E-cadherin binds to desmoglein to facilitate desmosome assembly

Omer Shafraz; Matthias Rübsam; Sara N Stahley; Amber L Caldara; Andrew P Kowalczyk; Carien M Niessen; Sanjeevi Sivasankar

doi:10.7554/eLife.37629

. 2018 Jul 12;7:e37629. doi: 10.7554/eLife.37629

E-cadherin binds to desmoglein to facilitate desmosome assembly

Omer Shafraz ^1,^†, Matthias Rübsam ^2,^†, Sara N Stahley ³, Amber L Caldara ³, Andrew P Kowalczyk ³, Carien M Niessen ², Sanjeevi Sivasankar ^1,^✉

Editors: William I Weis⁴, Anna Akhmanova⁵

PMCID: PMC6066328 PMID: 29999492

Abstract

Desmosomes are adhesive junctions composed of two desmosomal cadherins: desmocollin (Dsc) and desmoglein (Dsg). Previous studies demonstrate that E-cadherin (Ecad), an adhesive protein that interacts in both trans (between opposing cells) and cis (on the same cell surface) conformations, facilitates desmosome assembly via an unknown mechanism. Here we use structure-function analysis to resolve the mechanistic roles of Ecad in desmosome formation. Using AFM force measurements, we demonstrate that Ecad interacts with isoform 2 of Dsg via a conserved Leu-175 on the Ecad cis binding interface. Super-resolution imaging reveals that Ecad is enriched in nascent desmosomes, supporting a role for Ecad in early desmosome assembly. Finally, confocal imaging demonstrates that desmosome assembly is initiated at sites of Ecad mediated adhesion, and that Ecad-L175 is required for efficient Dsg2 and desmoplakin recruitment to intercellular contacts. We propose that Ecad trans interactions at nascent cell-cell contacts initiate the recruitment of Dsg through direct cis interactions with Ecad which facilitates desmosome assembly.

Research organism: None

Introduction

The formation, organization and maintenance of complex tissue structures are mediated by the cadherin superfamily of cell-cell adhesion proteins, a large protein group composed of four major subfamilies: classical cadherins, desmosomal cadherins, protocadherins and atypical cadherins (Brasch et al., 2012). Of these proteins, desmosomal cadherins and classical cadherins are essential for the maintenance of tissue integrity (Rübsam et al., 2017a). While classical cadherins mediate cell-cell adhesion in all soft tissue and play critical roles in tissue morphogenesis, desmosomal cadherins mediate robust cell-cell adhesion in tissues like the epidermis and heart that are exposed to significant levels of mechanical stress (Dusek et al., 2007). There are two types of desmosomal cadherins: desmoglein (Dsg) and desmocollin (Dsc), which are organized into four Dsg isoforms (Dsg1-4) and three Dsc isoforms (Dsc1-3) (Getsios et al., 2004). Of these isoforms, Dsg2 and Dsc2 are widely expressed in the epithelia and are also found in non-epithelial cells such as in the myocardium of the heart and lymph node follicles (Getsios et al., 2004). Loss of Dsg2 is embryonically lethal (Eshkind et al., 2002) while mutations in Dsg2 (Awad et al., 2006) and Dsc2 (Syrris et al., 2006) cause arrhythmogenic right ventricular cardiomyopathy (ARVC), a hereditary heart disease. In contrast, isoforms 1 and 3 are restricted to complex epithelial tissues (Green and Simpson, 2007) and their loss of function leads to epidermal fragility, such as in the autoimmune blistering disease pemphigus (Amagai and Stanley, 2012).

Dsc and Dsg associate with anchoring and signaling proteins to form robust intercellular junctions called desmosomes. Several studies have shown that the classical cadherin, E-cadherin (Ecad), promotes desmosome assembly. Immuno-electron micrographs demonstrate that Ecad localizes to the intercellular region of the bovine tongue epithelial desmosomes (Jones, 1988). Blocking Ecad adhesion with antibodies delay desmosome formation in MDCK cells (Gumbiner et al., 1988) and in human keratinocytes (Lewis et al., 1994; Wheelock and Jensen, 1992). Desmosome formation in keratinocytes requires junctional initiation by the classical cadherins, Ecad or P-cadherin (Pcad) (Amagai et al., 1995; Michels et al., 2009). Consequently, Ecad and Pcad deficient mice show defective desmosome assembly (Tinkle et al., 2008). However, the precise molecular mechanisms by which classical cadherins promote desmosome formation are unknown.

Since Ecad interacts laterally to form cis dimers on the same cell surface (Harrison et al., 2011) while Ecad molecules from opposing cells interact in a trans strand-swap dimer conformation (Boggon et al., 2002; Parisini et al., 2007; Vendome et al., 2011) and a trans X-dimer conformation (Ciatto et al., 2010; Harrison et al., 2010), we used mutants that specifically abolish either Ecad trans or cis interactions and tested their binding to either Dsg2 or Dsc2. We characterized these interactions at different stages of desmosome formation using an integrated structure/function analysis that combined single molecule force measurements of wild type (WT) and mutant cadherins with an atomic force microscope (AFM), super-resolution imaging of desmosomes in human keratinocytes and confocal fluorescence microscopy of Ecad-knockout, Pcad-knockdown mouse keratinocytes (E^KO/P^KD), transfected with WT and mutant cadherins. The data identify a novel Ca²⁺- independent direct interaction between Ecad and Dsg2 that is mediated by a conserved Leu 175 on Ecad. Previous structural studies have shown that L175 mediates homophilic Ecad cis dimerization (Harrison et al., 2011). Our data suggests that desmosome assembly is initiated in two stages: a first stage that requires stable Ecad trans-homodimerization and a second stage characterized by the direct heterophilic binding between Ecad and Dsg2 that facilitates further desmosome assembly. The interactions between Ecad and Dsg2 are short-lived and as desmosomes mature, Dsg2 dissociates from Ecad and forms stable bonds with Dsc2 to mediate robust adhesion.

Results

Ecad interacts with Dsg2 to form a Ca²⁺-independent heterodimer

We identified the binding partners for recombinant Dsc2, Dsg2 and Ecad using single molecule AFM force measurements. The complete extracellular region of Dsc2, Dsg2 and Ecad were expressed in mammalian cells and biotinylated at their C-terminus using a biotin ligase enzyme (Materials and methods). Identical concentrations of biotinylated cadherins were immobilized on AFM tips and glass coverslip (CS) substrates that were functionalized with polyethylene glycol (PEG) tethers and decorated with streptavidin protein (Figure 1A and B), (Materials and methods),(Lowndes et al., 2014; Manibog et al., 2014; Rakshit et al., 2012; Sivasankar et al., 2009; Zhang et al., 2009). Under similar experimental conditions, cadherin surface density was previously determined to be 65 ± 18 cadherins per µm², which corresponds to an average distance of 124 nm between neighboring cadherins (Zhang et al., 2009). Since the separation between neighboring cadherins is an order of magnitude larger than the radius of curvature of the AFM tip, the measured unbinding events correspond to the interaction of only a single cadherin immobilized on the surface and the AFM tip respectively. At the start of the experiment, the AFM cantilever and substrate were brought into contact to allow opposing cadherins to interact. The tip was then withdrawn from the substrate and the force required to rupture the adhesive complex was measured. Interaction of opposing cadherins resulted in unbinding events characterized by non-linear stretching of the PEG tethers (Figure 1C); PEG stretching served as a molecular fingerprint for single molecule unbinding since its extension under load has been extensively characterized (Oesterhelt et al., 1999). If the cadherins did not interact, no unbinding forces were measured (Figure 1D).

Figure 1. — (A) Schematic of specific interaction experiment. The AFM tip and substrate were functionalized with PEG linkers some of which were decorated with streptavidins. Biotinylated cadherin proteins were attached to streptavidin. (B) Schematic of nonspecific interaction experiment. The probability of interactions between the AFM tip functionalized with biotinylated cadherin proteins and the substrate lacking cadherins (left) and the binding probability of an AFM cantilever lacking cadherin and a substrate decorated with cadherins (right) was measured. Example force versus tip-surface distance traces showing (C) a single unbinding event with signature PEG stretching and (D) no interaction. (E) Specific binding probabilities for different combination of cadherins on the tip and substrate measured in Ca²⁺ (blue) and in EGTA (red), a Ca²⁺ chelator. Non-specific binding levels (hatched green) were determined from the average of measured binding probabilities between a cadherin functionalized AFM tip and a surface lacking cadherin and between an AFM tip lacking cadherin and surface functionalized with biotinylated cadherin proteins. Dsg2/Dsg2 data was from a total of 1666 (Ca²⁺) and 1849 (EGTA) measurements; Ecad/Ecad data was from a total of 1052 (Ca²⁺) and 2150 (EGTA) measurements; Ecad/Dsg2 data was from a total of 2215 (Ca²⁺) and 2051 (EGTA) measurements; Dsc2/Dsc2 data was from a total of 1658 (Ca²⁺) and 2025 (EGTA) measurements; Dsc2/Dsg2 data was from a total of 1850 (Ca²⁺) and 2025 (EGTA) measurements; Dsc2/Ecad data was from a total of 2122 (Ca²⁺) and 2098 (EGTA) measurements. Error bars are s.e. calculated using bootstrap with replacement.

To determine the adhesive properties of different cadherins, we measured the binding probabilities of various combinations of Dsg2, Dsc2 and Ecad in the presence of Ca²⁺ or in the presence of EGTA, a Ca²⁺ chelating agent. We characterized the levels of nonspecific interactions in every experiment by measuring both the binding probability of a cadherin functionalized AFM cantilever and a substrate lacking cadherin and also the binding probability of an AFM cantilever lacking cadherin and a cadherin-functionalized substrate (Figure 1B); nonspecific binding probabilities (shown as green hatched bars in Figure 1E) are the average of both these sets of measurements. In agreement with our previous results (Lowndes et al., 2014), the probability of homophilic Dsg2 interaction in either the presence (1.6 ± 0.3%) or absence (2.5 ± 0.4%) of Ca²⁺ was comparable to nonspecific adhesion (1.2 ± 0.2% in Ca²⁺ and 1.4 ± 0.2% in EGTA). In contrast, Dsc2/Dsc2 showed a Ca²⁺-dependent homophilic interaction (Lowndes et al., 2014), with a binding probability of 4.1 ± 0.5% in Ca²⁺ and 2.0 ± 0.3% in EGTA (corresponding nonspecific binding was 1.6 ± 0.2% and 1.3 ± 0.2% in Ca²⁺ and EGTA, respectively). Heterophilic interactions between Dsc2 (on AFM tip) and Dsg2 (on CS) were Ca²⁺-independent (Lowndes et al., 2014), with binding probabilities of 4.8 ± 0.5% in Ca²⁺ and 5.6 ± 0.5% in EGTA (nonspecific binding levels were 1.4 ± 0.1% and 1.3 ± 0.1% in Ca²⁺ and EGTA, respectively). As expected, Ecad also showed a Ca²⁺-dependent homophilic interaction with a binding probability of 5.4 ± 0.7% in Ca²⁺ and 2.0 ± 0.3% in EGTA (corresponding nonspecific binding was 0.5 ± 0.1% in Ca²⁺ and 0.9 ± 0.3% in EGTA) (Figure 1E).

Surprisingly, we also measured Ca²⁺-independent heterophilic interactions between Dsg2 (on CS) and Ecad (on AFM tip) with binding probabilities of 5.7 ± 0.5% and 8.4 ± 0.6% in the presence of Ca²⁺ and EGTA respectively (nonspecific binding: 0.9 ± 0.1% in Ca²⁺ and 1.1 ± 0.1% in EGTA). In contrast, Dsc2 on AFM tip and Ecad on the CS did not show any heterophilic binding either in Ca²⁺ or in EGTA; while a binding probability of 0.5 ± 0.2%, comparable to nonspecific adhesion of 1.1 ± 0.1%, was measured in Ca²⁺, the binding frequency of Dsc2/Ecad interactions in EGTA was 0.8 ± 0.2% similar to nonspecific binding of 1.1 ± 0.1%. Our binding probability measurements thus demonstrate that Dsg2 and Ecad form a Ca²⁺-independent heterophilic dimer while Dsc2 does not bind heterophilically to Ecad.

Ecad/Dsg2 and Dsc2/Dsc2 dimers have lower lifetimes than Dsc2/Dsg2 dimers

Next, we compared the dissociation rates for Dsc2/Dsc2, Dsc2/Dsg2 and Ecad/Dsg2 dimers, in the presence of Ca²⁺, using single molecule dynamic force spectroscopy (DFS). The cadherins were immobilized on the AFM tip and substrate as described above and the surface density of the protein was empirically adjusted such that the binding probability was ~6%. Under these conditions, Poisson statistics predicts that more than 97% of measured events are from the rupture of single bonds. The single molecule unbinding events, which were characterized by the non-linear stretching of the PEG tethers (Figure 1C), were fit to an extended freely-jointed chain model (Oesterhelt et al., 1999) using a total least squares fitting protocol. Specific unbinding events were unambiguously identified since they occurred at a distance corresponding to the contour length of two PEG tethers; only specific unbinding events were used in further analysis (see Materials and methods). The measurements were repeated several thousand times at six different rates of application of force (loading rates) and at different positions of the substrate.

The most probable unbinding force at the different loading rates were fit to the Bell-Evans model (Bell, 1978; Evans and Ritchie, 1997) to measure the intrinsic off-rate under zero force, k⁰_off and the width of energy barrier that inhibit protein dissociation, x_β (Figure 2A,B,C). We used cluster analysis to group the single molecule unbinding events for fitting (Yen and Sivasankar, 2018). We have previously shown that a K-means clustering algorithm greatly improves the estimation of kinetic parameters in DFS (Yen and Sivasankar, 2018). This analysis showed that the off-rate of Dsc2/Dsc2 dimers (Figure 2A) and Dsg2/Ecad dimers (Figure 2C) were comparable with a k⁰_off of 1.26 s⁻¹ and 1.24 s⁻¹ respectively. In contrast, the lifetime of the Dsc2/Dsg2 dimer (Figure 2B) was 6x longer with a smaller k⁰_off of 0.20 s⁻¹, demonstrating that the Dsg2/Dsc2 dimer was more stable than either the Dsc2/Dsc2 or the Dsg2/Ecad. These measurements suggest that, upon dissociation of Dsg2/Ecad and Dsc2/Dsc2 complexes, the free Dsg2 and Dsc2 would preferentially bind. In agreement with our data, recent solution binding affinity measurements of desmosomal cadherins have shown heterophilic interactions are stronger than homophilic binding (Harrison et al., 2016).

