plakoglobin is required for effective intermediate filament

Plakoglobin Is Required for Effective IntermediateFilament Anchorage to DesmosomesDevrim Acehan1, Christopher Petzold1, Iwona Gumper2, David D. Sabatini2, Eliane J. Muller3,Pamela Cowin2,4 and David L. Stokes1,2,5

Desmosomes are adhesive junctions that provide mechanical coupling between cells. Plakoglobin (PG) is amajor component of the intracellular plaque that serves to connect transmembrane elements to thecytoskeleton. We have used electron tomography and immunolabeling to investigate the consequences ofPG knockout on the molecular architecture of the intracellular plaque in cultured keratinocytes. Althoughknockout keratinocytes form substantial numbers of desmosome-like junctions and have a relatively normalintercellular distribution of desmosomal cadherins, their cytoplasmic plaques are sparse and anchoring ofintermediate filaments is defective. In the knockout, b-catenin appears to substitute for PG in the clustering ofcadherins, but is unable to recruit normal levels of plakophilin-1 and desmoplakin to the plaque. By comparingtomograms of wild type and knockout desmosomes, we have assigned particular densities to desmoplakin anddescribed their interaction with intermediate filaments. Desmoplakin molecules are more extended in wild typethan knockout desmosomes, as if intermediate filament connections produced tension within the plaque. Onthe basis of our observations, we propose a particular assembly sequence, beginning with cadherin clusteringwithin the plasma membrane, followed by recruitment of plakophilin and desmoplakin to the plaque, andending with anchoring of intermediate filaments, which represents the key to adhesive strength.

Journal of Investigative Dermatology (2008) 128, 2665–2675; doi:10.1038/jid.2008.141; published online 22 May 2008

INTRODUCTIONDesmosomes are large macromolecular complexes thatrepresent a major category of intercellular junction (Garrodet al., 2002). They impart mechanical strength to a widerange of tissues and their importance is underscored by theirprevalence in tissues that experience shear stress, such as theheart and skin. The cadherins (desmoglein and desmocollin)are transmembrane glycoproteins that form extracellularbonds with their counterparts from the neighboring cell. Onthe intracellular membrane surface, these cadherins nucleatea dense plaque of proteins designed to couple the extra-cellular bond with the network of intermediate filaments thatcourse throughout the cell. Plakoglobin (PG) and plakophilinare the major components of the outer dense plaque that isproximal to the membrane. Desmoplakin composes an inner

dense plaque that is further from the membrane and thatmediates the binding of intermediate filaments.

Desmoplakin is a member of the plakin family, which alsoincludes envoplakin, periplakin, bullous pemphigoid anti-gen-1, and plectin. Generally speaking, these proteins play arole in anchoring membrane-associated protein complexesto intermediate filaments (Leung et al., 2002). Sequencepredictions indicate that all family members have a long,a-helical rod flanked by globular domains at their N- andC-termini. In the case of desmoplakin, the N-terminal,globular domain targets to the outer dense plaque (Stappen-beck et al., 1993; Bornslaeger et al., 1996; Smith and Fuchs,1998), where it is proposed to cluster PG and cadherincomplexes into discrete patches (Kowalczyk et al., 1997;North et al., 1999). The C-terminal end has a characteristicplakin-repeat domain and is thought to interact directly withintermediate filaments (Green et al., 1992; Choi et al., 2002).

PG and plakophilin are members of the armadillo (ARM)protein family, which includes b-catenin and p120ctn that arefound in the cytoplasmic plaque of the related adherensjunction. Both PG and b-catenin associate with a conservedcytoplasmic region of various cadherins via the 12 ARMrepeats that characterize this family. Classical cadherins bindto ARM repeats 5–9 (Huber and Weis, 2001), whereasdesmosomal cadherins bind to repeats flanking this centralregion (Witcher et al., 1996). b-Catenin has an importantsecondary role in WNT signaling, which also involvesinteractions of the ARM repeats, this time with Lef/Tcftranscription factors and the elements of the so-called

& 2008 The Society for Investigative Dermatology www.jidonline.org 2665

ORIGINAL ARTICLE

Received 4 January 2008; revised 17 March 2008; accepted 4 April 2008;published online 22 May 2008

1Skirball Institute for Biomolecular Medicine, New York University School ofMedicine, New York, New York, USA; 2Department of Cell Biology,New York University School of Medicine, New York, New York, USA;3Vetsuisse Faculty, Institute of Animal Pathology, University of Bern, Bern,Switzerland; 4Department Dermatology, New York University School ofMedicine, New York, New York, USA and 5New York Structural BiologyCenter, New York, New York, USA

Correspondence: Dr David L. Stokes, Skirball Institute 3-13, New YorkUniversity School of Medicine, 540 First Avenue, New York, New York10012, USA.E-mail: [email protected]

Abbreviations: ARM, Armadillo; EM, electron microscopy; PG, plakoglobin

http://dx.doi.org/10.1038/jid.2008.141

http://www.jidonline.org

mailto:[email protected]

destruction complex (Nelson and Nusse, 2004). PG has highsequence homology both in the central ARM repeats and inthe N-terminal regulatory motif that is targeted by thedestruction complex (Hatsell et al., 2003), and recent studiessupport an analogous role in transcriptional regulation by PG(Miravet et al., 2003; Muller-Tidow et al., 2004; Williamsonet al., 2006). Plakophilin has a similar architecture, althoughwith fewer ARM repeats and a distinctive kink in the middle(Choi and Weis, 2005). Despite structural homology, nobinding partners have been identified for the plakophilinARM domain. Instead, a multitude of interactions involvethe globular N-terminal domain, which is unrelated to PG(Hatzfeld, 2007).

