principles of e-cadherin supramolecular organization in vivo · results: we use 3d superresolution...

Principles of E-Cadherin

Current Biology 23, 2197–2207, November 18, 2013 ª2013 Elsevier Ltd All rights reserved http://dx.doi.org/10.1016/j.cub.2013.09.015

Article

Supramolecular Organization In Vivo

Binh-An Truong Quang,1 Madhav Mani,2,3 Olga Markova,1

Thomas Lecuit,1 and Pierre-Francois Lenne1,*1Developmental Biology Institute of Marseilles,UMR 7288 CNRS, Aix-Marseille Universite, 13288 MarseilleCedex 9, France2Kavli Institute of Theoretical Physics, Santa Barbara,CA 93101, USA3University of California, Department of Physics,Santa Barbara, CA 93101, USA

Summary

Background: E-cadherin plays a pivotal role in tissue morpho-genesis by forming clusters that support intercellular adhesionand transmit tension. What controls E-cadherin mesoscopicorganization in clusters is unclear.Results: We use 3D superresolution quantitative microscopyin Drosophila embryos to characterize the size distributionof E-cadherin nanometric clusters. The cluster size followspower-law distributions over three orders of magnitude withexponential decay at large cluster sizes. By exploring the pre-dictions of a general theoretical framework including clusterfusion and fission events and recycling of E-cadherin, we iden-tify two distinct active mechanisms setting the cluster-sizedistribution. Dynamin-dependent endocytosis targets largeclusters only, thereby imposing a cutoff size. Moreover, inter-actions between E-cadherin clusters and actin filaments con-trol the fission in a size-dependent manner.Conclusions: E-cadherin clustering depends on key corticalregulators, which provide tunable and local control overE-cadherin organization. Our data provide the foundation fora quantitative understanding of how E-cadherin distributionaffects adhesion andmight regulate force transmission in vivo.

Introduction

Epithelia form barriers that are extensively remodeled duringdevelopment and in the adult, such as in the gut. Cell-celladhesion underlies tissue cohesion. Adhesion requires thecis and trans association of E-cadherin (E-cad) moleculesthrough their extracellular domains [1–7], and their stabiliza-tion by interactions with the cortical actin cytoskeleton atadherens junctions. Interactions between E-cadherin andF-actin requires b-catenin and a-catenin [8–10], along withother F-actin binding proteins such as Vinculin [11, 12] andEPLIN [13]. The morphogenesis of epithelia is accompaniedby the extensive remodeling of cell contacts. On one hand,this requires regulation of cell-cell adhesion, for instance,through regulation of levels of cadherins, and/or turnover byendocytic recycling. On the other hand, junction remodelingdepends on the local regulation of cortical tension associatedwith the junctional actomyosin network. For example, polar-ized actomyosin enrichment drives cell intercalation during tis-sue extension, or cell constriction during tissue invagination

*Correspondence: [email protected]

[14]. Remarkably, cortical tension and adhesion are not inde-pendent processes as they are both dependent on E-cadherincomplexes. Cortical actomyosin pulls on E-cadherin com-plexes via a-catenin [12, 15–17], and therefore the forces trans-mitted at cell contacts require E-cadherin to control contactshape [18, 19]. Thus, E-cadherin complexes have two inter-twined mechanical functions: they support the formation andstabilization of cell contacts, and they transmit tensile forcesat cell contacts during tissue remodeling [14, 20].An abundant literature documents the fact that regulation of

cadherin levels has a profound effect on tissue organizationand tissue dynamics. Rapid uptake of E-cadherin by endocy-tosis and strong reduction in E-cadherin levels are associatedwith a loss of columnar epithelial organization during epithe-lial-mesenchymal transition (EMT), an important develop-mental process (such as in the dermomyotome and lateralplate mesoderm of developing vertebrate embryos) and animportant step in cancer progression [21]. Differential expres-sion of adhesion molecules was also shown to drive cell-sort-ing phenomena in cell culture [22, 23] and in vivo [24, 25].E-cadherin recycling is also required for cell intercalation dur-ing germband extension in Drosophila [26–28] and zebrafishepiboly [29].Yet, the levels of E-cadherin molecules, although an impor-

tant parameter, are a too rough description to account forE-cadherin’s role in tissue morphogenesis. E-cadherins donot accumulate uniformly at adherens junctions and forminstead finite size clusters in many cell types even visiblewith diffraction-limited optics [12, 30–33]. These clusterstransmit cell tension [15, 31], and their existence leads oneto consider that cell mechanics is discretized at cell contacts.The size, distribution, and lateral dynamics of E-cadherin clus-ters are indeed expected to impact on adhesive forces andtensile force transmission locally. Remarkably, however, wedo not understand how such clusters form and what sets theirsize. To a large extent, this is due to the fact that such clustershave not been characterized quantitatively yet. Our observa-tion and quantification of a broad distribution of E-cadherinclusters stands in contrast to predictions for a passive systemand elicits the following questions: Which active processesdetermine the supramolecular organization of E-cadherinin vivo and to what extent? In particular, how do endocytosisand actin interactions regulate cluster distribution?By combining 3D superresolution optical imaging with theo-

retical modeling, we address here the mechanisms governingthe formation of E-cadherin clusters in vivo. We first analyzehow E-cadherin is organized at adherens junctions of intactepithelia of Drosophila embryos and find it forms dense nano-metric clusters, containing mixed cis- and trans-pairs. Sec-ond, we show that the size of E-cadherin clusters follows apower-law distribution over three orders of magnitude withexponential suppression of clusters above a certain size.Such polydispersed distributions are observed over a largerange of E-cadherin surface densities at junctions and at alltimes during 2 hr of epithelial morphogenesis. In an attemptto study the kinetics of cluster growth, we observed the forma-tion of new junctions and found that clusters grow rapidly(<5 min) up to a steady state along with E-cadherin junctional

http://dx.doi.org/10.1016/j.cub.2013.09.015


mailto:[email protected]

http://crossmark.crossref.org/dialog/?doi=10.1016/j.cub.2013.09.015&domain=pdf

Figure 1. E-Cadherin Distribution in Live

Epithelia of Drosophila Embryos

(A) Confocal images showing the apicolateral dis-

tribution of the endogenous E-cadherin fused to

GFP (endoE-cad::GFP) (left) and E-cad fused to

the photoconvertible protein EosFP (E-Cad::

EosFP) (right) in the ventrolateral region (stage 8).

The scale bar represents 5 mm.

(B) Schematic diagram of epithelial cells during

cellularization and gastrulation. AJ, adherens

junction; BJ, basal junction (left). E-cad intensity

during cellularization and gastrulation along the

apicobasal axis (right).

(C) Absolute level of E-cad revealed by endoE-

cad::GFP during early embryogenesis.

(D) Junctional surface density of endoE-cad::

GFP (black) and E-cad::EosFP (blue) at two

stages of tissue morphogenesis in live (FCS and

confocal imaging) and fixed embryos (orange,

PALM imaging). Error bars represent the SD.

See also Figure S1.