Figure 2. — Loading rates of the rupture events measured in Ca²⁺ at six different pulling velocities were grouped using K-means clustering method. Each clustered loading rate is shown by a different color, with each circle represent a single rupture event. The mean force and mean loading rates (black filled squares) for the groups were fit to Bell-Evans model (red line) using a nonlinear least-squares fitting with bisquare weights. Fits yielded the intrinsic off-rate (k⁰_off) and the width of the transition energy barrier (x_β). Error bars in force correspond to standard deviation. 95% confidence interval (CI) was calculated using bootstrap with replacement. Analysis shown for (A) Dsc2/Dsc2 (B) Dsc2/Dsg2 and (C) Ecad/Dsg2. The data shown in panels A, B and C correspond to 415 events, 988 events, and 725 events respectively.

Ecad is present in nascent desmosomes but not in mature desmosomes

Next, we used structured illumination microscopy (SIM) to test for the presence of Ecad at different stages of desmosome assembly in human keratinocytes. Keratinocytes were first cultured in a medium containing a low concentration of Ca²⁺ ions not conducive for desmosome formation (100 µM Ca²⁺) and the Ca²⁺ concentration was then increased to trigger desmosome assembly (550 µM Ca²⁺), (Materials and methods). At three time points following the Ca²⁺ switch (1 hr, 3 hr and 18 hr), keratinocytes were fixed and immunostained for Dsg2, Ecad and Desmoplakin (DP), a protein that links desmosomal proteins to the intermediate filament cytoskeleton. Since DP is an obligate desmosomal protein, its distribution allowed us to identify individual desmosomes on the keratinocyte surface with desmosomal junctions defined by regions of parallel DP ‘railroad tracks’ (Stahley et al., 2016a; Stahley et al., 2016b) (Figure 3A,B).

Figure 3. — (A) Analysis of cadherin localization within desmosomes. Structured illumination microscopy (SIM) is able to resolve the distance from plaque to plaque when desmosomes are stained with a C-terminal DP antibody and an N-terminal cadherin antibody, as shown in the example SIM image (Figure 3A). Desmosomes were defined by regions of parallel DP staining, or ‘railroad tracks’. DP (green) and cadherin (either Ecad or Dsg2, red) fluorescence intensity were measured within the desmosome region of interest (black rectangle). (B) Representative images of desmosomal regions in human keratinocytes cultured in high Ca²⁺ media for 1, 3 or 18 hr as indicated. Images are oriented with cell border horizontal. Scale bar, 0.5 µm. (C) Quantification of cadherin (Ecad or Dsg2) levels relative to DP in desmosomes at different time points after initiation of desmosome assembly with high Ca²⁺ culture conditions. AU, arbitrary units. Means ± SE, n = 25 desmosomes, *p<0.05.

Figure 3—figure supplement 1. — (A) Analysis of cadherin localization within desmosomes. Structured illumination microscopy (SIM) is able to resolve the distance from plaque to plaque when desmosomes are stained with a C-terminal DP antibody and an N-terminal cadherin antibody, as shown in the example SIM image (Figure 3A). Desmosomes were defined by regions of parallel DP staining, or ‘railroad tracks’. DP (green) and cadherin (either Ecad or Dsg2, red) fluorescence intensity were measured within the desmosome region of interest (black rectangle). (B) Representative images of desmosomal regions in human keratinocytes cultured in high Ca²⁺ media for 1, 3 or 18 hr as indicated. Images are oriented with cell border horizontal. Scale bar, 0.5 µm. (C) Quantification of cadherin (Ecad or Dsg2) levels relative to DP in desmosomes at different time points after initiation of desmosome assembly with high Ca²⁺ culture conditions. AU, arbitrary units. Means ± SE, n = 25 desmosomes, *p<0.05.

Comparison of relative Dsg2 and Ecad levels contained within the DP railroad tracks demonstrated that Ecad levels were enriched in nascent desmosomes, with relative levels decreasing as desmosomes matured (Figure 3B,C). Compared to Ecad levels in nascent desmosomes, Ecad fluorescence intensity decreased by 50% in mature desmosomes (Figure 3C). In contrast, the relative levels of Dsg2 stayed constant at all time points examined (Figure 3C). Importantly, relative levels of Ecad along the entire cell border did not change with time (Figure 3—figure supplement 1), confirming that the Ecad enrichment within DP railroad tracks at early time points is specific to desmosomal regions of the membrane, and was significant beyond that expected by random chance or because of changes in E-cadherin localization unrelated to junction maturation. Taken together, these data indicate that prior to desmosome maturation, Ecad is significantly enriched in nascent desmosomes.

Leu 175 mediates Ecad and Dsg2 interactions

Next, we proceeded to use single molecule AFM measurements to determine the precise molecular interactions that mediate Ecad/Dsg2 binding. We used mutants that specifically abolish either Ecad trans or cis interactions and tested their binding to Dsg2. Structural studies show that trans strand-swap dimer formation can be eliminated by mutating a conserved Trp2 to Ala (W2A) (Harrison et al., 2010). Similarly, mutating a conserved Lys14 to Glu (K14E) eliminates a key salt-bridge in the X-dimer interface and abolishes X-dimer formation (Harrison et al., 2010). We therefore tested the binding between the Ecad W2A-K14E double mutant (DM) and Dsg2. In agreement with previous results (Harrison et al., 2010; Rakshit et al., 2012), we confirmed that the DM cannot interact homophilically; the binding interaction between opposing DMs (1.4 ± 0.3% in Ca²⁺ and 1.8 ± 0.3% in EGTA) were comparable to nonspecific binding in Ca²⁺(1.4 ± 0.1%) and in EGTA (1.1 ± 0.1%) (Figure 4). However, when we measured the interactions between the DM on AFM tip and Dsg2 on CS, our data showed that the DM interacts with Dsg2 in a Ca²⁺-independent heterophilic fashion; while the DM/Dsg2 binding probability in Ca²⁺ and in EGTA was 6.1 ± 0.5% and 6.9 ± 0.5% respectively, the non-specific interaction in Ca²⁺ and EGTA was 1.3 ± 0.1% and 1.2 ± 0.1% respectively (Figure 4). This demonstrates that the Ecad/Dsg2 binding interface is different from the previously established interface for Ecad trans dimerization.

Figure 4. — Homophilic binding probability of Ecad W2A-K14E double mutant (DM); heterophilic binding probability of DM and Dsg2; homophilic binding probability of Ecad L175D *cis* dimer mutant (CM); and the heterophilic binding probability of CM and Dsg2 was measured in Ca²⁺ (blue) and in EGTA (red). Nonspecific binding probabilities determined from the average of measured binding probabilities between a cadherin functionalized AFM tip and a surface lacking cadherin and between an AFM tip lacking cadherin and surface functionalized with biotinylated cadherins are shown in shaded green. DM/DM data was from a total of 1898 (Ca²⁺) and 2122 (EGTA) measurements; DM/Dsg2 data was from a total of 2150 (Ca²⁺) and 2009 (EGTA) measurements; CM/CM data was from a total of 1970 (Ca²⁺) and 1906 (EGTA) measurements; CM/Dsg2 data was from a total of 2027 (Ca²⁺) and 2122 (EGTA) measurements. Error bars are s.e. calculated using bootstrap with replacement.

Next, we tested whether Ecad interacts with Dsg2 via its cis dimer interface. Since previous studies had shown that mutating a conserved Leu 175 to Asp (L175D) eliminates Ecad cis dimerization (Harrison et al., 2011), we measured the interaction of this Ecad cis mutant (CM) and Dsg2. First, we confirmed that CM was functional by measuring its trans binding probability. As shown previously (Harrison et al., 2011), our data confirmed that the CM forms Ca²⁺-dependent trans dimers; the binding probabilities in Ca²⁺ and in EGTA were 4.1 ± 0.5% and 0.8 ± 0.2% respectively (corresponding nonspecific binding probabilities in Ca²⁺ and in EGTA were 1.2 ± 0.2% and 1.4 ± 0.2% respectively) (Figure 4). However, the interaction of CM (on CS) and Dsg2 (on AFM tip) was comparable to the measured nonspecific binding; we measured binding probabilities of 1.8 ± 0.3% in Ca²⁺ and 1.5 ± 0.3% in EGTA which were similar to measured nonspecific binding levels of 1.2 ± 0.1% in Ca²⁺ and 1.4 ± 0.1% in EGTA (Figure 4). Previous simulations (Wu et al., 2010; Wu et al., 2011) and single molecule FRET experiment (Zhang et al., 2009) have shown that cis homo-dimerization of Ecad requires prior trans dimer formation. Thus, the failure to form CM/Dsg2 interactions is not due to the abrogation of Ecad cis dimers. Taken together, these experiments show that Ecad/Dsg2 dimerization is mediated by L175, which also mediates cis homo-dimerization in Ecad.

Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly

To test whether the Ecad L175D mutation can impede recruitment of Dsg2 in keratinocytes, we expressed either full-length Ecad WT or mutants in E^KO/P^KD keratinocytes and analyzed Dsg2 recruitment to sites of intercellular contacts. We have previously shown that the E^KO/P^KD keratinocytes are unable to assemble adherens junction (AJs) and desmosomes due to the loss of all classical cadherins (Michels et al., 2009). As we wanted to assess the ability of Ecad mutants to recruit desmosomal proteins early during junction formation, we performed confocal microscopy on keratinocytes immunostained for Ecad, Dsg2 and DP, at three time points following the Ca²⁺ switch (3 hr, 6 hr and 18 hr, Figure 5, Materials and methods). The ability of transfected keratinocytes to form AJs was first assessed by the formation of ‘zipper-like’ patterns of Ecad at intercellular contacts. We have previously shown that these zippers represent early AJs and recruit AJ marker proteins like vinculin (Rübsam et al., 2017b). Only intercellular interfaces with AJ zippers were examined for their ability to recruit Dsg2 and DP to these contacts (Figure 5—figure supplement 1A). Importantly, we never observed any enrichment of desmosomal components at intercellular contacts in the absence of AJ zippers (Michels et al., 2009).

Figure 5. — (A) Immunofluorescence analysis for transfected WT or mutant Ecad (myc) and Dsg2, 6 hr after allowing de novo junction formation in Ecad^KO/Pcad^KD keratinocytes. Note decreased localization of Dsg2 at intercellular contacts formed by Ecad-L175D. (B) Quantification of adherens junction (AJ) formation at intercellular contacts, judged by zipper-like enrichment of Ecad constructs. (C) Quantification of Dsg2 recruitment to AJ formed by WT or mutant Ecad. (D) Immunofluorescence analysis for transfected WT or mutant Ecad (myc) and desmoplakin (DP), 6 hr after allowing de novo junction formation in Ecad^KO/Pcad^KD keratinocytes. Note decreased localization of DP at intercellular contacts formed by Ecad-L175D. (E) Quantification of DP recruitment to AJ formed by WT or mutant Ecad. Numbers of quantified junctions (interface of two contacting, transfected cells) are shown above each bar and are derived from at least three independent experiments. Scale bar: 10 µm.

In agreement with previous results (Vasioukhin et al., 2000), our data showed that 3 hr after the Ca²⁺ switch, 93% of Ecad WT was enriched in zipper-like early AJs at sites of intercellular contacts. In contrast only 48% of Ecad-L175D transfected keratinocytes formed AJ zippers, likely due to impaired Ecad cis-dimer formation (Figure 5A,B) (Harrison et al., 2011). At these early time points following the Ca²⁺ switch, 66% of these Ecad WT zipper contacts were positive for Dsg2. Junctional localization of Ecad and Dsg2 increased to 97%, 18 hr after allowing cells to engage in Ca²⁺-dependent intercellular adhesion (Figure 5A,C). In contrast, only 39% of Ecad-L175D induced zipper contacts showed Dsg2 recruitment 3 hr after switching to high Ca²⁺ (Figure 5A,C), which increased to 65% at 18 hr, confirming that a direct interaction between Ecad and Dsg2 is required for efficient recruitment of Dsg2 to early intercellular contacts.

That there is no co-localization of DP and Ecad in the absence of zippers (Figure 5—figure supplement 1) suggests that Ecad trans interactions precede Ecad/Dsg2 interactions. This conclusion is further strengthened by our finding that DP was also not recruited upon transfection of the full-length Ecad-DM (W2A-K14E double mutant, Figure 5—figure supplement 1) in E^KO/P^KD keratinocytes, which abolishes Ecad trans adhesion, and thus zipper formation.

Finally, to examine whether L175D only interferes with Dsg2 recruitment or more generally delays desmosome assembly, we also examined the ability of Ecad-L175D to hinder DP recruitment. In E^KO/P^KD keratinocytes transfected with Ecad WT, DP was enriched in 91% of Ecad-positive intercellular contacts after 3 hr in high Ca²⁺ and in 100% of contacts after 18 hr in high Ca²⁺ (Figure 5D,E). In contrast, only 28% of Ecad-L175D established intercellular contacts were positive for DP after 3 hr, which increased to 72% after 18 hr. Importantly, after 3 hr, 73% of Ecad-K14E mutant contacts were DP positive, despite widespread defects in AJ formation (Figure 5B,D,E). Taken together these results confirm that amino acid L175 in Ecad mediates Dsg2 interactions and facilitates early desmosome complex formation in cells.

Discussion

Here, we integrate in vitro single molecule and cellular structure-function experiments to identify two critical events that initiate and promote efficient desmosome assembly: (i) stable trans-homodimerization of Ecad, and (ii) the direct heterophilic binding of Ecad and Dsg2 ectodomains. Our data demonstrates that desmosome assembly is initiated at sites of Ecad trans homodimerization. We also show that Ecad and Dsg2 bind via a conserved Leu 175 on the Ecad cis binding interface and form short-lived heterophilic complexes that localize to early desmosomes and efficiently recruit desmosomal proteins to sites of intercellular contact formation.

Previously, biochemical analysis identified Ecad/Dsg and Dsc/Dsg complexes upon removing Ca²⁺ from the cell culture (Troyanovsky et al., 1999). Our data builds upon these previous findings by showing a direct interaction between Ecad and Dsg that also occurs under conditions that promote intercellular adhesions and by identifying the Ecad interface responsible for the interaction. We also demonstrate that in a physiological setting, L175 on Ecad is important for recruitment of Dsg2 and DP and for efficient desmosome formation. It has also been suggested that cross-talk between classical and desmosomal cadherins is mediated by plakoglobin, a cytoplasmic signaling protein found in both AJs and desmosomes (Lewis et al., 1997). While we did not test the role of plakoglobin in desmosome formation, our data indicates that a direct physical interaction between Ecad and Dsg2 ectodomains is critical for timely desmosome assembly.

It is important to point out that Ecad/Dsg interactions reported here may not be unique to Dsg2 but may occur with other Dsg isoforms as well. In keratinocytes, Dsg3 is essential for desmosome assembly (Hartlieb et al., 2013; Koch et al., 1997) and a previous fluorescence co-localization and co-immunoprecipitation study reported interactions between Ecad and Dsg3 (Tsang et al., 2010). Since the E^KO/P^KD keratinocytes transfected with Ecad-L175D also express Dsg3, it is possible that poor junctional DP enrichment seen at early time-points, which was more impaired than Dsg2 recruitment, may result from the inability of Ecad-L175 to bind Dsg3.

