Professional Documents
Culture Documents
2023 07 04 547654v2 Full
2023 07 04 547654v2 Full
1 Somatic mutation of Afadin leads to anchorage independent survival and metastatic growth of
3 Max A.K. Ratze1, Lotte N.F.L. Enserink1, Noboru Ishiyama2, Christina H.J. Veltman1, Isaac J.
4 Nijman3, Rene Bernards4, Paul J. van Diest1, Matthias Christgen5, Patrick W.B. Derksen1#
5
6 1 Department of Pathology, University Medical Center Utrecht, Utrecht, The Netherlands.
8 3 Center for Molecular Medicine, Cancer Genomics Netherlands, Department of Genetics, University
10 4 Division of Molecular Carcinogenesis, Center for Biomedical Genetics and Cancer Genomics
13
14 Running title: Afadin is a tumor suppressor for ILC
15 Key words: Metastatic Breast Cancer, Afadin, E-cadherin, Adherens Junction, α-catenin
16
17 #Correspondence to:
21 p.w.b.derksen@umcutrecht.nl
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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.
22 ABSTRACT
23 Loss of E-cadherin (CDH1) and the adherens junction (AJ) drive development and progression of
24 invasive lobular breast cancer (ILC). However, approximately 40% retain wild type CDH1 alleles,
25 indicating that modulation of other genes attenuates the AJ during ILC etiology. To identify alternative
26 drivers, we performed targeted sequencing in CDH1 wild type samples, based on a defined set of 100
27 AJ, tight junction, and desmosome genes we designated as the ‘Adhesome’. In 146 ILC samples, we
28 identified 62 cases (43%) with wild type CDH1 alleles in which we detected a total of 284 mutations in
29 36 Adhesome genes. After selection based on occurrence and potential loss of function, we identified
31 Functional studies in E-cadherin-expressing breast cancer cells showed that Afadin knockout leads to
33 resistance, which are classical ILC hallmarks. Afadin reconstitutions show that F-actin organization
34 critically depends on the E-catenin binding CC domain. Afadin loss in intraductal xenograft mouse
35 breast cancer models leads to ILC-type morphologies and overt lung metastases. AFDN truncate
36 reconstitutions revealed that deletion of the C-terminal E-catenin binding CC domain is sufficient to
37 drive metastatic ILC. In conclusion, we identified and functionally coupled a somatic frameshift AFDN
38 mutation in breast cancer to destabilization the epithelial AJ and the development of ILC hallmarks
39 such as actomyosin-dependent anoikis resistance and single cell invasion. As such, Afadin
40 represents a candidate tumor suppressor for E-cadherin-positive ILC development and progression.
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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 INTRODUCTION
42 Invasive lobular carcinoma (ILC) constitutes up to 15% of all breast cancers, making it the second
43 most common breast cancer type. ILC is characterized by non-cohesive growth patterns and diffuse
44 dissemination to specific sites such as the gastrointestinal tract, ovaries and peritoneum (Rakha &
45 Ellis, 2010). ILC development and progression are caused by a functional inactivation of E-cadherin
46 (CDH1) (Berx et al, 1995; Derksen et al, 2011, 2006), a transmembrane molecule that represents the
47 principal component of the adherens junction (AJ) (Meng & Takeichi, 2009). A general hallmark of ILC
48 cells is the acquisition of anchorage independent survival, also called anoikis resistance (Christgen &
49 Derksen, 2015). Anoikis resistance in ILC cells is primarily driven by an autocrine activation of growth
50 factor receptors and their effectors such as PI3K/AKT/FOXO/BMF (Hornsveld et al, 2016; Nagle et al,
51 2018; Teo et al, 2018), and constitutive activation of Rock1-dependent actomyosin contraction
52 (Schackmann et al, 2011; Ven et al, 2015). Notwithstanding this, it is well established that
53 approximately 10% of ILC cases preserve membranous E-cadherin expression (Reed et al, 2016;
54 Grabenstetter et al, 2020). Moreover, somatic inactivating mutations in CDH1 do not fully account for
55 the observed E-cadherin loss in ILC, as approximately 30-40% retain the wild-type alleles (Ciriello et
56 al, 2015; Desmedt et al, 2016; Pareja et al, 2020). These findings imply that alternative drivers may
58 The epithelial AJ is essential for cell-cell connections between two neighboring cells. Extracellular E-
59 cadherin homotypic interactions in cis and in trans provide a mechanical signaling hub within the AJ
60 through linkage to the actin cytoskeleton (Yamada et al, 2005; Nelson, 2008; Kurita et al, 2011). Upon
61 cell-cell contact, the cytosolic E-cadherin tail interacts with p120-catenin (CTNND1), β-catenin
62 (CTNNB1) and αE-catenin (CTNNA1, α-catenin from hereon), proteins that are essential for the
63 stability of the AJ (Anastasiadis et al, 2000; Davis et al, 2003) and for mechanical and biochemical
64 signal transduction towards F-actin respectively (reviewed in: (Leckband & Rooij, 2014)). Linking of
65 the cytoskeleton occurs by indirect binding through either classical Cadherins or Nectins via their
66 associated accessory proteins. Connections to the F-actin cytoskeleton are dependent on binding to
67 β-catenin (CTNNB1) and the subsequent interaction with α-catenin, facilitating mechanically
68 transduced cues via the F-actin cytoskeleton and its dynamics (Meng & Takeichi, 2009; Shapiro &
69 Weis, 2009; Efimova & Svitkina, 2018). Interestingly, interaction of α-catenin with the F-actin
70 cytoskeleton is dependent on interactions with additional proteins such as Vinculin and Eplin to build
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
71 up the initial forces along the F-actin cytoskeleton and modulate the actin cytoskeleton in response to
72 forces applied to the AJ (Weiss et al, 1998; Dufour et al, 2013; Collins et al, 2015). Interestingly, α-
73 catenin has been advocated as an alternative mode for tumor suppression in malignancies that are
74 caused by E-cadherin dysfunction, such as ILC and diffuse gastric cancer (Majewski et al, 2013; Blair
75 et al, 2020). Functional follow-up studies in breast cancer subsequently showed that α-catenin loss
76 indeed leads to ILC-like phenotypes in culture and in mice, and acquisition of actomyosin-dependent
77 anoikis resistance (Hollestelle et al, 2010; Groot et al, 2018). Together, these findings indicate that
78 loss of components that function in cell-cell contacts (other than E-cadherin) can induce
80 E-cadherin and Nectin both control linkage to the F-actin cytoskeleton. Unlike E-cadherin, Nectin
81 mediates F-actin modulation primarily through Afadin (AFDN or MLLT4) (Mandai et al, 1997; Rikitake
82 & Takai, 2011) (Reviewed in: (Huxham et al, 2021)). The longer l-Afadin splice variant (hereafter
83 Afadin) controls these connections as a multi-domain scaffolding protein that interacts with various
84 tight junction (TJ) components, indirect connections through α-catenin and the AJ (Sakakibara et al,
85 2020), or direct interaction with F-actin via its Coiled Coil (CC) and/or F-Actin Binding (FAB) domains
86 (Mandai et al, 1997; Choi et al, 2016; Carminati et al, 2016). Moreover, binding of AFDN with LMO7
87 and ADIP allow for interaction of Nectin-based junctions with Cadherin-based AJs (Asada et al, 2003;
88 Ooshio et al, 2004). Its favorable localization due to its function to link the F-actin cytoskeleton to the
89 nectin-based AJ, enables it to support the connection between the F-actin cytoskeleton and the AJ
91 Cell-cell interactions are additionally reinforced by the formation of TJs, multiprotein junctional
92 complexes that are essential in the formation of a functional barrier in epithelium and endothelium and
93 correct apical polarization of cells (Zihni et al, 2016). Like the AJ, TJs facilitate cell-cell interactions
94 through a complex arrangement of transmembrane proteins that interact cytoplasmic adaptors such
95 as zona occludens proteins (Morita et al, 1999). Zona occludens 1 (ZO1) links the TJ to the AJ
96 through complex formation with α-catenin and recruitment of JAMA to the Nectin-based junctions
97 (Fukuhara et al, 2002; Maiers et al, 2013). As such, proteins within the TJ may be critical to the tumor
98 suppressive function of the AJ. Indeed, inactivation of the TJ polarity protein PAR3 induces
99 destabilization of AJs, leading to disrupted actin dynamics and invasion of breast cancer cell
101 Here, we have performed targeted sequencing on ILC samples to reveal alternative drivers of ILC. In
102 a cohort of clinical samples that had retained wild type CDH1 alleles, we identify Afadin as a
103 candidate tumor suppressor for the development and progression of ILC. Functionally, loss of Afadin
104 induces an E-cadherin positive ILC phenotype, and tumor progression through the acquisition of
106 RESULTS
108 To identify alternative drivers of ILC, we performed a focused gene set exome sequencing analysis on
109 86 genes that play an essential role in epithelial cell-cell adhesion. This gene set (from hereon: the
110 Adhesome) contains genes coding for AJ, TJ and desmosome (DS) proteins (Table 1). In total, 154