Current Biology Vol 23 No 222198

surface density. To understand the origins of these distribu-tions, we explore the falsifiable predictions of a general theo-retical kinetic framework, which incorporates clustering andrecycling mechanisms and propose experiments to challengedifferent models. Blocking endocytosis using a dynaminmutant alters only the cluster size above which cluster fre-quency is exponentially suppressed, while having no measur-able effect on the frequency of smaller clusters. This indicatesthat dynamin-dependent endocytosis preferentially targetslarge clusters, with rates increasing dramatically past a criticalcluster size (>100molecules). a-catenin-mediated interactionsof E-cadherin with actin influence the power-law regime.Our observations are consistent with actin-based active con-trol of cluster fission and passive/diffusive fusion dynamics.This work demonstrates that endocytosis and interactionsbetween E-cadherin and actin have antagonistic qualitativeand quantitative functions, providing tunable and local controlof E-cadherin clustering.

Results

E-Cadherin Surface Density in Live Drosophila Embryos

We first examined the distribution of E-cad::GFP in earlyepithelia, using a knockin line replacing the endogenousE-cadherin, called E-cad hereafter [34]. Similar to what hasbeen observed previously [31, 35], E-cad organizes nonuni-formly. Confocal images show diffraction-limited ‘‘spots’’(Figure 1A), whose distribution changes during tissuemorpho-genesis (Figure 1B and Figure S1 available online). AlthoughE-cad was mostly found basally during cellularization, ittended to organize apically, during gastrulation, forming theadherens junctions (AJs) [35]. E-cad fused to the photoconver-tible monomeric EosFP (E-cad::EosFP), which we used here-after for superresolution imaging, organized similarly at AJs(Figure 1A, right), and also exhibited a clear apicobasal polarity(Figures 1B and S1).

To determine the absolute numberof E-cad in single cells and at celljunctions, we then performed live quan-titative imaging, combining confocalmicroscopy with fluorescence correla-tion spectroscopy (FCS) [36] (Figures

1C and 1D). FCS was used to calibrate the photon count rateper GFP molecule and per EosFP molecule, which was thenused to translate the number of photons in the confocalimages into number of molecules (Figures S1D and S1E). Onaverage, 36 13 104 E-cad was detected at the plasma mem-brane during gastrulation (stage 7b, called ‘‘early’’ hereafter,n = 5 embryos, mean 6 SD) (Figure 1C), consistent with previ-ous reports using ELISA assays [37]. Forty-five minutes later(stage 9, called ‘‘late’’ stage hereafter), the number of E-cadincreased by about 50% (4.5 6 2 3 104 molecules per cell).At the AJ over a 1 mm thickness, the surface density of E-cadwas 370 6 130 and 630 6 150 molecules per mm2, at, respec-tively, early and late stages (Figure 1D). Quantification ofE-cad::EosFP by confocal microscopy using FCS calibrationindicates that it is 50% less abundant than the endogeneousE-cad and follows the same rate of increase (Figure 1D).Note that if E-cad were equally spaced in the two apposedmembranes, 600 molecules per mm2 would translate into adistance in the plane of a membrane of 60 nm, which is aboutten times larger than the lateral size of individual E-cad. Incontrast, if all E-cad molecules of a junction of 5 mm lengthwould condense in a tightly packed aggregate they wouldform a single macroscopic cluster of w200 nm diameter, thatis, 1/25 of the length of a junction.

Superresolution Imaging Reveals Packed NanometricE-Cad Clusters

To determine the supramolecular organization of E-cad atcell-cell contacts, we developed an optical setup based onphotoactivation localized microscopy [38] (PALM) and usedthe E-cad::EosFP to map E-cad distribution. Figure 2 showsthe density and single molecule maps of E-cad::EosFP atcell junctions. Single E-cad::EosFP molecules could belocalized with 30 nm precision in the plane of the epitheliumand 50–100 nm precision along the apicobasal direction(optical axis) [39] (Figures 2B, 2C, and S2A). Comparison

Figure 2. Superresolution Images of E-Cad in

Epithelia

(A) Sum of all images used for PALM analysis.

(B) PALM image showing the supramolecular

organization of E-cad::EosFP. Higher magnifica-

tion (bottom) shows uneven dense regions along

a cell junction. Scale bars represent 5 mm (top)

and 500 nm (bottom).

(C) Single molecule map and cluster organization

along junctions represented in projections (top

and middle) in an oblique view (bottom). x, y

coordinates are in the plane of the junctions,

z is along the apicobasal axis. Molecules are

color coded according to the cluster size (num-

ber of molecules). Localization precision: x-y,

30 nm; z, 50–100 nm.

(D) Density of E-cad::EosFP in clusters of size

larger than 50 molecules. The density of

E-cad::EosFP in cis-clusters (magenta curve)

was determined from a selection of clusters (36

clusters), detected on PALM image as paired-

clusters in separated membranes (inset). ‘‘All-

clusters’’ data correspond to all clusters found

in one epithelium (black curve, 128 clusters).

The scale bar represents 200 nm.

See also Figure S2.

E-Cadherin Supramolecular Organization In Vivo2199

with optical-diffraction limited image shows that high intensityaggregates, which were not possible to resolve with conven-tional optics, could be resolved into nanometric clusters inthe PALM images (compare Figures 2B with 2A, sum of allimages used for PALM analysis).

To quantify these further, we applied an automatic detectionalgorithm to identify clusters and classified them according tocluster size, i.e., the number of E-cad they contain (Figures 2Cand S2).

The density of E-cad::EosFP in clusters was found to berall = 7,400 6 3,200 per mm2 (mean 6 SD, 128 clusters, oneembryo, size >50 molecules, Figures 2D and S2B). This wassignificantly different from the density of E-cad::EosFP in cis-clusters, that we could unambiguously detect in occasionalcases where membranes were separated (rcis = 4,800 61,900 molecules per mm2, mean 6 SD, 36 clusters, p < 1023

Wilcoxon test, Figure 2D). Comparison between the densitiesrall and rcis suggests that in average about 25% molecules inclusters are in cis.

Given the density of E-cad::EosFP in cis-clusters and thefact that E-cad::EosFP molecules detected in PALM repre-sents about 30% 6 15% of total E-cad (endogeneous E-cadand E-cad::EosFP, Figure 1D), we therefore estimated the den-sity of adjacent E-cad in a membrane to be 16 (+16/210)3 103

per mm2. This would translate into an average distancebetween adjacent E-cad of 7.9 nm (+2.4/21.8). This value isconsistent with structural studies of mammalian cadherinsshowing that adjacent cadherins in a membrane are spacedby 7 nm [3, 40, 41] and confirms that we detect bona fide clus-ters of packed E-cad.

E-Cad Is Organized intoPolydispersed Clusters

Thus, E-cad forms dense nanometricclusters in vivo and their size can bestudied quantitatively. This led us toinvestigate the mechanisms deter-mining cluster distributions at cell junc-tions. We first quantified the distribution

of cluster size, i.e., the number of multimers of size n, cn, over alarge number of junctions in different embryos of the samestage (492 junctions, 13 embryos, stage 7b, ‘‘early,’’ Figure 3A).Remarkably, clusters were very polydispersed in size, rangingover three orders of magnitude. We ensured that PALMacquisition and our data analysis techniques preserve theinformation of molecular organizations (Figure S3; Supple-mental Experimental Procedures). The cluster-size distribu-tion is well approximated by a power law with an exponentialcutoff: cn = A naexp(2n/n*) with an exponent a = 21.94 60.03, a cutoff n* = 100 6 20 molecules and an amplitudeA = 33 6 2 (Figure 3A).The existence of a power-law distribution indicates that the

dynamics that shape the distribution do not single out anyone size, and as a consequence the system does not exhibita characteristic size. This ‘‘democracy’’ of cluster sizesbreaks down in the exponential cutoff regime of the distribu-tion for larger clusters. Furthermore, because a straightfor-ward clustering process ought to lead to ‘‘coarsening’’—theabsence of any intermediate scales—our observations indi-cate the presence of a removal mechanism. In the absenceof deposition, a removal mechanism would lead to a lossof cadherins at junctions. The observation of steady-statedistribution suggests the presence of both removal anddeposition, that is, recycling. Remarkably E-cad organizationis very different from that of other membrane components,such as GPI-anchored proteins, which form very smalloligomeric structures [42], integrin or syntaxin, which orga-nizes in clusters with characteristic size of a few tens ofmolecules [43, 44].

