosin II-B-ablated and mutated mice. Mol Biol Cell 18, 1121 2305-2312. 1122 55. Conti, M.A., and Adelstein, R.S. (2008). Nonmuscle myosin II moves in new 1123 directions. J Cell Sci 121, 11-18. 1124 56. Sellers, J.R., and Heissler, S.M. (2019). Nonmuscle myosin-2 isoforms. Curr Biol 29, 1125 R275-R278. 1126 57. Melli, L., Billington, N., Sun, S.A., Bird, J.E., Nagy, A., Friedman, T.B., Takagi, Y., and 1127 Sellers, J.R. (2018). Bipolar filaments of human nonmuscle myosin 2-A and 2-B have 1128 distinct motile and mechanical properties. Elife 7. 1129 58. Itoh, M., Tsukita, S., Yamazaki, Y., and Sugimoto, H. (2012). Rho GTP exchange 1130 factor ARHGEF11 regulates the integrity of epithelial junctions by connecting ZO-1 1131 and RhoA-myosin II signaling. Proc Natl Acad Sci U S A 109, 9905-9910. 1132 59. Otani, T., Ichii, T., Aono, S., and Takeichi, M. (2006). Cdc42 GEF Tuba regulates the 1133 junctional configuration of simple epithelial cells. J Cell Biol 175, 135-146. 1134 60. Van Itallie, C.M., Tietgens, A.J., Krystofiak, E., Kachar, B., and Anderson, J.M. 1135 (2015). A complex of ZO-1 and the BAR-domain protein TOCA-1 regulates actin 1136 assembly at the tight junction. Mol Biol Cell 26, 2769-2787. 1137 61. Bruckner, B.R., and Janshoff, A. (2018). Importance of integrity of cell-cell junctions 1138 for the mechanics of confluent MDCK II cells. Sci Rep 8, 14117. 1139 62. Hatte, G., Prigent, C., and Tassan, J.P. (2018). Tight junctions negatively regulate 1140 mechanical forces applied to adherens junctions in vertebrate epithelial tissue. J Cell 1141 Sci 131.

29 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1142 63. Vasileva, E., Sluysmans, S., Bochaton-Piallat, M.L., and Citi, S. (2017). Cell-specific 1143 diversity in the expression and organization of cytoplasmic plaque proteins of apical 1144 junctions. Ann. NY. Acad. Sci 1405, 160-166. 1145 64. Huo, L., Wen, W., Wang, R., Kam, C., Xia, J., Feng, W., and Zhang, M. (2011). 1146 Cdc42-dependent formation of the ZO-1/MRCKbeta complex at the leading edge 1147 controls cell migration. EMBO J 30, 665-678. 1148 65. Unbekandt, M., and Olson, M.F. (2014). The actin-myosin regulatory MRCK kinases: 1149 regulation, biological functions and associations with human cancer. J Mol Med (Berl) 1150 92, 217-225. 1151 66. Belardi, I., Hamkins-Indik, T., Harris, A.R., and D.A., F. (2019). A weak link with actin 1152 organizes tight junctions to control epithelial permeability. BioRXiv, 1153 doi.org/10.1101/805689. 1154 67. McNeil, E., Capaldo, C.T., and Macara, I.G. (2006). Zonula occludens-1 function in 1155 the assembly of tight junctions in Madin-Darby canine kidney epithelial cells. Mol Biol 1156 Cell 17, 1922-1932. 1157 68. King, J.M., Tan, C.J.P., Thomason, N.C., White, A.R., Shen, L., and Turner, J.R. 1158 (2016). Zonula occludens-1 ZU5 domain contributes essential stabilizing interactions 1159 at the tight junction. FASEB J. 30, 1250.1257. 1160 69. Miyake, Y., Inoue, N., Nishimura, K., Kinoshita, N., Hosoya, H., and Yonemura, S. 1161 (2006). Actomyosin tension is required for correct recruitment of adherens junction 1162 components and zonula occludens formation. Exp Cell Res 312, 1637-1650. 1163 70. Rubsam, M., Mertz, A.F., Kubo, A., Marg, S., Jungst, C., Goranci-Buzhala, G., 1164 Schauss, A.C., Horsley, V., Dufresne, E.R., Moser, M., et al. (2017). E-cadherin 1165 integrates mechanotransduction and EGFR signaling to control junctional tissue 1166 polarization and tight junction positioning. Nature communications 8, 1250. 1167 71. Guillemot, L., Hammar, E., Kaister, C., Ritz, J., Caille, D., Jond, L., Bauer, C., Meda, 1168 P., and Citi, S. (2004). Disruption of the cingulin gene does not prevent tight junction 1169 formation but alters gene expression. Journal of cell science 117, 5245-5256. 1170 72. Guillemot, L., Schneider, Y., Brun, P., Castagliuolo, I., Pizzuti, D., Martines, D., Jond, 1171 L., Bongiovanni, M., and Citi, S. (2012). Cingulin is dispensable for epithelial barrier 1172 function and tight junction structure, and plays a role in the control of claudin-2 1173 expression and response to duodenal mucosa injury. J. Cell Sci. 125, 5005-5012. 1174 73. Woodruff, J.B., Ferreira Gomes, B., Widlund, P.O., Mahamid, J., Honigmann, A., and 1175 Hyman, A.A. (2017). The Centrosome Is a Selective Condensate that Nucleates 1176 Microtubules by Concentrating Tubulin. Cell 169, 1066-1077 e1010. 1177 74. Alberti, S., Gladfelter, A., and Mittag, T. (2019). Considerations and Challenges in 1178 Studying Liquid-Liquid Phase Separation and Biomolecular Condensates. Cell 176, 1179 419-434. 1180 75. Sun, T., Li, Q., Xu, Y., Zhang, Z., Lai, L., and Pei, J. (2019). Prediction of liquid-liquid 1181 phase separation proteins using machine learning. BioRxiv, doi.org/10.1101/842336. 1182 76. Yano, T., Matsui, T., Tamura, A., Uji, M., and Tsukita, S. (2013). The association of 1183 microtubules with tight junctions is promoted by cingulin phosphorylation by AMPK. J 1184 Cell Biol 203, 605-614. 1185 77. Mangan, A.J., Sietsema, D.V., Li, D., Moore, J.K., Citi, S., and Prekeris, R. (2016). 1186 Cingulin and actin mediate midbody-dependent apical lumen formation during 1187 polarization of epithelial cells. Nature communications 7, 12426. 1188 78. Rouaud, F., Vasileva, E., Spadaro, D., Tsukita, S., and Citi, S. (2019). R40.76 binds 1189 to the alpha domain of ZO-1: role of ZO-1 (alpha+) in epithelial differentiation and 1190 mechano-sensing. Tissue barriers 7, e1653748. 1191 79. Paschoud, S., and Citi, S. (2008). Inducible overexpression of cingulin in stably 1192 transfected MDCK cells does not affect tight junction organization and gene 1193 expression. Mol. Membr. Biol. 25, 1-13. 1194 80. Paschoud, S., Jond, L., Guerrera, D., and Citi, S. (2014). PLEKHA7 modulates 1195 epithelial tight junction barrier function. Tissue barriers 2, e28755. 1196 81. Shah, J., Rouaud, F., Guerrera, D., Vasileva, E., Popov, L.M., Kelley, W.L., 1197 Rubinstein, E., Carette, J.E., Amieva, M.R., and Citi, S. (2018). A Dock-and-Lock 1198 Mechanism Clusters ADAM10 at Cell-Cell Junctions to Promote alpha-Toxin 1199 Cytotoxicity. Cell Rep 25, 2132-2147 e2137.