111 ILC samples from the RATHER cohort (Michaut et al, 2016) were sequenced using Adhesome-
112 targeted Next-Generation Sequencing, of which 146 samples passed the sequencing quality checks
113 (Fig. 1A). Deleterious CDH1 mutations were found in 84 samples (57.5% of total), which is in line with
114 previously published numbers for CDH1 status in ILC (Berx et al, 1995; Pareja et al, 2020). To enable
115 identification of alternative ILC drivers, CDH1 mutant samples were excluded. Also, mutations that
116 occurred more than three times were considered nucleotide polymorphisms and were omitted from
117 further analysis. In the remaining 62 samples, 284 non-synonymous mutations were identified in a
118 total of 78 different genes (Fig. 1B). We next focused on mutations with a potential major impact on
119 loss of protein function (start lost/stop gained, frameshift and missense mutations), which led to the
120 identification of Afadin (AFDN), which we prioritized based on its published established roles in cell-
121 cell adhesion. In this ILC sample, AFDN contained a heterozygous p.Lys630Stop frameshift mutation
122 that potentially leads to truncation and loss of Afadin protein expression (Fig. 1C). Next, we examined
123 the corresponding primary ILC sample (R-L1047-C), which featured small to medium-sized breast
124 cancer cells arranged in one- to two-cell thick trabeculae and single cell files (Fig. 1D). Foci with
125 dissociative growth were also evident. Mammary ducts were encircled by tumor cells in a targetoid
126 fashion and tumor cells accumulated at the border from connective to adipose tissue. Tumor cells
127 showed scanty cytosol, delicate nuclear chromatin and focal biconcave nuclear compressions. Mitotic
128 activity was low. The final diagnosis was ILC grade 2. Importantly, using immunofluorescence, we
129 observed that Afadin expression was lost in this sample, accompanied by a relatively weak and non-
131
132 Somatic AFDN mutations are associated with CDH1 wild-type lobular breast cancer
133 Based on our identification of an inactivating AFDN frameshift mutation, we analyzed the occurrence
134 and frequency of AFDN mutations in the publicly available breast cancer databases METABRIC and
135 TCGA. Overall, in TCGA and METABRIC combined, 1/142 tumors (0.7%) have had a reported
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
136 truncating mutation, which is similar to our study (1/146, 0.7%). In total, 79 unique somatic AFDN
137 mutations were reported in a total of n=88 mammary carcinomas, of which 70 (79.5%) were IDC-NST
138 (invasive ductal carcinoma, no special type), 10 (11.4%) ILC and 5 (5.6%) mixed Ductal and Lobular
139 carcinoma cases (Suppl. Table 1). The remaining 3 cases (3.3%) were reported without specified
140 histologic subtype. Among the 54 missense mutations, 34 (63.0%) were associated with decreased
141 Afadin mRNA expression (Suppl. Table 1). Of the 27 AFDN truncating (Frame-Shift and Nonsense)
142 mutations reported within METABRIC and TCGA, 24 mutations (88.9%) are predicted to cause loss of
143 the C-terminal domain of AFDN, leading to loss of both the α-catenin binding site and FAB domain
145 We next investigated if CDH1 and AFDN mutations co-occurred in the METABRIC and TCGA breast
146 cancer datasets and found significant overall mutual exclusivity for CDH1 and AFDN (p<0.0001; Fig.
147 2B), especially in the subtypes ILC and mixed ductal/lobular carcinoma (p<0.0001; Fig. 2B). In
148 contrast, and as expected, we found that CDH1 and AFDN mutations do not occur mutually exclusive
150 containing AFDN mutations to confirm the initial differential diagnosis. We revised one case that was
151 initially diagnosed as IDC-NST as a mixed Ductal and Lobular phenotype (TCGA-BH-A0DZ-01,
152 Suppl. Table 1). Of note, we revised two cases as apocrine breast cancer, a subtype that normally
153 has an occurrence of 1% (Dellapasqua et al. 2013), compared to 10% in this subset of AFDN mutant
154 tumors.
155
156 Inactivation of AFDN leads to loss of cell-cell contact and anoikis resistance in breast cancer
157 cells.
158 To investigate the consequences of Afadin loss in breast cancer, we performed CRISPR/Cas9-
159 mediated knockout of AFDN in MCF7 cells, an E-cadherin expressing and ER positive human breast
160 cancer cell line (Fig. 3A). After transduction, Afadin knockout cells (MCF7::∆AFDN) were selected,
161 expanded and subjected to phenotypical analysis. In contrast to wild type MCF7 cells, MCF7::∆AFDN
162 cells did not show a cobblestone epithelial phenotype, but instead grew as non-cohesive motile cells
163 (Fig. 3B). Because anchorage-independence is a hallmark of ILC cells (Derksen et al, 2011, 2006),
164 we next assessed if Afadin loss induces anoikis resistance. Indeed, MCF7::∆AFDN cells cultured in
166 cadherin loss in MCF7::∆CDH1 (Fig. 3C & D) (Hornsveld et al, 2016). Moreover, using inhibitors
167 targeting RhoA and Rock1 (C3 and Y27632), we showed that the anoikis resistance of MCF7::∆AFDN
169 MCF7::∆CDH1 cells (Fig. 3E), allowing to conclude that Afadin loss induces an actomyosin-
171
172 AFDN loss attenuates proper adherens junction formation and F-actin linkage
173 In contrast to MCF7 wild type cells, MCF7::∆AFDN cells show an aberrant expression of E-cadherin
174 on the plasma membrane, similar to the E-cadherin expression patterns in the AFDN mutant ILC
175 sample. Immunofluorescence (IF) for Afadin confirmed the successful knockout (Fig. 4A).
176 Concomitant IF for E-cadherin, β-catenin and α-catenin showed that an AJ was formed on the plasma
177 membrane in MCF7::∆AFDN cells (Fig. 4A, Suppl. Fig. 1A and 1B), but with a patchy, punctate
178 expression pattern (Fig. 4A). In agreement with published literature (Sakakibara et al, 2020), we found
179 that Zona Occludens 1 (ZO1) Fig. expression is reduced at the tri-cellular junctions compared to wild
180 type controls (Suppl. Fig. 1C). Importantly, F-actin IF expression analysis showed that cortical F-actin
181 distribution is less confined at the plasma membrane when compared to MCF wild type cells (Fig. 4A).
182 To quantify these differences, we examined the spatial E-cadherin/F-actin expression and distribution
183 at the plasma membrane using perpendicular line scans (Fig. 4B), and calculated the ratio of F-actin
184 over E-cadherin expression at the Full Width at Half Maximum Height (FWHM). These analyses
185 confirmed that loss of AFDN in MCF7::∆AFDN cells leads to diffused distribution of cortical F-actin
186 (Fig. 4C & D). Our findings therefore indicate that Afadin controls proper linkage from E-cadherin to
187 the F-actin cytoskeleton. Furthermore, our work suggests that Afadin loss in breast cancer cells
188 induces improper F-actin mechanical tension, leading to constitutive deregulation of actomyosin
189 contraction, which has been causally linked to anoikis resistance in ILC (Schackmann et al, 2011;
191
192 The CC domain of Afadin is essential for proper AJ formation
193 The identified AFDN mutations indicate that most truncating mutations lead to loss of the C-terminal
194 domain, causing impaiment of binding to α-catenin and F-actin respectively. We therefore
195 hypothesized that an Afadin dependent F-actin to AJ connection occurs through either binding of α-
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
196 catenin to the Afadin CC-domain, through the Afadin FAB-domain, or both. To investigate the
197 individual requirement, we used full length N-terminally GFP-tagged rat l-Afadin (FL) (Sakakibara et
198 al, 2018) and generated truncated constructs lacking either the CC-domain containing the α-catenin
199 binding region (∆CC) the FAB-domain (∆FAB), or the entire C-terminus from amino acid position 1391
200 onward (∆Cterm)(Fig. 5A). We next performed reconstitution experiments in MCF7::∆AFDN cells,
201 confirming expression at the correct molecular weight of either the full length or truncated constructs
202 using a GFP antibody (Fig. 5B). Expression of either the FL or ∆FAB constructs in MCF7::∆AFDN
203 cells fully rescued the non-cohesive phenotype, resulting in proper cell-cell adhesion and a
204 cobblestone morphology (Fig. 5C). In contrast, reconstitution with either the ∆CC or ∆Cterm truncates
205 did not restore the epithelial phenotype in MCF7::∆AFDN cells (Fig 5C). Although both truncates did
206 not rescue epithelial morphology, cells reconstituted with ∆CC constructs showed a hypomorphic
207 phenotype, i.e. a partial retainment of the migratory morphology and spindle-shaped cells (Fig. 5C). In
208 line with the cellular phenotypes, subsequent examination and quantification of the spatial E-
209 cadherin/F-actin expression at the plasma membrane confirmed that the FL and ∆FAB reconstitution
210 fully rescue proper co-localization of cortical F-actin to E-cadherin (Fig. 5D). In contrast to the FL and
211 ∆FAB rescue experiments, we observe that neither the ∆CC, nor the ∆Cterm reconstitution is able to
212 rescue Afadin loss, leading to higher F-actin/E-cadherin ratios, i.e. dispersed cortical F-actin (Fig. 5E).