Within the desmosomal plaque, these constituents engagein a number of complex interactions. In particular, thedesmoplakin N-terminal domain interacts with the cytoplas-mic tails of desmocollin ‘‘a’’ isoforms, the N-terminal head ofplakophilin-1, and the central ARM repeats of PG (Kapprellet al., 1988; Troyanovsky et al., 1996; Kowalczyk et al.,1997). Plakophilin and PG, in turn, interact with eachother and promote clustering of desmosomal cadherins(Bornslaeger et al., 2001; Chen et al., 2002; Bonne et al.,2003). Plakophilin is reported to bind to intermediatefilaments, but desmoplakin appears to be primarily respon-sible for linking the desmosomal plaque to the cytoskeleton(Kapprell et al., 1988; Smith and Fuchs, 1998; Bornslaegeret al., 2001). At first glance there appears to be redundancyand competition among cadherins, PG, and plakophilin fordesmoplakin binding. Nevertheless, a full complement ofinteractions seems to be required for full adhesive function(Bierkamp et al., 1996; Ruiz et al., 1996; McGrath et al.,1997; Gallicano et al., 1998; McKoy et al., 2000; Yin andGreen, 2004). To date, many questions remain about thephysical architecture of the desmosomal plaque, the specificnature of these diverse molecular interactions, and the waydesmosomal assembly is regulated.

Genetic knockout of the PG gene is lethal during theembryonic period, with severe heart and skin defects,presumably due to malfunction of desmosomes and theintercalated disk (Bierkamp et al., 1996; Ruiz et al., 1996).Unlike in wild-type animals, b-catenin was detected inepidermal desmosomes from these knockout embryos,suggesting that this highly homologous ARM family memberattempts to substitute for Pg, but has limited functionality(Bierkamp et al., 1999). In this study, we have used PG-knockout cells to shed light on the physical architecture andmolecular interactions within the intracellular plaque. Wehave used electron tomography to determine the three-dimensional structures of desmosomes and have usedimmunolabeling to document changes in the compositionof the plaque.

RESULTSCharacterization of cell cultures by PCR, immunoblotting,and immunofluorescence

To investigate the effects of PG on the cytoplasmic plaque ofdesmosomes, we have used long-term cell cultures of kera-tinocytes derived from mice that were wild type (PGþ /þ )

for the PG gene, or homozygous (PG�/�) for the knockoutallele. Cell cultures were initially seeded at low Ca2þ

concentrations (0.05 mM) and, after reaching confluence,were switched to 1.4 mM Ca2þ for periods up to 48 hours toinduce desmosome formation (Hennings and Holbrook,1983).

Genotypes were confirmed by PCR (Figure 1a). DNA fromPGþ /þ cells produced an B650-bp band correspondingto the wild-type allele, whereas DNA from PG�/� cellsproduced an B450-bp band corresponding to the targetedPG allele. The expression levels of PG in the cultured cellswere assessed by immunoblotting of extracts from culturedcells prepared 15 hours after inducing desmosome formationby increasing the Ca2þ concentration to 1.4 mM (Figure 1b).The resulting data verified that PG was not present inhomozygous, knockout cells.

The subcellular distributions of various desmosomalproteins in wild type and knockout cells were characterizedby immunofluorescence 15 hours after inducing desmosomeformation. As expected, PG�/� cells were unlabeled(Figure 2b), demonstrating the specificity of the PG anti-bodies. The same antibodies labeled cell borders in PGþ /þ

cells (Figure 2a), reflecting a concentration of PG in bothdesmosomes and adherens junctions. Immunolabeling forplakophilin-1 (Figure 2c and d) and desmoplakin (Figure 2eand f) was concentrated at the periphery of both cell types.Desmoplakin produced a punctate distribution in wild-typecells, characteristic of its desmosomal localization, butproduced a diffuse component of staining both around theplasma membrane and within the cytoplasm in culturedPG�/� cells, as seen previously in epidermis (Bierkamp et al.,1999). Keratin-14 antibody labeling (Figure 2g and h)revealed intermediate filaments coursing through the cyto-plasm of cells regardless of their PG genotype. A transcellular

750

500

300(bp)

(kDa)

2,000

1,000

1,500

M +/– +/+ +/++/– –/– –/– –/– M

20013085

M +/+ –/–

+/+ –/–

+/+ –/–

PG E-cadβ-Cat

Figure 1. Characterization of cell cultures. (a) PCR amplified distinct bands

for DNA from wild type and PG-knockout alleles. As expected, DNA from

heterozygous cells (PGþ /�) produced both bands. Size markers are shown in

lanes marked ‘‘M’’. (b) Immunoblotting verified expression of PG in wild type,

but not in knockout cells. Control experiments involved blotting the same

protein extracts with antibodies to either b-catenin or E-cadherin.

2666 Journal of Investigative Dermatology (2008), Volume 128

D Acehan et al.Tomography of Plakoglobin-Knockout Desmosomes

network was observed frequently in wild-type cells, but onlyoccasionally in mutant cells. The distribution of E-cadherin, acomponent of adherens junctions but not of desmosomes,was comparable in both knockout and wild-type cells(Figure 2i and j), and b-catenin produced a somewhat morepunctate labeling in knockout cells (Figure 2k) as comparedwith PGþ /þ cells (Figure 2l). These results suggest that theabsence of PG had significant effects on the distributionof certain associated proteins, although an ultrastructuralstudy was necessary to evaluate the specific effects on thedesmosomal architecture.

The number and size of desmosomesWhen Transwell filters were used to support cells throughoutthe rapid freezing/freeze-substitution protocol used for ourbest electron microscopy (EM) preparations, parallel cellborders were punctuated by numerous desmosomes in bothPGþ /þ and PG�/� cultures. On the other hand, when fixedcells were scraped off culture dishes for more conventionalEM preparations, few desmosomes remained and thecorresponding cell contacts were largely disrupted in PG�/�

cultures (Figure S1). High-magnification images revealed thatPG�/� desmosomes were frequently torn out of one cell afterthese scraping procedures (Figure 3), but not in cells thatwere consistently supported on Transwell filters. Wild-typedesmosomes appeared normal in both the preparations.These results illustrate an innate susceptibility of knockoutcells to shear stress, as previously documented by indirectmeasurements of adhesive strength (Yin et al., 2005b), butindicate that this susceptibility is not simply due to a lack ofdesmosome formation in the absence of PG.