30 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1200 82. Wahl, M., Koberling, F., Patting, M., Rahn, H., and Erdmann, R. (2004). Time- 1201 resolved confocal fluorescence imaging and spectrocopy system with single molecule 1202 sensitivity and sub-micrometer resolution. Curr Pharm Biotechnol 5, 299-308. 1203

31 Figure1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A Detector and B C feedback electronics 5 0.015 **** WT **** CGN-KO CGNL1-KO **** Photodiode 4 DOUBLE-KO Laser Hertz model ) 0.010 3 N

2 Cantilever Force, n 0.005

Tip oung's modulus (MPa Y 1

MDCK cells 0 0.000 0.0 0.5 1.0 1.5 2.0 WT MDCK Distance, μm MDCK CGN-KO CGNL1-KO DOUBLE KO

D MDCK-WT E MDCK-CGN-KO

F G 4.8 **** 5.0 * **** 4.8 4.6 ) ) 4.6 ) (ns τ ) (ns τ 4.4 4.4 ime ( ime ( T T Slope Life

Life 4.2 WT 0.182 4.2 CGN-KO 0.067 CGNL1-KO 0.098 4.0 DOUBLE-KO 0.081

4.0 3.8 8 10 12 MDCK WT MDCK Distance, μm Apical CGN-KO CGNL1-KO DOUBLE KO

Figure 1 Figure2 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

cingulin β-catenin WT CGN-KO CGNL1-KO DOUBLE KO C **** A B 1.8 **** **** ns **** **** 1.6 ex ind 1.4 ag MDCK-transwells z MDCK

Occludin Occludin Occludin Occludin ig Z 1.2

D 1.8 **** E **** 1.0

1.6 WT

ex CGN-KO

ind CGNL1-KO 1.4 DOUBLE KO ag GFP (GFP-cCGN) Occludin z ig ns Z 1.2 MDCK CGN-KO + rescue 1.0 CGN-KO GFP-cCGN -+ -- -- + GFP-msCGN -- -+ -- rescue GFP GFP (GFP-mCGN) Occludin -- -- -+ MDCK CGN-KO + rescue MDCK CGN-KO+rescue GFP-cCGN J rescue

GFP Occludin NM2B

MDCK WT+ CGN-KO KO NM2B GFP (GFP-cCGN) ZO-2 F K

WT rescue NM2A Phalloidin

KO NM2A CGN ZO-2 KO Phalloidin GFP (GFP-cCGN) ZO-2 G L

WT rescue NM2B ARHGEF18 KO NM2B CGN ZO-2 KO ARHGEF18 GFP (GFP-cCGN) ZO-2 H O 250 250 WT γ-actin CGN-KO γ-actin KO 200 PLEKHA7 200 PLEKHA7 y, a.u.

y, a.u. 150 150

Phalloidin 100 100 Intensit WT Phalloidin CGN ZO-2 Intensit 50 50 I 0 0 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -2.5 -1.5 -0.5 0.5 1.5 2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 Distance, μm Distance, μm WT

P 150 WT β-actin 150 CGN-KO β-actin

ARHGEF18 PLEKHA7 PLEKHA7 KO ARHGEF18 CGN ZO-2 100 100 y, a.u. y, a.u. mCCD WT+ CGN-KO 50 50

M Intensit Intensit 0 0 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 KO -2.5 -1.5 -0.5 0.5 1.5 2.5 -2.5 -1.5 -0.5 0.5 1.5 2.5 Distance, μm Distance, μm 5 -actin 5 ns γ Q **** WT γ-actin CGN PLEKHA7 R 4 4 N 3 3

WT 2 2 -actin/PLEKHA7 -actin/PLEKHA7 -actin FWHM ratio, a.u. FWHM ratio, a.u. γ β β 1 1

WT WT KO β-actin CGN PLEKHA7 CGN-KO CGN-KO

Figure 2 Figure3 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B C Input (prey) GST GST-ZU5GSTGST-ZU5GSTGST-ZU5 GFP + - - + + - - - - Input GST GST-ZU5GST GST-ZU5 G-CGN-FL - + - - - + + - - HA-CGNL1-FL + - + + - - G-CGN-Δ1-70 - - + - - - - + + HA-CGNL1-Δ1-80 - + - - + + 150 controlGFP G-888-1619G-1550-1619G-888-1748G-1150-1748G-1619-1748 (ZU5) 150 150

100 IB:CGNL1 50 50 75 IB:GFP 25 35 50 37 Baits IB:GFP Bait=GST-CGN(1-70) (Ponceau S) 35 25 150

50 IB:GFP 25 25 25 50 Bait=GST(Ponceau S) 35 Baits

25

D Eph4 ZO-1-KO + ZO-1 rescue E MDCK WT overexpression F MDCK WT overexpression myc-ZO-1-FL

CGN myc PLEKHA7 HA (ZO-1) CGN HA (ZO-1) CGNL1 U5 ΔΖ

myc-ZO-1- CGN myc PLEKHA7 HA (ZO-1-ΔZU5) CGN HA (ZO-1-ΔZU5) CGNL1

G Eph4 WT+ siZO-2 untreated I Eph4 CGN-KO GFP si-control * * * GFP ZO-2 DbpA DbpA ZO-2 PLEKHA7 GFP-CGN 1-70

* * * siZO-2 * * * GFP ZO-2 DbpA DbpA ZO-2 PLEKHA7

H Eph4 WT+ siZO-2+BLEBBISTATIN J G-CGN 1-70 K GFP input (ZO-1)no lysatecontrolμl 5 10 50 100 GFPG-CGN 1-70 * * 250 * IB:ZO-1 37 IB: 25 GFP 75 Bait INPUT 3rd protein

50 GST-DbpA GFP ZO-2 DbpA GFP-CGN 1-70 GFP L

input (ZO-1)no lysatecontrolμl 5 10 50 100 250 IB:ZO-1 input GST(HA-DbpA)G-CGNG-GEFH1-C1 1-70 50 IB:HA * * * 75 37 Bait GFP ZO-2 DbpA 25 Baits

50 GST-DbpA

Figure 3 Figure4 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A ZO-1 CGN PLEKHA7 E Eph4 CGN-FL-myc

CGN-KO+WT mix CGN-KO+WT ZO-1 myc MERGE PLEKHA7 B 1-70-myc MDCK Δ

CGN-KO+WT mix CGN-KO+WT ZO-1 CGN Occludin ZO-1 myc MERGE PLEKHA7 CGN-

ZO-1 CGNL1 PLEKHA7 C Eph4 CGN-KO + rescue Eph4 GFP-myc ZO-1 myc MERGE PLEKHA7 CGNL1-KO+WT mix CGNL1-KO+WT D F MDCK

CGNL1-KO+WT mix CGNL1-KO+WT ZO-1 CGNL1 Occludin CGN-FL-myc WT *** ZO-1 myc MERGE PLEKHA7 G Eph4 CGN-KO ns ** **

. 2.0 ns

ns

1.5 ns 1-70-myc * Δ ZO-1 myc MERGE PLEKHA7 CGN- 1.0 ns ns , ZO-1/PLEKHA7, a.u Eph4 WT + overexpression Eph4 WT y

0.5 GFP-myc

Relative intensit 0.0 ZO-1 myc MERGE PLEKHA7

-GFP 1-70 1-70 -GFP 1-70 1-70 +GFP Δ Δ +GFP Δ Δ -CGN+CGN FL FL -CGN+CGN FL FL -CGN+CGN -CGN+CGN

Figure 4 Figure5 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A A’ A’’ B B’ B’’ 0 sec 4 sec 24 sec 28 sec

0 sec 4 sec 42 sec 44 sec 8 sec 12 sec 32 sec 34 sec

GFP-CGN GFP-ZO-1 8 sec 12 sec 48 sec 52 sec

C cCGN in WT D mZO-1 in WT E mZO-1 in CGN-KO GFP (GFP-cCGN) NM2B GFP (GFP-mZO-1) GFP (GFP-mZO-1) NM2B

NM2B NM2B

GFP (GFP-cCGN) Phalloidin GFP (GFP-mZO-1) Phalloidin Phalloidin GFP (GFP-mZO-1) Phalloidin GFP (GFP-cCGN) ARHGEF18 GFP (GFP-mZO-1) ARHGEF18 GFP (GFP-mZO-1) ARHGEF18 ARHGEF18

F hZO-1 in WT G hZO-1 in CGN-KO H ns **** 1.0 ns **** 0.8 A, a.u. Phalloidin 0.6 Full-length HA (hZO1-FL-HA) Phalloidin HA (hZO1-FL-HA) Phalloidin 0.4 Relative intensity Phalloidin/H ZO-1 HA (hZO1-ΔZU5-HA) Phalloidin HA (hZO1-ΔZU5-HA) 0.2

0.0

-ZU5 -ZU5 -ZU5 Δ Δ Δ ZO1 FL ZO1 FL ZO1- ZO1- Phalloidin Phalloidin WT CGN-KO

Figure 5 Figure6 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A WT B CGN-KO

TJ

ZA

C

claudin, occludin cadherin nonmuscle myosin 2B

ZO-1 (ZU5 “open”) JAM-A, nectin NM2B force catenins ZO-1 (ZU5 “closed”) nonmuscle myosin 2A afadin, PLEKHA7 vinculin NM2A force cingulin paracingulin γ-actin plasma membrane β-actin