213 Our reconstitution experiments in AFDN knockout cells thus show that the coiled-coil region of Afadin
214 that contains the α-catenin binding domain, is required for proper AJ formation and linkage to the F-
215 actin cytoskeleton in E-cadherin expressing breast cancer cells. We conclude that attenuated cortical
216 F-actin distribution may underpin the acquisition of Rho-Rock-dependent anoikis resistance upon
218
219 Loss of AFDN leads to an ILC phenotypical features and lung metastasis phenotype in a
221 To assess the functional impact of AFDN loss in breast cancer, we performed mammary intraductal
222 injections, using MCF7::∆AFDN cells that had been stably reconstituted with either the ∆CC or
223 ∆Cterm GFP-tagged AFDN truncate. In this setup we monitored longitudinal tumor volumes and
224 observed no significant differences in primary tumor size between the tumor cohorts, indicating that
225 Afadin loss does not impact primary tumor growth (Fig. 6A). All cohorts had retained membranous
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
226 expression of E-cadherin (Fig. 6B, middle panels) with carcinomas in the rescue cohorts expressing
227 GFP, confirming proper expression of the constructs after transplantation (Fig. 6B, bottom panels).
228 Moreover, carcinomas in the wild type MCF7 tumor cohort were characterized by expansive,
229 circumscribed growth with occasional collective invasion of cells into the duct (Fig. 6C). In contrast,
230 mice injected with MCF7::∆AFDN or the ∆αCAT/∆Cterm reconstituted cells showed typical ILC-like
231 features, including overt single cell invasion and the formation of trabeculae and single or Indian files,
232 most notably at the invasive front (Fig. 6C). Interestingly, we observed a significantly shorter overall
233 survival of mice that were injected with either the MCF7::∆AFDN knockout cells or the ∆Cterm
234 truncate reconstituted cells (p<0.05)(Fig. 6D). This was supported by single cell intravasation of
235 cancer cells in the MCF7::∆AFDN knockout cells and both the ∆CC and ∆Cterm reconstitution
236 conditions (Fig. 6E). Examination of the lung tissues showed that the decreased survival in these
237 mice was due to the development of overt metastasis of cells expressing human GATA3, used as a
238 marker for human cells (Fig. 6F and 6G). Although reconstitution with ∆CC did not result in a
239 significant survival decrease, we detected an increase in metastatic burden after quantification of the
240 GATA3 positive cells, similar to the increase observed after reconstitution with the ∆Cterm truncate
242
243 We conclude that loss of Afadin expression in E-cadherin expressing breast cancer cells induces the
244 formation of ILC-type metastatic tumors. Our data reveal an essential tumor suppressive role for
245 Afadin in breast cancer progression. Afadin is essential to control a proper and matured connection of
246 E-cadherin to the F-actin cytoskeleton to prevent loss of cell-cell contacts, the acquisition of anoikis
247 resistance, and the progression towards ILC-like metastatic tumors (see Fig. 7).
248 In sum, we have identified Afadin as a candidate tumor suppressor for ILC progression.
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
249 DISCUSSION
250 Loss of E-cadherin underpins ILC biology and etiology (reviewed in: (Christgen et al, 2016)). The
251 biochemical consequences of E-cadherin loss are converging on the activation of survival pathways
252 driven by growth factor receptor activation (Bajrami et al, 2018; Nagle et al, 2018; Teo et al, 2018)
253 and constitutive actomyosin contraction (Schackmann et al, 2011; Ven et al, 2015; Groot et al, 2018).
254 Complete loss of E-cadherin expression in ILC appears to drive actomyosin contraction through loss
255 of cell contacts and autocrine activation of RhoA through the suppression of the Rock antagonist
256 MRIP (Schackmann et al, 2011). However, it remained less clear what propels F-actin contraction in
257 the 10% of ILC cases that have retained E-cadherin expression.
258 Previous data have connected loss of α-catenin to E-cadherin expressing ILC cell lines (Hazan et al,
259 1997; Hollestelle et al, 2010), and provided functional evidence that loss of α-catenin leads to an ILC-
260 like phenotype and actomyosin-dependent anoikis resistance (Groot et al, 2018). In short, the
261 aforementioned studies indicated that an imbalanced actomyosin-dependent force distribution due to
262 loss of proper linkage of E-cadherin to the F-actin cytoskeleton, may be sufficient to drive single cell
263 survival and dissemination during ILC progression. In the current study we have used a combination
264 of genomics, cell biology and mouse experiments to identify new alternative drivers of ILC. Based on
265 the strong dependency on a balanced E-cadherin to F-actin connection, we have performed targeted
266 next generation DNA sequencing on ILC cases that retained the wild type CDH1 alleles and identified
267 the AFDN p.Lys630stop frameshift mutation. Afadin is linked to the metastasis free survival of breast
268 cancer (Letessier et al, 2007; Fournier et al, 2011), and recent retrospective phylogenetic analyses
269 identified a CDH1 wild type ILC case that presented with an AFDN deletion (Fimereli et al, 2022).
270 Interestingly, we found that 82% of the identified AFDN mutations in the TCGA and METABRIC
271 databases (0.4% of total) occur in IDC-NST, which suggests that some cases were potentially
272 misdiagnosed and may in fact histologically be ILC or mixed IDC/ILC. Indeed, from the twenty AFDN
273 mutant TCGA cases that had available histology data, we could identify one case being misdiagnosed
274 as IDC-NST.
275 Experimentally, loss of Afadin in E-cadherin expressing luminal-type MCF7 cells leads to less
276 confined cortical localization of F-actin at the plasma membrane and subsequent colocalization with
277 the E-cadherin based AJ, indicative of incomplete maturation of the AJ. With the residual but aberrant
278 localization of wild type E-cadherin complexes on the membrane, we conclude that the tumor
279 suppressor function of Afadin is derived from the attenuated association and incomplete maturation of
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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 the E-cadherin-F-actin cytoskeleton link. In the case of Afadin loss, we found that the coiled-coil
281 domain containing an α-catenin binding region is necessary and sufficient to confer proper
282 actomyosin organization at the AJ. Because decreased cortical F-actin localization and contraction
283 are caused by reduced buildup of forces along the F-actin fibers connecting the cortical F-actin to the
284 AJ (Vasioukhin et al, 2000; Verma et al, 2004; Sumida et al, 2011; Taguchi et al, 2011), we conclude
285 that this lack of pulling forces is strongly exacerbated in the absence of Afadin, where insufficient
286 coupling of E-cadherin to F-actin prevents apical constriction. Consequently, Rho Kinase may be
287 released from Shroom3 and drive actomyosin contraction (Kimura et al, 1996; Uehata et al, 1997), a
288 process that will eventually drive ILC formation and metastasis. One of the causes of the reduction in
289 mechano-transduction upon Afadin loss could be the forces of F-actin exerted on Vinculin during
290 initial AJ formation, which is essential for establishing the AJ and F-actin connection (Duc et al, 2010).
291 It has become clear that α-catenin is the rate-limiting tension sensor upon stretch activation between
292 E-cadherin and F-actin (Yonemura et al, 2010), whereby Vinculin interacts with α-catenin only after
293 unfolding of α-catenin (Yao et al, 2014). Upon AJ maturation following a balanced interplay of forces,
294 localized expression levels of Vinculin are reduced in the junction, rendering a stabilized AJ including
295 Afadin (Rooij, 2014). Previous work by the Takai laboratory demonstrated that Afadin is required for
296 proper AJ formation, but that the FAB domain is dispensable for this event (Sakakibara et al, 2020).