Desmosome assembly is Ca²⁺ dependent and previous models using Dsc2 and Dsg2 propose that desmosomal cadherins assemble in two phases (Burdett and Sullivan, 2002; Lowndes et al., 2014). The first phase of assembly is believed to involve the clustering of Dsc into nucleation sites. Since Dsc2 homo-dimerization is Ca²⁺ dependent (Figure 1E), it follows that desmosome assembly requires Ca²⁺, even though Dsc2/Dsg2 heterodimers are Ca²⁺ independent. In the second phase of desmosome formation, Dsg2 is recruited to the clustered Dsc2 nucleation sites through an unknown mechanism that is Ca²⁺ and W2 independent and that relies on heterophilic interactions with other proteins (Lowndes et al., 2014). Our data suggests that Ecad mediates Ca²⁺ independent recruitment of Dsg2, and perhaps other Dsgs, to nascent Dsc clusters. Previous studies have also shown that initial cell-cell contacts subsequently trigger multiple phases of DP recruitment to finally assemble desmosomes (Godsel et al., 2005).

Due to the transient nature of Ecad/Dsg2 interactions, desmosomal proteins recruited to junctions are expected to rapidly segregate from Ecad. Consequently, images of intercellular interfaces are expected to have both overlapping and separate Ecad and desmosomal protein staining patterns. In agreement, although Dsg2 and DP show some co-localization at Ecad-positive zippers (Figure 5), significant non-overlapping junctional staining is observed, even within the first 3 hr. This suggests that AJs and desmosomes rapidly segregate into distinct intercellular junctions.

Although our biophysical experiments demonstrate that Ecad-DM directly interact with Dsg2, imaging of keratinocytes transfected with Ecad-K14E and Ecad-DM show that stable trans Ecad homodimerization is required for DP and Dsg2 recruitment (Figure 5—figure supplement 1). While the K14E mutants are strongly impaired in AJ zipper formation, they only localize with Dsg2 or DP when AJ zippers are present. Similarly, cells expressing Ecad-DM which do not form AJs, also do not recruit DP or Dsg2. These results indicate that Ecad trans binding serves as the initial spatial cue for the subsequent recruitment of Dsg2 and consequently for desmosome assembly. These results are thus in line with the observation that combined loss of both E-, and Pcad, the two main classical cadherins expressed in keratinocytes, prevents desmosome assembly (Michels et al., 2009). Mechanistically, the cross talk between Ecad trans dimerization and Ecad/Dsg2 binding may be analogous to Ecad cis homodimerization via L175, which occurs only when conformational entropy is reduced by prior Ecad trans binding (Wu et al., 2010; Wu et al., 2011). Alternatively, Ecad trans dimers may be needed to bring opposing cells closer together to initiate desmosome formation. In support of this possibility, a previous study has suggested that when opposing HeLa cell membranes are brought into close proximity, by the interaction of protein zero neuronal adhesion molecules, desmosomes are formed (Doyle et al., 1995).

Based on these previous studies and our new data, we propose a model (Figure 6) whereby desmosome assembly is facilitated both by the direct cis interactions of classical cadherins and Dsg and the trans binding of opposing classical cadherins. In our model, the trans homodimerization of Ecad from opposing cells serves as a spatial cue to coordinate desmosome assembly. Ecad subsequently forms cis dimer complexes with Dsg though the results presented here do not rule out Ecad/Dsg binding in a trans conformation. Furthermore, the data suggest that once localized within the native desmosome, Dsg dissociates from Ecad and binds to Dsc to form mature desmosomes. Since the Dsc2/Dsg2 complex has a longer lifetime than both the Ecad/Dsg2 complex and Dsc2/Dsc2 complex, the Dsg/Dsc interactions likely promote robust cell-cell adhesion and permit the mature desmosome to withstand mechanical force.

Materials and methods

Purification of cadherin ectodomains

The generation of HEK293T cells stably expressing Ecad WT and Ecad W2A-K14E fused to a C-terminal Avi-tag and His-tag have been described previously (Rakshit et al., 2012; Zhang et al., 2009). Plasmids for full length Dsc2 and Dsg2 ectodomains fused to a C-terminal Avi-tag and His-tag were a kind gift from Prof. W. James Nelson (Stanford University) while HEK293T cells stably expressing full-length Ecad mutant L175D fused to a C-terminal Avi-tag and His-tag were a kind gift from Dr. Yunxiang Zhang (Stanford University). As described previously (Lowndes et al., 2014), the Dsc2 and Dsg2 plasmids were transiently transfected into HEK 293 T cells using lipofactamine 2000 (Life Technologies) following the manufacturer’s protocol. Two days post transfection, the conditioned media was collected for protein purification. The transfected cells expressing WT/mutant Ecads were grown to confluence in DMEM containing 10% FBS and 200 µg/ml of Genecitin (G418; Invitrogen) and exchanged into serum free DMEM with 400 µg/ml of Genecitin. Conditioned media was collected 4 days after media exchange.

Purification and biotinylation of His-tagged Dsc2, Dsg2, Ecad WT, W2A-K14E, and L175D followed a protocol that has been described previously (Lowndes et al., 2014; Rakshit et al., 2012; Zhang et al., 2009). Media containing cadherin was incubated overnight, at 4°C, with Ni NTA agarose beads (Qiagen). The beads were loaded onto a glass chromatography column (BioRad) and washed with buffer at pH 7.5 (20 mM NaH₂PO₄, 500 mM NaCl and 1 mM CaCl₂) containing 50 mM imidazole. The bound protein was eluted with the same buffer containing 250 mM imidazole. Following purification, the protein was exchanged into a pH 7.5 buffer containing 25 mM Hepes, 5 mM NaCl, and 1 mM CaCl₂ and biotinylated with BirA enzyme (BirA500 kit; Avidity). After biotinylation for 1 hr at 30°C, free biotins were removed using either a spin column (Millipore) or superdex 200 10/300 GL column.

Single molecule AFM force measurements

Purified cadherins were immobilized on coverslips (CS) and AFM cantilevers (Olympus, model TR400PSA) using a previously described method (Manibog et al., 2016). Briefly, the CS and cantilevers were cleaned with 25% H₂O₂:75% H₂SO₄ and washed with DI water. The CS was then cleaned with 1 M KOH and washed with DI water. Both the CS and cantilevers were washed with acetone and functionalized using 2% (v/v) 3-aminopropyltriethoxysilane (Sigma) dissolved in acetone. Next, N-hydroxysuccinimide ester functionalized PEG spacers (MW 5000, Lysan Bio) were covalently attached to the silanized AFM tip and coverslip; 7% of the PEG spacers were decorated with biotin groups. Prior to a measurement, the functionalized AFM cantilever and coverslip were incubated overnight with BSA (1 mg/ml) to further reduce non-specific binding. The tip and surface were then incubated with 0.1 mg/ml streptavidin for 30 min and biotinylated cadherins were attached to the streptavidin. Finally, the surfaces were incubated with 0.02 mg/ml biotin for 10 min to block the free biotin binding sites on streptavidin.

Force measurements were performed using an Agilent 5500 AFM with a closed loop scanner. The spring constants of the cantilevers were measured using the thermal fluctuation method (Hutter and Bechhoefer, 1993). All the experiments were performed in a pH 7.5 buffer containing 10 mM Tris-HCl, 100 mM NaCl and 10 mM KCl with either 2.5 mM Ca²⁺ or 2 mM EGTA. The region of PEG stretching in each unbinding force curve was fit to an extended freely jointed chain model using a total least squares fitting protocol. The contour length $L_{c}$ of the PEG tethers was determined from the fits. The histogram of $L_{c}$ for each experiment was fit to a Gaussian distribution and only force curves that had an $L_{c}$ within one standard deviation from the center were accepted for further analysis. Loading rates were calculated as described elsewhere (Ray et al., 2007). K-means clustering method was used to group loading rates (Yen and Sivasankar, 2018). Mean force $F^{*}$ and mean loading rate $r_{f}$ were calculated for each group and plots of $F$ * vs. $r_{f}$ were fit using nonlinear least-squares fitting with bisquare weights to the Bell-Evans model (Bell, 1978; Evans and Ritchie, 1997) $.$ Confidence intervals (CIs) for $k_{o f f}^{0}$ and $x_{β}$ were determined using bootstrap-with-replacement, as described previously (Yen et al., 2016).

SIM imaging and analysis of cadherin localization within desmosomes

Primary human keratinocytes (HKs, passage 2) were isolated from neonatal foreskin as previously described (Calkins et al., 2006) and cultured in KBM-Gold basal medium (100 µM calcium) supplemented with KGM-Gold Single-Quot Kit (Lonza, Walkersville, MD). HKs were cultured to 70% confluence on glass coverslips, switched to 550 µM calcium to induce junction assembly for the various time points indicated and then processed for structured illumination microscopy (SIM) as described below.

HKs were fixed in methanol and immunostained with primary antibodies for 1 hr and secondary antibodies for 30 min, both at 37°C. The following primary antibodies were used in the SIM experiments: mouse anti-Ecad antibody (HECD-1, Abcam); rat anti-uvomorulin (DECMA-1, Sigma); mouse anti-Dsg2 antibody (AH12.2, a kind gift from Dr. Asma Nusrat, Emory University); desmoplakin antibody (NW6, a kind gift from Dr. Kathleen Green, Northwestern University). Secondary antibodies conjugated to Alexa Fluorophore were purchased from Invitrogen. SIM was performed using the Nikon N-SIM system on an Eclipse Ti-E microscopy system equipped with a 100x/1.49 NA oil immersion objective and 488 and 561 nm solid-state lasers. 3D SIM images were captured with an EM charge-coupled device camera (DU-897, Andor Technology) and reconstructed using NIS-Elements software with the N-SIM module (version 3.22, Nikon).

SIM is able to resolve the distance from plaque to plaque when desmosomes are stained with a C-terminal DP antibody and an N-terminal cadherin antibody, as shown in the example SIM image (Figure 3A). For analysis of cadherin localization within desmosomes, desmosomes were first defined by regions of parallel DP staining, or ‘railroad tracks’. Using ImageJ, a desmosome region of interest (black rectangle) was identified via DP staining (Figure 3A). Once a DP and ‘railroad track’-positive region of interest was identified, cadherin (red) fluorescence intensity levels were then independently measured along with DP (green) levels. Pairwise multiple comparisons were performed via a Tukey test with a significance level of α = 0.05.

Isolation, culture, transfection and confocal imaging of primary keratinocytes

Spontaneously immortalized primary keratinocytes, isolated from newborn mice, were cultured in DMEM/HAM’s F12 (FAD) medium with low Ca²⁺ (50 μM) (Biochrom) supplemented with 10% FCS (chelated), penicillin (100 U ml⁻¹), streptomycin (100 μg ml⁻¹, Biochrom A2212), adenine (1.8 × 10⁻⁴ M, Sigma A3159), L-glutamine (2 mM, Biochrom K0282), hydrocortisone (0.5 μg ml⁻¹, Sigma H4001), EGF (10 ng ml⁻¹, Sigma E9644), cholera enterotoxin (10⁻¹⁰ M, Sigma C-8052), insulin (5 μg ml⁻¹, Sigma I1882), and ascorbic acid (0.05 mg ml⁻¹, Sigma A4034). Keratinocytes were kept at 32°C and 5% CO₂. Ecad^KO/Pcad^KD cells were generated by lentiviral transduction of Ecad-deficient keratinocytes using C14 shRNA directed against Pcad (Michels et al., 2009). Cultured cells were regularly monitored for mycoplasma contamination and discarded in case of positive results. Cellular identity was validated by PCR genotyping from genomic DNA and western blot analysis of Dsg 1-2 as markers for keratinocyte identity. In addition, loss of Ecad and efficient knockdown of Pcad was confirmed on the RNA level by RT-PCR as well as on the protein level through western blot analysis.

Ecad-K14E and Ecad-L175D mutants were generated using WT mouse Ecad cDNA in a pcDNA3 backbone including a C-terminal 6myc-tag. Mutations were carried out using ‘QuikChange Lightning Site-Directed Mutagenesis Kit’ (Agilent). Keratinocytes were transfected at 80–100% confluency with ViromerRed (lipocalyx) according to the manufacturer’s protocol. In brief 1.5 µg DNA were diluted in 100 µl Buffer, added to 1.25 µl ViromerRED and incubated for 15 min at room temperature. Approximately 33 µl transfection mix were used per well (24 well plate).

Confocal images were obtained with a Leica TCS SP8, equipped with a white light laser and gateable hybrid detectors (HyDs) and a PlanApo 63x, 1.4 NA objective. Epifluorescence images were obtained with a Leica DMI6000 with a PlanApo 63x, 1.4 NA objective. The following primary antibodies were used in this study: rabbit monoclonal against Dsg2 (1:500, Abcam #ab150372); mouse monoclonal against DP1/2 (1:200, Progen #61003); mouse monoclonal against c-myc (IF 1:2000, Cell Signaling #2276); Secondary antibodies were species-specific antibodies conjugated with either AlexaFluor 488, 594 or 647, used at a dilution of 1:500 for immunofluorescence (Molecular Probes, Life Technologies)

Acknowledgements

This research was supported in part by the American Heart Association (12SDG9320022) and National Institute of General Medical Sciences of the National Institutes of Health (R01GM121885) to SS; by the Deutsche Forschungsgemeinschaft (DFG SFB 829 A1, DFG SFB 829 Z2, DFG NI 1234/6–1, DFG SPP 1782) to CMN; and by the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health (R01AR048266) to APK.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Sanjeevi Sivasankar, Email: sivasank@iastate.edu.

William I Weis, Stanford University Medical Center, United States.

Anna Akhmanova, Utrecht University, Netherlands.

Funding Information

This paper was supported by the following grants:

National Institute of Arthritis and Musculoskeletal and Skin Diseases R01AR048266 to Andrew P Kowalczyk.
Deutsche Forschungsgemeinschaft DFG SFB 829 A1 to Carien M Niessen.
Deutsche Forschungsgemeinschaft DFG SFB 829 Z2 to Carien M Niessen.
Deutsche Forschungsgemeinschaft DFG NI 1234/6-1 to Carien M Niessen.
Deutsche Forschungsgemeinschaft DFG SPP 1782 to Carien M Niessen.
American Heart Association 12SDG9320022 to Sanjeevi Sivasankar.
National Institute of General Medical Sciences R01GM121885 to Sanjeevi Sivasankar.

Additional information

Competing interests

No competing interests declared.

Author contributions

Formal analysis, Investigation, Methodology, Writing—original draft.

Formal analysis, Investigation, Methodology, Writing—review and editing.

Investigation, Methodology.

Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing—review and editing.

Conceptualization, Formal analysis, Supervision, Funding acquisition, Writing—original draft, Project administration.

Additional files

Transparent reporting form

elife-37629-transrepform.docx^{(244.3KB, docx)}

DOI: 10.7554/eLife.37629.010

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.

References

Amagai M, Fujimori T, Masunaga T, Shimizu H, Nishikawa T, Shimizu N, Takeichi M, Hashimoto T. Delayed assembly of desmosomes in keratinocytes with disrupted classic-cadherin-mediated cell adhesion by a dominant negative mutant. Journal of Investigative Dermatology. 1995;104:27–32. doi: 10.1111/1523-1747.ep12613462. [DOI] [PubMed] [Google Scholar]
Amagai M, Stanley JR. Desmoglein as a target in skin disease and beyond. Journal of Investigative Dermatology. 2012;132:776–784. doi: 10.1038/jid.2011.390. [DOI] [PMC free article] [PubMed] [Google Scholar]
Awad MM, Dalal D, Cho E, Amat-Alarcon N, James C, Tichnell C, Tucker A, Russell SD, Bluemke DA, Dietz HC, Calkins H, Judge DP. DSG2 mutations contribute to arrhythmogenic right ventricular dysplasia/cardiomyopathy. The American Journal of Human Genetics. 2006;79:136–142. doi: 10.1086/504393. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bell GI. Models for the specific adhesion of cells to cells. Science. 1978;200:618–627. doi: 10.1126/science.347575. [DOI] [PubMed] [Google Scholar]
Boggon TJ, Murray J, Chappuis-Flament S, Wong E, Gumbiner BM, Shapiro L. C-cadherin ectodomain structure and implications for cell adhesion mechanisms. Science. 2002;296:1308–1313. doi: 10.1126/science.1071559. [DOI] [PubMed] [Google Scholar]
Brasch J, Harrison OJ, Honig B, Shapiro L. Thinking outside the cell: how cadherins drive adhesion. Trends in Cell Biology. 2012;22:299–310. doi: 10.1016/j.tcb.2012.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
Burdett ID, Sullivan KH. Desmosome assembly in MDCK cells: transport of precursors to the cell surface occurs by two phases of vesicular traffic and involves major changes in centrosome and Golgi location during a Ca(2+) shift. Experimental Cell Research. 2002;276:296–309. doi: 10.1006/excr.2002.5509. [DOI] [PubMed] [Google Scholar]
Calkins CC, Setzer SV, Jennings JM, Summers S, Tsunoda K, Amagai M, Kowalczyk AP. Desmoglein endocytosis and desmosome disassembly are coordinated responses to pemphigus autoantibodies. Journal of Biological Chemistry. 2006;281:7623–7634. doi: 10.1074/jbc.M512447200. [DOI] [PubMed] [Google Scholar]
Ciatto C, Bahna F, Zampieri N, VanSteenhouse HC, Katsamba PS, Ahlsen G, Harrison OJ, Brasch J, Jin X, Posy S, Vendome J, Ranscht B, Jessell TM, Honig B, Shapiro L. T-cadherin structures reveal a novel adhesive binding mechanism. Nature Structural & Molecular Biology. 2010;17:339–347. doi: 10.1038/nsmb.1781. [DOI] [PMC free article] [PubMed] [Google Scholar]
Doyle JP, Stempak JG, Cowin P, Colman DR, D'Urso D. Protein zero, a nervous system adhesion molecule, triggers epithelial reversion in host carcinoma cells. The Journal of Cell Biology. 1995;131:465–482. doi: 10.1083/jcb.131.2.465. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dusek RL, Godsel LM, Green KJ. Discriminating roles of desmosomal cadherins: beyond desmosomal adhesion. Journal of Dermatological Science. 2007;45:7–21. doi: 10.1016/j.jdermsci.2006.10.006. [DOI] [PubMed] [Google Scholar]
Eshkind L, Tian Q, Schmidt A, Franke WW, Windoffer R, Leube RE. Loss of desmoglein 2 suggests essential functions for early embryonic development and proliferation of embryonal stem cells. European Journal of Cell Biology. 2002;81:592–598. doi: 10.1078/0171-9335-00278. [DOI] [PubMed] [Google Scholar]
Evans E, Ritchie K. Dynamic strength of molecular adhesion bonds. Biophysical Journal. 1997;72:1541–1555. doi: 10.1016/S0006-3495(97)78802-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Getsios S, Huen AC, Green KJ. Working out the strength and flexibility of desmosomes. Nature Reviews Molecular Cell Biology. 2004;5:271–281. doi: 10.1038/nrm1356. [DOI] [PubMed] [Google Scholar]
Godsel LM, Hsieh SN, Amargo EV, Bass AE, Pascoe-McGillicuddy LT, Huen AC, Thorne ME, Gaudry CA, Park JK, Myung K, Goldman RD, Chew TL, Green KJ. Desmoplakin assembly dynamics in four dimensions: multiple phases differentially regulated by intermediate filaments and actin. The Journal of Cell Biology. 2005;171:1045–1059. doi: 10.1083/jcb.200510038. [DOI] [PMC free article] [PubMed] [Google Scholar]
Green KJ, Simpson CL. Desmosomes: new perspectives on a classic. Journal of Investigative Dermatology. 2007;127:2499–2515. doi: 10.1038/sj.jid.5701015. [DOI] [PubMed] [Google Scholar]
Gumbiner B, Stevenson B, Grimaldi A. The role of the cell adhesion molecule uvomorulin in the formation and maintenance of the epithelial junctional complex. The Journal of Cell Biology. 1988;107:1575–1587. doi: 10.1083/jcb.107.4.1575. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harrison OJ, Bahna F, Katsamba PS, Jin X, Brasch J, Vendome J, Ahlsen G, Carroll KJ, Price SR, Honig B, Shapiro L. Two-step adhesive binding by classical cadherins. Nature Structural & Molecular Biology. 2010;17:348–357. doi: 10.1038/nsmb.1784. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harrison OJ, Brasch J, Lasso G, Katsamba PS, Ahlsen G, Honig B, Shapiro L. Structural basis of adhesive binding by desmocollins and desmogleins. PNAS. 2016;113:7160–7165. doi: 10.1073/pnas.1606272113. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harrison OJ, Jin X, Hong S, Bahna F, Ahlsen G, Brasch J, Wu Y, Vendome J, Felsovalyi K, Hampton CM, Troyanovsky RB, Ben-Shaul A, Frank J, Troyanovsky SM, Shapiro L, Honig B. The extracellular architecture of adherens junctions revealed by crystal structures of type I cadherins. Structure. 2011;19:244–256. doi: 10.1016/j.str.2010.11.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hartlieb E, Kempf B, Partilla M, Vigh B, Spindler V, Waschke J. Desmoglein 2 is less important than desmoglein 3 for keratinocyte cohesion. PLoS ONE. 2013;8:e53739. doi: 10.1371/journal.pone.0053739. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hong S, Troyanovsky RB, Troyanovsky SM. Cadherin exits the junction by switching its adhesive bond. The Journal of Cell Biology. 2011;192:1073. doi: 10.1083/jcb.201006113. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hutter JL, Bechhoefer J. Calibration of atomic‐force microscope tips. Review of Scientific Instruments. 1993;64:1868–1873. doi: 10.1063/1.1143970. [DOI] [Google Scholar]
Jones JC. Characterization of a 125K glycoprotein associated with bovine epithelial desmosomes. Journal of Cell Science. 1988;89:207–216. doi: 10.1242/jcs.89.2.207. [DOI] [PubMed] [Google Scholar]
Koch PJ, Mahoney MG, Ishikawa H, Pulkkinen L, Uitto J, Shultz L, Murphy GF, Whitaker-Menezes D, Stanley JR. Targeted disruption of the pemphigus vulgaris antigen (desmoglein 3) gene in mice causes loss of keratinocyte cell adhesion with a phenotype similar to pemphigus vulgaris. The Journal of Cell Biology. 1997;137:1091. doi: 10.1083/jcb.137.5.1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lewis JE, Jensen PJ, Wheelock MJ. Cadherin function is required for human keratinocytes to assemble desmosomes and stratify in response to calcium. Journal of Investigative Dermatology. 1994;102:870–877. doi: 10.1111/1523-1747.ep12382690. [DOI] [PubMed] [Google Scholar]
Lewis JE, Wahl JK, Sass KM, Jensen PJ, Johnson KR, Wheelock MJ. Cross-talk between adherens junctions and desmosomes depends on plakoglobin. The Journal of Cell Biology. 1997;136:919–934. doi: 10.1083/jcb.136.4.919. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lowndes M, Rakshit S, Shafraz O, Borghi N, Harmon RM, Green KJ, Sivasankar S, Nelson WJ. Different roles of cadherins in the assembly and structural integrity of the desmosome complex. Journal of Cell Science. 2014;127:2339–2350. doi: 10.1242/jcs.146316. [DOI] [PMC free article] [PubMed] [Google Scholar]
Manibog K, Li H, Rakshit S, Sivasankar S. Resolving the molecular mechanism of cadherin catch bond formation. Nature Communications. 2014;5:3941. doi: 10.1038/ncomms4941. [DOI] [PubMed] [Google Scholar]
Manibog K, Sankar K, Kim SA, Zhang Y, Jernigan RL, Sivasankar S. Molecular determinants of cadherin ideal bond formation: conformation-dependent unbinding on a multidimensional landscape. PNAS. 2016;113:E5711–E5720. doi: 10.1073/pnas.1604012113. [DOI] [PMC free article] [PubMed] [Google Scholar]
Michels C, Buchta T, Bloch W, Krieg T, Niessen CM. Classical cadherins regulate desmosome formation. Journal of Investigative Dermatology. 2009;129:2072–2075. doi: 10.1038/jid.2009.17. [DOI] [PubMed] [Google Scholar]
Oesterhelt F, Rief M, Gaub HE. Single molecule force spectroscopy by AFM indicates helical structure of poly(ethylene-glycol) in water. New Journal of Physics. 1999;1:6. doi: 10.1088/1367-2630/1/1/006. [DOI] [Google Scholar]
Parisini E, Higgins JM, Liu JH, Brenner MB, Wang JH. The crystal structure of human E-cadherin domains 1 and 2, and comparison with other cadherins in the context of adhesion mechanism. Journal of Molecular Biology. 2007;373:401–411. doi: 10.1016/j.jmb.2007.08.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rakshit S, Zhang Y, Manibog K, Shafraz O, Sivasankar S. Ideal, catch, and slip bonds in cadherin adhesion. PNAS. 2012;109:18815–18820. doi: 10.1073/pnas.1208349109. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ray C, Brown JR, Akhremitchev BB. Rupture force analysis and the associated systematic errors in force spectroscopy by AFM. Langmuir. 2007;23:6076–6083. doi: 10.1021/la070131e. [DOI] [PubMed] [Google Scholar]
Rübsam M, Broussard JA, Wickström SA, Nekrasova O, Green KJ, Niessen CM. Adherens junctions and desmosomes coordinate mechanics and signaling to orchestrate tissue morphogenesis and function: an evolutionary perspective. Cold Spring Harbor Perspectives in Biology. 2017a:a029207. doi: 10.1101/cshperspect.a029207. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rübsam M, Mertz AF, Kubo A, Marg S, Jüngst C, Goranci-Buzhala G, Schauss AC, Horsley V, Dufresne ER, Moser M, Ziegler W, Amagai M, Wickström SA, Niessen CM. E-cadherin integrates mechanotransduction and EGFR signaling to control junctional tissue polarization and tight junction positioning. Nature Communications. 2017b;8:1250. doi: 10.1038/s41467-017-01170-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sivasankar S, Zhang Y, Nelson WJ, Chu S. Characterizing the initial encounter complex in cadherin adhesion. Structure. 2009;17:1075–1081. doi: 10.1016/j.str.2009.06.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stahley SN, Bartle EI, Atkinson CE, Kowalczyk AP, Mattheyses AL. Molecular organization of the desmosome as revealed by direct stochastic optical reconstruction microscopy. Journal of Cell Science. 2016a;129:2897–2904. doi: 10.1242/jcs.185785. [DOI] [PMC free article] [PubMed] [Google Scholar]
Stahley SN, Warren MF, Feldman RJ, Swerlick RA, Mattheyses AL, Kowalczyk AP. Super-Resolution microscopy reveals altered desmosomal protein organization in tissue from patients with Pemphigus vulgaris. Journal of Investigative Dermatology. 2016b;136:59–66. doi: 10.1038/JID.2015.353. [DOI] [PMC free article] [PubMed] [Google Scholar]
Syrris P, Ward D, Evans A, Asimaki A, Gandjbakhch E, Sen-Chowdhry S, McKenna WJ. Arrhythmogenic right ventricular dysplasia/cardiomyopathy associated with mutations in the desmosomal gene desmocollin-2. The American Journal of Human Genetics. 2006;79:978–984. doi: 10.1086/509122. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tinkle CL, Pasolli HA, Stokes N, Fuchs E. New insights into cadherin function in epidermal sheet formation and maintenance of tissue integrity. PNAS. 2008;105:15405–15410. doi: 10.1073/pnas.0807374105. [DOI] [PMC free article] [PubMed] [Google Scholar]
Troyanovsky RB, Klingelhöfer J, Troyanovsky S. Removal of calcium ions triggers a novel type of intercadherin interaction. Journal of Cell Science. 1999;112:4379–4387. doi: 10.1242/jcs.112.23.4379. [DOI] [PubMed] [Google Scholar]
Tsang SM, Liu L, Teh MT, Wheeler A, Grose R, Hart IR, Garrod DR, Fortune F, Wan H. Desmoglein 3, via an interaction with E-cadherin, is associated with activation of Src. PLoS One. 2010;5:e14211. doi: 10.1371/journal.pone.0014211. [DOI] [PMC free article] [PubMed] [Google Scholar]
Vasioukhin V, Bauer C, Yin M, Fuchs E. Directed actin polymerization is the driving force for epithelial cell-cell adhesion. Cell. 2000;100:209–219. doi: 10.1016/S0092-8674(00)81559-7. [DOI] [PubMed] [Google Scholar]
Vendome J, Posy S, Jin X, Bahna F, Ahlsen G, Shapiro L, Honig B. Molecular design principles underlying β-strand swapping in the adhesive dimerization of cadherins. Nature Structural & Molecular Biology. 2011;18:693–700. doi: 10.1038/nsmb.2051. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wheelock MJ, Jensen PJ. Regulation of keratinocyte intercellular junction organization and epidermal morphogenesis by E-cadherin. The Journal of Cell Biology. 1992;117:415–425. doi: 10.1083/jcb.117.2.415. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wu Y, Jin X, Harrison O, Shapiro L, Honig BH, Ben-Shaul A. Cooperativity between trans and cis interactions in cadherin-mediated junction formation. PNAS. 2010;107:17592–17597. doi: 10.1073/pnas.1011247107. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wu Y, Vendome J, Shapiro L, Ben-Shaul A, Honig B. Transforming binding affinities from three dimensions to two with application to cadherin clustering. Nature. 2011;475:510–513. doi: 10.1038/nature10183. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yen CF, Harischandra DS, Kanthasamy A, Sivasankar S. Copper-induced structural conversion templates prion protein oligomerization and neurotoxicity. Science Advances. 2016;2:e1600014. doi: 10.1126/sciadv.1600014. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yen CF, Sivasankar S. Improving estimation of kinetic parameters in dynamic force spectroscopy using cluster analysis. The Journal of Chemical Physics. 2018;148:123301. doi: 10.1063/1.5001325. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang Y, Sivasankar S, Nelson WJ, Chu S. Resolving cadherin interactions and binding cooperativity at the single-molecule level. PNAS. 2009;106:109–114. doi: 10.1073/pnas.0811350106. [DOI] [PMC free article] [PubMed] [Google Scholar]

eLife. doi: 10.7554/eLife.37629.012

Decision letter

Editor: William I Weis¹

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "E-cadherin binds to desmoglein and facilitates desmosome assembly" for consideration by eLife. Your article has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

This paper addresses the mechanism of cadherin-mediated regulation of desmosome assembly. While the reviewers agree that the work is potentially exciting, they raised several significant issues. First, all agree that the model must be tested by examining co-localization in cells expressing the cadherin mutant. There are also concerns about the reliability of the co-localization data in Figure 5; a more rigorous analysis is needed to demonstrate that the co-localization of E-cadherin and desmoplakin occurs much more than would be expected by chance. Second, the modeling assumes that L175 forms part of the interface, but without assessment of stoichiometry it cannot be ruled out that the loss of binding is an indirect effect of loss of cis dimerization. The modeling is also viewed as weak since no complementary interfaces on Dsg2 were tested. Without such data the molecular modeling needs to be considered highly speculative. Finally, as noted by reviewer 3, even if supported by these additions, the model and Discussion section do not really explain how E-cadherin regulates desmosomes.