Figure 3. E-Cad Cluster-Size Distribution and the Effects of E-Cad Junctional Surface Density and Junction Maturation

(A) Cluster-size distribution at early stage (n = 492 junctions analyzed). Green dots are experimental data. Blue squares show the same distribution using a

logarithmic binning. Solid red line is a fit of the experimental data by a power law with a power exponent 21.94 6 0.03 and an exponential cutoff (100 6 20

molecules). Error bars represent SD.

(B) E-cad junctional surface density at early and late stages.

(C) Cumulative distribution functions at two junctional surface densities (s1 = 175 6 25 molecules/mm2 and s2 = 275 6 25 molecules/mm2). Comparison

between early and late stages (KS test).

(D–F) Power-law exponent (D), cutoff size n* (E) and amplitude (F) of the cluster-size distribution as a function of junctional surface density.

(G) PALM images of AJs at early and late stages. The scale bar represents 5 mm.

See also Figure S3.


Are these features of the cluster-size distribution particularto junctions at an early stage of maturation or are they pre-served over a large range of surface densities and ‘‘age’’ ofjunctions? This question is particularly relevant as E-cadsurface densities change during development and impactprofoundly on sorting behaviors between different cell popula-tions and tissues [45]. To address this question, we first inves-tigated how clustering amplitude and distribution changed asa function of E-cad surface density that varies over a broadrange at a given stage (Figure 3B). The functional form of thedistribution is preserved at all observed E-cad junctionalsurface densities (Figure 3C, cumulative distributions for twosurface densities s1 = 175 and s2 = 275 molecules/mm2), witha power-law exponent a and cutoff size n*, which increaseswith E-cad junctional surface density (Figures 3D and 3E,492 junctions in total). In contrast, the amplitude A remainsrelatively constant over a large range of E-cad junctional

surface densities (Figure 3F). As elaborated upon in the Sup-plemental Experimental Procedures, the constancy in mono-mer concentration could be a consequence of two possiblelimiting considerations: strong monomer-monomer bindinginto a trans-E-cad bond or alternatively rapid clustering oftrans-E-cad bonds relative to their dissociation into mono-mers. Regardless, at a phenomenological level, at a givenE-cad junctional surface density, the two parameters a andn*, characterize the cluster-size distributions andwill be exten-sively used hereafter to investigate the mechanisms that regu-late E-cad clustering.Strikingly, the clustering depends strongly on the local

junctional surface density but not on the age of junctions(Figure 3C). Even if on average the E-cad junctional surfacedensity increased by 70% between early and late stages (Fig-ure 3B), we found that junctions at the two stages having thesame E-cad junctional surface density exhibited the same

Figure 4. Kinetics of E-Cad Clustering in New

Forming Junctions

(A) PALM images of E-cad::EosFP during the

formation of a new junction after cell division,

t0 = 0–20 s after the formation of the interface

(left), t1 = 120 6 30 s (middle), and t2 = 240 6

30 s (right). Scale bars represent 5 mm (top) and

1 mm (bottom).

(B) Cluster-size distribution and E-cad junction

surface density (inset) at three time points of junc-

tion formation after cell division (17, 16, and 16

analyzed junctions for t0, t1, and t2, respectively).

(C) Comparison between new formed junctions

and more mature ones.

See also Figure S4.


cluster-size distributions (Figure 3C, Kolmogorov-Smirnov[KS] test p > 0.2), indicating that we are observing steady-statedistributions. Despite earlier reports using conventional micro-scopy indicating that E-cad seems to form a continuous belt atthese later stages, PALM observations clearly show that AJsare still composed of many polydisperse clusters (Figure 3G).

Equilibration of E-Cad Clusters in New Forming Junctions

During tissue morphogenesis, cells need to dynamicallymodulate their adhesion as junctions disappear and reform[26, 46]. It is therefore important to determine the timescalesover which changes in E-cad surface density can potentiallyimpact on clustering at AJs. To this end, we monitored howE-cad organizes at cell membranes forming new junctionsafter cell division, using PALM at different time points.

After cytokinesis, a new interface is created between theapposed membranes of daughter cells (Figure 4A, top, t =0–200 0) [47–49]. At this time point, the E-cad junctional surfacedensity is low, 20 molecules per mm2 (17 junctions, Figure 4B,inset). The cluster-size distribution is narrow with only a smallfraction of E-cad engaged in clusters (<10%) (Figure 4B).

By about 2 min later, the number of clusters increased; clus-ters with more than ten molecules were detected. Strikingly, atthis time point, the distribution was the same as the oneobserved inmoremature junctions having the same E-cad sur-face density (Figure 4C), indicating that clusters have hadsufficient time to fuse and the distribution has adjusted tothe increase in E-cad surface density. After about 4 min, thejunctional surface density increased by 4-fold and clustersweremore numerous (Figure 4B). Again the distribution agreedwith more mature junctions at the same surface density (Fig-ure 4C), indicating that the cluster distribution adjusts rela-tively rapidly to alterations in E-cad junction surface density.As detailed in the Supplemental Information, the characteristictime required for the distribution to equilibrate to the increasedjunctional pool of cadherin is determined by the rate of clusterfusion: if diffusion limited, it is on the order of 100 s (typical

junction length 4 mm, cadherin diffusionconstant 0.01 mm2/s [31]), consistentwith the above observations.

Equilibration aside, an increase inE-cad junction surface density at a newcell junction can result from two distinctsources: accumulation of free mono-mers diffusing from other regions of thecell membrane and by vesicular traf-ficking. By determining the number ofE-cad::EosFP found in vesicles close to

the junctions (6.66 2, mean6 SD, 344 vesicles), we estimatedthat a net flux of 15 vesicles per minute would be required at aminimum to bring E-cad to new forming junctions and accountfor the observed changes in E-cad surface density (Figure S4).Future dynamic data will allow us to quantify the relativecontribution of these two E-cad sources.