297 Our work is in full agreement with these findings, showing that reconstitution of an Afadin truncate
298 lacking the C-terminal FAB domain fully rescues correct formation of the E-cadherin to F-actin
299 connection, and the restoration of an epithelial phenotype. In contrast, we show that reconstitution
300 with deletion truncates lacking the CC domain prevents cortical actin colocalization with the E-
301 cadherin based AJ, which shows that direct association of the CC domain and Afadin upon AJ
303 It is nonetheless striking that the C-terminal FAB domain is dispensable for proper AJ formation and
304 function in breast cancer cells. Previous studies in Drosophila have demonstrated that while the FAB
305 domain is not essential for AJ formation, it supports the stability of AJ under tension (Perez-Vale et al,
306 2021). In non-malignant mammalian cells, loss the FAB domain appears to delay the formation of
307 adherens and tight junctions (Sakakibara et al, 2018). Interestingly, the CC domain of Afadin contains
308 four predicted alpha helices, of which the first N-terminal helix overlaps with the mapped binding site
309 of Afadin to α-catenin (Maruo et al, 2018). Recent analyses by the Peifer lab indicate that the
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
310 predicted C-terminal helices show a near complete overlap with the F-actin binding site identified by
311 Carminati et al. (Carminati et al, 2016), suggesting that the CC domain might not only confer α-
312 catenin binding, but also F-actin interaction, or both (preprint Gurley et al, 2023). In short, based on
313 our work and the aforementioned studies we propose that either: (i) the FAB domain binds F-actin
314 and enforces AJ stability but is not solely responsible for F-actin linkage, (ii) all major functions related
315 to AJ stability are deployed by Afadin through α-catenin dependent F-actin linkage, or (iii) the
316 predicted F-actin binding sites in the CC intrinsically disordered region (IDR) are dominant in
317 facilitating AJ formation and stability together with α-catenin linkage. Because we observe that
318 deletion of the CC domain is sufficient to disturb epithelial integrity in breast cancer cells through
319 improper AJ formation, induce actomyosin dependent anoikis resistance, and drive invasive tumor
320 growth and metastatic dissemination, it remains currently unclear how interplay of the scenarios
321 controls the phenotypes observed. Future work will have to delineate the exact mechanisms that
322 underpin the role of Afadin in AJ formation and function through fine-mapping of the predicted F-actin
324 Regardless of the precise mode of action, we envision that in parallel to the Afadin-α-catenin bond,
325 the concomitant linkage of Afadin to F-actin serves to further stabilize the AJ and allow subsequent
326 build-up of forces that can then reach a threshold for the reduction of Vinculin levels. In this scenario,
327 Afadin would function as a gatekeeper of AJ maturation through the exposure of force dependent
328 unfolding of the Afadin binding MIII subdomain/region of α-catenin (Sakakibara et al, 2020). While the
329 exact balance of forces and domain requirements are still unclear, our work indicates that the
330 necessity of Afadin for proper maturation of the epithelial AJ acts as a potent tumor suppressor
331 mechanism for the development and progression of ILC. Ultimately, preventing formation of a mature
332 AJ and the accompanying mechanical transduction may lead to constitutive activation of RhoA and
333 Rock1 dependent actomyosin contraction through the disturbance of physiological feedback. Our data
334 indicate that these events lead to anoikis resistance, a feature that underpins metastatic
336
337 In conclusion, loss of AFDN leads to an ILC-like phenotype in E-cadherin expressing breast cancer
338 cells, due to its binding to α-catenin and the subsequent linkage to, and stabilization of, the F-actin
339 cytoskeleton. Similar to the consequence of α-catenin inactivation in breast cancer cells (Groot et al,
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
340 2018), we demonstrate that Afadin loss leads to an F-actin dependent non-cohesive cellular
341 phenotype, founded in detrimental cortical actomyosin dynamics that result in anchorage
342 independence and metastatic dissemination. As such, we reaffirm the notion that carcinomas driven
343 by loss of the AJ or the loss of the F-actin cytoskeleton connection with the AJ should be considered
344 ‘actin’ driven diseases (Bruner & Derksen, 2018). Delineating the disruption of proper AJ formation
345 will be key in the correct diagnosis and future treatment of malignancies driven by functional loss of E-
349 A. Dataflow for the Adhesome analysis in ILC. Indicated are the percentages CDH1 wild type versus
350 mutant ILC cases. B. Distribution of Adhesome mutation types. Pie charts depicting the mutation
351 types found in CDH1 wild-type and mutant ILC. Numbers represent % of total. C. Schematic depiction
352 of the Afadin protein. Indicated are the functional domains, the identified frameshift mutation, and the
353 predicted truncating stop codon in patient R-L1047-C. Domains and regions: RA, Ras associated;
354 FHA, forkhead associated; DIL, Dilution; PR, proline rich; CC, coiled coil; FAB, F-actin binding. The
355 binding sites for the effector proteins are indicated. Partly adapted from (Mandai et al, 2013) D.
356 Histology from sample R-L1047-C (top), a healthy mammary duct with ILC infiltrate from sample R-
357 L1047-C (middle), accompanied by a representative Invasive Ductal Carcinoma-Non Specific Type
358 (IDC-NST) sample (bottom) E. Afadin protein expression is lost in ILC sample R-L1047-C containing
359 the identified p.Phe629fs mutation. Shown are immunofluorescence images for E-cadherin (left) and
360 Afadin (middle) expression. Shown are ILC sample R-L1047-C (top), a healthy mammary duct with
361 ILC infiltrate from sample R-L1047-C (middle), accompanied by a representative IDC-NST sample
362 (bottom). Right panels show the merged composite. Size bar = 100 µM.
363
364 Figure 2. The occurrence of AFDN mutations in breast cancer predispose towards C-terminal
365 protein loss and occur mutually exclusively from somatic CDH1 mutations.
366 A. METABRIC and TCGA cBioportal AFDN mutations (Pereira et al, 2016; Berger et al, 2018).
367 Lollipop graph depicting distribution of AFDN mutations from the TCGA and METABRIC databases.
368 Domains and regions: RA, Ras associated; FHA, forkhead associated; DIL, Dilution; PR, proline rich;
369 CC, coiled coil; FAB, F-actin binding. B. TCGA and METABRIC mutual exclusivity of AFDN and CDH1
370 mutations in breast cancer. Somatic mutations in AFDN and CDH1 occur mutually exclusive in ILC
371 and ILC-like breast cancer. * p-value calculated by Fisher’s Exact Test.
372
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
373 Figure 3. Loss of AFDN in E-cadherin expressing breast cancer cells MCF7 cells leads to loss
375 A. Western blot showing the effect of CRISPR-Cas9 mediated knockout of Afadin or E-cadherin in
376 MCF7. AKT was used as loading control. *, unspecific protein (loading). B. Representative DIC
377 images showing MCF7 wild type, MCF7::∆AFDN, and MCF7::∆CDH1 cells, revealing a phenotypical
378 change in cellular morphology upon CDH1 or AFDN knockout. Bottom row shows zoom-ins of top
379 row. Size bar = 100 µM top row, size bar = 20 µM bottom row. C and D. Afadin loss induces anoikis
380 resistance. Contour plots (C) showing the anoikis resistance profiles of MCF7, MCF7::∆CDH1 and
381 MCF7::∆AFDN cells. Samples were stained with FITC-conjugated Annexin-V and Propidium Iodide
382 and analyzed by FACS. Quantifications are shown in (D) E. Anoikis resistance in MCF7::∆AFDN cells
383 is RhoA and Rock1 dependent. Cells were cultured in suspended conditions and treated with C3
384 transferase (C3, Rho inhibitor) or Y-27632 (Rock inhibitor). Anoikis resistance was assessed as in (D).
386
387 Figure 4. Afadin loss attenuates proper AJ formation and cortical F-actin distribution.
388 A. Immunofluorescence for Afadin (green), E-cadherin (red) and F-actin (white), showing the
389 attenuation of proper AJ formation. The right image is a composite of the left panels. Size bar = 5 µm.
390 B – D. Requirement of Afadin for proper F-actin organization at the epithelial AJ. Cartoon (B) showing
391 the setup of the line scan quantifications shown in (C and D). Line scans comparing the localization of
392 E-cadherin and F-actin were done and quantified perpendicular to the plasma membrane (between
393 the arrows indicated in the composite image depicted in (A)). Quantification of colocalization was
394 done based on the ratio of the Full Width at Half Maximum (FWHM) of F-actin over E-cadherin from
396 Figure 5. The α-catenin binding domain of Afadin is essential for epithelial cell-cell adhesion
398 A. Schematic overview of the GFP-tagged full length and truncation Afadin. B and C. Reconstitution
399 of Afadin full length or truncates in MCF7::∆AFDN cells. Western blots (B) showing expression based
400 on the GFP tag (top, GFP). The impact of Afadin loss or truncation on E-cadherin expression was
401 probed (middle), and Akt was used as loading control (bottom). The predicted molecular protein
402 weight is indicated below the GFP blot (Pred. MW). HEK293Ts were used as a transfection control
403 with the FL construct. Representative DIC images (C) showing the phenotypical impact of the Afadin
404 truncates in MCF7::∆AFDN cells. Similar cell numbers were plated and photographed after 18 hours.
405 Size bar = 10 µm. D and E. Immunofluorescence images showing MCF7::∆AFDN cells reconstituted
406 with the indicated truncates (see Fig. 5A for details). E-cadherin, F-actin and GFP-Afadin were
407 stained (D) and line scans were performed between the arrow heads (indicated in the merged
408 composites). The ratio of the full width at half maximum (FWHM) of the F-actin over E-cadherin
409 expression signals were quantified, normalized to 1, and compared in (E). ***p<0.001; ****p<0.0001.
410 Note that absence of the Afadin F-actin binding domain is not sufficient to disrupt E-cadherin to F-
411 actin linkage, while deletion of the α-catenin domain is required for proper AJ formation.