Given these concerns, the paper cannot be accepted in its current form. However, if these issues can be addressed, we would be happy to consider a resubmission.

Reviewer #1:

In this study Shafraz et al., examine the heterodimerization of E-cadherin and Desmoglein 2 (Dsg2) and its possible role in desmosome formation. They find that E-cadherin can bind to Dsg2 but not Desmocollin, that binding to Dsg2 is blocked by a mutation in E-cadherin that blocks E-cadherin cis homodimerization (L175D), but not blocked by an E-cadherin mutation that blocks E-cadherin dimerization in trans. These observations, plus computational docking results, support a model in which E-cadherin and Dsg2 interact via a binding interface that is somewhat similar to that used by protocadherins. Fixed-cell SIM immunocytochemistry images of primary keratinocytes are interpreted to indicate that E-cadherin is present in early desmosomes but reduced in mature desmosomes.

Summary recommendation.

The single-molecule biophysical measurements in this paper are for the most part well done, but other data are weak relative to the model that the authors propose. For this reason, the manuscript may not be appropriate for eLife in its current form.

Essential revisions:

1) The single-molecule AFM experiments are the most solid part of the paper. However, in my reading the authors do not exclude the possibility that the binding events they observe may be due to interactions between pre-formed E-cadherin cis-dimers. The stoichiometry of the interaction is important to establish, since if binding to Dsg2 involved required a E-cadherin cis-dimer, it would change the interpretation of the result that the L175D mutation blocks the Dsg2/E-cad interaction. Such a scenario would also challenge the validity of the docking simulations. In my reading, the low probability of binding per contact of the AFM tip with the surface is weak evidence that binding necessarily reflects the interaction of two monomers, nor is the low average density of cadherin molecules on the surface. Further, the cadherins are immobilized with streptavidin, which has multiple biotin binding sites.

For these reasons, I think that proving that heterodimerization occurs between monomers is important. To do so, I would recommend systematically measuring the interaction probability of the tip with the surface as a function of the areal surface density of E-cadherin on the coverslip surface. If the interaction indeed requires monomeric E-cadherin this measurement should scale linearly with added E-cadherin, at least at low densities. However, this specific measurement is not required-any measurement that establishes the stoichiometry will do.

2) The docking simulations are interesting, but in my view offer weak evidence for the specific binding geometry that the authors propose. I would strongly recommend that the authors be more cautious in the interpretation and presentation of these results, or alternatively present additional data, for example EM images, that support or refute the computational result.

3) Unfortunately, the SIM images are not convincing as currently presented. In particular, it is not clear what counts as a "desmosome" in the authors' analysis (Figure 5B). A more sophisticated treatment of colocalization would be required to establish the point that E-cadherin and desmoplakin colocalize, and that this colocalization decreases as junctions mature.

4) I leave this final point at the discretion of the editor. In my view, the colocalization data, even if taken as true, do not establish functional relevance. It would be very nice to see a knockdown/reconstitution experiment with L175D E-cadherin. The prediction is that this molecule would never colocalize with desmoplakin, regardless of the junction age. Violation of this prediction would cast doubt on the authors' favored model.

As a general principle I don't like to add too many additional experiments during the review process. If the editor decides that this or conceptually similar experiments are not required for publication, I would urge the authors to be much more cautious about their interpretive claims.

Reviewer #2:

This is an interesting paper that presents AFM and other data on the interaction of E-cadherin with the desmosomal cadherins Dsg2 and Dsc2, as well as homophilic and heterophilic interaction data for Dsg2 and Dsc2. The authors propose possible mechanisms with relevance to the assembly of desmosomes and suggest models for E-cadherin/Dsg2 interaction, where it is proposed that a novel trans (cell-to-cell) interaction between these two proteins may initiate junction formation. Despite high interest in this question, there are issues that diminish my enthusiasm for the work as currently presented, which should be addressed by the authors.

1) The finding that E-cadherin residue L175 is critical for interaction with Dsg2 is intriguing. However, no interacting partner residues are identified. In this circumstance, predicting a conformation for E-cadherin/Dsg2 interaction by docking seems precarious; it seems uncertain that the conformation predicted relates to the true bound conformation.

2) There is only a tenuous connection the proposed Dsg2/E-cadherin interaction to the biology of desmosome assembly. Yet the authors seem to have a clear way forward. They argue that the presence of E-cadherin in nascent desmosomes (assessed experimentally by co-staining for desmoplakin in Figure 5) is evidence for the role of the Dsg2/E-cadherin association they have identified. Now that they have identified a mutant that prevents that association, the experiments of Figure 5 performed with the mutant should lead to different results. This experiment would substantially add to the rigor of this paper.

Reviewer #3:

This manuscript identifies and characterizes a molecular interaction between E-cadherin and Desmoglein 2 using isolated proteins and shows increased colocalization between these two molecules in keratinocytes early during desmosome assembly. These are potentially very interesting observations that may explain how classical cadherins control the assembly of desmosomes. Although this is well documented by several groups in the literature, the underlying mechanisms are still mostly unknown. However, their conclusion "Our integrated single molecule experiments, protein-protein docking predictions and cell based super resolution imaging show that Ecad/Dsg2 interactions play an important role in early desmosome assembly" stated at the start of the Discussion section is a rather strong overstatement given that the interaction is based on characterization on isolated extracellular domains and static albeit high resolution microscopy. No experiment is done to show that when one perturbs this interaction desmosome assembly is indeed impaired, which in my opinion would be the minimal to make this set of data really interesting and relevant for a broad audience, also in light of the fact that the interaction is calcium independent whereas classical cadherin dependent early desmosome assembly does require the presence of calcium (which is not discussed at all).

As this is a fully new interaction of low affinity, it is probably hard to do any interaction assays in the context of cells in which one cell expresses E-cadherin and the other Dsg2 to examine whether there is any interaction. However, as their data are partially based on modeling, the authors should also generate a mutant in which the A125 and A124 are mutated to provide much better evidence for the existence of the proposed of y-dimers.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for resubmitting your work entitled "E-cadherin binds to desmoglein to facilitate desmosome assembly" for further consideration at eLife. Your revised article has been favorably reviewed by Anna Akhmanova (Senior Editor), a Reviewing Editor, and three reviewers.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

The reviewers feel that while the biophysical data are convincing, more rigorous quantitative analysis is needed to support conclusions made from the data in Figures 3 and 5. In principle these concerns can be addressed with additional data analysis rather than experiments, and in this case, we would consider a revised manuscript.

1) The conclusion that E-cadherin-Dsg interaction is needed for initial desmosome formation is based on colocalization of Dsg and E-cadherin decreasing as the junction matures. However, in Figure 3 it is not clear whether the observed colocalization with DP is significant beyond that expected by chance or arises from time-dependent changes in E-cadherin sub cellular localization unrelated to junction maturation. In Figure 3C, control experiments shown in the right panel highlight co-localization of Dsg2 and DP – known Desmosome components – and show a close correspondence in fluorescence signals at all three time points. In contrast, while the expression of E-cadherin and DP in the left panels does appear to overlap to some degree, they do not show the correspondence in signals observed for the control experiments. The response to the original review points on this subject (e.g. reviewer 1, point 3) focuses on defining a desmosome, but does not address the above point.

2) Similar concerns were raised regarding Figure 5, which is supposed to show that Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly. The images in panel A indeed show a dramatic difference in Dsg2 and DP localization relative to the WT or mutant cadherin. While Dsg2 and DP co-localize with WT or K14E Ecad at junctions, they remain in separate puncta flanking the L175D Ecad junction and do not co-localize with it. However, it is not clear what the authors quantified to test the significance of this phenotype. In B, what is "% of junction forming contacts"? How is a contact defined? What junctions are they measuring? AJ or desmosomes? In C and E, what do they mean by% of Dsg2 or DP positive junctions? How were junctions defined?

There are also modifications to the Introduction and Discussion section needed to address the points above as well as the accessibility of the paper:

1) Related to be points above, the link between in vitro binding experiments and the proposed biological role seems tenuous based on localization studies that reveal overlapped but not truly co-localized staining patterns. The authors need to explain this finding; perhaps co-localization is dynamic, and/or only a subset of E-cadherin molecules is bound to Dsg2. This point at least needs to be discussed to rationalize the results of the localization experiments with the proposed model.

2. In the beginning of the Discussion section the authors claim that they: "identify two critical events that initiate and promote efficient desmosome assembly: (i) stable trans homodimerization of Ecad, and (ii) the direct heterophilic binding of Ecad and Dsg2 ectodomains. Our data demonstrates that desmosome assembly is initiated at sites of Ecad trans homodimerization. Subsequently, Ecad and Dsg2 bind via a conserved Leu 175 on the Ecad cis binding interface and form short-lived heterophilic complexes that localize to early desmosomes and efficiently recruit desmosomal proteins to sites of intercellular contact formation. As desmosomes mature, Dsg2 dissociates from Ecad and forms stable bonds with Dsc2 to mediate robust adhesion." – This paragraph mixes observations, interpretation, and hypothetical models and as such is misleading. They should separate the summary of results from interpretation and their model.

3. The authors write in the Abstract and Introduction that "Ecad interacts with Dsg2 via a conserved Leu 175 on the Ecad cis binding interface." – The authors shouldn't assume that readers are familiar with E-cad homophilic interactions and should explain this explicitly.

Reviewer #1:

The core advances of this paper – the demonstration that E-cadherin and Dsg2 bind to each other in AFM experiments, that the interaction is Ca²⁺-independent, and the binding is regulated by the cis-interface region of E-cadherin, specifically by residue L175 – are interesting and novel. The model for desmosome assembly based on this finding is interesting, with support from the new SIM experiments (although other conceivable models could also be consistent with these data).

While the new experiments presented in this revised manuscript are squarely aimed at addressing criticisms of biological relevance for the original manuscript, there are issues that remain and should be addressed – at least in the Discussion section. Co-localization data using structured illumination microscopy has been employed to demonstrate E-cadherin and DP co-localization on the cell surface and its changes as desmosomes mature. However, the co-localization is not highly convincing. Specifically, in Figure 3C, control experiments shown in the right panel highlight co-localization of Dsg2 and DP – known Desmosome components – and show a close correspondence in fluorescence signals at all three time points. In contrast, while the expression of E-cadherin and DP in the left panels does appear to overlap to some degree, they do not show the correspondence in signals observed for the control experiments (compare traces in Figure 3C). Changes in E-cadherin/DP localization upon desmosome maturation described in the text are also not clearly evident in the Figure 3.

In Figure 5, the data show a dependence of E-cadherin and Dsg2 on DP localization in transfected keratinocytes. X-dimer and cis-interface mutant E-cadherins were tested in this assay and the cis-mutant appears to interfere with co-localization of E-cadherin and Dsg2 or DP. Again, though, co-localization of wild-type E-cadherin with Dsg2 does not seem convincing (Figure 5A top panels). The change caused by the L175D mutation is striking (Figure 5A bottom panel), but does not necessarily implicate binding to Dsg2, as this mutation is expected to disrupt E-cadherin cis interactions, which could also have phenotypic effects.

Figure 2 could easily be supplementary, since it is hard for a non-specialist to glean any information from the plots. The numbers quoted in the text are sufficient. In addition, the authors mention that the k_off rate is 10 fold weaker for the Dsc2/Dsg2 dimer, yet the values quoted in the text only suggest a 6-fold difference. This should be clarified.

Overall, this paper presents an interesting biophysical study of E-cadherin and Dsg2 binding, the link between in vitro binding experiments and the proposed biological role seems tenuous based on localization studies that reveal overlapped but not truly co-localized staining patterns. The authors need to explain this finding; perhaps co-localization is dynamic, and/or only a subset of E-cadherin molecules is bound to Dsg2. This point at least needs to be discussed to rationalize the results of the localization experiments with the proposed model.

Reviewer #2:

The authors have addressed many of my prior concerns. My remaining concern is that they do not show that the localization of E-cadherin to DP puncta is more (or less) than what would be expected by chance. The trend with time that they report is not sufficient, as in principle the local abundance of E-cadherin could be changing in time.

Reviewer #3:

In their paper, "E-cadherin binds to desmoglein to facilitate desmosome assembly", Shafraz et al., use single molecule AFM to show that E-cadherin forms a direct, calcium independent, interaction with Dsg2 that has a lifetime similar to Dsc2/Dsc2 dimers but is ten times weaker than Dsg2/Dsc2 dimers. On the E-cadherin side, this interaction depended on Leu175, which also mediates Ecad cis interactions, but was independent of the residues that mediate trans Ecad interactions. The authors used SIM imaging of human keratinocytes, following a switch from low to high calcium, to show that Ecad levels in Desmoplakin positive structures are higher in early desmosome than in mature desmosomes. Finally, using a model of EcadKO/PcadKD mouse keratinocytes, in which they reintroduce WT or mutant Ecad, the authors show using conventional microscopy that Ecad(L175D) is highly ineffective in recruiting Dsg2 at early time points and in desmosome formation, and this defect is not explained by its compromised ability to form AJs.