We quantified the numbers and sizes of desmosomes inTranswell cultures (Table 1) and found that PG�/� cells haveB60% the total number of desmosomes observed in wild-type cells. For both cell types, the number and size ofdesmosomes steadily increased over the first 48 hours and

then leveled off. Notably, PG�/� cells had a significantpopulation of abnormally large (4640 nm diameter) desmo-somes that were not observed in PGþ /þ cells (Figure S2). Thenumber and size distribution of desmosomes from PGþ /þ

cells were comparable to that of primary cultures ofkeratinocytes isolated from newborn mice (not shown). Onthe other hand, the relative number of desmosomes observedin cultured PG�/� keratinocytes after scraping was consider-ably lower, close to the 5% value reported by Bierkamp et al.(1999) for the epidermis of PG�/� embryos compared withwild type. They also reported desmosomes torn from theplasma membrane of adjacent cells in tissue from knockoutanimals. These lower percentages and tissue damage mostlikely reflect increased mechanical challenge within live

PG

–/–

PG

+/+

PG PP1 DP K14 E-cad β-Cat

a

b

c

d

e

f

g

h

i

j

k

l

Figure 2. Immunofluorescence images from PGþ /þ (top row) and PG�/� (bottom row) cell cultures. The primary antibody target is indicated at the top

margin. (a, b) As expected, PG is only present in PGþ /þ cells and is concentrated at the cell boundaries. (c, d) Plakophilin-1 staining is similar for both cell types.

(e, f) Desmoplakin is concentrated at cell boundaries with a less punctate appearance and greater cytoplasmic distribution in PG�/� cells. (g, h) Keratin-14

labeling is similar in both cell types, although the transcell network visible in PG�/� cells was a relatively rare occurrence. (i, j) E-cadherin served as a control

and, as expected, produces a similar labeling pattern in both cell types. (k, l) b-Catenin is concentrated along the cell surface in both cell types, but has a

more punctate appearance in PG�/� cells. Bar¼ 20 mm.

a b

Figure 3. Broken desmosomes from PG�/� cells. (a,b) Ruptured cells are

frequently observed when PG�/� cells are scraped from culture dishes

during the EM preparation. Given the consequent mechanical stress,

PG-deficient desmosomes frequently break off from one cell, leaving a

fragment of the cell membrane. This observation suggests that the intercellular

bond provided by the cadherin molecules is intact but that the lack of

intermediate filament attachments produces a vulnerability in cell adhesion

that ultimately leads to cell rupture. Bars¼100 nm.

www.jidonline.org 2667



tissue or during specimen preparation for EM and reflect aweakness in the desmosomal structure.

Ultrastructure of desmosomes

The fine structures of wild-type and PG-deficient desmo-somes were compared by conventional EM and electrontomography of freeze-substituted samples. Whereas conven-tional projection images suffer from superposition of featuresthroughout the thickness of the section, tomography allowscomputational slicing through the three-dimensional struc-tures, thus allowing assessment of molecular interactions in adense macromolecular assembly like the desmosome. Thecharacteristic features include a dense group of strand-likedensities (cadherin molecules) extending across the inter-cellular gap and associating with one another within theso-called midline that appears halfway between the twomembranes, a dense intracellular plaque, and bundles ofintermediate filaments. In projection images, it is generallydifficult to distinguish individual components or their specificassociations, due either to superposition or to excessivefixation and staining. By using freeze substitution for speci-men preservation and tomography for imaging, we have beenable to delineate individual cadherin molecules in theintercellular gap (He et al., 2003) as well as individualintermediate filament associations with the intracellularplaque (see below). A potential alternative is to imagefrozen-hydrated sections, as in a recent report addressingcadherin organization in epidermis (Al-Amoudi et al.,2007). However, the resulting images have very low contrastand extensive image averaging is therefore required toreveal features within the intercellular gap. Althoughfrozen, unstained tissue provides better specimen preserva-tion, thorough characterization of the intracellular plaque

is required before applying this method to knockoutdesmosomes.

Projection images show that desmosomes either inprimary keratinocyte cultures (Figure 4b) or in long-termcultures from wild type and PGþ /� keratinocytes (Figure 4c–f)were structurally similar to those in newborn mouse skin(Figure 4a) (He et al., 2003). In contrast, PG knockoutresulted in distinct changes in desmosomal architecture(Figure 4g–i). Although intercellular densities associatedwith cadherins appeared to be similar in projection images(Figure 4f and i) and tomographic slices (Figure 5c and f) fromboth cell types, the outer dense plaque was narrower and lessdensely packed than in wild-type desmosomes (Figures 4 and5, and see more tomographic images in Figure S3). Inaddition, tomographic slices clearly show that the innerdense plaque, which in wild-type desmosomes comprises anetwork of densities, was either disorganized or completelyabsent in knockout cells (Figure 5d and e; Figures S2 and S3;and Movie S2).