412
413 Figure 6. Loss of Afadin leads to metastatic breast cancer with invasive lobular features in
414 mice.
415 A. Inhibition of proper AJ formation does not impact tumor growth The indicated MCF cells were
416 injected intraductally in mice, and tumor volumes quantified over time. Error bars depict standard
417 deviations. B. E-cadherin expression is retained in cells lacking Afadin or Afadin C-terminal truncates.
418 Shown are immunohistochemistry panels for E-cadherin (middle) and GFP-Afadin (bottom) in the
419 indicated primary tumors. Left panels are H&E stains. Size bar = 100 µm. C. Afadin loss of function
420 induces ILC features. Shown are representative examples of primary tumors that developed in the
421 mice from (A). Note the acquisition of invasive trabecular single cell files (arrows) and individual cells
422 invading (arrowheads). Size bar = 100µm. D. Loss of the Afadin C-terminal α-catenin and F-actin
423 binding domains results in reduced survival. Shown is a Kaplan Meijer tumor-free survival plot for the
424 indicated tumor cohorts. Censored cases are marked by ticks on the plots. E. Examples of
425 intravasation in the periphery of primary tumors that developed in the mice from (A). Arrowheads
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
426 indicate individual tumor cells. Size bar = 50µm. F and G. Loss of the Afadin C-terminal α-catenin is
427 sufficient to drive metastatic disease. Shown are H&E stains (top) and immunohistochemistry for
428 human GATA3 (bottom) on lung sections (D). Note the overt increase in metastatic potential upon
429 loss of the α-catenin binding domain of Afadin. Size bar = 50 µm. G. Quantification of metastases,
431
432 Figure 7. Working model for the consequences of Afadin loss on the interaction of the
433 Adherens Junction and the F-actin cytoskeleton. In CDH1 wild type conditions, the AJ can form
434 proper interactions with the F-actin cytoskeleton, leading to a balanced outside-in force distribution
435 from the AJ to the F-actin cytoskeleton and complete maturation of the AJ. Tumors with an intact and
436 balanced AJs will not acquire full anoikis resistance and display low metastatic potential (left panel,
437 Wild type). Upon mutational loss of E-cadherin, cells are unable to form E-cadherin based AJs,
438 leading to overt loss of mechanical forces from the AJ to the F-actin cytoskeleton due to attenuated
439 feed-back signals (middle panel, ∆E-cadherin). E-cadherin loss results in constitutive actomyosin
440 contraction, Rho-Rock dependent anoikis resistance, and characteristic ILC-type phenotypes like
441 trabecular growth patterns, single cell invasion, and distant metastasis. Upon Afadin loss, E-cadherin
442 based-AJ are formed, but the lack of stabilization that occurs from inability of Afadin to bind the α-
443 catenin domain containing coiled-coil region, causes a disbalance of forces towards the F-actin
444 cytoskeleton, which prevents proper formation of a mature AJ (right panel, ∆Afadin). The classical C-
445 terminal F-actin binding domain is dispensable for this function. As a result of unbalanced mechanical
446 forces, adequate feedback from the F-actin cytoskeleton is prevented, and homotypic E-cadherin
447 based cell-cell junction cannot mature properly, resulting in dispersed cortical F-actin distribution.
448 Consequently, cells display and depend on actomyosin contraction driven anoikis resistance, leading
450
451 TABLE HEADING
454 Supplemental Figure 1. A-C Immunofluorescence images showing single and merged images (right
455 panels) of E-cadherin and α-catenin (A), E-cadherin and β-catenin (B) and Afadin and ZO-1 (C) in
456 MCF7 wild type (Wild type) and MCF7::∆AFDN cells (∆AFDN). Note the loss of ZO-1 expression at
458
459 Supplemental Figure 2. Afadin function does not impact Estrogen receptor (ER) expression in
460 primary tumors. Shown are H&E (left panels) and ER expression (right panels) in primary tumors that
461 developed after injection with either MCF7 wild type, MCF7::∆AFDN, or MCF7::∆AFDN cells,
463
464 SUPPLEMENTAL TABLE HEADINGS
465 Supplemental Table 1. TCGA & METABRIC AFDN status. Full overview of all AFDN mutations
466 reported in TCGA and METABRIC with revised diagnoses and CDH1 status.
467
468 Supplemental Table 2. gRNAs and primers used for the CRISPR/Cas9 mediated knockout of AFDN
469 and subsequent TIDE analysis and primers designed to generate the Afadin Full Length, ∆FAB, ∆CC
472 M.A.K.R. and P.W.B.D. designed the study. M.A.K.R., N.F.L.E, and K.V performed experiments.
473 N.F.L.E. and M.A.K.R. performed preclinical mouse experiments. P.W.B.D., I.N. and J.L. designed the
474 targeted sequencing capture strategy and performed subsequent analyses. R.B. provided patient
475 data, tissue samples and DNA samples. N.I. provided conceptual input in Afadin form, function, and
476 cloning of Afadin truncate constructs. M.C. and P.D. performed human and mouse histological tumor
477 diagnostics. All authors read the manuscript and were given opportunity to provide input.
478
479 ACKNOWLEDGEMENTS
480 We acknowledge the Utrecht Sequencing Facility (USEQ) for providing sequencing service and data.
481 USEQ is subsidized by the University Medical Center Utrecht and The Netherlands X-omics Initiative
482 (NWO project 184.034.019). We thank Astrid Bosma for help with patient samples and Yoshimi Takai
483 for the full length rat I-Afadin construct. Financial support was obtained from the Dutch Cancer
484 Society (KWF 10456), Breast Cancer Now (2018NovPCC1297), and the European Union’s Horizon
485 2020 FET Proactive program under the grant agreement No. 731957 (MECHANO-CONTROL). This
486 publication is also based upon work from COST action LOBSTERPOT (CA19138), supported by
490 DNA extracted from ILC samples was kindly provided by the RATHER consortium (Michaut et al,
491 2016). Processed DNA was captured using a custom Agilent SurePrint array as previously described,
492 based on the Adhesome as listed in Table 1. Sequencing of enriched samples was performed on an
493 Illumina Miseq sequencer to a depth of >50x average target base coverage. All sequencing data were
494 analyzed for mutations using in-house developed and validated pipelines (Mokry et al, 2010; Nijman
495 et al, 2010; Harakalova et al, 2011; Hoogwater et al, 2010). All mutations were mapped to GRCh37.
496
497 Cell Culture
498 MCF7 cells were cultured in DMEM-F12 (Invitrogen) containing 12% fetal calf serum, 100 IU/mL
499 penicillin, and 100 μg/mL streptomycin. MCF7 was obtained from American Type Culture Collection
500 and profiled using short-tandem repeat (STR) profiling (LGC Standards).
501
502 CRISPR/Cas9-generated AFDN KO Baculovirus
503 Baculoviruses expressing Cas9 were produced as described previously (Hindriksen et al, 2017). Two
504 gRNAs were tested and after TIDE analysis the highest efficiency gRNA was expanded (Brinkman et
505 al, 2014). gRNAs and primers used to perform TIDE analysis are listed in Supplemental Table 2.
506
507 Anoikis assay
508 Anoikis resistance was quantified as described previously (Schackmann et al, 2013). Samples were
509 treated with: 10 µM Y-27632 (Selleckchem, Huissen, The Netherlands) and 0.02 µg/mL C3-
510 transferase inhibitor (CT04-A cytoskeleton Inc, Heerhugowaard, The Netherlands), co-cultured with
511 cells over 4 days in anchorage-independent conditions using ultra-low cluster 24 well plates
512 (Corning). Error bars depict standard deviation. A two-sided Students’ t-test was performed to
513 calculate statistical significance; p-values less than 0.05 were considered significant.