A Framework for Kinetics of Cluster Formation

To uncover the mechanisms of clustering that account forobserved distributions, we employ a general theoretical frame-work, which was introduced by Smoluchowski [50], and thathas been extended [51] and applied in various contexts.Although this framework recovers power-law distributionsand can straightforwardly account for deviations from it, itsutility is manifest in its predictions of qualitatively distinctdistributional outcomes of contrasting molecular clusteringmechanisms [52, 53].Cadherin clusters can fuse to each other generating larger

clusters and undergo fission, thereby producing smaller clus-ters as depicted in Figure 4. Accounting for these simpledynamics, Equation 1 represents the time evolution of the con-centration of clusters of size n, cn:

dcn

dt= sðnÞ|ffl{zffl}

recycling

+Xm2 1

264 k2

n; mcn+m|fflfflfflfflfflffl{zfflfflfflfflfflffl}fission

2 k +n; mcncm|fflfflfflfflfflffl{zfflfflfflfflfflffl}

fusion

375

+1

2

Xn2 1

m=1

264 k +

n2m; mcn2mcm|fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl}fusion

2 k2m; n2mcn|fflfflfflfflfflfflffl{zfflfflfflfflfflfflffl}

fission

375:

(Equation 1)

The recycling term, elaborated upon later, characterizes thecluster-size-dependent flux of E-cad on and off junctions. Inparticular, flux into the population will account for monomer-monomer association. Higher-order monomer interactionswith clusters are neglected, as is direct dissociation of clustersinto monomers. The rates, k+n,m and k2n,m, parameterize the

Figure 5. Regulation of Cluster-Size Distribution by Endocytosis

(A) Schematics of the kinetic model for clustering: Number and size of clusters result from the balance of fusion and fission events at the cell surface and

recycling. Recycling consists in delivery of E-cad monomers and endocytosis/removal of clusters.

(B) PALM images of E-cad::EosFP in shi-ts mutant and control at 32�C. Higher magnifications show a cell junction in shi-ts mutant and in control.

(C) Junctional surface density in shi-ts mutant and in control (32�C).(D) Cumulative distribution of the cluster size at the surface density of E-cad::EosFP = 200 molecules/mm2 (p value shown, KS test).

(E and F) Power-law exponent (E) and cutoff size (F) as a function of junctional surface density in shi-ts mutant and control.

(G and G0) Transmission electron micrograph of AJs in wild-type embryos. The plasma membranes of contacting cells are shown, together with spot

adherens junctions (SAJs) (white arrowheads), and clathrin-coated pits (black arrow). Higher magnification (G0) corresponds to the square box in (F).

Scale bars represent 5 mm in (B) and 250 nm in (G). See also Figure S5.


rate at which clusters of size n and m fuse into a cluster of size(n + m) and the reverse fission process, respectively. Theseeffects are illustrated in Figure 5A (details in SupplementalInformation). We allow the rates to depend on the size of inter-acting clusters. Although fusion and fission rates would berelated by detailed balance in a passive process, here we donot posit such a constraint. We have neglected several, poten-tially important, effects: transport of clusters along the junctionas well as distinctions between cis and trans clusters. Neglectof these details reflects the inaccessibility to these featuresexperimentally. As alluded to previously, rates that do notsingle out any particular size, for example, k+n,m w (nb + mb)or k+n,m w (nm)b, result in power-law distributions. Asan example, were a particular scale singled out, N in thecase k+n,m w exp[(n + m)/N], polydispersed distributionswould not be observed.

The process of fusion alone ensures that at sufficiently longtimes all available cadherins will reside within large clusters, aprocess referred to as coarsening. This is avoided through theprocess of endocytosis that depletes the junction of cadherinsand recycles them back onto interfaces: here representedthrough a recycling term in Equation 1, to be discussed later.

Observations suggest an investigation of the steady-statesolutions to these equations and, in particular, an explorationof which cellular processes and molecular players influencefusion, fission, and recycling as described within this frame-work. We begin by perturbing endocytosis.

Regulation of Cluster-Size Distribution by Endocytosis

Endocytosis has been shown to modulate the level of E-cad atcell junctions in mammalian cell cultures [54] and in vivo [27,

29]. Although endocytosis has an effect on E-cad junctionalsurface density, E-cad clusters could be directly targeted forendocytosis [55] and thereby have a significant impact onclusters organization. We investigated the role of endocytosison the supramolecular organization of E-cad by blocking theendocytic pathway mediated by Shibire (Shi), the Drosophilaortholog of Dynamin, a GTPase involved in vesicle scission,which is concentrated at AJs in remodeling epitheliaDrosophila [27]. As expected, in shimutants (using a tempera-ture-sensitive [ts] allele, shi-ts and placing embryos at therestrictive temperature 32�C for 20 min), the junctional sur-face density of E-cad increased by 80% (12 embryos, p value <10210, Figures 5B and 5C). We then compared the cluster-sizedistribution between shi-ts mutants and control (at the restric-tive temperature 32�C) at the same junctional E-cad surfacedensity s* (Figures 5C and 5D, s* = 200 molecules per mm2).Remarkably, cumulative distribution plots show that the largerclusters (size > 100 of E-cad::EosFP, i.e., size > 300 for E-cad)were more abundant when endocytosis was blocked. Consis-tent with this, the observed cluster distribution (exponential)cutoff scale n* was systematically larger in shi-ts mutantscompared to control (Figure 5F). However, the exponent ofthe power-law a was unchanged (Figure 5E). We thereforecan rule out that endocytosis tunes clustering by simplyremoving E-cad monomers and thereby tuning overall E-cadsurface density. Instead, our observation indicates thatendocytosis targets large E-cad clusters, while not perturb-ing the mechanisms of clustering at scales smaller than n* <100. This is consistent with electron micrographs showingnumerous clathrin-coated pits in the vicinity or within spotAJs (Figures 5G and 5G0). This also supports the recent


proposal that E-cad-regulated clustering in cis accelerates itsendocytosis in vivo [27]. Taken together, our observationssuggest a recycling term of the form,

sðnÞ= jondn; 1 2 joff exp�n=n*

�cn: (Equation 2)

Equation 2 represents the strong depletion of clusters fromjunctions of sizes larger than n* at a rate joff (exponentialcluster-size dependence ensures that endocytosis of smallclusters, n < n*, is negligible). Furthermore, we make the bio-logically plausible assumption that cadherins are recycledand delivered to the interface in monomeric form at a rate jon(dn,1 is the kronecker delta function that equals 1 if n = 1 and0 otherwise). Taken together, endocytosis prevents the accu-mulation of large clusters—coarsening—thereby ensuring asteady state.

Cluster-Size Distribution at Steady-State andSize-Dependent Clustering Events

By targeting large clusters, endocytosis produces a cutoff butleaves unchanged the power-law regime of the cluster-sizedistribution (at a given E-cad junctional surface density, Fig-ure 5D). Interestingly, our observations over a wide range ofE-cad junctional surface densities have exponents that liein the interval a <21.5 (Figure 5E). We note that passive fusionand fission of clusters, with E-cad delivery and endocytosisensuring a steady state, recovers a distribution with an expo-nent a = 21.5 [56] (details in Supplemental Information). Thissuggests that cluster distributions in vivo are a consequenceof actively regulated fission and fusion rates. In particular,we are interested in whether interactions between E-cad andactin influence fission and fusion of clusters, and if so, how.

Alterations in fusion versus fission of E-cad clusters pro-duce qualitatively distinct effects on E-cad distributions(mathematical details can be found in the Supplemental Infor-mation) (Figures 6A, 6B, and S5). In the size-dependent fusionmodel (Figure 6A), the steady-state concentration of a clusterof a given size is determined by the balance of fluxes in, ofsmaller fusing clusters, and a flux out. If smaller clustersfuse slower than larger clusters fuse, then the distributionsteepens. As an example, were the size dependence offusion rates described by k+n,m w (nb+mb ), the exponent ofthe cluster distribution would follow a = 2b 21.5, indicatinga steepening a < 21.5 when larger clusters fuse more effi-ciently b > 0.