These inner plaque densities generally associate with frayedintermediate filament bundles that lie in close proximity to theinner plaque of wild-type desmosomes (Figures 5a, b and 7;Figure S2; and Movies S1 and S3). Although such intermediatefilament bundles were occasionally visible in the generalvicinity of desmosomes in knockout cells, they rarelyassociated with these desmosomes 15–48 hours after inducingtheir formation with 1.4 mM Ca2þ medium. In addition to theexamples shown in Figures 4 and 5 and Figure S3, we havescored B50 desmosomes from each cell type. The results(Table 1) indicate almost no intermediate filament connectionsto knockout desmosomes at 15 and 30 hours time points, with10% of knockout desmosomes having only limited intermedi-ate filament connections after 48 hours. In many wild-type

Table 1. Characterization of desmosomes in wild type and knockout cells1

Number2 Size distribution3 Intermediate filament associations4

Desmo./cell 4160 nm (%) 160–320 nm (%) 320–560 nm (%) 4640 nm (%) 0 1 2 3

PG+/+

15 hours 3.3 51 47 2 0 0 2% 45% 53%

30 hours 5.8 37 51 12 0 0 0 39% 61%

PG�/�

15 hours 2.3 71 2 10 0 82% 14% 4% 0%

30 hours 3.4 36 0 15 6 96% 4% 0% 0%

48 hours 4.3 26 55 15 4 79% 11% 10% 0%

PG, plakoglobin.1Desmosomes were identified and characterized in low-magnification projection images of cultured cells representing various time intervals in high Ca2+

medium.2Average number of desmosomes per cell. The presence of the cell nucleus was used to quantify the number of cells represented in any given field. About100 desmosomes were counted for each category.3Length of individual desmosomes were measured and binned into four size categories. Percentages refer to total population from a given row (that is, celltype and time point) and each represents an average of about 100 desmosomes.4Each desmosome was assigned a score from 0–3, where 0 corresponds to no intermediate filaments visible near the desmosome, 1 corresponds toambiguous density near the intracellular plaque, 2 corresponds to a small number of intermediate filament bundles, and 3 corresponds to strong intermediatefilament bundles associated with the plaque. Each population included B50 desmosomes.



desmosomes, the intermediate filament bundles were morestrongly visible on one side of a given desmosome, undoubt-edly because they run at right angles to the plane of the sectionin the opposing cell. This observation suggests that there maybe coordination between the intermediate filament networksin adjacent cells, which might be mediated in some, as yetundefined, way by the desmosomal architecture. Althoughmiscellaneous cellular constituents are excluded from theimmediate vicinity of wild-type desmosomes, microtubulesand ribosomes were seen in close proximity to the desmo-somes in knockout cells (Figure 5; Figures S2, S3), probablyreflecting their failure to build an organized inner plaque.

Immunolabeling of desmosomal components

Immunoelectron microscopy was used to localize individualproteins in wild type and PG-knockout desmosomes. First,several different antibodies were used to localize PG in wild-type desmosomes (Figure 6a–c) and to verify its absencein PG�/� cells (Figure 6e–g). Second, we confirmed that,in these long-term keratinocyte cultures, antibodies againstb-catenin labeled desmosomes in PG�/� but not PGþ /þ cells(Figure 6d and h). In adherens junctions, PG and b-catenincoexist (Peifer et al., 1992) and presumably compete forbinding to the cytoplasmic tail of the relevant cadherins. In

desmosomes, PG normally wins this competition andexcludes b-catenin (Lewis et al., 1997; Bierkamp et al.,1999), but in the absence of PG, b-catenin apparently hassufficient affinity for the cytoplasmic tails of desmosomalcadherins to account for their observed clustering even in theabsence of PG.

We next addressed whether absence of PG affected thedisposition of desmoplakin within the desmosomal plaques.In particular, we used antibodies specific for the desmoplakinN- or C-terminus to determine not only the relative amount ofdesmoplakin, but also its orientation within the plaque. InPGþ /þ keratinocytes, the desmoplakin N-terminus waslocalized to the outer plaque (Figure 6k), whereas C-terminallabeling was found in the inner plaque where intermediatefilaments join the desmosome (Figure 6j), as expected fromearlier studies (North et al., 1999). Interestingly, we foundthat this general orientation was maintained in PG�/�

desmosomes (Figure 6n and o) even though they lackedconnections to the intermediate filament network. Specifi-cally, in a statistical analysis of the distribution of desmopla-kin labels (Figure 6l), we found that the mean distance of thedesmoplakin N-terminus from the central midline wasB17 nm in both PGþ /þ (n¼100, SEM¼ 0.77) and PG�/�

(n¼41, SEM¼1.3) desmosomes. Although the C-terminus

a b c

fed

g h i

PG

–/–

PG

+/+

Figure 4. Projection images of desmosomes. (a) Newborn mouse skin. (b) Primary culture of keratinocytes from newborn mouse skin. (c) PGþ /� cell culture.

(d–f) PGþ /þ cell culture. (g–i) PG�/� cell culture. Note that magnifications are variable, but that all scale bars correspond to 100 nm. The outer dense plaque

(ODP, large arrowheads) and inner dense plaque (IDP, small arrowheads) are labeled in panel (f) and indicated in panels (d), (e), and (i). Intermediate filaments

are indicated by arrows in panel (d). The greater section thickness used for tomography and the retention of soluble cytoplasmic components in freeze-

substituted samples cause features to appear less distinct relative to projection images in some previous publications. Similarly, the appearance of cadherins

in the intercellular region depends on the precise angle of view, with a better defined midline produced when the section is cut precisely normal to the

membrane plane. Taking this effect into account, the intercellular cadherin network appears similar in all these cells, but the PG�/� desmosomes are

characterized by a narrower, less dense cytoplasmic plaque with severely reduced connections to the intermediate filament network.




was decidedly further from the membrane in both the cases,there was a significant difference in the distance measured inwild type versus knockout (Po0.02): 70 nm in PGþ /þ

(n¼83, SEM¼2.6) and 60 nm in PG�/� (n¼59, SEM¼ 3.6)desmosomes, respectively. Furthermore, we observed that thenumber of desmoplakin molecules localized to PG�/�

desmosomes was 2–3 times lower compared with PGþ /þ

desmosomes, which is consistent with previous biochemicaldata from whole-cell lysates (Caldelari et al., 2001; Yin et al.,2005a). These findings indicate that, in PG-knockout desmo-somes, cadherins, plakophilin, and/or b-catenin continue toorient desmoplakin correctly within the cytoplasmic plaque.However, the location of the desmoplakin C-terminus isaltered, perhaps reflecting the defect in anchoring intermedi-ate filaments to PG�/� desmosomes. Specifically, an apparentshortening of the desmoplakin molecule may be due to a lackof tension within this inner dense plaque that normallyaccompanies its linkage to the cytoskeleton.