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
514 Immunofluorescence
515 Cells were allowed to adhere to glass coverslips overnight, washed with PBS containing Ca2+ and
2+
516 Mg (serial 14080055, Gibco, Paisley, Scotland, UK) and subsequently fixed using 4%
517 paraformaldehyde. Samples were blocked with 4% BSA in PBS for 1 hr. Samples were incubated with
518 primary antibodies overnight at 4°C. The following primary antibodies were used: mouse anti-p120
519 6H11 (1:500; Santa Cruz Biotechnology, Santa Cruz, California, United States), rabbit anti-GFP
520 (1:1000; FL Santa Cruz), rabbit anti-l-Afadin (1:500, Sigma-Aldrich, Zwijndrecht, the Netherlands),
522 Biosciences, Becton, United States, clone98/pp120), mouse anti-α-catenin(1:100, Enzo Life Sciences,
523 Farmingdale, New York, United States). For detection, samples were incubated with secondary
524 antibodies in blocking buffer for 1 hr. at room temperature. The following secondary antibodies were
525 used: (1:600; Invitrogen, Waltham, Massachusetts, United States): goat anti-mouse Alexa488
526 (#A11029), goat anti-rabbit Alexa488 (#A11034), goat anti-mouse Alexa568 (#A11031), goat anti-
527 rabbit Alexa568 (#A11036), goat anti-rabbit Alexa647 (#A21245) and goat anti-rat Alexa647
528 (#A21247). Afterwards, cells were incubated with Alexa633-phalloidin to visualize F-actin (1:300;
529 #A22284; Life Technologies, Carlsbad, California, United States) or DAPI to stain DNA, washed and
530 mounted using Prolong Diamond Antifade Mountant (P36961; Thermo Fisher Scientific, Waltham,
531 Massachusetts, United States). Imaging was performed on a LSM700 confocal microscope (Zeiss,
532 Jena, Germany). Image processing was performed using ImageJ, Illustrator and Photoshop (Adobe).
533
534 DNA Constructs
535 pEGFP-l-Afadin containing the full rat Afadin cDNA was a kind gift from Yoshimi Takai (Sakakibara et
536 al, 2018). N-terminally GFP conjugated Afadin truncates were cloned into a pLV lentiviral backbone
537 (pLV.bc.PURO, (Schackmann et al, 2011)) by In-Fusion PCR cloning using the pEGFP-l-Afadin
538 plasmid as template. In short, pLV.BC.PURO was linearized using MluI and NheI digestion and the
539 amplified GFP-Afadin fragments were incorporated using In-Fusion HD Cloning Plus (#638920;
540 Takara, Kusatsu, Shiga, Japan). In-Fusion compatible complementary ends to the target vector were
542
543 Transfections
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
544 Cells were plated at a density of 1*10^6 per 10cm plate and transfected on the following day using
545 Fugene HD (Promega, Madison, Wisconsin, United States, E2311). A PGK-GFP vector was
546 transfected in parallel as control for transfection efficiency. Three days after transfection, cells were
547 selected for two days using 2 µg/mL Puromycin (Sigma Aldrich, P8833).
548
549 Intraductal injections and mouse studies
550 MCF7::∆AFDN cells were transduced with lentiviral pLV.BC.PURO particles with the different
551 truncates as described (Schackmann et al, 2011). Two days after transduction, cells were put on
552 Puromycin selection (2 µg/mL) for 2 weeks and expression was confirmed on Western Blot. 250,000
553 cells suspended in 25 µL were injected intraductally in the fourth mammary gland of recipient Rag2-/-
554 ;IL2cγR-/- mice ((Gimeno et al, 2004), Envigo, Indianapolis, Indiana, United States) Tumor onset and
555 growth was monitored longitudinally and tumor volume was determined by digital pressure-sensitive
556 caliper measurements (Mitutoyo, Veenendaal, The Netherlands) using the following formula:
557
ߨሺ/
ሻ. Animals that developed tumors were euthanized if the tumors reached a size of > 1000
3
558 mm or in cases of severe discomfort otherwise.
559
560 Histological analysis
561 Formaldehyde-fixed, paraffin-embedded tissues were sectioned at 4 μm and stained with hematoxylin
562 and eosin (H&E). For immunohistochemical staining, fixed sections were rehydrated and incubated
563 with primary antibodies against ER (SP1; 1:1; Roche, Basel, Swiss), GATA3 (L50-823; 1:1; Roche),
564 E-cadherin (612130; 1:150; BD Biosciences), and GFP (SC-8334;1:150; Santa Cruz). Endogenous
565 peroxidases were blocked with 3% H2O2and biotin-conjugated secondary antibodies were used,
566 followed by incubation with HRP-conjugated streptavidin–biotin complex (DAKO). Substrate was
567 developed with DAB (DAKO)and pictures were produced using a Nikon Eclipse E800microscope with
568 a Nikon DXM1200 digital camera (Nikon, Amsterdam, The Netherlands). Appropriate positive
569 (expressing tissues) and negative controls (matched isotype and omitting primary antibody) were
571
572 Biochemistry
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
573 Cells were lysed in sample buffer and proteins were separated and western blotted as described
574 previously (Schackmann et al, 2013). The following primary antibodies were used: Rabbit anti-l-Afadin
575 (1:1.000, A0349; Sigma-Aldrich), rabbit anti-GFP (1:500; sc-8334, Santa Cruz Biotechnology), and
576 goat anti-AKT (1:1,000, C-20/sc-1618; Santa Cruz Biotechnology). All blots were incubated for 30
577 min. with either rabbit anti-goat-PO (DAKO, Heverlee, Belgium; P160), goat anti-rabbit-PO (Bio-Rad,
578 Veenendaal, The Netherlands; 170-6515) or goat anti-mouse-PO (Bio-Rad; 170-6516) secondary
579 anti-bodies. Images were acquired on an Amersham Imager 600 (Amersham, ‘s-Hertogenbosch, The
580 Netherlands).
581
582 cBioportal analysis and data extraction
583 Data for the METABRIC and TCGA datasets were retrieved from the cBioportal public databases
584 (https://www.cbioportal.org/) (Pereira et al, 2016; Cerami et al, 2012; Gao et al, 2013)
585
586 Statistical Analysis
587 All statistical tests were performed using Prism 9 Version 9.4.1.
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
588 REFERENCES
589 Anastasiadis PZ, Moon SY, Thoreson MA, Mariner DJ, Crawford HC, Zheng Y & Reynolds a B (2000)
590 Inhibition of RhoA by p120 catenin. Nature Cell Biology 2: 637–644
591 Asada M, Irie K, Morimoto K, Yamada A, Ikeda W, Takeuchi M & Takai Y (2003) ADIP, a Novel
592 Afadin- and α-Actinin-Binding Protein Localized at Cell-Cell Adherens Junctions*. J Biol Chem
593 278: 4103–4111
594 Bajrami I, Marlow R, Ven M van de, Brough R, Pemberton HN, Frankum J, Song F, Rafiq R, Konde A,
595 Krastev DB, et al (2018) E-Cadherin/ROS1 Inhibitor Synthetic Lethality in Breast Cancer. Cancer
596 Discov 8: 498–515
597 Berger AC, Korkut A, Kanchi RS, Hegde AM, Lenoir W, Liu W, Liu Y, Fan H, Shen H, Ravikumar V, et
598 al (2018) A Comprehensive Pan-Cancer Molecular Study of Gynecologic and Breast Cancers.
599 Cancer Cell 33: 690-705.e9
600 Berx G, CletonJansen AM, Nollet F, Leeuw WJ, Vijver M, Cornelisse C & Roy F (1995) Ecadherin
601 is a tumour/invasion suppressor gene mutated in human lobular breast cancers. Embo J 14:
602 6107–6115
603 Blair VR, McLeod M, Carneiro F, Coit DG, D’Addario JL, Dieren JM van, Harris KL, Hoogerbrugge N,
604 Oliveira C, Post RS van der, et al (2020) Hereditary diffuse gastric cancer: updated clinical
605 practice guidelines. Lancet Oncol 21: e386–e397
606 Brinkman EK, Chen T, Amendola M & Steensel B van (2014) Easy quantitative assessment of
607 genome editing by sequence trace decomposition. Nucleic Acids Research 42: e168–e168
608 Bruner HC & Derksen PWB (2018) Loss of E-Cadherin-Dependent Cell-Cell Adhesion and the
609 Development and Progression of Cancer. Cold Spring Harbor Perspectives in Biology 10: 1–20
610 Carminati M, Gallini S, Pirovano L, Alfieri A, Bisi S & Mapelli M (2016) Concomitant binding of Afadin
611 to LGN and F-actin directs planar spindle orientation. Nat Struct Mol Biol 23: 155–163
612 Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML,
613 Larsson E, et al (2012) The cBio Cancer Genomics Portal: An Open Platform for Exploring
614 Multidimensional Cancer Genomics Data. Cancer Discov 2: 401–404
615 Choi W, Acharya BR, Peyret G, Fardin M-A, Mège R-M, Ladoux B, Yap AS, Fanning AS & Peifer M
616 (2016) Remodeling the zonula adherens in response to tension and the role of afadin in this
617 response. J Cell Biol 213: 243–260
618 Christgen M & Derksen PW (2015) Lobular breast cancer: molecular basis, mouse and cellular
619 models. Breast Cancer Res 17: 491–9
620 Christgen M, Steinemann D, Kühnle E, Länger F, Gluz O, Harbeck N & Kreipe H (2016) Lobular
621 breast cancer: Clinical, molecular and morphological characteristics. Pathology - Res Pract 212:
622 583–597
623 Ciriello G, Gatza ML, Beck AH, Wilkerson MD, Rhie SK, Pastore A, Zhang H, McLellan M, Yau C,
624 Kandoth C, et al (2015) Comprehensive Molecular Portraits of Invasive Lobular Breast Cancer.
625 Cell 163: 506–519
626 Collins RJ, Jiang WG, Hargest R, Mason MD & Sanders AJ (2015) EPLIN: a fundamental actin
627 regulator in cancer metastasis? Cancer Metast Rev 34: 753–764
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
628 Davis MA, Ireton RC & Reynolds AB (2003) A core function for p120-catenin in cadherin turnover. J
629 Cell Biology 163: 525–534
630 Derksen PWB, Braumuller TM, Burg E van der, Hornsveld M, Mesman E, Wesseling J, Krimpenfort P
631 & Jonkers J (2011) Mammary-specific inactivation of E-cadherin and p53 impairs functional gland
632 development and leads to pleomorphic invasive lobular carcinoma in mice. Dis Model Mech 4:
633 347–358
634 Derksen PWB, Liu X, Saridin F, Gulden HVD, Zevenhoven J, Evers B, Beijnum JR van, Griffioen AW,
635 Vink J, Krimpenfort P, et al (2006) Somatic inactivation of E-cadherin and p53 in mice leads to
636 metastatic lobular mammary carcinoma through induction of anoikis resistance and angiogenesis.