Figure 6 KeyResourceTablebioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Mouse monoclonal anti-GFP Roche Cat# 11814460001; RRID:AB_390913 Rabbit polyclonal anti-GFP Thermo Fisher Cat# A-11122; Scientific RRID:AB_221569 Mouse monoclonal anti-HA Thermo Fisher Cat# 32-6700; Scientific RRID:AB_2533092 Rabbit polyclonal anti-HA Santa Cruz Cat# sc-805; RRID:AB_631618 Mouse monoclonal anti-myc Gerard Evan, MRC Cat# 9E10; LMB RRID:AB_2266850 Rabbit polyclonal anti-cingulin Citi Laboratory C532 Mouse monoclonal anti-cingulin Citi Laboratory 22BD5A1 Rabbit polyclonal anti-paracingulin Citi Laboratory 20893 Mouse monoclonal anti-paracingulin Santa Cruz Cat# sc-377525 Mouse monoclonal anti-ZO-1 Thermo Fisher Cat# 33-9100; RRID: Scientific AB_2533147 Rabbit polyclonal anti-ZO-1 Thermo Fisher Cat# 61-7300; Scientific RRID:AB_2533938 Rabbit Polyclonal anti-ZO-1 (ZU5) [1] R3 Rat monoclonal anti-ZO-1 Prof. D. Goodenough, R40.76; Harvard Medical RRID:AB_2205518 School Guinea pig polyclonal anti-PLEKHA7 Citi Laboratory GP2737 Rabbit polyclonal anti-PLEKHA7 Citi Laboratory R30388 Rat polyclonal anti-PLEKHA6 Citi Laboratory RtSZR127 (unpublished) Rabbit polyclonal anti-DbpA/ZONAB Thermo Fisher Cat# 40-2800; Scientific RRID:AB_2533460 Goat polyclonal anti-ZO-2 Santa Cruz Cat# sc-8148; RRID:AB_2271821 Rabbit polyclonal anti-ZO-2 Thermo Fisher Cat# 71-1400; Scientific RRID:AB_2533976 Rabbit polyclonal anti-ZO-3 Thermo Fisher Cat# 36-4100; Scientific RRID:AB_148475 Rabbit polyclonal anti-occludin Thermo Fisher Cat# 71-1500; Scientific RRID:AB_2533977 Mouse monoclonal anti-occludin Thermo Fisher Cat# 33-1500; Scientific RRID:AB_87033 Mouse monoclonal anti-E-cadherin BD Biosciences Cat# BD 610181; RRID:AB_397580 Rabbit polyclonal anti-α-catenin Sigma-Aldrich/Merck Cat# C2081; RRID:AB_476830 Rabbit polyclonal anti-β-catenin Sigma-Aldrich/Merck Cat# C2206; RRID:AB_476831 Rabbit polyclonal anti-afadin Sigma-Aldrich/Merck Cat# A0224; RRID:AB_257871 Rabbit polyclonal anti-claudin-1 Thermo Fisher Cat# 51-9000; Scientific RRID:AB_2533916 Mouse monoclonal anti-β-tubulin Thermo Fisher Cat# 32-2600; Scientific RRID:AB_2533072 Rabbit polyclonal anti-NM2A Biolegend Cat# 909801; RRID:AB_291638 Rabbit polyclonal anti-NM2B Biolegend Cat# 909901; RRID:AB_291639

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Mouse monoclonal anti-b-actin [2] RRID:AB_2571580

Mouse monoclonal anti-g-actin [2] RRID:AB_2571583

Rabbit polyclonal anti-ARHGEF18 Sigma-Aldrich/Merck Cat# HPA042689; RRID:AB_10794385 Alexa Fluor 488-AffiniPure Donkey Anti-Rabbit IgG Jackson Laboratory Cat# 711-545-152; RRID:AB_2313584 Alexa Fluor 488-AffiniPure Donkey Anti-Mouse IgG Jackson Laboratory Cat# 715-546-150; RRID: AB_2340849 Alexa Fluor 488-AffiniPure Donkey Anti-goat IgG Jackson Laboratory Cat# 705-546-147; RRID:AB_2340430 Alexa Fluor 488-AffiniPure Donkey Anti-rat IgG Jackson Laboratory Cat# 712-546-150; RRID: AB_2340685 Cy3-AffiniPure Donkey Anti-Rabbit IgG Jackson Laboratory Cat# 711-165-152; RRID:AB_2307443 Cy3-AffiniPure Donkey Anti-Mouse IgG Jackson Laboratory Cat# 715-165-151; RRID:AB_2315777 Cy3-AffiniPure Donkey Anti-Rat IgG Jackson Laboratory Cat# 712-166-150; RRID:AB_2340668 Cy3-AffiniPure Donkey Anti-Goat IgG Jackson Laboratory Cat# 705-166-147; RRID:AB_2340413 FITC-phalloidin Sigma Cat# P5282 TRITC-phalloidin Thermo Fisher Cat# R415; Scientific RRID:AB_2572408 Alexa Fluor 647-AffiniPure Donkey Anti-Guinea Pig IgG Jackson Laboratory Cat# 706-605-148; RRID:AB_2340476 Alexa Fluor 647-AffiniPure Donkey Anti-Rabbit IgG Jackson Laboratory Cat# 711-605-152; RRID:AB_2492288 Alexa Fluor 647-AffiniPure Donkey Anti-Goat IgG Jackson Laboratory Cat# 705-606-147; RRID:AB_2340438 Cy5-AffiniPure Donkey Anti-Mouse IgG Jackson Laboratory Cat# 715-175-150; RRID:AB_2340819 Cy5-AffiniPure Donkey Anti-Rat IgG Jackson Laboratory Cat# 712-175-153; RRID:AB_2340672 Anti-Mouse IgG (H+L), HRP Conjugate Promega Cat# W4021; RRID:AB_430834 Anti-Rabbit IgG (H+L), HRP Conjugate Promega Cat# W4011; RRID:AB_430833 Anti-Rat IgG (H+L), HRP Conjugate Thermo Fisher Cat# 62-9520; RRID: Scientific AB_2533965 Anti-Goat IgG (H+L), HRP Conjugate Promega Cat# V8051; RRID:AB_430838 Bacterial and Virus Strains BL21 Competent E.coli NEB Cat# C2530H DH5 alpha Competent E.coli Thermo Fisher Cat# 18265017 Scientific DH10B Competent cells E.coli Thermo Fisher Cat# 18297010 Scientific Chemicals, Peptides, and Recombinant Proteins Blebbistatin Sigma-Aldrich Cat# B0560 Hygromycin B Gold InvivoGen Cat# ant-hg-2 FliptR Roux Laboratory, N/A University of Geneva; [3] GST-DbpA Balda-Matter N/A Laboratory, (UCL) GST-cCGN(1-70) This paper S1851

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

GST-hZU5 This paper S1789 GST-GEF-H1 Balda-Matter N/A Laboratory, UCL; [4] GST-hCGN(1-121) This paper S96 GST-hCGN(1-353) This paper S97 GST-hCGNL1(1-119) This paper S1024 Critical Commercial Assays Lipofectamine RNAiMAX Thermo Fisher Cat# 13778030 Scientific Lipofectamine 2000 Thermo Fisher Cat# 11668027 Scientific jetOPTIMUS® Polyplus Cat# 117-15 Q5 High fidelity Polymerase NEB Cat# M0491L T4 DNA Ligase Promega Cat# M1801 DNeasy Blood & Tissue kit QIAGEN Cat# 69504 Glass coverslips 12 mm diameter Thermo Scientific Cat# Menzel CBAD00120RAC20 MNZ#0 24 mm Transwell with 0.4 µm pore polyester membrane Corning Cat# 3450 insert Falcon 12-well tissue culture plates Corning Cat# 353043 Falcon 24-well tissue culture plates Corning Cat# 351147 CELLSTAR 96-well tissue culture plates Greiner Bio-One Cat# 655 180 35-mm Glass Bottom Culture dishes MatTek corporation Cat# P35G-1.5-14-C Pierce protease inhibitor Thermo Fisher A32963 Scientific cOmplete protease inhibitor cocktail Roche 11697498001 Glutathione Sepharose 4B beads GE Healthcare Cat# 0756-01 Pierce glutathione magnetic agarose beads Thermo Fisher Cat# 78602 Scientific Vectashield with DAPI Vector Laboratories Cat# H-1200 Fluoromount-G SouthernBiotech Cat# 0100-01 Counting chamber (hemocytometer) Marienfeld Superior Cat# 0640010 Experimental Models: Cell Lines Mouse mammary epithelial cell line Eph4 WT Reichmann http://web.expasy Laboratory, University .org/cellosaurus/ of Zurich; [5] CVCL_0073 Mouse mammary epithelial cell line Eph4 ZO-1-KO Tsukita Laboratory, N/A Osaka University; [6] Mouse mammary epithelial cell line Eph4 CGN-KO This paper N/A Mouse mammary epithelial cell line Eph4 CGNL1-KO This paper N/A Human embryonic kidney HEK 293T ATCC N/A SF9 insect cell lines Louise Fairall, MRC N/A LMB Mouse Cortical Collecting Duct Cell Line (mCCD) WT Eric Féraille, N/A University of Geneva; [7] Mouse Cortical Collecting Duct Cell Line (mCCD) CGN- This paper N/A KO Mouse Cortical Collecting Duct Cell Line (mCCD) This paper N/A CGNL1-KO MDCKII (Madin–Darby Canine Kidney) Tet-off Alan Fanning, Clontech University of North Carolina at Chapel Hill MDCKII (Madin–Darby Canine Kidney) Tet-off CGN-KO This paper N/A