Finally, we investigated the distribution of plakophilin-1 inwild type and knockout cells. Although labeling was present indesmosomes in both cell types, we observed an B6-folddecrease in the number of plakophilin-1 labels associatedwith PG�/� desmosomes compared with PGþ /þ desmosomes(Figure 6i and m). We also calculated the ratio of plakophilinlabels in desmosomal versus non-desmosomal regions of theplasma membrane. In wild-type cells, this ratio was 1.5,indicating that plakophilin preferentially associated withdesmosomes. In contrast, the ratio was 0.17 in knockout cells,reflecting a comparable density of plakophilin molecules

associating with non-desmosomal regions of the membrane,but a much lower density at the desmosomes. This result isconsistent with the membrane localization of plakophilinimmunofluorescence in both cell types and suggests thatplakophilin associates with the membrane prior to incorpora-tion into the desmosomal plaque.

DISCUSSIONIn this report, we have compared desmosomes produced bylong-term keratinocyte cultures derived from wild type andPG-knockout mice. Although this knockout has beencharacterized in previous studies (Bierkamp et al., 1996,1999; Ruiz et al., 1996; Caldelari et al., 2001; Yin et al.,2005a), electron tomography allowed us to evaluate thespecific structural defect in desmosomal architecture. Inparticular, we documented that 15–48 hours after inducingdesmosome formation with the Ca2þ switch, cadherin–cad-herin interactions in the intercellular gap of PG�/� junctionsappear to be similar to those in wild-type desmosomes,indicating that cadherin molecules are clustered normallywithin the cell membrane even in the absence of PG.However, we found that this knockout results in a thinner,sparser outer dense plaque and furthermore that the innerdense plaque seems to be disorganized or missing altogether.In addition, the desmosomes seen by electron tomographyare defective in anchoring the intermediate filament networkto the membrane and the cohesion of corresponding cellcultures is more susceptible to mechanical stress. Immuno-gold labeling confirmed previous reports that b-catenin was

PG

–/–

PG

+/+

a b c

d e f

IDPODP

Figure 5. Tomographic imaging of desmosomes. (a–c) PGþ /þ . (d–f) PG�/�. Grayscale images correspond to 7-A thick slices through the tomographic

volume and color has been applied to objects representing the three-dimensional segmentation of selected components of the desmosome. The outer dense

plaque (ODP) and inner dense plaque (IDP) are indicated in panel (a) and appear to be substantially more developed in wild type desmosomes compared to the

knockout. Cell membranes are shown in cyan, desmoplakin in yellow and intermediate filaments in blue. In knockout desmosomes, ribosomes (red) and

microtubules (purple) are seen more frequently in the region of the IDP. Close-up views (c, f) show that extracellular cadherin interactions are comparable

in the two cell types, although the corresponding densities are not as well preserved as in epidermal tissue (for example, Figures 3a and 7). Animated

representations of these tomographic reconstructions are shown in Movies S1 and S2.



present in knockout desmosomes (Bierkamp et al., 1999) andpresumably responsible for cadherin clustering. However, wealso documented substantially reduced levels of desmoplakinand plakophilin-1, suggesting that recruitment of plaquecomponents is secondary to cadherin clustering and depen-dent on unique properties of PG. The consequent defect inanchoring intermediate filaments provides a definitivestructural explanation for the failure of PG�/� tissue adhesion.Thus, our observations suggest an orderly sequence fordesmosome assembly initiated by cadherin clustering,mediated by either PG or b-catenin, followed by recruitmentof plakophilin and desmoplakin to the intracellular plaque,

and consummated by coupling to the intermediate filamentnetwork.

Interactions in the outer dense plaque

In wild-type desmosomes, the outer dense plaque iscomposed of PG, plakophilin, and the N-terminal domainsof desmoplakin. Specifically, the ARM repeats of PG interactwith the N-terminal domain of desmoplakin, thus anchoringit to the desmosome (Kowalczyk et al., 1997). In PG�/�

desmosomes, b-catenin is not expected to bind desmoplakin;instead, the presence of desmoplakin most likely reflects itsinteraction with desmocollin-1a (Troyanovsky et al., 1994;

PG 11E4 PG 1407 PG 1408 β-CateninP

G +

/+P

G –

/–P

G –

/–P

G +

/+

Plakophilin1 DP-C DP-N

PG+/+ PG–/–

5040302010

10

5 15 25 35DP N-term

DP C-term

nm

nm

%

504030

50 70 90 120

20

30

10

%

a b c d

e f g h

i j k l

m n o

Figure 6. Immunogold localization of desmosomal proteins. Cell type is indicated along the left-hand margin, whereas antibody targets are indicated at the

top and bottom. Lower frequency gold labels are marked with arrowheads. (a–c) Three different PG antibodies resulted in robust labeling in PGþ /þ

desmosomes, but did not label PG�/� desmosomes at all (e–g). (d, h) b-Catenin was absent from PGþ /þ desmosomes, but present in PG�/� desmosomes.