Overall, this work is novel and interesting and the combination of biophysical and cell biological approaches is very powerful and provides molecular mechanistic insight into age-old observations of interaction between classical and desmosomal cadherins. After addressing the points outlined below, I think this work will be most appropriate for publication in eLife.

1) The most burning question I was left with after reading the paper, and which I think the authors can and should answer before publication is whether the interaction between Ecad and Dsg2 is a cis or trans interaction. While reading the paper, I had imagined a trans interaction and it was only when I reached Figure 6 and saw their model that I realized the authors assume it to be a cis interaction, though they say in the discussion their data is equally consistent with a trans interaction. I feel this is an important point that can't be left open. Two color STORM imaging or biochemical cross-linking experiments could be employed to tease apart these two possibilities.

2) Figure 3 – If E-cad is being replaced by Dsc2, then Dsc2 level should go up over time. Can the authors show this?

3) Given the focus of this paper is on the interaction between Ecad and Dsg2, it seems to me to be important to image both endogenous E-cad and Dsg2 with SIM and show their transient co-localization in desmosomes.

4) Figure 5 is supposed to show that Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly. The images in panel A indeed show a dramatic difference in Dsg2 and DP localization relative to the WT or mutant cadherin. While Dsg2 and DP co-localize with WT or K14E Ecad at junctions, they remain in separate puncta flanking the L175D Ecad junction and do not co-localize with it. However, it is not clear what the authors quantified to test the significance of this phenotype. In B, what is "% of junction forming contacts"? How is a contact defined? What junctions are they measuring? AJ or desmosomes? In C and E, what do they mean by% of Dsg2 or DP positive junctions? How were junctions defined?

Neither of these quantifications seems to take advantage of the rich 2D (or 3D) information in the images to quantify co-localization.

5) In the beginning of the Discussion section the authors claim that they: "identify two critical events that initiate and promote efficient desmosome assembly: (i) stable trans homodimerization of Ecad, and (ii) the direct heterophilic binding of Ecad and Dsg2 ectodomains. Our data demonstrates that desmosome assembly is initiated at sites of Ecad trans homodimerization. Subsequently, Ecad and Dsg2 bind via a conserved Leu 175 on the Ecad cis binding interface and form short-lived heterophilic complexes that localize to early desmosomes and efficiently recruit desmosomal proteins to sites of intercellular contact formation. As desmosomes mature, Dsg2 dissociates from Ecad and forms stable bonds with Dsc2 to mediate robust adhesion." – This paragraph mixes observations, interpretation, and hypothetical models and as such is misleading. They should separate the summary of results from interpretation and their model.

eLife. 2018 Jul 12;7:e37629. doi: 10.7554/eLife.37629.013

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

1) This paper addresses the mechanism of cadherin-mediated regulation of desmosome assembly. While the reviewers agree that the work is potentially exciting, they raised several significant issues. First, all agree that the model must be tested by examining co-localization in cells expressing the cadherin mutant.

We thank the reviewers for suggesting these important experiments. As described in our detailed response to the reviewers, we have now tested the role of the different Ecad mutations in impeding recruitment of Dsg2 in keratinocytes. We expressed full-length Ecad WT, and Ecad mutants (L175D, K14E, and W2A-K14E) in Ecad-knockout, Pcad-knockdown mouse keratinocytes (E^KO/P^KD) and analyzed Dsg2 recruitment to sites of intercellular contacts. These results confirm that amino acid L175 on Ecad, mediates Dsg2 interactions and facilitates early desmosome complex formation in cells. Our experiments also reveal that desmosome assembly is initiated at sites of Ecad trans homodimerization.

2) There are also concerns about the reliability of the co-localization data in Figure 5; a more rigorous analysis is needed to demonstrate that the co-localization of E-cadherin and desmoplakin occurs much more than would be expected by chance.

In the field, desmosomes are identified by the size and rail road track appearance of DP staining in SIM images. As outlined in our response to the reviewers, we used this established approach to locate desmosomes. In addition, we now stress that we are not measuring colocalization but rather the recruitment of different cadherins (Ecad and Dsg) to regions of the membrane that display this pattern.

3) Second, the modeling assumes that L175 forms part of the interface, but without assessment of stoichiometry it cannot be ruled out that the loss of binding is an indirect effect of loss of cis dimerization. The modeling is also viewed as weak since no complementary interfaces on Dsg2 were tested. Without such data the molecular modeling needs to be considered highly speculative.

As described in our detailed response to the reviewers, previous computational simulations and biophysical experiments show that Ecad cis-dimer formation requires prior trans dimerization. Consequently, Ecads cannot bind to the surface as stand-alone, pre-formed, cis-dimers. This makes it exceedingly unlikely that the measured binding interactions occur between Ecad cis-dimers and Dsg2. Nonetheless, we agree with the reviewers that the molecular docking results are quite weak. We have therefore eliminated these modeling results from the manuscript.

4) Finally, as noted by reviewer 3, even if supported by these additions, the model and Discussion section do not really explain how E-cadherin regulates desmosomes.

Based on our new cellular structure-function data, we now propose a model whereby desmosome assembly is facilitated both by the direct cis interactions of Ecad and Dsg2 and the trans binding of opposing Ecad. In our model, the trans homodimerization of Ecad from opposing cells serves as a spatial cue to coordinate desmosome assembly. Ecad subsequently forms cis-dimer complexes with Dsg2 to initiate desmosome assembly. The model is presented and discussed in the new manuscript.

Reviewer #1:

1) The single-molecule AFM experiments are the most solid part of the paper. However, in my reading the authors do not exclude the possibility that the binding events they observe may be due to interactions between pre-formed E-cadherin cis-dimers. The stoichiometry of the interaction is important to establish, since if binding to Dsg2 involved required a E-cadherin cis-dimer, it would change the interpretation of the result that the L175D mutation blocks the Dsg2/E-cad interaction. Such a scenario would also challenge the validity of the docking simulations. In my reading, the low probability of binding per contact of the AFM tip with the surface is weak evidence that binding necessarily reflects the interaction of two monomers, nor is the low average density of cadherin molecules on the surface. Further, the cadherins are immobilized with streptavidin, which has multiple biotin binding sites.

For these reasons, I think that proving that heterodimerization occurs between monomers is important. To do so, I would recommend systematically measuring the interaction probability of the tip with the surface as a function of the areal surface density of E-cadherin on the coverslip surface. If the interaction indeed requires monomeric E-cadherin this measurement should scale linearly with added E-cadherin, at least at low densities. However, this specific measurement is not required-any measurement that establishes the stoichiometry will do.

Previous computer simulations (Wu et al., 2010; Wu et al., 2011) have shown that Ecad cis homodimerization requires prior Ecad trans-dimer formation. Similarly, previous single molecule FRET measurements have shown that it is not possible to form stand-alone Ecad cis-dimers, even when Ecad monomers are placed in close proximity in a cis orientation (Zhang et al., 2009). Taken together, these results suggest that in our experiments, Ecads cannot be immobilized on the AFM tip or substrate as pre-formed cis-dimers. Consequently, the failure of the Ecad-L175D mutant to interact with Dsg2 is not due to the absence of Ecad cis dimerization. We now discuss this in subsection “Leu 175 mediates Ecad and Dsg2 interactions” of the manuscript.

Unfortunately, technical limitations do not allow the protein stoichiometry experiment that the reviewer suggests. In our experiments, we use protein surface densities that allow us to unambiguously identify single binding events from the corresponding stretching of PEG tethers. If the cadherin surface density is increased, as proposed by the reviewer, the probability of simultaneous multiple Ecad binding events increases which makes a quantitative analysis of protein binding probability unreliable. For this reason (along with the intrinsic weakness of our docking analysis), we have excluded the protein docking simulations from the manuscript.

Finally, in the revised manuscript, we now use cluster analysis to group the single molecule unbinding events for Dynamic Force Spectroscopy (DFS). We have recently shown that the K-means clustering algorithm that we employed, greatly improves the estimation of kinetic parameters in DFS (Yen and Sivasankar, 2018). Consequently, the lifetimes we now report for Ecad/Dsg2, Dsc2/Dsg2, and Dsc2/Dsc2 interactions are more reliable.

2) The docking simulations are interesting, but in my view offer weak evidence for the specific binding geometry that the authors propose. I would strongly recommend that the authors be more cautious in the interpretation and presentation of these results, or alternatively present additional data, for example EM images, that support or refute the computational result.

We agree with the reviewer that the protein docking data is weak. We have therefore eliminated these results from the manuscript.

3) Unfortunately, the SIM images are not convincing as currently presented. In particular, it is not clear what counts as a "desmosome" in the authors' analysis (Figure 5B). A more sophisticated treatment of colocalization would be required to establish the point that E-cadherin and desmoplakin colocalize, and that this colocalization decreases as junctions mature.

In previous studies (Stahley et al., 2016a, 2016b) we have utilized the “rail road track” appearance of DP as revealed by super-resolution imaging to define desmosomes by immunofluorescence (as depicted in Figure 3A). Others in the field have also used ‘railroad’ track morphology to identify desmosomes in SIM images (see for instance: Chen et al., (2012); Ungewiß et al., (2017)). The size and organization of these structures, as revealed by SIM, is fully consistent with classical EM definitions of desmosomes. We measured pixel intensity for the two cadherins (Ecad or Dsg2) within these structures using standard image analysis. Importantly, we are not attempting to claim co-localization. Rather, we are measuring recruitment of cadherin to membrane domains that display DP rail road track staining patterns. We do appreciate that it is sometimes difficult to discern rail-road tracks at early time points. For this reason, the earliest time we felt that we could identify bona fine rail road track staining was at 1 hour. Furthermore, we are only making measurements at locations where we can observe rail road track patterns, rather than DP puncta. Lastly, our finding that Ecad is present in nascent desmosomes is consistent with classic immuno-EM experiments showing that Ecad can localize to desmosomes (Jones, 1988). Therefore, we feel confident in our ability to identify desmosomes and to measure relative levels of the two cadherins at different times.

4) I leave this final point at the discretion of the editor. In my view, the colocalization data, even if taken as true, do not establish functional relevance. It would be very nice to see a knockdown/reconstitution experiment with L175D E-cadherin. The prediction is that this molecule would never colocalize with desmoplakin, regardless of the junction age. Violation of this prediction would cast doubt on the authors' favored model.

Based on the reviewer’s suggestion, we have now performed experiments to directly test the role of the different Ecad extracellular domain mutations in impeding recruitment of DP and Dsg2 in keratinocytes. This data is described in subsection “Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly”. The corresponding images are shown in Figure 5 and in Figure 5—figure supplement 1. The methods used are described in subsection “Isolation, culture, transfection and confocal imaging of primary keratinocytes”. For these experiments, we used E^KO/P^KD mouse keratinocytes, which express virtually no classical cadherins. We have previously shown that these E^KO/P^KD keratinocytes are unable to assemble AJs and desmosomes due to the loss of all classical cadherins (Michels et al., 2009). These cells thus allow us to directly assess the ability of the Ecad mutants to i) initiate AJ formation and ii) assess their ability to recruit desmosomal components to sites of initial cell-cell contact formation.

Since we were interested in assessing the ability of Ecad mutants to recruit desmosomal proteins early during junction formation, we expressed either full-length Ecad WT or mutants in E^KO/P^KD keratinocytes and analyzed DP and Dsg2 recruitment to sites of intercellular contacts using confocal microscopy. To examine recruitment at different stages of junction formation and maturation, keratinocytes were fixed and immunostained for DP, Dsg and Ecad at three time points following the Ca²⁺ switch (3 hours, 6 hours and 18 hours).

Our data showed that 3 hours after the Ca²⁺ switch, 93% of WT-Ecad was enriched in zipper-like early AJs at sites of intercellular contacts. In contrast only 48% of L175D-Ecad transfected keratinocytes formed AJ zippers, likely due to impaired Ecad cis-dimer formation (Figure 5 A, B). At these early time points following the Ca²⁺ switch, 66% and 91% of the WT-Ecad zipper contacts were positive for Dsg2 and DP respectively (Figure 5 A, C, D, E). When the cells were allowed to engage in Ca²⁺-dependent intercellular adhesion for 18 hour, the junctional localizations of Ecad-Dsg2 and Ecad-DP increased to 97% and 100% respectively (Figure 5 A, C, D, E). In contrast, only 39% of L175-induced zipper contacts showed Dsg2 recruitment 3 hours after switching to high Ca²⁺ which increased to 65% at 18 hours (Figure 5 A, C), confirming that a direct interaction between Ecad and Dsg2 is required for efficient recruitment of Dsg2 to early intercellular contacts. Similarly, after 3 hours, only 28% of L175D-Ecad mutant established intercellular contacts were positive for DP, which was increased to 72% after 18 hours, suggesting a compensatory mechanism at later stages (Figure 5 D, E).

To confirm that the observed effect of the L175D mutation was not due to hindered AJ formation, we transfected the E^KO/P^KD keratinocytes with the full-length Ecad-K14E mutant that abolishes X-dimer formation and traps Ecad in a strand-swap dimer conformation. Our data showed that only 4% of the transfected contacting cells formed zippers at early (3 hours) timepoints after switching to high Ca²⁺ with only 24% of the transfected contacting cells showing zippers at late (18 hours) time points (Figure 5 A, B, Figure 5—figure supplement 1), confirming that K14 is essential for effective AJ formation, and thus intercellular contact establishment. However, the few K14E AJs that were formed, recruited Dsg2, and especially DP, more efficiently than L175D (Figure 5 A, C, E, Figure 5—figure supplement 1). This result confirmed that delayed recruitment of Dsg2 to inter-cellular contacts was not the result of the inability of the Ecad L175D to efficiently form AJs but rather due to absence of direct interaction between Ecad and Dsg2. As there was never any co-localization of DP and Ecad in the absence of zippers (Figure 5—figure supplement 1), the data also suggest that Ecad trans interactions precede Ecad/Dsg2 interactions. This conclusion is further strengthened by our finding that no DP recruitment was observed upon abolishing Ecad trans adhesion, and thus zippers, by transfecting the full-length Ecad-DM (W2A-K14E double mutant, Figure 5—figure supplement 1) in E^KO/P^KD keratinocytes. Taken together these results confirm that amino acid L175 mediates Dsg2 interactions and facilitates early desmosome complex formation in cells.

Reviewer #2:

1) The finding that E-cadherin residue L175 is critical for interaction with Dsg2 is intriguing. However, no interacting partner residues are identified. In this circumstance, predicting a conformation for E-cadherin/Dsg2 interaction by docking seems precarious; it seems uncertain that the conformation predicted relates to the true bound conformation.