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

MDCKII (Madin–Darby Canine Kidney) Tet-off CGNL1- This paper N/A KO MDCKII (Madin–Darby Canine Kidney) Tet-off CGNL1- This paper N/A KO-YFP-myc MDCKII (Madin–Darby Canine Kidney) Tet-off This paper N/A CGN/CGNL1-dKO Oligonucleotides siRNA target sequence: mouse ZO-2 [8] N/A ctcctatcacgaagcttat siRNA negative control Sigma-Aldrich Cat# N. SIC001 CRISPR target sequence mCGN: This paper N/A GGCAGTGCGGACCATGTGAA CRISPR target sequence cCGN: This paper N/A CCACGGATACCCTTGATGAACG CRISPR target sequence mCGNL1: This paper N/A TTCCGGCGACAAGATTCCGC CRISPR target sequence cCGNL1: This paper N/A TTCCAGCAACGGGTCCGTGC Forward primer for mCGN genotyping: 5’- This paper N/A GAACCGCGGGTCTGCTAGGTC-3’ Reverse primer for mCGN genotyping: 5’- This paper N/A GAGCTCGAGGCATCTTCCCTGTC -3’ Forward primer for cCGN genotyping: 5’- This paper N/A GAGTCTAGATGCTAGAGCCAATACCTATGGA -3’ Reverse primer for cCGN genotyping: 5’- This paper N/A GAGGAATTCGCTGTGAAGGCAGATTTCATAGTT -3’ Forward primer for mCGNL1 genotyping: 5’- This paper N/A GAGTCTAGAACGCAAAGACGGGC -3’ Reverse primer for mCGNL1 genotyping: 5’- This paper N/A ACCAGAATTCGATGTGGCAG -3’ Forward primer for cCGNL1 genotyping: 5’- This paper N/A GAGGGATCCGAGGTCATGCCAAAGGTGTCAG -3’ Reverse primer for cCGNL1 genotyping: 5’- This paper N/A GAGGTCGACGTGGATGGCACATTCCTGAT -3’ Recombinant DNA pSpCas9(BB)-2A-GFP (PX458) [9] Addgene plasmid 48138 pBluescript II KS(+) Stratagene Cat# 212207 pCDNA3.1(+)-GFP-hZO-1(888-1619) [8] S1807 pCDNA3.1(+)-GFP-hZO-1(1150-1619) [8] S1808 pCDNA3.1(+)-GFP-hZO-1(888-1748) [8] S1809 pCDNA3.1(+)-GFP-hZO-1(1150-1748) [8] S1810 pCDNA3.1(+)-GFP-hZO-1(1619-1748) [8] S1811 pCDNA3.1(+)-GFP-hZO-1(1698-1748) This paper S1874 pCDNA3.1(+)-GFP-hZO-1(1550-1650) This paper S1875 pACEBac1-2xStrep-Spy-hZO-1(optimized for insect cell [8] S1928 expression)-AVI pCDNA3.1(+)-HA-DbpA [10] S1575 pCDNA3.1(+)-CFP-HA [10] S1150 pCDNA3.1(+)-myc-hZO-1-HA [8] S1947 pCDNA3.1(+)-myc-hZO-1-DZU-5(1-1619)-HA [1] S2161 pCDNA3.1(-)-GFP-mZO-1-myc-his [1] S2474 pCDNA3.1(-)-mCGN-myc-his This paper S2407 pCDNA3.1(-)-mCGN-D1-70-myc-his This paper S2408 pCDNA3.1(-)-GFP-myc-his [11] S1166 pTRE2Hyg-GFP-mCGN-myc This paper S2363

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

pTRE2Hyg-GFP-mCGNL1-myc This paper S2386 pCDNA3.1(-)-GFP-hCGN This paper S2508 pCDNA3.1(-)-GFP-hCGN-D1-70 This paper S2509 pCDNA3.1(+)-hCGN-HA This paper S2411 pTRE2Hyg-GFP-cCGN-myc [12] S2294/S1052 pTRE2Hyg-YFP-cCGNL1-myc [13] S1137 pTRE2Hyg-YFP-myc [13] S1152 pTRE2Hyg-GFP-cCGN(1-70) This paper S1821 pCDNA3.1(+)-hCGNL1-myc-HA This paper S2442 pCDNA3.1(+)-hCGNL1-D1-80-HA This paper S2510 pTRE2Hyg-GFP-myc [14] S1210 pCDNA3.1(-)-cCGN-HA This paper S1136 pCDNA3.1(+)-cCGNL1-HA This paper S2432 pCDNA3.1(+)-chimera cCGNL1head(1-579) +cCGNrod-tail(339- This paper S2433 1190) -HA pCDNA3.1(+)-chimera cCGNhead(1-338) +cCGNL1rod-tail(580- This paper S2434 1295) -HA Software and Algorithms Image J N/A imagej.nih.gov/ij/; RRID:SCR_003070 Adobe Photoshop CS6 N/A adobe.com Adobe Illustrator CS6 N/A adobe.com GraphPad Prism 8 N/A graphpad.com/scient ific-software/prism/ SnapGene N/A snapgene.com Zhang Lab CRISPR design tool Zhang Lab, MIT www.crispr.mit.edu

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

REFERENCES FOR KEY RESOURCE TABLE

1. Rouaud, F., Vasileva, E., Spadaro, D., Tsukita, S., and Citi, S. (2019). R40.76 binds to the alpha domain of ZO-1: role of ZO-1 (alpha+) in epithelial differentiation and mechano-sensing. Tissue barriers 7, e1653748. 2. Dugina, V., Zwaenepoel, I., Gabbiani, G., Clement, S., and Chaponnier, C. (2009). Beta and gamma-cytoplasmic actins display distinct distribution and functional diversity. J Cell Sci 122, 2980-2988. 3. Colom, A., Derivery, E., Soleimanpour, S., Tomba, C., Molin, M.D., Sakai, N., Gonzalez-Gaitan, M., Matile, S., and Roux, A. (2018). A fluorescent membrane tension probe. Nat Chem 10, 1118-1125. 4. Nie, M., Aijaz, S., Leefa Chong San, I.V., Balda, M.S., and Matter, K. (2009). The Y-box factor ZONAB/DbpA associates with GEF-H1/Lfc and mediates Rho-stimulated transcription. EMBO Rep 10, 1125-1131. 5. Fialka, I., Schwarz, H., Reichmann, E., Oft, M., Busslinger, M., and Beug, H. (1996). The estrogen-dependent c-JunER protein causes a reversible loss of mammary epithelial cell polarity involving a destabilization of adherens junctions. J Cell Biol 132, 1115-1132. 6. Umeda, K., Matsui, T., Nakayama, M., Furuse, K., Sasaki, H., Furuse, M., and Tsukita, S. (2004). Establishment and characterization of cultured epithelial cells lacking expression of ZO-1. J Biol Chem 279, 44785-44794. 7. Wang, Y.B., Leroy, V., Maunsbach, A.B., Doucet, A., Hasler, U., Dizin, E., Ernandez, T., de Seigneux, S., Martin, P.Y., and Feraille, E. (2014). Sodium transport is modulated by p38 kinase-dependent cross-talk between ENaC and Na,K-ATPase in collecting duct principal cells. J Am Soc Nephrol 25, 250- 259. 8. Spadaro, D., Le, S., Laroche, T., Mean, I., Jond, L., Yan, J., and Citi, S. (2017). Tension-Dependent Stretching Activates ZO-1 to Control the Junctional Localization of Its Interactors. Curr Biol 27, 3783-3795 e3788. 9. Ran, F.A., Hsu, P.D., Wright, J., Agarwala, V., Scott, D.A., and Zhang, F. (2013). Genome engineering using the CRISPR-Cas9 system. Nat Protoc 8, 2281-2308. 10. Spadaro, D., Tapia, R., Jond, L., Sudol, M., Fanning, A.S., and Citi, S. (2014). ZO Proteins Redundantly Regulate the Transcription Factor DbpA/ZONAB. J Biol Chem 289, 22500-22511. 11. Guerrera, D., Shah, J., Vasileva, E., Sluysmans, S., Mean, I., Jond, L., Poser, I., Mann, M., Hyman, A.A., and Citi, S. (2016). PLEKHA7 Recruits PDZD11 to Adherens Junctions to Stabilize Nectins. J Biol Chem 291, 11016-11029. 12. Paschoud, S., and Citi, S. (2008). Inducible overexpression of cingulin in stably transfected MDCK cells does not affect tight junction organization and gene expression. Mol. Membr. Biol. 25, 1-13. 13. Paschoud, S., Guillemot, L., and Citi, S. (2012). Distinct domains of paracingulin are involved in its targeting to the actin cytoskeleton and regulation of apical junction assembly. Journal of Biological Chemistry 287, 13159-13169. 14. Paschoud, S., Jond, L., Guerrera, D., and Citi, S. (2014). PLEKHA7 modulates epithelial tight junction barrier function. Tissue barriers 2, e28755.