(i, m) Plakophilin-1 labeling was observed for both cells types, although levels were considerably lower for the knockout cells. (j, k, n, o) Two different

antibodies targeted the N- and C-terminal domains of desmoplakin. Although both localized to the cytoplasmic plaque, the N-terminus was significantly closer

to the membrane than the C-terminus. Also, fewer desmoplakin antibodies were observed in knockout cells. (l) Histograms illustrate the distribution of

antibodies for the two ends of desmoplakin. Student’s t-test indicated a significant difference between N- and C-terminal locations in both cell lines (Po0.01)

and also a significant difference between C-terminal locations in PGþ /þ and PG�/� cells (Po0.02).




Smith and Fuchs, 1998), albeit at the lower levels as indicatedby our immunolabeling studies and supported by previousanalyses of protein levels in cell lysates (Caldelari et al.,2001; Yin et al., 2005a). Plakophilin-1 is reported to binddesmoglein-1, desmocollin-1, and desmoplakin (Kapprellet al., 1988; Smith and Fuchs, 1998; Kowalczyk et al.,1999; Hatzfeld et al., 2000), but our observation of a sixfoldreduction of plakophilin-1 in PG�/� desmosomes suggeststhat the desmoplakin N-terminal domain plays a larger rolethan desmosomal cadherins in recruiting plakophilin-1 to theouter dense plaque.

Linkage with the inner dense plaque

Linkage of cadherins, PG, and plakophilin in the outer denseplaque to the intermediate filament network occurs via aninner plaque, which is composed primarily of desmoplakin.We have extended previous immunolabeling of desmoplakinin wild-type desmosomes (North et al., 1999) to demonstratethe disposition of desmoplakin in knockout desmosomes. Inboth wild type and knockout cells, the N-terminal domainof desmoplakin lies close to the membrane, whereas theC-terminal domain is 40–50 nm away at the inner margin ofthe cytoplasmic plaque. This disposition is consistent with therole of this C-terminal domain in anchoring intermediatefilaments (Stappenbeck et al., 1993; Kouklis et al., 1994;Meng et al., 1997; Smith and Fuchs, 1998; Choi et al., 2002),which localize to the same area. In our tomograms fromwild-type keratinocytes, desmoplakin appears as strands ofdensity running across this region. In PG�/� desmosomes,these strands are either absent or much attenuated, which isconsistent with the 2–3� reduction in immunolabelingof desmoplakin in the knockout. Figure 7 shows three-dimensional segmentation of a desmosome from wild-typemouse epidermis. Several consecutive slices are shown inFigure 7b–e in which individual strands are visible connectingthe lower membrane to the intermediate filaments running atthe very bottom of the panels (see also Movie S3) and thesegmentation in Figure 7f gives a general sense of the three-dimensional distribution of these desmoplakin densities. Onthe basis of their size and shape, as well as previous evidencefor a dimeric assembly (Green et al., 1992), we hypothesizethat individual strands correspond to desmoplakin dimers.These appear to interact with the lateral edge of individualintermediate filaments frayed from a larger bundle that runsroughly parallel to the membrane surface.

We found that the number of desmosomes observed inknockout cells depends strongly on the preparation methodused for electron microscopy in a way that reflects avulnerability of these mutant desmosomes to mechanicalstress. Even though immunofluorescence images indicate anintact intermediate filament network in PG�/� cells, electrontomography reveals that desmosomal connections are defec-tive. Indeed, under conditions of increased mechanical stress,PG�/� cells had a preponderance of desmosome-associatedmembrane fragments torn from an adjacent cell (Figure 3),graphically illustrating the consequences of this defect evenwhen the cadherin-mediated intercellular bond remainsintact. Failure of intermediate filament coupling offers a

compelling explanation for the susceptibility of this knockoutcell line to mechanical stress and for the embryonic lethalityof PG and desmoplakin knockouts in mice, although theearlier and more severe phenotype of the latter (Gallicanoet al., 1998) suggests additional critical roles of desmoplakinin desmosome assembly.

Our work extends previous observations of desmosomes inepidermis of PG-knockout mice (Bierkamp et al., 1999).These authors described increased intercellular spacesbetween cells in the upper and lower spinous layers as wellas missing cytoplasmic plaques and few associated inter-mediate filament bundles, although their conventionalelectron micrographs were not able to define the detaileddesmosomal architecture. In addition, they reported brokendesmosomes with fragments left attached to the neighboringcell, analogous to the images we present in Figure 3. On theother hand, transcellular networks of intermediate filamentsand immunoprecipitation of desmoglein by keratin-14antibodies in knockout cell cultures seem to reflect intactconnections between intermediate filaments and cadherins(Caldelari et al., 2001). Although these latter observationsappear to be inconsistent with our observations by tomo-graphy, it is possible that longer periods of desmosomeformation produce a subpopulation of knockout desmosomesthat form limited associations with the intermediate filamentnetwork (c.f., Table 1).

Regulation of desmoplakin binding to intermediate filaments

Our data are consistent with a sequential model of desmo-some assembly: clustering of cadherins by PG followed by

IDP

ODP

ODP

IDP

a

b c d e f

Figure 7. Segmentation of the inner dense plaque of a desmosome from

wild-type mouse epidermis. (a) Overview of this desmosome, which has

relatively low density of cadherins and intracellular plaque proteins, making

segmentation easier. Segmented area is framed; the inner dense plaque (IDP)

and outer dense plaque (ODP) are indicated. (b–e) Sections from lower,

middle and upper portions of the tomogram framed in panel (a).