F-Actin Regulates Clustering by Reducing Fission EventsThe cytoplasmic tail of E-cad links E-cad complexes to theactin cytoskeleton. It binds to b-catenin, which, in turn, bindsa-catenin (a-Cat) [12, 57]. a-Catmediatesmechanical couplingto actin filaments [31, 58]. RNAi-mediated reduction of a-Catcaused a 40% reduction of E-cad at junctions (Figures 6Cand 6D), consistent with previous reports [31]. Remarkably,the cluster-size distribution shows that a-Cat also has animpact on E-cad clustering independent from E-cad surfacedensity. At a given E-cad junction surface density (s* = 75mol-ecules per mm2), the level of clustering is reduced significantlyin a-Cat RNAi embryos (p < 10210, Figure 6E). Cluster-size dis-tributions remain power laws but with a steeper slope thancontrol, aa-CatRNAi < aWT < 21.5 (Figure 6F).

According to the size-dependent fusion model, this wouldsuggest that a-Cat reduces the fusion of larger clusters relativeto smaller ones (Figure 6A). This seems implausible given therole that a-Cat plays in mediating actin/E-cad interactions.

Alternatively, the observed steepness of cluster-size distri-butions could be explained by actin controlling the fissionrates in a size-dependent manner by reducing the fission ofsmaller clusters (Figure 6B). Indeed, were smaller clustersmore efficient at breaking up, the rate at which E-cad wouldpopulate larger clusters would be hindered. This slowingdown of clustering would produce a steepening of the clus-ter-size distribution. This fission model would suggest that,in a-Cat RNAi embryos, the fission of smaller clusters wouldbe enhanced with respect to wild-type (compare Figures 6Band 6E) Thus, our observations suggest that the role of actinis to stabilize cadherin clusters to possible fission events(Figure 6B).Our model predicts that depolymerizing or stabilizing

F-actin ought to produce, respectively, a steepening or a flat-tening of the distribution relative to the wild-type, which iswhat we observe using, respectively, the F-actin depolymeriz-ing drug Latrunculin A (aLatA < aWT) or the stabilizing drug jas-plakinolide (aJaspla > aWT) [59] (Figure 6F).The fact that E-cad clustering is often confined to the most

apical part of cell-cell contacts suggests that specific polarityproteins in addition to cytoskeletal regulation [35, 60] mightplay an important role. We therefore investigated the role ofpolarity cues on the organization principles of E-cad at AJsin the light of our model.The polarity protein Par3 was reported to be a positional

landmark for AJs [35, 60]. Moreover, Par3 colocalizes withE-cad complexes in spots AJs [31, 61] and coimmunoprecipi-tates with b-Cat [62]. However, Par3 is not required for thenucleation of E-cad clusters [37]. We addressed the mecha-nisms by which Par3 might regulate/position E-cad clusters.RNAi-mediated reduction of Par3 yielded a significant reduc-tion of E-cad at junctions (Figure 6D) and also a strongdecrease of E-cad clustering independent of its effect onE-cad surface density (p < 10213, Figure 6E). The effect onthe cluster-size distribution was remarkably similar to thatobserved for a-Cat: aa-CatRNAi = aPar3-RNAi (Figure 6F, KS test,p > 0.9). This suggests that Par-3 might regulate the positionand organization of E-cad clusters using actin-dependentmechanisms controlling the fission rates like a-Cat.

Discussion

E-Cad Supramolecular Organization In Vivo

We characterized with PALM microscopy the precise molecu-lar densities and cluster organization of E-cad in Drosophilaembryos and proposed a general theoretical framework thatallows us determine the essential mechanisms of E-cad clus-tering in vivo. Superresolution quantification of E-cad clustersreveals a mesoscopic organization of polydisperse nanomet-ric clusters for different E-cad surface density levels as wellas mutant backgrounds. In particular, the distribution can bewell approximated by a truncated power-law distribution thatis precisely regulated by the distinct activity of E-cad/actincoupling and E-cad endocytosis. So far, assaying whetherthese processes altered the mesoscopic organization ofE-cad has been unexplored and quantifying the phenotypesassociated with perturbing E-cad/actin coupling and endocy-tosis has not been possible. The accuracy of our measure-ments make both these possible and allows us to assignmechanistic roles to key cortical regulators involved in bothE-cad/actin coupling and endocytosis. The developmentalconsequences of this precise control of E-cad organizationare now made addressable by our work.

Figure 6. Actin/E-Cadherin Interactions Control Fission Mechanisms

(A andB) Predicted cluster-size distributions in the size-dependent fusion (A) and size-dependent fission (B)models. In the size-dependent fusionmodel (A),

the rates of fusion are power laws with exponent b: k+n,m = k+0(nb + mb). In the size-dependent fission model (B), the rates of fission are power laws with

exponent u: k-n,m = k-0(nu + mu). Deviations from the slope a = 21.5 allow discrimination between models.

(C–E) a-Cat and Par3 control of cluster-size distribution. PALM images in a-Cat RNAi-injected embryo and control (C), junctional surface density (D), and

cumulative distribution at the surface density of 75 molecules/mm2 (E).

(F) Power-law exponents as a function of E-cad junctional surface density in a-Cat RNAi (red squares), Par3-RNAi (blue hexagons), Latrunculin-A-treated

(magenta triangles), jasplakinolide-treated (green triangles), and control (black circles) embryos. The scale bar represents 5 mm.


Endocytosis Controls E-Cadherin Surface Density by

Targeting Large ClustersOur data show that endocytosis controls E-cad at AJs in quan-titative and qualitative ways. It limits the growth of largeclusters by their targeted removal (Figure 7A) and as aconsequence reduces E-cad surface density. We showedthat E-cad removal by dynamin-dependent endocytosis isonly significant above a cluster-size threshold of about300 E-cad molecules, which corresponds to clusters largerthan 60 nm in diameter. One possible mechanism is thatlarge clusters could favor the assembly of the endocyticmachinery and then be endocytosed with a higher frequencycompared to smaller ones. This conclusion is in agreement

with the recent proposal that E-cad clustering leads to therecruitment of the adaptator protein AP2 and clathrin [27]. Inaddition to clathrin vesicles, caveolae or pinocytic vesiclesthat could accommodate more E-cad than clathrin vesiclesmight be also involved in E-cad endocytosis. Importantly,endocytosis prevents the formation of macroscopic clusters,which could freeze tissue dynamics by affecting the actomy-osin networks.

Actin Regulation of E-Cad ClusteringIt is largely accepted that E-cad-mediated adhesion relieson the trans-homodimerization and cis-interactions, whichproduce lateral clustering. Previous reports relying on key

Figure 7. Regulation of Cluster Size and Number by Endocytosis and Actin-

Dependent Mechanisms

(A) A schematic model.

(B) E-cad clustering phase diagram. Supramolecular organization of

E-cad characterized by two components, power-law exponent and cutoff

size (a, n*) in control and perturbed conditions.


features of E-cad structure had shown that trans- and cis-interactions of E-cad are important for clustering [63–66].Although E-cad can cluster without actin [66], our data indicatethat, in living epithelia, the maintenance of E-cad clusters re-quires E-cad interactions with actin. More specifically, werule out the possibility that interaction with actin enhancesthe fusion of cadherin clusters; instead, our work suggeststhat actin-based regulation inhibits the breakup of cadherinclusters (Figure 7A). We can therefore propose that thesteady-state organization of clusters in vivo results from thebalance between actin-independent fusion processes medi-ated by cis- and trans-interactions, as reported in vitro,and actin-dependent fission mechanisms. This reconcilesopposing views emphasizing either the role of actin-depen-dent mechanisms in clustering or the intrinsic self-assemblyproperties of E-cad.