We agree with the reviewer that the protein docking data is speculative and have therefore eliminated these results from the manuscript.

2) There is only a tenuous connection the proposed Dsg2/E-cadherin interaction to the biology of desmosome assembly. Yet the authors seem to have a clear way forward. They argue that the presence of E-cadherin in nascent desmosomes (assessed experimentally by co-staining for desmoplakin in Figure 5) is evidence for the role of the Dsg2/E-cadherin association they have identified. Now that they have identified a mutant that prevents that association, the experiments of Figure 5 performed with the mutant should lead to different results. This experiment would substantially add to the rigor of this paper.

As described in response 4 to reviewer 1, we expressed full-length Ecad WT, and Ecad mutants in E^KO/P^KD keratinocytes and analyzed Dsg2 and DP recruitment to sites of intercellular contacts. These results confirm that amino acid L175 mediates Dsg2 interactions and facilitates early desmosome complex formation in cells. This data is described in subsection “Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly” of the manuscript. The corresponding images are shown in Figure 5 and in Figure 5—figure supplement 1. The methods used are described in subsection “Isolation, culture, transfection and confocal imaging of primary keratinocytes”.

Reviewer #3:

1) This manuscript identifies and characterizes a molecular interaction between E-cadherin and Desmoglein 2 using isolated proteins and shows increased colocalization between these two molecules in keratinocytes early during desmosome assembly. These are potentially very interesting observations that may explain how classical cadherins control the assembly of desmosomes. Although this is well documented by several groups in the literature, the underlying mechanisms are still mostly unknown. However, their conclusion "Our integrated single molecule experiments, protein-protein docking predictions and cell based super resolution imaging show that Ecad/Dsg2 interactions play an important role in early desmosome assembly" stated at the start of the Discussion section is a rather strong overstatement given that the interaction is based on characterization on isolated extracellular domains and static albeit high resolution microscopy. No experiment is done to show that when one perturbs this interaction desmosome assembly is indeed impaired, which in my opinion would be the minimal to make this set of data really interesting and relevant for a broad audience.

As described in response 4 to reviewer 1, we have now expressed full-length Ecad WT, and Ecad mutants in E^KO/P^KD keratinocytes and shown that amino acid L175 mediates Dsg2 interactions and facilitates early desmosome formation in cells as assessed by DP recruitment. This data is described in subsection “Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly” of the manuscript. The corresponding images are shown in Figure 5 and in Figure 5—figure supplement 1. The methods used are described in subsection “Isolation, culture, transfection and confocal imaging of primary keratinocytes”.

2) Also, in light of the fact that the interaction is calcium independent whereas classical cadherin dependent early desmosome assembly does require the presence of calcium (which is not discussed at all).

We thank the reviewer for this insightful comment. As described in subsection “Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly”, confocal imaging of mutant Ecad expressed in mouse keratinocytes now reveals that desmosome assembly is initiated at sites of Ca²⁺ dependent Ecad trans homodimerization. Based on our data, we propose a model whereby desmosome assembly is facilitated both by the direct cis interactions of Ecad and Dsg2 and the trans binding of opposing Ecad. In our model, the trans homodimerization of Ecad from opposing cells serves as a spatial cue to coordinate desmosome assembly. Ecad subsequently forms cis dimer complexes with Dsg2. Once localized within the native desmosome, Dsg2 dissociates from Ecad and binds to Dsc2 to form mature desmosomes. Since the Dsc2/Dsg2 complex has a longer lifetime than both the Ecad/Dsg2 complex and Dsc2/Dsc2 complex, this likely allows for robust cell-cell adhesion and permits the mature desmosome to withstand mechanical force. This model is described in Figure 6, and in the Discussion section.

3) As this is a fully new interaction of low affinity, it is probably hard to do any interaction assays in the context of cells in which one cell expresses E-cadherin and the other Dsg2 to examine whether there is any interaction. However, as their data are partially based on modeling, the authors should also generate a mutant in which the A125 and A124 are mutated to provide much better evidence for the existence of the proposed of y-dimers.

Since the protein docking results are not robust, we have now eliminated these data from the manuscript. However, our experiments with E^KO/P^KD keratinocytes showing that mutation of amino acid L175 on Ecad impairs recruitment of Dgs2 and especially DP to sites of early contact formation, demonstrates that Ecad-L175 facilitates desmosome formation in cells, confirms our biophysical results.

[Editors' note: the author responses to the re-review follow.]

Summary:

The reviewers feel that while the biophysical data are convincing, more rigorous quantitative analysis is needed to support conclusions made from the data in Figures 3 and 5. In principle these concerns can be addressed with additional data analysis rather than experiments, and in this case, we would consider a revised manuscript.

We thank the editor for this encouraging review. As described below, our revised manuscript addresses all the highlighted issues.

1) The conclusion that E-cadherin-Dsg interaction is needed for initial desmosome formation is based on colocalization of Dsg and E-cadherin decreasing as the junction matures. However, in Figure 3 it is not clear whether the observed colocalization with DP is significant beyond that expected by chance or arises from time-dependent changes in E-cadherin sub cellular localization unrelated to junction maturation. In Figure 3C, control experiments shown in the right panel highlight co-localization of Dsg2 and DP – known Desmosome components – and show a close correspondence in fluorescence signals at all three time points. In contrast, while the expression of E-cadherin and DP in the left panels does appear to overlap to some degree, they do not show the correspondence in signals observed for the control experiments. The response to the original review points on this subject (e.g. reviewer 1, point 3) focuses on defining a desmosome, but does not address the above point.

We would like to clarify that we are not measuring co-localization but instead the recruitment of either Ecad or Dsg2 to regions of the membrane that display a DP ‘railroad track’ pattern. To make this clearer, we have now modified the revised manuscript to emphasize that Ecad is ‘enriched in nascent desmosomes’ rather than being ‘excluded from mature desmosomes’.

To directly address the reviewers concerns regarding our measurement of Ecad enrichment in desmosomes, we now show that the relative levels of Ecad along the entire cell border (Ecad:DP ratio at cell borders) remain unchanged over time. This data confirms that the Ecad enrichment within DP railroad tracks at early time points is specific to desmosomal regions of the membrane and is significant beyond that expected by random chance or because of changes in E-cadherin localization unrelated to junction maturation. The data is shown in Figure 3—figure supplement 1 and is described in subsection “Ecad is present in nascent desmosomes but not in mature desmosomes”.

Finally, since the line-scans shown in Figure 3 were confusing, we have eliminated the line-scans from the revised manuscript. Instead, in Figure 3B, we now show several representative images of desmosomal regions in human keratinocytes cultured in high Ca²⁺ media for 1, 3 or 18 hours. These results emphasize the main ‘take-away’ message of Figure 3C: Ecad levels are enriched in nascent desmosomes, with relative levels decreasing as desmosomes mature.

2) Similar concerns were raised regarding Figure 5, which is supposed to show that Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly. The images in panel A indeed show a dramatic difference in Dsg2 and DP localization relative to the WT or mutant cadherin. While Dsg2 and DP co-localize with WT or K14E Ecad at junctions, they remain in separate puncta flanking the L175D Ecad junction and do not co-localize with it. However, it is not clear what the authors quantified to test the significance of this phenotype. In B, what is "% of junction forming contacts"? How is a contact defined? What junctions are they measuring? AJ or desmosomes? In C and E, what do they mean by% of Dsg2 or DP positive junctions? How were junctions defined?

We have now added a figure panel (Figure 5—figure supplement 1A) to illustrate the quantification criteria and junctional categories that were used in our analysis. This panel shows examples of cells transfected with Ecad mutant that do not (left panel) or do (right panel) show intercellular AJ formation. For Dsg2 or DP recruitment, we only quantified those intercellular interfaces in which AJ zippers were observed. In the insets of the figure panel, we show examples of AJ positive contacts that are either positive for DP or negative for DP.

In subsection “Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly” of the manuscript, we now state that the ability of transfected keratinocytes to form AJs was first assessed by the formation of ‘zipper-like’ patterns of Ecad at intercellular contacts. We have previously shown that these zippers represent early AJs and recruit AJ marker proteins like vinculin (Rübsam et al., 2017). We only examined intercellular interfaces with AJ zippers, for their ability to recruit Dsg2 and DP to these contacts. Importantly, we have never observed any enrichment of desmosomal components at intercellular contacts in the absence of AJ zippers (Michels et al., 2009).

As described in response 3a below, we now discuss in the Discussion section that although Dsg2 and DP do show some co-localization at Ecad-positive zippers within 3 hours, there is also non-overlapping junctional staining suggesting that AJs and desmosomes are already starting to segregate into distinct intercellular junctions. We did not quantify this co-localization in our manuscript, since the main point of our experiments were to show the relevance of L175 for recruiting Dsg2 to intercellular interfaces and facilitating desmosome formation (as illustrated by DP recruitment to intercellular contacts). We believe that the delay in recruitment between WT, L175 mutant, and the K14 mutant, that is quantified in Figure 5, illustrates this point.

3) There are also modifications to the Introduction and Discussion section needed to address the points above as well as the accessibility of the paper:

a) Related to be points above, the link between in vitro binding experiments and the proposed biological role seems tenuous based on localization studies that reveal overlapped but not truly co-localized staining patterns. The authors need to explain this finding; perhaps co-localization is dynamic, and/or only a subset of E-cadherin molecules is bound to Dsg2. This point at least needs to be discussed to rationalize the results of the localization experiments with the proposed model.

Due to the transient nature of Ecad/Dsg2 interactions, desmosomal proteins recruited to junctions are expected to rapidly segregate from Ecad. Consequently, images of intercellular interfaces are expected to have both overlapping and separate Ecad and desmosomal protein staining patterns. In agreement, although Dsg2 and DP show some co-localization at Ecad-positive zippers (Figure 5), significant non-overlapping junctional staining is observed, even within the first 3 hours. This suggests that AJs and desmosomes rapidly segregate into distinct intercellular junctions. We now state this in the Discussion section.

b) In the beginning of the Discussion section the authors claim that they: "identify two critical events that initiate and promote efficient desmosome assembly: (i) stable trans homodimerization of Ecad, and (ii) the direct heterophilic binding of Ecad and Dsg2 ectodomains. Our data demonstrates that desmosome assembly is initiated at sites of Ecad trans homodimerization. Subsequently, Ecad and Dsg2 bind via a conserved Leu 175 on the Ecad cis binding interface and form short-lived heterophilic complexes that localize to early desmosomes and efficiently recruit desmosomal proteins to sites of intercellular contact formation. As desmosomes mature, Dsg2 dissociates from Ecad and forms stable bonds with Dsc2 to mediate robust adhesion." – This paragraph mixes observations, interpretation, and hypothetical models and as such is misleading. They should separate the summary of results from interpretation and their model.

As suggested by the reviewers, we have eliminated the interpretations from the first paragraph of the Discussion section and now only summarize the results.

c) The authors write in the Abstract and Introduction "Ecad interacts with Dsg2 via a conserved Leu 175 on the Ecad cis binding interface." – The authors shouldn't assume that readers are familiar with E-cad homophilic interactions and should explain this explicitly.

In the Abstract, we now state that ‘Previous studies demonstrate that E-cadherin (Ecad), an adhesive protein that interacts in both trans and cis conformations, facilitates desmosome assembly via an unknown mechanism’.

Furthermore, in the Introduction, we now state: ‘Since Ecad interacts laterally to form cis-dimers on the same cell surface (Harrison et al., 2011) while Ecad molecules from opposing cells interact in a trans strand-swap dimer conformation (Boggon et al., 2002; Parisini et al., 2007; Vendome et al., 2011) and a trans X-dimer conformation (Ciatto et al., 2010; Harrison et al., 2010), we used mutants that specifically abolish either Ecad trans or cis interactions and tested their binding to either Dsg2 or Dsc2’. Finally, in the introduction, we now state that ‘Previous structural studies have shown that L175 mediates homophilic Ecad cis dimerization (Harrison et al., 2011)’.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Transparent reporting form

elife-37629-transrepform.docx^{(244.3KB, docx)}

DOI: 10.7554/eLife.37629.010

Data Availability Statement

All data generated or analyzed during this study are included in the manuscript and supporting files.

[bib1] Amagai M, Fujimori T, Masunaga T, Shimizu H, Nishikawa T, Shimizu N, Takeichi M, Hashimoto T. Delayed assembly of desmosomes in keratinocytes with disrupted classic-cadherin-mediated cell adhesion by a dominant negative mutant. Journal of Investigative Dermatology. 1995;104:27–32. doi: 10.1111/1523-1747.ep12613462. [DOI] [PubMed] [Google Scholar]

[bib2] Amagai M, Stanley JR. Desmoglein as a target in skin disease and beyond. Journal of Investigative Dermatology. 2012;132:776–784. doi: 10.1038/jid.2011.390. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib3] Awad MM, Dalal D, Cho E, Amat-Alarcon N, James C, Tichnell C, Tucker A, Russell SD, Bluemke DA, Dietz HC, Calkins H, Judge DP. DSG2 mutations contribute to arrhythmogenic right ventricular dysplasia/cardiomyopathy. The American Journal of Human Genetics. 2006;79:136–142. doi: 10.1086/504393. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Bell GI. Models for the specific adhesion of cells to cells. Science. 1978;200:618–627. doi: 10.1126/science.347575. [DOI] [PubMed] [Google Scholar]

[bib5] Boggon TJ, Murray J, Chappuis-Flament S, Wong E, Gumbiner BM, Shapiro L. C-cadherin ectodomain structure and implications for cell adhesion mechanisms. Science. 2002;296:1308–1313. doi: 10.1126/science.1071559. [DOI] [PubMed] [Google Scholar]

[bib6] Brasch J, Harrison OJ, Honig B, Shapiro L. Thinking outside the cell: how cadherins drive adhesion. Trends in Cell Biology. 2012;22:299–310. doi: 10.1016/j.tcb.2012.03.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Burdett ID, Sullivan KH. Desmosome assembly in MDCK cells: transport of precursors to the cell surface occurs by two phases of vesicular traffic and involves major changes in centrosome and Golgi location during a Ca(2+) shift. Experimental Cell Research. 2002;276:296–309. doi: 10.1006/excr.2002.5509. [DOI] [PubMed] [Google Scholar]

[bib8] Calkins CC, Setzer SV, Jennings JM, Summers S, Tsunoda K, Amagai M, Kowalczyk AP. Desmoglein endocytosis and desmosome disassembly are coordinated responses to pemphigus autoantibodies. Journal of Biological Chemistry. 2006;281:7623–7634. doi: 10.1074/jbc.M512447200. [DOI] [PubMed] [Google Scholar]