(f) Segmentation of the densities for the membrane (cyan), cadherins

(purple and pink), desmoplakin (yellow and orange), and intermediate

filaments (blue).



recruitment of plaque proteins followed by linkage ofintermediate filaments. To coordinate this assembly, bindingof desmoplakin to intermediate filaments must be regulatedin order to prevent binding at non-desmosomal sites andperhaps also to control desmosome assembly/disassemblyduring cell division or migration. Although the mechanismshave not been precisely defined, PG appears to play a pivotalrole. The observed distance between desmoplakin N- andC-termini in desmosomes is considerably smaller than thelength of isolated, metal-shadowed desmoplakin molecules(up to 180 nm, O’Keefe et al., 1989). This discrepancysuggests that the central a-helical rod of desmoplakin (Greenet al., 1992) is either highly flexible or folded up within theintracellular plaque. Thus, one mechanism for regulatingassembly could be an autoinhibitory interaction betweenN- and C-terminal domains, as seen for example in smooth-muscle myosin (Burgess et al., 2007). Recruitment to thedesmosome could disrupt this interaction, either by sterichindrance of PG binding to the N-terminal domain or byPG-mediated phosphorylation. With respect to phosphory-lation, serine phosphorylation of the C-terminal GSR domainof desmoplakin has been implicated in the regulation ofintermediate filament binding (Stappenbeck et al., 1994); alsoepidermal growth factor stimulation results in tyrosinephosphorylation of PG, loss of desmoplakin from desmo-somes, and decreased adhesive strength (Yin et al., 2005a).Furthermore, evidence is building for the participation of PGin a signaling pathway that mediates the response to theautoimmune disease pemphigus, which results in dramati-cally weakened desmosomal adhesion (Caldelari et al., 2001;Williamson et al., 2006; de Bruin et al., 2007).

A rather different view is that productive binding ofintermediate filaments requires a threshold concentration ofdesmoplakin molecules. It is clear from our tomograms thatthe inner dense plaque in wild-type cells consists of a proteinnetwork that is dense enough to exclude extraneous cellularconstituents such as ribosomes and microtubules, reflecting avery high local concentration of desmoplakin within thisregion. In PG�/� desmosomes, this network is absent andimmunolabeling indicated reduced amounts of desmoplakinconsistent with a considerably lower local concentration thatappears to be ineffective in binding intermediate filaments.Such a threshold requirement could reflect weak binding ofdesmoplakin to intermediate filaments, which is overcome bya binding cooperativity manifested only in a large ensemble.Thus, although transient, individual interactions could occurin the bulk cytoplasm, only after clustering molecules withinthe desmosomal plaque would desmoplakin be effective inanchoring the cytoskeleton. A similar mechanism has beenproposed as the basis for cadherin–cadherin selectivity, withthe sum of numerous weak interactions explaining theirspecificity as well as the adhesive strength of the overallensemble (Chen et al., 2005).

MATERIALS AND METHODSCell culture, genotyping, and immunoblotting

Long-term cultures of PGþ /þ and PG�/� keratinocytes have been

described previously (Caldelari et al., 2000, 2001). Primary cultures

of wild-type keratinocytes (Hennings and Holbrook, 1983) were

established by treating mouse skin with dispase for 4–12 hours on

ice, which allowed separation of the epidermis from the dermis. Use

of mice was approved by the authors’ Institutional Review Board.

The epidermis was digested with 0.5% trypsin at 37 1C, treated with

25 mg ml�1 DNAse, and sheared by vigorous pipetting with a Pasteur

pipette. The resulting solution was filtered through several layers of

gauze and washed with DMEM and centrifuged at 1,000 r.p.m. to

collect primary keratinocytes. Long-term cultures were grown in

CnT2 growth medium (CellnTec Advanced Cell Systems, Bern,

Switzerland) and primary keratinocytes were grown in keratinocyte

growth medium (Cambrex Bio Science, Walkersville MD) for 3–5

days in order to reach confluence in the absence of calcium. In the

case of the knockout, cells are hyperproliferative and, to compen-

sate, they were seeded at a lower concentration than that used for

wild-type cells (Williamson et al., 2006). Polycarbonate Transwell

filters (Corning Inc., Corning, NY) were used for electron tomo-

graphy and plastic Petri dishes were used for other experiments.

After reaching confluence, desmosome formation was initiated by

adding 1.4 mM CaCl2 to the culture media.

For PCR, primers A(neo)f 50-GCCTTCTATCGCCTTCTTGAC-30

and B(ex5)r 50-CTCAGGAGTTAGGAGCACTG-30 were used to

recognize the neomycin cassette in the targeted allele; C(ex3)f

50-AGCTGCTCAACGATGAGGAC-30 and D(ex4)r 50-AGCATTCGG

ACTAGGGCAGG-30 were used to recognize exon-3 and exon-4

of the wild-type gene. For western blots, cells were washed with

phosphate-buffered saline buffer, scraped off of Petri dishes, and

collected by centrifugation. The resulting cell pellet was resus-

pended in radioimmunoprecipitation buffer (50 mM Tris-HCl,

150 mM NaCl, 1 mM PMSF, 1 mM EDTA, 5mg ml�1 Aprotinin,

5mg ml�1 Leupeptin, 1% Triton X-100, 1% sodium deoxycholate,

0.1% SDS, pH 7.4), pipetted vigorously through a 22-G needle, and

incubated on ice for 1 hour. Cell lysates were centrifuged at 15,000 g

for 20 minutes at 4 1C and the resulting supernatant was run on 10%

SDS polyacrylamide gels, transferred to polyvinylidene difluoride

membranes, and blotted with antibodies for PG (5172, mouse

monoclonal), E-cadherin (mouse monoclonal; BD Biosciences, San

Jose, CA) and b-catenin (mouse monoclonal from BD Biociences,

Franklin Lakes, NJ).

Immunofluorescence and immunogold labeling

For immunofluorescence light microscopy, cells were grown on

glass slides, fixed with methanol at �20 1C, and incubated for 1 hour

on ice with primary antibodies listed above as well as the following:

desmoplakin (2.15, mouse monoclonal), plakophilin (mouse mono-

clonal; Abcam Inc., Cambridge, MA), and keratin-14 (rabbit

polyclonal; Covance Inc., Princeton, NJ). After washing with

phosphate-buffered saline buffer, cells were incubated with fluores-

cently labeled anti rabbit (cy3) or anti mouse (Alexa488, Texas Red)

secondary antibodies for 1 hour on ice and visualized at room

temperature using a � 63 Plano Apochromat objective with a Zeiss

Axoplan 2 model LSM510 confocal fluorescence microscope.