E-Cad Clustering Phase Diagram

Quantitative measurements of E-cad cluster-size distributionsas a function of E-cad surface density in WT, mutant, andtreated scenarios can be projected onto the clustering phasediagram in Figure 7B. The empirical fact that two parameters,

a and n*, accurately characterize these distributions permitssuch a projection.Each genetic condition tracks out a curve in Figure 7B.

Comparing these different curves and how they map ontothe other is insightful. (1) The shi-ts curve is a vertical shift ofthe WT curve, indicating the dynamin-dependent endocytosisleaves the power-law part of the cluster distribution un-changed and simply alters the scale at which endocytosisbecomes significant. (2) The a-Cat RNAi curve is a shift ofthe WT curve horizontally to the left, indicating that a-Catsolely influences the power-law part of the distribution.Dynamin-dependent endocytosis and a-Cat therefore movecurves along orthogonal directions in this phase diagram.Naturally, if perturbing expression levels of a molecule led toan a and an n* shift, it would indicate its joint influence onboth endocytosis and E-cad cluster stability.Reduction of E-cad/actin coupling in a Par3 mutant leads to

similar transformations relative to WT as a-Cat RNAi suggest-ing a similar role played by the two in E-cad organization. Simi-larly, depolymerizing F-actin by Latrunculin A translates theWT curve to the left. Conversely, promoting actin polymeriza-tion by jasplakinolide translates the WT curve to the right sug-gesting an increased stability of cadherin clusters.We believe that the approach proposed here could be help-

ful to address the role of endocytosis and more generally traf-ficking in the regulation of E-cad organization at cell junctions.

Consequences for Tissue DynamicsWe found that the monomeric concentration of E-cad at junc-tion is immune to changes in total junctional surface densityand that the distribution of clusters is quantitatively controlledover a large range of surface densities. This provides a discre-tized mode of cell-cell adhesion, which supports corticalforces during junction remodeling [16] and cell sorting in vivo[18]. The role of actin crosslinking and endocytosis reportedhere may provide local and fast control over cluster formationand dynamics. This has strong implications for tissuemorpho-genesis. Indeed, Par3 distribution is spatially regulated inearly Drosophila embryos [67] and affects the distribution ofE-cad at cells junctions [26, 27]. In addition, E-cad endocytosisis locally controlled and affects E-cad planar polarity atcell junctions. When both endocytosis of E-cad and Par3 dis-tribution are affected, junction remodeling is perturbed, sug-gesting the major impact of controlling local E-cad clusterorganization.Although in vitro data indicate that E-cadherin transmits

force between the cytoskeleton and the cell environment[17], the question of how E-cadherin clustering modulatesforce transmission in vivo is not solved yet and will requirefurther work combiningmechanical measurements and super-resolution imaging as presented above.Our study of E-cad supramolecular organization in vivo

paves the way for a quantitative understanding of adhesionand force transmission and how these might impact on celland tissue dynamics in living organisms.

Experimental Procedures

Imaging

PALM images were performed using a home-built optical setup with three

laser lines, 405, 488, and 561 nm, for activation and imaging the green

and red forms of EosFP, respectively (see the Supplemental Experimental

Procedures). Drosophila embryos were imaged under oblique illumination.

Total magnification was 3160 and astigmatic detection was performed as

described in [39]. Single molecules were detected using a generalized


likelihood ratio test as described in MTT algorithm [68] and localized using

a bidimensional Gaussian approximating the microscope point spread

function (PSF). FCS and confocal imaging were performed with a LSM780

(Zeiss) in the photon-counting mode (see the Supplemental Experimental

Procedures).

Transmission electron micrographs were obtained on ultrathin sections

using a Zeiss EM 912 electron microscope, equipped with a camera Gatan

Bioscan, as described in [27].

Data analysis

Clusters detection was performed using first an intermolecular distance cri-

terion and second a mean-shift algorithm adapted from [69]. Clusters were

further analyzed using inertia matrices to determine the density of mole-

cules (see the Supplemental Experimental Procedures).

Fly Stocks and Sample Preparation

Sqh-E-Cad::EosFP(III) was obtained by combining the E-cad::mEosFP1

fusion used previously [31] to the promoter of spaghetti-squash. Embryo

fixation, staining, and injection of RNAi probes were performed as

described in [60].

Supplemental Information

Supplemental Information includes Supplemental Experimental Procedures

and five figures and can be found with this article online at http://dx.doi.org/

10.1016/j.cub.2013.09.015.

Author Contributions

P.-F.L. planned the project with T.L. and analyzed the experiments together

with B.-A.T.Q and T.L.. B.-A.T.Q. conducted all experiments except FCS

measurements, which were performed together with P.-F.L. M.M. devel-

oped the theoretical model and analyzed its predictions together with

P.-F.L. O.M. performed the numerical simulations with B.-A.T.Q. P.-F.L.

and B.-A.T.Q. developed the PALM setup and quantification procedures.

All authors discussed the data. The manuscript was written by M.M. and

P.-F.L., and all authors commented on it.

Acknowledgments

We thank R. Levayer and J.M. Philippe for preparing the E-cad::EosFP

construct, A. Pelissier-Monier for TEM micrographs, and everyone who

provided us with reagents. We also thank all members of the Lenne and

Lecuit laboratories for discussions and comments on the manuscript. We

thank Jitu Mayor for critical reading of the manuscript. This work was sup-

ported by the CNRS, the Fondation de la Recherche Medicale (FRM), the

HFSP, the ANR-Blanc and the national infrastructure France Bio-Imaging.

B.-A.T.Q. was supported by a PhD fellowship by the Region PACA-Andor

Technology and Association pour la Recherche sur le Cancer (ARC).

Received: May 9, 2013

Revised: July 29, 2013

Accepted: September 5, 2013

Published: October 31, 2013

References

1. Nose, A., Tsuji, K., and Takeichi, M. (1990). Localization of specificity

determining sites in cadherin cell adhesion molecules. Cell 61, 147–155.

2. Takeichi, M. (1991). Cadherin cell adhesion receptors as a morphoge-

netic regulator. Science 251, 1451–1455.

3. Shapiro, L., Fannon, A.M., Kwong, P.D., Thompson, A., Lehmann, M.S.,

Grubel, G., Legrand, J.F., Als-Nielsen, J., Colman, D.R., and

Hendrickson, W.A. (1995). Structural basis of cell-cell adhesion by

cadherins. Nature 374, 327–337.

4. Yap, A.S., Brieher, W.M., and Gumbiner, B.M. (1997). Molecular and

functional analysis of cadherin-based adherens junctions. Annu. Rev.

Cell Dev. Biol. 13, 119–146.

5. Zhu, B., Chappuis-Flament, S., Wong, E., Jensen, I.E., Gumbiner, B.M.,

and Leckband, D. (2003). Functional analysis of the structural basis of

homophilic cadherin adhesion. Biophys. J. 84, 4033–4042.

6. Patel, S.D., Ciatto, C., Chen, C.P., Bahna, F., Rajebhosale, M., Arkus, N.,

Schieren, I., Jessell, T.M., Honig, B., Price, S.R., and Shapiro, L. (2006).

Type II cadherin ectodomain structures: implications for classical cad-

herin specificity. Cell 124, 1255–1268.