6 Table bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Table S1. Young’s modulus values for MDCK lines.

Line genotype Young’s modulus (MPa) WT 0.0037 ± 0.0014 CGN-KO 0.0015 ± 0.0007 CGNL1-KO 0.0021 ± 0.0019 CGN/CGNL1-dKO 0.0014 ± 0.0005 SupplementalbioRxiv Data preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S1-1 made available under aCC-BY-NC-ND 4.0 International license.

A Exon 2 E G Eph4 MDCK CGN M. musculus WT WT WT allele .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATGTGAAGGCCACCATCTATGGCATCC...... CGN PLEKHA7 MERGE CGN ZO-1 MERGE Clone

allele-1 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATGATGAAGGCCACCATCTATGGCATCC...... <1 bp (+)> cl. 1 1B5 allele-2 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATG:<2bp(-)>:AAGGCCACCATCTATGGCATCC...... 1B5 CGN PLEKHA7 MERGE CGN ZO-1 MERGE

Eph4 allele-1 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATGTTGAAGGCCACCATCTATGGCATCC......

<1 bp (+)> CGN-KO .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATG:<2bp(-)>:AAGGCCACCATCTATGGCATCC...... 1D6 allele-2 CGN-KO cl. 1H9 1D6 CGN PLEKHA7 MERGE CGN ZO-1 MERGE allele-1 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATGCTGAAGGCCACCATCTATGGCATCC...... <1 bp (+)> cl. 3 allele-2 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCAT:<14bp(-)>:CTATGGCATCC...... F mCCD

.....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATGGTGAAGGCCACCATCTATGGCATCC...... cl. 1G8 allele-1 CGN ZO-1 MERGE <1 bp (+)> WT cl. 4 allele-2 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATG:<2bp(-)>:AAGGCCACCATCTATGGCATCC...... mCCD CGN PLEKHA7 MERGE allele-1 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATG:<1bp(-)>:GAAGGCCACCATCTATGGCATCC.....

cl. 8 .....AGAGAGAGACAGCGCCACCAGGCAGTGCGGACCATG:<2bp(-)>:AAGGCCACCATCTATGGCATCC...... allele-2 I CGN-KO clone 3 CGN PLEKHA7 MERGE B Exon 1 kDa WT clone 3clone clone4 8 150 IB:CGN CGN clone 4 50 C. l. familiaris CGN-KO IB:β-tubulin CGN PLEKHA7 MERGE mCCD

WT allele ...... AAGCCTGGACAACCGGCTCCCACGGGATACCCTTGATGAACGAGAACATCAGTTCCCTAC......

Clone clone 8 allele-1 .....AAGCCTGGACAACCGGCTCCCACG:<2bp(-)>:ATACCCTTGATGAACGAGAACATCAGTTCCCTAC...... CGN PLEKHA7 MERGE

cl. 1 allele-2 .....TCCCACGTGATACTGTCCTAAGAATG...AGATCAAGGTTGGATGTTGGATACCCTTGATGAACGAG...... <415 bp (+)> H CGN-KO J CGN-KO allele-1 .....TCCCACGCTCTAGCTTCCCGGCAAC...GACCACTTCTGCGCTCGGGGATACCCTTGATGAACGAG...... <72 bp (+)> MDCK allele-2 .....AAGCCTGGACAACCGGCTCCCACG:<2bp(-)>:ATACCCTTGATGAACGAGAACATCAGTTCCCTAC...... kDa WT clone clone1B5 1D6 cl. 1H9 kDa WT clone clone1 clone1H9 1G8 150 IB:CGN 150 IB:CGN allele-1 .....AAGCCTGGACAACCGGCTCCCACGAGGATACCCTTGATGAACGAGAACATCAGTTCCCTAC...... <1 bp (+)>

allele-2 .....AAGCCTGGACAACCGGCTCCCACGG:<1bp(-)>:ATACCCTTGATGAACGAGAACATCAGTTCCCTAC...... Eph4 cl. 1G8 50 IB:β-tubulin 50 IB:β-tubulin MDCK

C Exon 2 K M Eph4 MDCK CGNL1 M. musculus WT WT

CGNL1 PLEKHA7 MERGE CGNL1 ZO-1 MERGE WT allele .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGATTCCGCAGGACCCATCCTGGATGGAG......

Clone .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGATA:<2bp(-)>:CGCAGGACCCATCCTGGATGGAG...... 1A4 allele-1 cl. 2-2 <1 bp (+)> CGNL1 PLEKHA7 MERGE CGNL1 ZO-1 MERGE allele-2 .....CCTTCCGGCGACAAGATTCATCCAGCAGGACCATGTGATC...GTCTTTGCTCAGGGCGGACTGTCGCAG..... CGNL1-KO 1A4 Eph4 <58 bp (+)> mCCD

L CGNL1-KO

allele-1 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGATTCCCGCAGGACCCATCCTGGATGGAG...... cl. 2-5 <1 bp (+)> CGNL1 ZO-1 MERGE WT

cl. 7 allele-2 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGAT:<2bp(-)>:CGCAGGACCCATCCTGGATGGAG...... CGNL1 PLEKHA7 MERGE allele-1 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGAT:<2bp(-)>:CGCAGGACCCATCCTGGATGGAG...... allele-2 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAG:<4bp(-)>:CGCAGGACCCATCCTGGATGGAG...... cl. 8 mCCD O CGNL1-KO

allele-1 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGAT:<2bp(-)>:CGCAGGACCCATCCTGGATGGAG...... clone 7 .....ACAGGCCAGATGTGCTACCCTTCCGGCGACAAGATT:<2bp(-)>:GCAGGACCCATCCTGGATGGAG...... CGNL1 PLEKHA7 MERGE cl. 9 allele-2 kDa WT clone 7clone clone8 9

150 IB:CGNL1

Exon 1 clone 8 D 50 IB:β-tubulin CGNL1 PLEKHA7 MERGE mCCD CGNL1-KO CGNL1 C. l. familiaris clone 9

WT allele .....AGAAGTAACCTGCCTCTGCATTCCAGCAACGGGTCCGTGCTGGAGGAGAGCAGCGCAGAG...... CGNL1 PLEKHA7 MERGE CGNL1-KO Clone N CGNL1-KO P allele-1 .....AGAAGTAACCTGCCTCTGCATTCCAGCAACGGGTCCGGTGCTGGAGGAGAGCAGCGCAGAG...... <1 bp (+)> .....AGAAGTAACCTGCCTCTGCATTCCAGCAACGGGTCCGTTGCTGGAGGAGAGCAGCGCAGAG...... WT clone clone2-2 2-5 cl. 2-2 allele-2 kDa <1 bp (+)> kDa WT clone 1A4 IB:CGNL1

MDCKII 150 allele-1 .....AGAAGTAACCTGCCTCTGCATTCCAGCAACGGGTCCGTTGCTGGAGGAGAGCAGCGCAGAG...... 150 IB:CGNL1 <1 bp (+)> cl. 2-5 50 50 IB:β-tubulin MDCKII Eph4 IB:β-tubulin bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S1-2 made available under aCC-BY-NC-ND 4.0 International license.