[bib9] Ciatto C, Bahna F, Zampieri N, VanSteenhouse HC, Katsamba PS, Ahlsen G, Harrison OJ, Brasch J, Jin X, Posy S, Vendome J, Ranscht B, Jessell TM, Honig B, Shapiro L. T-cadherin structures reveal a novel adhesive binding mechanism. Nature Structural & Molecular Biology. 2010;17:339–347. doi: 10.1038/nsmb.1781. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib10] Doyle JP, Stempak JG, Cowin P, Colman DR, D'Urso D. Protein zero, a nervous system adhesion molecule, triggers epithelial reversion in host carcinoma cells. The Journal of Cell Biology. 1995;131:465–482. doi: 10.1083/jcb.131.2.465. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Dusek RL, Godsel LM, Green KJ. Discriminating roles of desmosomal cadherins: beyond desmosomal adhesion. Journal of Dermatological Science. 2007;45:7–21. doi: 10.1016/j.jdermsci.2006.10.006. [DOI] [PubMed] [Google Scholar]

[bib12] Eshkind L, Tian Q, Schmidt A, Franke WW, Windoffer R, Leube RE. Loss of desmoglein 2 suggests essential functions for early embryonic development and proliferation of embryonal stem cells. European Journal of Cell Biology. 2002;81:592–598. doi: 10.1078/0171-9335-00278. [DOI] [PubMed] [Google Scholar]

[bib13] Evans E, Ritchie K. Dynamic strength of molecular adhesion bonds. Biophysical Journal. 1997;72:1541–1555. doi: 10.1016/S0006-3495(97)78802-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib14] Getsios S, Huen AC, Green KJ. Working out the strength and flexibility of desmosomes. Nature Reviews Molecular Cell Biology. 2004;5:271–281. doi: 10.1038/nrm1356. [DOI] [PubMed] [Google Scholar]

[bib15] Godsel LM, Hsieh SN, Amargo EV, Bass AE, Pascoe-McGillicuddy LT, Huen AC, Thorne ME, Gaudry CA, Park JK, Myung K, Goldman RD, Chew TL, Green KJ. Desmoplakin assembly dynamics in four dimensions: multiple phases differentially regulated by intermediate filaments and actin. The Journal of Cell Biology. 2005;171:1045–1059. doi: 10.1083/jcb.200510038. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Green KJ, Simpson CL. Desmosomes: new perspectives on a classic. Journal of Investigative Dermatology. 2007;127:2499–2515. doi: 10.1038/sj.jid.5701015. [DOI] [PubMed] [Google Scholar]

[bib17] Gumbiner B, Stevenson B, Grimaldi A. The role of the cell adhesion molecule uvomorulin in the formation and maintenance of the epithelial junctional complex. The Journal of Cell Biology. 1988;107:1575–1587. doi: 10.1083/jcb.107.4.1575. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Harrison OJ, Bahna F, Katsamba PS, Jin X, Brasch J, Vendome J, Ahlsen G, Carroll KJ, Price SR, Honig B, Shapiro L. Two-step adhesive binding by classical cadherins. Nature Structural & Molecular Biology. 2010;17:348–357. doi: 10.1038/nsmb.1784. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib19] Harrison OJ, Brasch J, Lasso G, Katsamba PS, Ahlsen G, Honig B, Shapiro L. Structural basis of adhesive binding by desmocollins and desmogleins. PNAS. 2016;113:7160–7165. doi: 10.1073/pnas.1606272113. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] Harrison OJ, Jin X, Hong S, Bahna F, Ahlsen G, Brasch J, Wu Y, Vendome J, Felsovalyi K, Hampton CM, Troyanovsky RB, Ben-Shaul A, Frank J, Troyanovsky SM, Shapiro L, Honig B. The extracellular architecture of adherens junctions revealed by crystal structures of type I cadherins. Structure. 2011;19:244–256. doi: 10.1016/j.str.2010.11.016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib21] Hartlieb E, Kempf B, Partilla M, Vigh B, Spindler V, Waschke J. Desmoglein 2 is less important than desmoglein 3 for keratinocyte cohesion. PLoS ONE. 2013;8:e53739. doi: 10.1371/journal.pone.0053739. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] Hong S, Troyanovsky RB, Troyanovsky SM. Cadherin exits the junction by switching its adhesive bond. The Journal of Cell Biology. 2011;192:1073. doi: 10.1083/jcb.201006113. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Hutter JL, Bechhoefer J. Calibration of atomic‐force microscope tips. Review of Scientific Instruments. 1993;64:1868–1873. doi: 10.1063/1.1143970. [DOI] [Google Scholar]

[bib24] Jones JC. Characterization of a 125K glycoprotein associated with bovine epithelial desmosomes. Journal of Cell Science. 1988;89:207–216. doi: 10.1242/jcs.89.2.207. [DOI] [PubMed] [Google Scholar]

[bib25] Koch PJ, Mahoney MG, Ishikawa H, Pulkkinen L, Uitto J, Shultz L, Murphy GF, Whitaker-Menezes D, Stanley JR. Targeted disruption of the pemphigus vulgaris antigen (desmoglein 3) gene in mice causes loss of keratinocyte cell adhesion with a phenotype similar to pemphigus vulgaris. The Journal of Cell Biology. 1997;137:1091. doi: 10.1083/jcb.137.5.1091. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib26] Lewis JE, Jensen PJ, Wheelock MJ. Cadherin function is required for human keratinocytes to assemble desmosomes and stratify in response to calcium. Journal of Investigative Dermatology. 1994;102:870–877. doi: 10.1111/1523-1747.ep12382690. [DOI] [PubMed] [Google Scholar]

[bib27] Lewis JE, Wahl JK, Sass KM, Jensen PJ, Johnson KR, Wheelock MJ. Cross-talk between adherens junctions and desmosomes depends on plakoglobin. The Journal of Cell Biology. 1997;136:919–934. doi: 10.1083/jcb.136.4.919. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib28] Lowndes M, Rakshit S, Shafraz O, Borghi N, Harmon RM, Green KJ, Sivasankar S, Nelson WJ. Different roles of cadherins in the assembly and structural integrity of the desmosome complex. Journal of Cell Science. 2014;127:2339–2350. doi: 10.1242/jcs.146316. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Manibog K, Li H, Rakshit S, Sivasankar S. Resolving the molecular mechanism of cadherin catch bond formation. Nature Communications. 2014;5:3941. doi: 10.1038/ncomms4941. [DOI] [PubMed] [Google Scholar]

[bib30] Manibog K, Sankar K, Kim SA, Zhang Y, Jernigan RL, Sivasankar S. Molecular determinants of cadherin ideal bond formation: conformation-dependent unbinding on a multidimensional landscape. PNAS. 2016;113:E5711–E5720. doi: 10.1073/pnas.1604012113. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Michels C, Buchta T, Bloch W, Krieg T, Niessen CM. Classical cadherins regulate desmosome formation. Journal of Investigative Dermatology. 2009;129:2072–2075. doi: 10.1038/jid.2009.17. [DOI] [PubMed] [Google Scholar]

[bib32] Oesterhelt F, Rief M, Gaub HE. Single molecule force spectroscopy by AFM indicates helical structure of poly(ethylene-glycol) in water. New Journal of Physics. 1999;1:6. doi: 10.1088/1367-2630/1/1/006. [DOI] [Google Scholar]

[bib33] Parisini E, Higgins JM, Liu JH, Brenner MB, Wang JH. The crystal structure of human E-cadherin domains 1 and 2, and comparison with other cadherins in the context of adhesion mechanism. Journal of Molecular Biology. 2007;373:401–411. doi: 10.1016/j.jmb.2007.08.011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] Rakshit S, Zhang Y, Manibog K, Shafraz O, Sivasankar S. Ideal, catch, and slip bonds in cadherin adhesion. PNAS. 2012;109:18815–18820. doi: 10.1073/pnas.1208349109. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib35] Ray C, Brown JR, Akhremitchev BB. Rupture force analysis and the associated systematic errors in force spectroscopy by AFM. Langmuir. 2007;23:6076–6083. doi: 10.1021/la070131e. [DOI] [PubMed] [Google Scholar]

[bib36] Rübsam M, Broussard JA, Wickström SA, Nekrasova O, Green KJ, Niessen CM. Adherens junctions and desmosomes coordinate mechanics and signaling to orchestrate tissue morphogenesis and function: an evolutionary perspective. Cold Spring Harbor Perspectives in Biology. 2017a:a029207. doi: 10.1101/cshperspect.a029207. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib37] Rübsam M, Mertz AF, Kubo A, Marg S, Jüngst C, Goranci-Buzhala G, Schauss AC, Horsley V, Dufresne ER, Moser M, Ziegler W, Amagai M, Wickström SA, Niessen CM. E-cadherin integrates mechanotransduction and EGFR signaling to control junctional tissue polarization and tight junction positioning. Nature Communications. 2017b;8:1250. doi: 10.1038/s41467-017-01170-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib38] Sivasankar S, Zhang Y, Nelson WJ, Chu S. Characterizing the initial encounter complex in cadherin adhesion. Structure. 2009;17:1075–1081. doi: 10.1016/j.str.2009.06.012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib39] Stahley SN, Bartle EI, Atkinson CE, Kowalczyk AP, Mattheyses AL. Molecular organization of the desmosome as revealed by direct stochastic optical reconstruction microscopy. Journal of Cell Science. 2016a;129:2897–2904. doi: 10.1242/jcs.185785. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib40] Stahley SN, Warren MF, Feldman RJ, Swerlick RA, Mattheyses AL, Kowalczyk AP. Super-Resolution microscopy reveals altered desmosomal protein organization in tissue from patients with Pemphigus vulgaris. Journal of Investigative Dermatology. 2016b;136:59–66. doi: 10.1038/JID.2015.353. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib41] Syrris P, Ward D, Evans A, Asimaki A, Gandjbakhch E, Sen-Chowdhry S, McKenna WJ. Arrhythmogenic right ventricular dysplasia/cardiomyopathy associated with mutations in the desmosomal gene desmocollin-2. The American Journal of Human Genetics. 2006;79:978–984. doi: 10.1086/509122. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Tinkle CL, Pasolli HA, Stokes N, Fuchs E. New insights into cadherin function in epidermal sheet formation and maintenance of tissue integrity. PNAS. 2008;105:15405–15410. doi: 10.1073/pnas.0807374105. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] Troyanovsky RB, Klingelhöfer J, Troyanovsky S. Removal of calcium ions triggers a novel type of intercadherin interaction. Journal of Cell Science. 1999;112:4379–4387. doi: 10.1242/jcs.112.23.4379. [DOI] [PubMed] [Google Scholar]

[bib44] Tsang SM, Liu L, Teh MT, Wheeler A, Grose R, Hart IR, Garrod DR, Fortune F, Wan H. Desmoglein 3, via an interaction with E-cadherin, is associated with activation of Src. PLoS One. 2010;5:e14211. doi: 10.1371/journal.pone.0014211. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib45] Vasioukhin V, Bauer C, Yin M, Fuchs E. Directed actin polymerization is the driving force for epithelial cell-cell adhesion. Cell. 2000;100:209–219. doi: 10.1016/S0092-8674(00)81559-7. [DOI] [PubMed] [Google Scholar]

[bib46] Vendome J, Posy S, Jin X, Bahna F, Ahlsen G, Shapiro L, Honig B. Molecular design principles underlying β-strand swapping in the adhesive dimerization of cadherins. Nature Structural & Molecular Biology. 2011;18:693–700. doi: 10.1038/nsmb.2051. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib47] Wheelock MJ, Jensen PJ. Regulation of keratinocyte intercellular junction organization and epidermal morphogenesis by E-cadherin. The Journal of Cell Biology. 1992;117:415–425. doi: 10.1083/jcb.117.2.415. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib48] Wu Y, Jin X, Harrison O, Shapiro L, Honig BH, Ben-Shaul A. Cooperativity between trans and cis interactions in cadherin-mediated junction formation. PNAS. 2010;107:17592–17597. doi: 10.1073/pnas.1011247107. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib49] Wu Y, Vendome J, Shapiro L, Ben-Shaul A, Honig B. Transforming binding affinities from three dimensions to two with application to cadherin clustering. Nature. 2011;475:510–513. doi: 10.1038/nature10183. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] Yen CF, Harischandra DS, Kanthasamy A, Sivasankar S. Copper-induced structural conversion templates prion protein oligomerization and neurotoxicity. Science Advances. 2016;2:e1600014. doi: 10.1126/sciadv.1600014. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib51] Yen CF, Sivasankar S. Improving estimation of kinetic parameters in dynamic force spectroscopy using cluster analysis. The Journal of Chemical Physics. 2018;148:123301. doi: 10.1063/1.5001325. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib52] Zhang Y, Sivasankar S, Nelson WJ, Chu S. Resolving cadherin interactions and binding cooperativity at the single-molecule level. PNAS. 2009;106:109–114. doi: 10.1073/pnas.0811350106. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

E-cadherin binds to desmoglein to facilitate desmosome assembly

Omer Shafraz

Matthias Rübsam

Sara N Stahley

Amber L Caldara

Andrew P Kowalczyk

Carien M Niessen

Sanjeevi Sivasankar

Roles

Abstract

Introduction

Results

Ecad interacts with Dsg2 to form a Ca2+-independent heterodimer

Figure 1. Ecad interacts with Dsg2 to form Ca2+-independent dimers.

Ecad/Dsg2 and Dsc2/Dsc2 dimers have lower lifetimes than Dsc2/Dsg2 dimers

Figure 2. Lifetimes of the Ecad/Dsg2 dimer and the Dsc2/Dsc2 dimer are shorter than the lifetime of the Dsg2/Dsc2 complex.

Ecad is present in nascent desmosomes but not in mature desmosomes

Figure 3. Ecad and Dsg2 are both localized in nascent desmosomes.

Figure 3—figure supplement 1. Relative Ecad levels remain unchanged over a calcium switch time-course.

Leu 175 mediates Ecad and Dsg2 interactions

Figure 4. Ecad interacts with Dsg2 via Leu 175.

Ecad L175 is essential for efficient intercellular Dsg2 recruitment and desmosome assembly

Figure 5. Ecad-L175 promotes desmosome assembly in cells.

Figure 5—figure supplement 1. Impaired junction formation in Ecad-K14E, Ecad-L175D, and Ecad-W2A/K14E (DM) mutants.

Discussion

Figure 6. Proposed model for the role of Ecad in desmosome assembly.

Materials and methods

Purification of cadherin ectodomains

Single molecule AFM force measurements

SIM imaging and analysis of cadherin localization within desmosomes

Isolation, culture, transfection and confocal imaging of primary keratinocytes

Acknowledgements

Funding Statement

Contributor Information

Funding Information

Additional information

Competing interests

Author contributions

Additional files

Data availability

References

Decision letter

Roles

Author response

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Ecad interacts with Dsg2 to form a Ca²⁺-independent heterodimer

Figure 1. Ecad interacts with Dsg2 to form Ca²⁺-independent dimers.