7. Hong,S., Troyanovsky,R.B., andTroyanovsky,S.M. (2010). Spontaneous

assembly and active disassembly balance adherens junction homeosta-

sis. Proc. Natl. Acad. Sci. USA 107, 3528–3533.

8. Kofron, M., Spagnuolo, A., Klymkowsky, M., Wylie, C., and Heasman, J.

(1997). The roles of maternal alpha-catenin and plakoglobin in the early

Xenopus embryo. Development 124, 1553–1560.

9. Torres, M., Stoykova, A., Huber, O., Chowdhury, K., Bonaldo, P.,

Mansouri, A., Butz, S., Kemler, R., and Gruss, P. (1997). An alpha-E-

catenin gene trapmutation defines its function in preimplantation devel-

opment. Proc. Natl. Acad. Sci. USA 94, 901–906.

10. Vasioukhin, V., Bauer, C., Yin, M., and Fuchs, E. (2000). Directed actin

polymerization is the driving force for epithelial cell-cell adhesion. Cell

100, 209–219.

11. Weiss, E.E., Kroemker, M., Rudiger, A.H., Jockusch, B.M., and Rudiger,

M. (1998). Vinculin is part of the cadherin-catenin junctional complex:

complex formation between alpha-catenin and vinculin. J. Cell Biol.

141, 755–764.

12. Yonemura, S., Wada, Y., Watanabe, T., Nagafuchi, A., and Shibata, M.

(2010). alpha-Catenin as a tension transducer that induces adherens

junction development. Nat. Cell Biol. 12, 533–542.

13. Abe, K., and Takeichi, M. (2008). EPLINmediates linkage of the cadherin

catenin complex to F-actin and stabilizes the circumferential actin belt.

Proc. Natl. Acad. Sci. USA 105, 13–19.

14. Lecuit, T., Lenne, P.F., and Munro, E. (2011). Force generation, trans-

mission, and integration during cell and tissue morphogenesis. Annu.

Rev. Cell Dev. Biol. 27, 157–184.

15. Martin, A.C., Kaschube,M., andWieschaus, E.F. (2009). Pulsed contrac-

tions of an actin-myosin network drive apical constriction. Nature 457,

495–499.

16. Rauzi, M., Lenne, P.F., and Lecuit, T. (2010). Planar polarized actomy-

osin contractile flows control epithelial junction remodelling. Nature

468, 1110–1114.

17. Borghi, N., Sorokina, M., Shcherbakova, O.G., Weis, W.I., Pruitt, B.L.,

Nelson, W.J., and Dunn, A.R. (2012). E-cadherin is under constitutive

actomyosin-generated tension that is increased at cell-cell contacts

upon externally applied stretch. Proc. Natl. Acad. Sci. USA 109,

12568–12573.

18. Maıtre, J.L., Berthoumieux, H., Krens, S.F., Salbreux, G., Julicher, F.,

Paluch, E., and Heisenberg, C.P. (2012). Adhesion functions in cell sort-

ing by mechanically coupling the cortices of adhering cells. Science

338, 253–256.

19. Levayer, R., and Lecuit, T. (2013). Oscillation and polarity of E-cadherin

asymmetries control actomyosin flow patterns during morphogenesis.

Dev. Cell 26, 162–175.

20. Wang, Y.C., Khan, Z., Kaschube, M., and Wieschaus, E.F. (2012).

Differential positioning of adherens junctions is associated with initia-

tion of epithelial folding. Nature 484, 390–393.

21. Thiery, J.P., Acloque, H., Huang, R.Y., and Nieto, M.A. (2009). Epithelial-

mesenchymal transitions in development and disease. Cell 139,

871–890.

22. Steinberg, M.S. (1963). Reconstruction of tissues by dissociated cells.

Some morphogenetic tissue movements and the sorting out of embry-

onic cells may have a common explanation. Science 141, 401–408.

23. Steinberg, M.S., and Takeichi, M. (1994). Experimental specification of

cell sorting, tissue spreading, and specific spatial patterning by quanti-

tative differences in cadherin expression. Proc. Natl. Acad. Sci. USA 91,

206–209.

24. Godt, D., and Tepass, U. (1998). Drosophila oocyte localization is medi-

ated by differential cadherin-based adhesion. Nature 395, 387–391.

25. Hayashi, T., and Carthew, R.W. (2004). Surface mechanics mediate

pattern formation in the developing retina. Nature 431, 647–652.

26. Blankenship, J.T., Backovic, S.T., Sanny, J.S., Weitz, O., and Zallen, J.A.

(2006). Multicellular rosette formation links planar cell polarity to tissue

morphogenesis. Dev. Cell 11, 459–470.

27. Levayer, R., Pelissier-Monier, A., and Lecuit, T. (2011). Spatial regula-

tion of Dia and Myosin-II by RhoGEF2 controls initiation of E-cadherin

endocytosis during epithelial morphogenesis. Nat. Cell Biol. 13,

529–540.

28. Tamada, M., Farrell, D.L., and Zallen, J.A. (2012). Abl regulates planar

polarized junctional dynamics through b-catenin tyrosine phosphoryla-

tion. Dev. Cell 22, 309–319.




29. Song, S., Eckerle, S., Onichtchouk, D., Marrs, J.A., Nitschke, R., and

Driever, W. (2013). Pou5f1-dependent EGF expression controls E-cad-

herin endocytosis, cell adhesion, and zebrafish epiboly movements.

Dev. Cell 24, 486–501.

30. Tepass, U., and Hartenstein, V. (1994). The development of cellular junc-

tions in the Drosophila embryo. Dev. Biol. 161, 563–596.

31. Cavey, M., Rauzi, M., Lenne, P.F., and Lecuit, T. (2008). A two-tiered

mechanism for stabilization and immobilization of E-cadherin. Nature

453, 751–756.

32. Kametani, Y., and Takeichi, M. (2007). Basal-to-apical cadherin flow at

cell junctions. Nat. Cell Biol. 9, 92–98.

33. Hong, S., Troyanovsky, R.B., and Troyanovsky, S.M. (2013). Binding to

F-actin guides cadherin cluster assembly, stability, and movement.

J. Cell Biol. 201, 131–143.

34. Huang, J., Zhou, W., Dong, W., Watson, A.M., and Hong, Y. (2009). From

the Cover: Directed, efficient, and versatile modifications of the

Drosophila genome by genomic engineering. Proc. Natl. Acad. Sci.

USA 106, 8284–8289.

35. Harris, T.J., and Peifer, M. (2004). Adherens junction-dependent

and -independent steps in the establishment of epithelial cell polarity

in Drosophila. J. Cell Biol. 167, 135–147.

36. Kim, S.A., Heinze, K.G., and Schwille, P. (2007). Fluorescence correla-

tion spectroscopy in living cells. Nat. Methods 4, 963–973.

37. McGill, M.A., McKinley, R.F., and Harris, T.J. (2009). Independent

cadherin-catenin and Bazooka clusters interact to assemble adherens

junctions. J. Cell Biol. 185, 787–796.

38. Betzig, E., Patterson, G.H., Sougrat, R., Lindwasser, O.W., Olenych, S.,

Bonifacino, J.S., Davidson, M.W., Lippincott-Schwartz, J., and Hess,

H.F. (2006). Imaging intracellular fluorescent proteins at nanometer res-

olution. Science 313, 1642–1645.

39. Huang, B., Wang, W., Bates, M., and Zhuang, X. (2008). Three-dimen-

sional super-resolution imaging by stochastic optical reconstruction

microscopy. Science 319, 810–813.