Q R MDCK MDCK MDCK MDCK CGN-KO CGNL1-KO CGN/CGNL1-dKO

WT IB: kDa MDCK WT MDCK WT #1G8 #1H9 #2-2 #2-5 #11C9 #21D3 CGN CGNL1 ZO-2 MERGE 150 CGN

150 CGNL1

cl. 11C9 CGN CGNL1 ZO-2 MERGE * * ** * * * * 250 ZO-1

CGN-CGNL1-dKO 150 ZO-2 cl. 21D3 CGN CGNL1 ZO-2 MERGE ZO-3 100 S 16 occludin 50 WT 14 25 claudin-1 CGN-KO 1G8 CGN-KO 1H9 12 CGNL1-KO 2-2 E-cadherin 100 ) CGNL1-KO 2-5 5 DOUBLE KO 11C9 100 10 α-catenin DOUBLE KO 21D3 r (x1*10 8 250 afadin

6 50 β−tubulin Cell numbe 4

2

0 0 1 2 3 4 5 6 Day

Figure S1 (Related to Figures 1-5). Generation and phenotypic characterization of knock-out (KO) clonal lines for either cingulin, paracingulin or both in Eph4, mCCD and MDCK cells.

(A-D) Generation of CRISPR-KO clones of either CGN (A-B) or CGNL1 (C-D) in Eph4 (A, C), mCCD (A, C) and MDCK (B, D) cells. Each scheme shows exon-intron structure, WT alleles sequence, guide RNA target sequence (red), mutations, insertions (”+”) and deletions (”-”)(bold) in each allele . Number of inserted or deleted nucleotides are indicated for each allele. Sequencing of the second allele for MDCK CGNL1-KO clone 2-5 was not successful probably because of huge insertion since there were two bands on the gel after amplification of the target region.

(E-J) Phenotypic characterization of CGN-KO clonal lines of Eph4 (E, H), mCCD (F, I) or MDCK (G, J) cells, either by immunofluorescence (E-G) or immunoblotting (I-J). Cells were labeled with antibodies against CGN and either PLEKHA7 or ZO-1, as reference junctional markers. Identifier for each clonal line is indicated on the left. Nuclei are stained with DAPI. Arrows indicate junctional localization, arrowheads indicate undetectable/decreased labeling. For immunoblots total RIPA lysates were used (STAR methods), and β-tubulin was used as a loading control. Numbers on the left indicate apparent size in kDa, based on the migration of prestained molecular weight markers.

(K-P) Phenotypic characterization of CGNL1-KO clonal lines of Eph4 (K, N), mCCD (L, O) or MDCK (M,P) cells, either by immunofluorescence (K-M) or immunoblotting (N-P).

(Q-R) Phenotypic characterization of double-KO (CGN+CGNL1) clonal lines of MDCK cells, either by immunoflu- orescence (Q) or immunoblotting (R). The asterisk in (R) indicates a non-specific polypeptide labelled by anti-CGNL1 antibodies.

(S) Proliferation curves of clonal KO lines of MDCK cells. bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Figure S2-1

**** MDCK CGN-WT A WT B 1.6 CGN-KO **** **** u

1.4 ns CFP-HA

Occludin HA MERGE

1.2

Zigzag index, a. ns ns

1.0 hCGN FL-HA Occludin HA MERGE

-CFP-HA+CFP-HA -CFP-HA+CFP-HA

-hCGN+hCGN FL-HA FL-HA -hCGN+hCGN FL-HA FL-HA -myc-hZO1+myc-hZO1 FL-HA FL-HA -myc-hZO1+myc-hZO1 FL-HA FL-HA hZO-1 FL-HA Occludin HA MERGE

C MDCK CGN-KO D MDCK WT CFP-HA

Occludin HA MERGE CGN HA PLEKHA6 hCGN FL-HA Occludin HA MERGE hZO-1 FL-HA CFP-HA CGN HA PLEKHA6 hZO-1 FL-HA Occludin HA MERGE

E MDCK WT+ CGNL1-KO(+YFP) F MDCK CGN-KO+rescue GFP *

NM2A GFP ZO-2 NM2B GFP ZO-2 *

NM2B GFP ZO-2 Phalloidin GFP ZO-2

Phalloidin GFP ZO-2 ARHGEF18 GFP ZO-2 ARHGEF18 GFP ZO-2 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Figure S2-2

G Eph4 WT+CGN-KO H Eph4 WT+CGNL1-KO

NM2A CGN PLEKHA7 NM2A CGNL1 PLEKHA7

NM2B CGN PLEKHA7 NM2B CGNL1 PLEKHA7

Phalloidin CGN ZO-2 Phalloidin CGNL1 ZO-2

-actin CGN ZO-2 -actin CGNL1 ZO-2

β-actin CGN ZO-2 β-actin CGNL1 ZO-2

I Eph4 CGN-KO+Rescue J Eph4 CGN-KO+Rescue

NM2B GFP-mCGN ZO-2 NM2B HA PLEKHA7 cCGNL1-FL-HA cCGN-FL-HA NM2B GFP ZO-2 NM2B HA PLEKHA7 cCGNL1(head) +cCGN(R+T)-HA NM2B HA PLEKHA7 Phalloidin GFP-mCGN ZO-2 cCGN(head)

+cCGNL1(R+T)-HA NM2B HA PLEKHA7 Phalloidin GFP ZO-2 CFP-HA

NM2B HA PLEKHA7 bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. Figure S2-3

K mCCD WT+CGN-KO Transwells L mCCD WT+CGNL1-KO Transwells

NM2A CGN PLEKHA7 NM2A CGNL1 PLEKHA7

NM2B CGN PLEKHA7 NM2B CGNL1 PLEKHA7

Phalloidin CGN PLEKHA7 Phalloidin CGNL1 PLEKHA7

-actin CGNL1 PLEKHA7

β-actin CGNL1 PLEKHA7

Figure S2 (Related to Figure 2). Cingulin controls membrane tortuosity in MDCK cells and recruits myo- sin2B (NM2B) to tight junctions of Eph4 and mCCD cells. (A) Quantification of zig-zag index in either CGN-KO or WT MDCK cells rescued with the indicated constructs. (B-C) Immunofluorescent (IF) localization of occludin and the indicated HA-tagged rescue constructs in either WT (B) or CGN-KO (C) MDCK cells. Lines in the occludin channel trace membranes to highlight shape. (D) IF localization of cingulin in MDCK WT cells transfected with the indicated HA-tagged constructs, showing increase in junctional cingulin induced by overexpression of ZO-1. (E) IF localization of NM2A, NM2B, actin filaments (phalloidin) and ARHGEF18 in mixed cultures of WT and CGNL1-KO MDCK cells. NB: CGNL1-KO cells were identified by stable expression of YFP, since endogenous levels of CGNL1 were too low to be reliably detected in WT-KO mixed cultures. Asterisks in (E) indicate diffuse apical (terminal web) labeling (F) IF localization of NM2B, F-actin (phalloidin), and ARHGEF18 in CGN-KO MDCK cells rescued with GFP alone. Either PLEKHA6 or ZO-2 were used as junctional reference marker in MDCK cells, that express low levels of PLEKHA7. (G-H) IF localization of NM2A, NM2B, actin filaments (phalloidin), -actin and β-actin in mixed cultures of either WT or CGN-KO Eph4 cells (G) or WT and CGNL1-KO Eph4 cells (H). Either PLEKHA7 or ZO-2 were used as junctional reference marker. (I-J) IF localization of NM2B and actin filaments (phalloidin) in CGN-KO Eph4 cells transfected with GFP-msC- GN-myc and GFP-myc (I) or localization of NM2B after transfection with indicated CGN/CGNL1 full length or chimaeric HA-tagged constructs (J). (K-L) IF localization of NM2A, NM2B, actin filaments (phalloidin) in mixed cultures of either WT or CGN-KO mCCD cells (K) or WT and CGNL1-KO mCCD cells (L). -actin and β-actin were examined also in WT-CGNL1-KO mixed cultures of mCCD (L). PLEKHA7 was used as junctional reference marker. Arrows= normal labeling, double arrows=increased labeling, arrowheads=reduced/undetectable/cytoplasmic label- ing. Scale bars, 10 μm. bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A G PDZ1 PDZ2PDZ3SH3U5 GUK U6 ABR ZU5 H I ZO-1