637 Cancer Cell 10: 437–449
638 Desmedt C, Zoppoli G, Gundem G, Pruneri G, Larsimont D, Fornili M, Fumagalli D, Brown D, Rothé
639 F, Vincent D, et al (2016) Genomic Characterization of Primary Invasive Lobular Breast Cancer. J
640 Clin Oncol 34: 1872–1881
641 Douma S, Laar TV, Zevenhoven J, Meuwissen R, Garderen EV & Peeper DS (2004) Suppression of
642 anoikis and induction of metastasis by the neurotrophic receptor TrkB. Nature 430: 1034–1039
643 Duc Q le, Shi Q, Blonk I, Sonnenberg A, Wang N, Leckband D & Rooij J de (2010) Vinculin
644 potentiates E-cadherin mechanosensing and is recruited to actin-anchored sites within adherens
645 junctions in a myosin II–dependent manner. J Cell Biol 189: 1107–1115
646 Dufour S, Mège R-M & Thiery JP (2013) α-catenin, vinculin, and F-actin in strengthening E-cadherin
647 cell–cell adhesions and mechanosensing. Cell Adhes Migr 7: 345–350
648 Efimova N & Svitkina TM (2018) Branched actin networks push against each other at adherens
649 junctions to maintain cell–cell adhesion. J Cell Biol 217: 1827–1845
650 Fimereli D, Venet D, Rediti M, Boeckx B, Maetens M, Majjaj S, Rouas G, Marchio C, Bertucci F,
651 Mariani O, et al (2022) Timing evolution of lobular breast cancer through phylogenetic analysis.
652 Ebiomedicine 82: 104169
656 Fukuhara A, Irie K, Nakanishi H, Takekuni K, Kawakatsu T, Ikeda W, Yamada A, Katata T, Honda T,
657 Sato T, et al (2002) Involvement of nectin in the localization of junctional adhesion molecule at
658 tight junctions. Oncogene 21: 7642–7655
659 Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Sun Y, Jacobsen A, Sinha R,
660 Larsson E, et al (2013) Integrative Analysis of Complex Cancer Genomics and Clinical Profiles
661 Using the cBioPortal. Sci Signal 6: pl1
662 Gimeno R, Weijer K, Voordouw A, Uittenbogaart CH, Legrand N, Alves NL, Wijnands E, Blom B &
663 Spits H (2004) Monitoring the effect of gene silencing by RNA interference in human CD34+ cells
664 injected into newborn RAG2-/- γc-/- mice: functional inactivation of p53 in developing T cells.
665 Blood 104: 3886–3893
666 Grabenstetter A, Mohanty AS, Rana S, Zehir A, Brannon AR, D’Alfonso TM, DeLair DF, Tan LK &
667 Ross DS (2020) E-cadherin immunohistochemical expression in invasive lobular carcinoma of the
668 breast: correlation with morphology and CDH1 somatic alterations. Hum Pathol 102: 44–53
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
669 Groot JS de, Ratze MA, Amersfoort M van, Eisemann T, Vlug EJ, Niklaas MT, Chin S, Caldas C,
670 Diest PJ van, Jonkers J, et al (2018) αEcatenin is a candidate tumor suppressor for the
671 development of Ecadherinexpressing lobulartype breast cancer. J Pathol 245: 456–467
672 Gurley NJ, Szymanski RA, Dowen RH, Butcher TA, Ishiyama N & Peifer M (2023) Exploring the
673 evolution and function of Canoe’s intrinsically disordered region in linking cell-cell junctions to the
674 cytoskeleton during embryonic morphogenesis. bioRxiv: 2023.03.06.531372 [PREPRINT]
675 Harakalova M, Mokry M, Hrdlickova B, Renkens I, Duran K, Roekel H van, Lansu N, Roosmalen M
676 van, Bruijn E de, Nijman IJ, et al (2011) Multiplexed array-based and in-solution genomic
677 enrichment for flexible and cost-effective targeted next-generation sequencing. Nat Protoc 6:
678 1870–1886
679 Hazan RB, Kang L, Roe S, Borgen PI & Rimm DL (1997) Vinculin Is Associated with the E-cadherin
680 Adhesion Complex*. J Biol Chem 272: 32448–32453
681 Hindriksen S, Bramer AJ, Truong MA, Vromans MJM, Post JB, Verlaan-Klink I, Snippert HJ, Lens
682 SMA & Hadders MA (2017) Baculoviral delivery of CRISPR/Cas9 facilitates efficient genome
683 editing in human cells. PLoS ONE 12: e0179514
684 Hollestelle A, Elstrodt F, Timmermans M, Sieuwerts AM, Klijn JGM, Foekens JA, Bakker MA den &
685 Schutte M (2010) Four human breast cancer cell lines with biallelic inactivating α-catenin gene
686 mutations. Breast Cancer Res Tr 122: 125–133
687 Hoogwater FJH, Nijkamp MW, Smakman N, Steller EJA, Emmink BL, Westendorp BF, Raats DAE,
688 Sprick MR, Schaefer U, Houdt WJV, et al (2010) Oncogenic K-Ras Turns Death Receptors Into
689 Metastasis-Promoting Receptors in Human and Mouse Colorectal Cancer Cells. Gastroenterology
690 138: 2357–2367
691 Hornsveld M, Tenhagen M, Ven RA van de, Smits AMM, Triest MH van, Amersfoort M van, Kloet
692 DEA, Dansen TB, Burgering BM & Derksen PWB (2016) Restraining FOXO3-dependent
693 transcriptional BMF activation underpins tumour growth and metastasis of E-cadherin-negative
694 breast cancer. Cell Death Differ 23: 1483–1492
695 Huxham J, Tabariès S & Siegel PM (2021) Afadin (AF6) in cancer progression: A multidomain
696 scaffold protein with complex and contradictory roles. BioEssays 43: e2000221-17
697 Kimura K, Ito M, Amano M, Chihara K, Fukata Y, Nakafuku M, Yamamori B, Feng J, Nakano T,
698 Okawa K, et al (1996) Regulation of Myosin Phosphatase by Rho and Rho-Associated Kinase
699 (Rho-Kinase). Science 273: 245–248
700 Kurita S, Ogita H & Takai Y (2011) Cooperative Role of Nectin-Nectin and Nectin-Afadin Interactions
701 in Formation of Nectin-based Cell-Cell Adhesion*. J Biol Chem 286: 36297–36303
702 Leckband DE & Rooij J de (2014) Cadherin Adhesion and Mechanotransduction. Annu Rev Cell Dev
703 Bi 30: 1–25
707 Maiers JL, Peng X, Fanning AS & DeMali KA (2013) ZO-1 recruitment to α-catenin – a novel
708 mechanism for coupling the assembly of tight junctions to adherens junctions. J Cell Sci 126:
709 3904–3915
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
710 Majewski IJ, Kluijt I, Cats A, Scerri TS, Jong D de, Kluin RJC, Hansford S, Hogervorst FBL, Bosma
711 AJ, Hofland I, et al (2013) An α-E-catenin (CTNNA1) mutation in hereditary diffuse gastric cancer.
712 J Pathol 229: 621–629
713 Mandai K, Nakanishi H, Satoh A, Obaishi H, Wada M, Nishioka H, Itoh M, Mizoguchi A, Aoki T,
714 Fujimoto T, et al (1997) Afadin: A Novel Actin Filament–binding Protein with One PDZ Domain
715 Localized at Cadherin-based Cell-to-Cell Adherens Junction. J Cell Biology 139: 517–528
716 Mandai K, Rikitake Y, Shimono Y & Takai Y (2013) Chapter Nineteen Afadin/AF-6 and Canoe Roles
717 in Cell Adhesion and Beyond. Prog Mol Biol Transl 116: 433–454
718 Maruo T, Sakakibara S, Miyata M, Itoh Y, Kurita S, Mandai K, Sasaki T & Takai Y (2018) Involvement
719 of l-afadin, but not s-afadin, in the formation of puncta adherentia junctions of hippocampal
720 synapses. Mol Cell Neurosci 92: 40–49
721 McCaffrey LM, Montalbano J, Mihai C & Macara IG (2012) Loss of the Par3 polarity protein promotes
722 breast tumorigenesis and metastasis. arXiv 22: 601–614
723 Meng W & Takeichi M (2009) Adherens Junction: Molecular Architecture and Regulation. Csh
724 Perspect Biol 1: a002899
725 Michaut M, Chin S-F, Majewski I, Severson TM, Bismeijer T, Koning L de, Peeters JK, Schouten PC,
726 Rueda OM, Bosma AJ, et al (2016) Integration of genomic, transcriptomic and proteomic data
727 identifies two biologically distinct subtypes of invasive lobular breast cancer. Sci Rep-uk 6: 18517
728 Mokry M, Feitsma H, Nijman IJ, Bruijn E de, Zaag PJ van der, Guryev V & Cuppen E (2010) Accurate
729 SNP and mutation detection by targeted custom microarray-based genomic enrichment of short-
730 fragment sequencing libraries. Nucleic Acids Res 38: e116–e116
731 Morita K, Furuse M, Fujimoto K & Tsukita S (1999) Claudin multigene family encoding four-
732 transmembrane domain protein components of tight junction strands. Proc National Acad Sci 96:
733 511–516
734 Nagle AM, Levine KM, Tasdemir N, Scott JA, Burlbaugh K, Kehm J, Katz TA, Boone DN, Jacobsen
735 BM, Atkinson JM, et al (2018) Loss of E-cadherin Enhances IGF1–IGF1R Pathway Activation and
736 Sensitizes Breast Cancers to Anti-IGF1R/InsR Inhibitors. Clin Cancer Res 24: 5165–5177
737 Nelson WJ (2008) Regulation of cell–cell adhesion by the cadherin–catenin complex. Biochem Soc T
738 36: 149–155
739 Nijman IJ, Mokry M, Boxtel R van, Toonen P, Bruijn E de & Cuppen E (2010) Mutation discovery by
740 targeted genomic enrichment of multiplexed barcoded samples. Nat Methods 7: 913–915
741 Ooshio T, Irie K, Morimoto K, Fukuhara A, Imai T & Takai Y (2004) Involvement of LMO7 in the
742 Association of Two Cell-Cell Adhesion Molecules, Nectin and E-cadherin, through Afadin and α-
743 Actinin in Epithelial Cells*. J Biol Chem 279: 31365–31373
744 Pareja F, Ferrando L, Lee SSK, Beca F, Selenica P, Brown DN, Farmanbar A, Paula ADC, Vahdatinia
745 M, Zhang H, et al (2020) The genomic landscape of metastatic histologic special types of invasive
746 breast cancer. Npj Breast Cancer 6: 53
747 Pereira B, Chin S-F, Rueda OM, Vollan H-KM, Provenzano E, Bardwell HA, Pugh M, Jones L, Russell
748 R, Sammut SJ, et al (2016) The somatic mutation profiles of 2,433 breast cancers refines their
749 genomic and transcriptomic landscapes. Nature Communications: 1–16
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
750 Perez-Vale KZ, Yow KD, Johnson RI, Byrnes AE, Finegan TM, Slep KC & Peifer M (2021) Multivalent
751 interactions make adherens junction–cytoskeletal linkage robust during morphogenesis. J Cell Biol
752 220: e202104087
753 Rakha EA & Ellis IO (2010) Lobular breast carcinoma and its variants. Semin Diagn Pathol 27: 49–61
754 Reed AEM, Kutasovic JR, Vargas AC, Jayanthan J, AlMurrani A, Reid LE, Chambers R, Silva LD,
755 Melville L, Evans E, et al (2016) An epithelial to mesenchymal transition programme does not
756 usually drive the phenotype of invasive lobular carcinomas. J Pathol 238: 489–494
757 Rikitake Y & Takai Y (2011) Chapter three Directional Cell Migration Regulation by Small G Proteins,
758 Nectin-like Molecule-5, and Afadin. Int Rev Cel Mol Bio 287: 97–143
759 Rooij J de (2014) Cadherin adhesion controlled by cortical actin dynamics. Nat Cell Biol 16: 508–510
760 Sakakibara S, Maruo T, Miyata M, Mizutani K & Takai Y (2018) Requirement of the Factinbinding
761 activity of lafadin for enhancing the formation of adherens and tight junctions. Genes Cells 23:
762 185–199
763 Sakakibara S, Mizutani K, Sugiura A, Sakane A, Sasaki T, Yonemura S & Takai Y (2020) Afadin
764 regulates actomyosin organization through αE-catenin at adherens junctions. J Cell Biol 219: 1–16
765 Schackmann RCJ, Amersfoort M van, Haarhuis JHI, Vlug EJ, Halim VA, Roodhart JML, Vermaat JS,
766 Voest EE, Groep P van der, Diest PJV, et al (2011) Cytosolic p120-catenin regulates growth of
767 metastatic lobular carcinoma through Rock1-mediated anoikis resistance. J Clin Invest 121: 3176–
768 3188
769 Schackmann RCJ, Klarenbeek S, Vlug EJ, Stelloo S, Amersfoort M van, Tenhagen M, Braumuller TM,
770 Vermeulen JF, Groep P van der, Peeters T, et al (2013) Loss of p120-Catenin Induces Metastatic
771 Progression of Breast Cancer by Inducing Anoikis Resistance and Augmenting Growth Factor
772 Receptor Signaling. Cancer Research 73: 4937–4949
773 Shapiro L & Weis WI (2009) Structure and Biochemistry of Cadherins and Catenins. Csh Perspect
774 Biol 1: a003053
775 Sumida GM, Tomita TM, Shih W & Yamada S (2011) Myosin II activity dependent and independent
776 vinculin recruitment to the sites of E-cadherin-mediated cell-cell adhesion. BMC Cell Biology 12:
777 48
780 Takai Y, Ikeda W, Ogita H & Rikitake Y (2008) The Immunoglobulin-Like Cell Adhesion Molecule
781 Nectin and Its Associated Protein Afadin. Annu Rev Cell Dev Bi 24: 309–342
782 Teo K, Gómez-Cuadrado L, Tenhagen M, Byron A, Rätze M, Amersfoort M van, Renes J, Strengman
783 E, Mandoli A, Singh AA, et al (2018) E-cadherin loss induces targetable autocrine activation of
784 growth factor signalling in lobular breast cancer. Scientific Reports 8: 15454–14
785 Uehata M, Ishizaki T, Satoh H, Ono T, Kawahara T, Morishita T, Tamakawa H, Yamagami K, Inui J,
786 Maekawa M, et al (1997) Calcium sensitization of smooth muscle mediated by a Rho-associated
787 protein kinase in hypertension. Nature 389: 990–994
788 Vasioukhin V, Bauer C, Yin M & Fuchs E (2000) Directed Actin Polymerization Is the Driving Force for
789 Epithelial Cell–Cell Adhesion. Cell 100: 209–219
bioRxiv preprint doi: https://doi.org/10.1101/2023.07.04.547654; this version posted July 5, 2023. 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.
790 Ven RAH van de, Tenhagen M, Meuleman W, Riel JJG van, Schackmann RCJ & Derksen PWB
791 (2015) Nuclear p120-catenin contributes to anoikis resistance of Lobular Breast Cancer through
792 Kaiso-dependent Wnt11 expression. Disease Models & Mechanisms 8: 373–384
793 Verma S, Shewan AM, Scott JA, Helwani FM, Elzen NR den, Miki H, Takenawa T & Yap AS (2004)
794 Arp2/3 Activity Is Necessary for Efficient Formation of E-cadherin Adhesive Contacts*. J Biological
795 Chem 279: 34062–34070
796 Weiss EE, Kroemker M, Rüdiger A-H, Jockusch BM & Rüdiger M (1998) Vinculin Is Part of the
797 Cadherin–Catenin Junctional Complex: Complex Formation between α-Catenin and Vinculin. J
798 Cell Biology 141: 755–764
799 Xue B, Krishnamurthy K, Allred DC & Muthuswamy SK (2013) Loss of Par3 promotes breast cancer
800 metastasis by compromising cell-cell cohesion. Nature Cell Biology 15: 189–200
801 Yamada S, Pokutta S, Drees F, Weis WI & Nelson WJ (2005) Deconstructing the Cadherin-Catenin-
802 Actin Complex. Cell 123: 889–901
803 Yao M, Qiu W, Liu R, Efremov AK, Cong P, Seddiki R, Payre M, Lim CT, Ladoux B, Mège R-M, et al
804 (2014) Force-dependent conformational switch of α-catenin controls vinculin binding. Nat Commun
805 5: 4525
806 Yonemura S, Wada Y, Watanabe T, Nagafuchi A & Shibata M (2010) α-Catenin as a tension
807 transducer that induces adherens junction development. Nat Cell Biol 12: 533–542
808 Zihni C, Mills C, Matter K & Balda MS (2016) Tight junctions: from simple barriers to multifunctional
809 molecular gates. Nat Rev Mol Cell Bio 17: 564–580
available under aCC-BY-NC-ND 4.0 International license.