40. Al-Amoudi, A., Dıez, D.C., Betts, M.J., and Frangakis, A.S. (2007). The

molecular architecture of cadherins in native epidermal desmosomes.

Nature 450, 832–837.

41. Harrison, O.J., Jin, X., Hong, S., Bahna, F., Ahlsen, G., Brasch, J.,Wu, Y.,

Vendome, J., Felsovalyi, K., Hampton, C.M., et al. (2011). The extracel-

lular architecture of adherens junctions revealed by crystal structures of

type I cadherins. Structure 19, 244–256.

42. Sharma, P., Varma, R., Sarasij, R.C., Ira, Gousset, K., Krishnamoorthy,

G., Rao, M., and Mayor, S. (2004). Nanoscale organization of multiple

GPI-anchored proteins in living cell membranes. Cell 116, 577–589.

43. van Zanten, T.S., Cambi, A., Koopman, M., Joosten, B., Figdor, C.G.,

and Garcia-Parajo, M.F. (2009). Hotspots of GPI-anchored proteins

and integrin nanoclusters function as nucleation sites for cell adhesion.

Proc. Natl. Acad. Sci. USA 106, 18557–18562.

44. Sieber, J.J., Willig, K.I., Kutzner, C., Gerding-Reimers, C., Harke, B.,

Donnert, G., Rammner, B., Eggeling, C., Hell, S.W., Grubmuller, H.,

and Lang, T. (2007). Anatomy and dynamics of a supramolecular mem-

brane protein cluster. Science 317, 1072–1076.

45. Krens, S.F., and Heisenberg, C.P. (2011). Cell sorting in development.

Curr. Top. Dev. Biol. 95, 189–213.

46. Bertet, C., Sulak, L., and Lecuit, T. (2004). Myosin-dependent junction

remodelling controls planar cell intercalation and axis elongation.

Nature 429, 667–671.

47. Founounou, N., Loyer, N., and Le Borgne, R. (2013). Septins regulate the

contractility of the actomyosin ring to enable adherens junction remod-

eling during cytokinesis of epithelial cells. Dev. Cell 24, 242–255.

48. Guillot, C., and Lecuit, T. (2013). Adhesion disengagement uncouples

intrinsic and extrinsic forces to drive cytokinesis in epithelial tissues.

Dev. Cell 24, 227–241.

49. Herszterg, S., Leibfried, A., Bosveld, F., Martin, C., and Bellaiche, Y.

(2013). Interplay between the dividing cell and its neighbors regulates

adherens junction formation during cytokinesis in epithelial tissue.

Dev. Cell 24, 256–270.

50. Smoluchowski, M.V. (1916). Drei Vortrage uber Diffusion, Brownsche

Molekularbewegung und Koagulation von Kolloidteilchen. Physik.

Zeit. 17, 557–585.

51. Family, F., Meakin, P., and Deutch, J.M. (1986). Kinetics of coagulation

with fragmentation: Scaling behavior and fluctuations. Phys. Rev. Lett.

57, 727–730.

52. Goldstein, M.L., Morris, S.A., and Yen, G.G. (2004). Problems with fitting

to the power-law distribution. Eur. Phys. J. B 41, 255–258.

53. Clauset, A.S., Shalizi, C.R., and Newman, M.E.J. (2009). Power-law dis-

tributions in empirical data. SIAM Rev. 51, 661–703.

54. de Beco, S., Gueudry, C., Amblard, F., and Coscoy, S. (2009).

Endocytosis is required for E-cadherin redistribution at mature adhe-

rens junctions. Proc. Natl. Acad. Sci. USA 106, 7010–7015.

55. Troyanovsky, R.B., Sokolov, E.P., and Troyanovsky, S.M. (2006).

Endocytosis of cadherin from intracellular junctions is the driving force

for cadherin adhesive dimer disassembly. Mol. Biol. Cell 17, 3484–3493.

56. Turner, M.S., Sens, P., and Socci, N.D. (2005). Nonequilibrium raftlike

membrane domains under continuous recycling. Phys. Rev. Lett. 95,

168301.

57. Bershadsky, A. (2004). Magic touch: how does cell-cell adhesion trigger

actin assembly? Trends Cell Biol. 14, 589–593.

58. Desai, R., Sarpal, R., Ishiyama, N., Pellikka, M., Ikura, M., and Tepass, U.

(2013). Monomeric a-catenin links cadherin to the actin cytoskeleton.

Nat. Cell Biol. 15, 261–273.

59. Bubb, M.R., Senderowicz, A.M., Sausville, E.A., Duncan, K.L., and Korn,

E.D. (1994). Jasplakinolide, a cytotoxic natural product, induces actin

polymerization and competitively inhibits the binding of phalloidin to

F-actin. J. Biol. Chem. 269, 14869–14871.

60. Pilot, F., Philippe, J.M., Lemmers, C., and Lecuit, T. (2006). Spatial con-

trol of actin organization at adherens junctions by a synaptotagmin-like

protein Btsz. Nature 442, 580–584.

61. Muller, H.A., andWieschaus, E. (1996). Armadillo, bazooka, and stardust

are critical for early stages in formation of the zonula adherens and

maintenance of the polarized blastoderm epithelium in Drosophila.

J. Cell Biol. 134, 149–163.

62. Wei, S.Y., Escudero, L.M., Yu, F., Chang, L.H., Chen, L.Y., Ho, Y.H., Lin,

C.M., Chou, C.S., Chia, W., Modolell, J., and Hsu, J.C. (2005). Echinoid is

a component of adherens junctions that cooperates with DE-Cadherin

to mediate cell adhesion. Dev. Cell 8, 493–504.

63. Troyanovsky, R.B., Laur, O., and Troyanovsky, S.M. (2007). Stable and

unstable cadherin dimers: mechanisms of formation and roles in cell

adhesion. Mol. Biol. Cell 18, 4343–4352.

64. Zhang, Y., Sivasankar, S., Nelson, W.J., and Chu, S. (2009). Resolving

cadherin interactions and binding cooperativity at the single-molecule

level. Proc. Natl. Acad. Sci. USA 106, 109–114.

65. Wu, Y., Jin, X., Harrison, O., Shapiro, L., Honig, B.H., and Ben-Shaul, A.

(2010). Cooperativity between trans and cis interactions in cadherin-

mediated junction formation. Proc. Natl. Acad. Sci. USA 107, 17592–

17597.

66. Wu, Y., Vendome, J., Shapiro, L., Ben-Shaul, A., and Honig, B. (2011).

Transforming binding affinities from three dimensions to two with appli-

cation to cadherin clustering. Nature 475, 510–513.

67. Simoes, Sde.M., Blankenship, J.T., Weitz, O., Farrell, D.L., Tamada, M.,

Fernandez-Gonzalez, R., and Zallen, J.A. (2010). Rho-kinase directs

Bazooka/Par-3 planar polarity during Drosophila axis elongation. Dev.

Cell 19, 377–388.

68. Serge, A., Bertaux, N., Rigneault, H., and Marguet, D. (2008). Dynamic

multiple-target tracing to probe spatiotemporal cartography of cell

membranes. Nat. Methods 5, 687–694.

69. Tarjan, R. (1972). Depth-first search and linear graph algorithms. SIAM

J. Comput. 1, 146–160.

principles of e-cadherin supramolecular organization in vivo · results: we use 3d superresolution...

Documents