888-1619 Input =GFP tag G (GST)G-CGNG-CGN (1-121)G-CGNL1 (1-353)G G-CGN (1-119)G-CGN (1-121)G-CGNL1 (1-353) (1-119) 1150-1619 =GST tag GFP + - + + + + - - - - 888-1748 CGN(1-70) GFP-ZU5 - + - - - - + + + + =bait 1150-1748 1619-1748 =preys kDa CGN CGNL1 B 35 IB:GFP

GFP 888-1748 888-1619 1150-1748 25 150 1150-1619 1619-1748 50 100 35 50 Eph4 WT+ZO-1-KO % binding ZO-1 ZO-1 ND ND ND 0 Ponceau-S C 25 PDZ1 PDZ2PDZ3SH3U5 GUK U6 ABR ZU5 ZO-1 J

=GFP tag 1150-1619 u 2.5 **** 1550-1650 =GST tag 2.0 CGN(1-70) 1698-1748 PLEKHA7 PLEKHA7 =preys 1.5 =bait 1619-1748 1.0 D E 0.5 Relative intensity 0.0

CGNL1/PLEKHA7, a. WT

GFP (G)G-1150-1619G-1550-1650G-1698-1748G-1619-1748 100 controlGFP G-1150-1619(G) G-1550-1650G-1698-1748G-1619-1748 ZO1-KO 100 75 M 75 K GST-CGN 1-70 50 1 50 beads IB:GFP

25 IB:GFP 25 no beads INPUT-prey 37 50 IB: GFP 0.75 0 13 26 66 132 165 198 231 264 Bait=GST-CGN(1-70) (Ponceau S) GFP-ZO-1 1619-1748aa 100 0.5 F GFP 1698-1748 150 1150-1619 1619-1748 75 Kd= 40.4 nM L beads 1550-1650 50 0.25 100 FRACTION BOUND GFP-ZO-1 1619-1748 25 IB:GFP beads 50 50 IB: % binding 25 13 26 165 231 GFP 0 50 100 150 200 250 300 0 NDND Bait=GST(Ponceau S) beads GST-CGN 1-70 (nM)

Figure S3 (Related to Figure 3). Binding of cingulin and paracingulin to the ZU5 domain of ZO-1 (A-B) Relative binding of ZO-1 fragments to CGN(1-70). (A) Schematic diagram of ZO-1 with its structural mas sl1, S Srcml, , alae ase, , , acb- ing-region), and ZU5. Bait and prey proteins are shown with either red (GST) or green (GFP) tags, respectively. Numbers correspond to the amino acid residues comprised in each construct. (B) Histogram, based on densitometry analysis, showing relative binding of ZO-1 fragments to the CGN(1-70) bait, taking binding of ZU5 (residues 1619-1748) as 100%. (C-F) The integrity of ZU5 domain is required for interaction with cingulin. (C) Schematic diagram of GFP-tagged ZO-1 prey constructs and GST-CGN(1-70) bait used in pulldown. (D) Prey normalization. (E) Immunoblot analy- sis, using anti-GFP, of pulldowns using either GST-CGN(1-70)(top) or GST (bottom) as baits (PonceauS), and normalized GFP-tagged ZO-1 fragments as preys. (F) Histogram, based on densitometry analysis, showing relative binding of ZO-1 fragments to the CGN(1-70) bait, taking binding of the ZU5 as 100%.

(G) Interaction of ZU5 with fragments of cingulin and paracingulin globular head regions. Immunoblot analysis, using anti-GFP, of pulldowns using either GST or the indicated fragment of either CGN or CGNL1 as baits, and GFP-tagged ZU5 as a preys. Numbers correspond to the amino acid residues comprised in each construct. (H-J) Immunofluorescent localization of cingulin (H) and paracingulin (I) in mixed cultures of WT and ZO-KO Eph4 cells. Arrows and arrowheads indicate normal and undetectable junctional labeling, respectively. Scale bars, 10 μm. ss uantification of CGNL1 labeling with respect to the reference junctional marker PLEKHA7. (K-M) Measurement of the affinity of interaction between CGN(1-70) and ZU5. (K) Immunoblot analysis of supernatant depletion assay (STAR Methods). Depletion is achieved by adding increasing amounts of GST-CGN(1-70) beads to the supernatant. Supernatant depletion by beads alone is shown in (L). Numbers below each lane (K-L) indicate concentration (nM) of recombinant proteins. (M) Plots of equilibrium binding isotherms, where fraction of protein bound (total minus remaining) is plotted against the concentration of the GST-CGN(1-70) used for depletion, using either the ZU5 of ZO-1, or beads. bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S4-1 made available under aCC-BY-NC-ND 4.0 International license.

* A B 1.5 . 15 . * y y 1.0 10

0.5 5 ZO-1 rb CGN MERGE PLEKHA7 Relative intensit rb ZO-1/PLEKHA7, a.u

0.0 Relative intensit rb ZO-1 R3/PLEKHA7, a.u 0

WT WT

CGN-KO * CGN-KO 1.5 . y ZO-1 rb R3 CGN MERGE PLEKHA7 1.0 Eph4 CGN KO+WT mix Eph4 CGN KO+WT

0.5 Relative intensit ms ZO-1/PLEKHA7, a.u 0.0 ZO-1 ms CGN MERGE PLEKHA7 WT CGN-KO C D ** 2.0 1.5

**** . . y

y 1.5 1.0

ZO-1 rat CGN MERGE PLEKHA7 1.0 0.5 0.5 Relative intensit rb ZO-1/PLEKHA7, a.u Relative intensit rat ZO-1/PLEKHA7, a.u 0.0 0.0

WT WT

CGN-KO CGN-KO ZO-1 rb CGN MERGE PLEKHA7

. **** 2.5 1.5 *** .

y 2.0 y mCCD CGN KO+WT mix mCCD CGN KO+WT 1.0 1.5 ZO-1 rb R3 CGN MERGE PLEKHA7 1.0 0.5 0.5 Relative intensit ms ZO-1/PLEKHA7, a.u Relative intensit rb ZO-1 R3/PLEKHA7, a.u 0.0 0.0

WT WT

CGN-KO CGN-KO ZO-1 ms CGN MERGE PLEKHA7

ns ns E F 3 2.0 1.5 2 1.0 1 ZO-1 rb CGNL1 MERGE PLEKHA7 0.5

Relative intensity Rb ZO-1/PLEKHA7,a.u. 0 0.0

WT Relative intensity Rb ZO-1 R3/PLEKHA7,a.u. WT

CGNL1-KO CGNL1-KO

2.0 ns

ZO-1 rb R3 CGNL1 MERGE PLEKHA7 1.5

Eph4 CGNL1-KO+WT mix Eph4 CGNL1-KO+WT 1.0

0.5

Relative intensity ms ZO-1/PLEKHA7,a.u. 0.0

WT

ZO-1 ms CGNL1 MERGE PLEKHA7 CGNL1-KO bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S4-2 made available under aCC-BY-NC-ND 4.0 International license.

G H

ns . 4 2.0 ns

y 3 1.5

1.0 ZO-1 rat CGNL1 MERGE PLEKHA7 2

1 0.5 Relative intensity Rb ZO-1/PLEKHA7,a.u. Relative intensit rat ZO-1/PLEKHA7, a.u 0 0.0

WT WT

CGNL1-KO CGNL1-KO

ZO-1 rb CGNL1 MERGE PLEKHA7

ns 3 4 ns

mCCD CGNL1-KO+WT mix mCCD CGNL1-KO+WT 3 2

ZO-1 rb R3 CGNL1 MERGE PLEKHA7 2 1 1

0 Relative intensity ms ZO-1/PLEKHA7,a.u. 0 Relative intensity Rb ZO-1 R3/PLEKHA7,a.u. WT WT

CGNL1-KO CGNL1-KO

ZO-1 ms CGNL1 MERGE PLEKHA7

I J MDCKII Eph4 mCCD MDCKII Eph4 mCCD

kDa kDa WT CGN-KO 1 WT CGN-KO 1B5CGN-KO 1D6WT CGN-KO 3 CGN-KO 4 CGN-KOIB: 8 WT CGNL1-KO 2-2CGNL1-KO WT2-5 CGNL1-KO WT1A4 CGNL1-KO CGNL1-KO7 CGNL1-KO 8 IB: 9 ZO-1 Rat R4076 ZO-1 Rat R4076 250 250

250 ZO-1 Rb R3 ZO-1 Rb R3 250

250 ZO-1 Rb 617300 250 ZO-1 Rb 617300

250 ZO-1 Ms 339100 250 ZO-1 Ms 339100

150 CGN 150 CGNL1

50 βTubulin 50 βTubulin

K L Eph4 Eph4

ZO-2 CGN PLEKHA7 ZO-2 CGNL1 PLEKHA7 mCCD mCCD

ZO-2 CGNL1 PLEKHA7 CGN-KO+WT mix CGN-KO+WT ZO-2 CGN PLEKHA7 CGNL1-KO+WT mix CGNL1-KO+WT MDCKII MDCKII

ZO-2 CGN Occludin ZO-2 CGNL1 Occludin

MDCKII Eph4 mCCD M MDCKII Eph4 mCCD N

kDa kDa WT CGN KO 1 WT CGN KO 1B5CGN KO 1D6WT CGN KO 3 CGN KO 4 CGN KO 8 WT CGNL1-KO 2-2CGNL1-KO WT2-5 CGNL1-KO WT1A4 CGNL1-KO CGNL1-KO7 CGNL1-KO 8 9

150 IB:ZO-2 150 IB:ZO-2

150 IB:CGN 150 IB:CGNL1

50 IB:βTubulin 50 IB:βTubulin bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S4-3 made available under aCC-BY-NC-ND 4.0 International license.

O Q mCCD mCCD

ZO-3 CGN MERGE PLEKHA7 mix CGNL1 KO+WT ZO-3 CGNL1 PLEKHA7 CGN-KO+WT mix CGN-KO+WT WT MDCKII

ZO-3 CGN MERGE PLEKHA6 ZO-3 Occludin MERGE MDCKII CGNL1-KO ZO-3 Occludin MERGE

mCCD MDCK mCCD MDCK .

P R . ns

u ns u **** 0.8 2.0 **** 1.0 1.5

0.8 0.6 1.5 1.0 0.6 0.4 1.0 ZO-3/PLEKHA6, a. , ZO-3/Occludin, a.u , ZO-3/PLEKHA7, a.u ZO-3/PLEKHA7, a. y y y, y, 0.4 0.5 0.2 0.5 0.2

0.0 0.0 0.0 0.0 Relative intensit Relative intensit Relative intensit Relative intensit WT WT WT WT

CGN-KO CGN-KO CGNL1-KO CGNL1-KO

S MDCKII mCCD T MDCKII mCCD

kDa WT CGN-KO 1 CGN-KO 1H9CGN-KO WT1G8 CGN-KO 3 CGN-KO 4 CGN-KO 8 kDa WT CGNL1-KO 2-2CGNL1-KO 2-5WT CGNL1-KO CGNL1-KO7 CGNL1-KO 8 9 150 150 IB:ZO-3 IB:ZO-3 100 100 150 IB:CGN 150 IB:CGNL1

50 IB:βTubulin 50 IB:βTubulin

U MDCK CGN-KO+rescue V

4 u

CFP-HA **** ZO-3 HA MERGE PLEKHA6 3

ZO-3/PLEKHA6, a. 2 y, ns * hCGN_FL-HA ZO-3 HA MERGE PLEKHA6 1 Relative intensit 0

-CFP-HA +CFP-HA

hZO-1_FL-HA -hCGN_FL-HA -hZO1_FL-HA+hZO1_FL-HA ZO-3 HA MERGE PLEKHA6 +hCGN_FL-HA bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.Figure It is S4-4 made available under aCC-BY-NC-ND 4.0 International license.

Figure S4 (Related to Figure 4). Cingulin but not paracingulin promote ZO-1 and ZO-3 accumulation at tight junctions.

(A-H, K-L, O-R) IF analysis of the junctional accumulation of ZO-1 (A, C, E, G), ZO-2 (K, L), ZO-3 (O,Q) and quantifications of ZO-1 labeling (B, D, F, H), ZO-3 labeling (P, R) with respect to the reference junctional marker PLEKHA7 or occludin (for MDCK) either in mixes of CGN-KO and WT cells (A-D, K, O-P), or in mixes of CGNL1-KO and WT cells (E-H, L, Q-R), using indicated antibodies against ZO-1, ZO-2 or ZO-3. Experiments were done using Eph4, MDCK and mCCD cells, as indicated. Separate cultures of WT and CGNL1-KO MDCK were used to examine ZO-3 labeling (Q-R).

(I-J, M-N, S-T). Immunoblotting analysis of lysates of WT and CGN-KO lines (I, M, S), and WT and CGNL1-KO lines (J, N, T) using different antibodies against ZO-1 (I-J), ZO-2 (M-N) or ZO-3 (S-T). Antibodies against either CGN (I, M, S) or CGNL1 (J, N, T), and β-tubulin were used as line phenotype and loading controls, respectively.

(U) IF analysis of the junctional accumulation of ZO-3 in MDCK CGN-KO after rescue with either HA-tagged human CGN, ZO-1 or CFP. PLEKHA6 was used as a junctional reference.

(V) Quantification of ZO-3 junctional IF signal in MDCK CGN-KO cells after rescue with either CFP-HA or full-length human CGN or ZO-1 constructs (see panels U).

Scale bars r , 10 μm. aa ls reresee as meaS. Sal eses are s arbrar s (a.u.). bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B GFP-cCGN GFP-cCGNL1 GFP

GFP-CGNL1 A’ 0 sec 4 sec ZO-1 ZO-1 ZO-1

8 sec 12 sec

CGNL1 in WT GFP in WT C D NM2B GFP-cCGNL1 NM2B GFP NM2B Phalloidin GFP-cCGNL1 Phalloidin GFP Phalloidin GFP-cCGNL1 ARHGEF18 GFP ARHGEF18 ARHGEF18

Figure S5 (Related to Figure 5). Cingulin and paracingulin condensates do not recruit ZO-1.

(A) CGNL1 form droplets that fuse into larger structures over time (A’), suggesting the formation of phase-separat- ed condensates.

(B) IF analysis of ZO-1 in condensates of either GFP-CGN or GFP-CGNL1 or after overexpression of GFP (nega- tive control) in MDCK WT cells. Arrows indicate condensates, arrowheads indicate lack of accumulation of ZO-1.

(C-D) IF analysis of NM2B, F-actin (TRITC-phalloidin) and ARHGEF18 in cells overexpressing either GFP-CGNL1 (C) or GFP alone (D). Arrows indicate condensates, arrows and arrowheads indicate co-accumulation or lack of detectable co-accumulation in condensates, and double arrowheads indicate redistributed subcortical labe- ling, respectively.

Scale bars, 10 μm, ece r 0.0 μm. bioRxiv preprint doi: https://doi.org/10.1101/2020.05.14.095364; this version posted May 15, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

A B

C E ZO-1 domains ZO-1 ZU5 domain

ABR (actin-binding region) CGN-KO

ZO-2, ZO-3, other interactors and LLPS inducers

claudin occludin JAM cingulin

-actin filaments +CGN NM2B filaments D

WT

ZU5 unfolded

Figure S6 (Related to Figure 2, Figure 3, Figure 5). Scheme linking conformational states of ZO-1 to junc- tion association and interactions with cingulin, actin filaments and other proteins

(A) Cytoplasmic ZO-1 (soluble, folded). ZU5 undergoes intramolecular interactions with ZPSG (Spadaro et al, 2017) and ZU5 domains. (B) Junction-associated ZO-1 in the absence of ZO-2 and cingulin, and with disrupted actomyosin organiza- tion/tension interacts with claudins and JAM-A, but does not recruit either occludin or DbpA (Figure 3, (Spadaro et al., 2017)). NB: in the presence of cingulin the ZU5 domain would unfold, but this is not sufficient to induce full ZO-1 stretching. (C) Junction-associated ZO-1 in the presence of interactors of the PDZ2-PDZ3-SH3-GUK and C-terminal domains, some of which may indirectly link ZO-1 to actin and myosin, and additional factors (phosphorylation (Beutel et al, 2019)) that induce liquid-liquid phase separation (LLPS) becomes unfolded and phase separates. However in the absence of cingulin (CGN-KO) the ZU5 domain is folded onto the ABR domain, inhibiting its interaction with actin filaments. (D) The ZO-1 configuration shown in C (CGN-KO) plus the addition of cingulin allows unfolding of the ZU5 domain and organization of the NM2B--actin web. (E) Key for identification of graphical elements.