For immunogold electron microscopy, we used primary anti-

bodies NW6 (desmoplakin C-terminal, rabbit polyclonal), NW161

(desmoplakin N-terminal, rabbit polyclonal), 11E4 (PG, mouse

monoclonal), 1407, and 1408 (PG chicken polyclonal), which were

a gift from Dr Kathleen J Green Laboratory (Northwestern University,

Feinberg School of Medicine). In addition, b-catenin (BD Biosciences)




and plakophilin-1 (Abcam Inc.) antibodies were obtained commer-

cially. Cells were grown on Petri dishes and then fixed for 2 hours

with 2% paraformaldehyde and 0.2% glutaraldehyde. Cells were

scraped with a Teflon spatula and centrifuged to form a pellet that

was infused with 2.3 M sucrose in phosphate-buffered saline and

frozen on a microtome chuck prior to cryosectioning. Sections

(80-nm thick) were cut with a cryo-ultramicrotome and collected on

nickel EM grids. For labeling, sections were incubated overnight at

4 1C with primary antibodies diluted in 1% BSA in phosphate-

buffered saline. After several washes to remove unbound primary

antibodies, sections were incubated with 10 nm gold-conjugated

protein-A for 30 minutes at room temperature. Monoclonal antibodies

required additional 30 minutes of incubation with rabbit or chicken

IgG to bridge between primary antibody and protein-A gold. In the

case of b-catenin, the primary antibody was directly labeled with

10 nm gold-conjugated antimouse IgG antibody. After labeling

reactions and washes, samples were further fixed with 1%

glutaraldehyde, stained with uranyl acetate at neutral pH, and

embedded in a mixture of 0.4% uranyl acetate and 1.8% methyl

cellulose.

Electron tomography

For electron tomography, cells were grown on Transwell membranes

and prepared by high-pressure freezing followed by freeze substitu-

tion. High-pressure freezing planchettes were loaded with 2� 2-mm

pieces of Transwell membranes, packed with 15% dextran in culture

medium, and frozen within 2 minutes after removal from growth

medium. For freeze substitution, the planchettes were incubated in

acetone containing 1% OsO4 at �90 1C for 48–72 hours, followed by

�60 1C for 12 hours, and �30 1C for 12 hours; transitions between

these temperatures were made at 5 1C hour�1. After warming

samples to room temperature, they were washed with pure acetone,

infiltrated with LX112 Epon resin, and polymerized at 60 1C.

Sections were stained with 1–3% uranyl acetate and Sato Lead

stain, and finally coated with 5 or 10 nm colloidal gold particles for

use as fiducial markers.

Electron micrographs of samples tilted between �701 and þ 701

at 1–21 intervals were recorded at B1mm defocus with 25–68 k

magnification on a 4 k� 4 k charge-coupled device camera (Tietz

Video Imaging Processing System GmbH, Gauting Germany) using

SerialEM (Mastronarde, 2005). A second tilt series of the same area

was collected after manually rotating the microscope grid by B901.

Images were binned 2� to produce sampling intervals of

0.25–0.68 nm, depending on magnification. Prior to data collection,

sections were stabilized by pre-irradiation with B105 electrons nm�2

and the cumulative dose for the dual-axis tilt series was B105 elec-

trons nm�2. Dual-axis tomographic reconstructions were calculated

using IMOD (Kremer et al., 1996) and features in selected

tomograms were segmented for presentation using Amira (Mercury

Computer Systems Inc., Chelmsford, MA).

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSWe thank Dr Kathleen Green for kindly providing numerous antibodies usedfor this work. We would like to acknowledge the New York Structural BiologyCenter for providing electron microscopy facilities used in this work. Fundingfor this study was provided by NIH Grant R01 GM071044 to DLS.

SUPPLEMENTARY MATERIAL

Figure S1. Influence of different methods for EM preparation.

Figure S2. Tomographic images of desmosomes.

Figure S3. A gallery of tomographic slices from PGþ /þ and PG�/� cells.

Movie S1. Animated series of slices through the tomographic volume inFigure 5a.

Movie S2. Animated series of slices through the tomographic volume inFigure 5d.

Movie S3. Animated series of slices through the tomographic volume inFigure 7a.

REFERENCES

Al-Amoudi A, Diez DC, Betts MJ, Frangakis AS (2007) The moleculararchitecture of cadherins in native epidermal desmosomes. Nature450:832–7

Bierkamp C, McLaughlin KJ, Schwarz H, Huber O, Kemler R (1996)Embryonic heart and skin defects in mice lacking plakoglobin. Dev Biol180:780–5

Bierkamp C, Schwarz H, Huber O, Kemler R (1999) Desmosomal localizationof beta-catenin in the skin of plakoglobin null-mutant mice. Develop-ment 126:371–81

Bonne S, Gilbert B, Hatzfeld M, Chen X, Green KJ, van Roy F (2003) Definingdesmosomal plakophilin-3 interactions. J Cell Biol 161:403–16

Bornslaeger EA, Corcoran CM, Stappenbeck TS, Green KJ (1996) Breaking theconnection: displacement of the desmosomal plaque protein desmo-plakin from cell–cell interfaces disrupts anchorage of intermediatefilament bundles and alters intercellular junction assembly. J Cell Biol134:985–1001

Bornslaeger EA, Godsel LM, Corcoran CM, Park JK, Hatzfeld M, KowalczykAP et al. (2001) Plakophilin 1 interferes with plakoglobin binding todesmoplakin, yet together with plakoglobin promotes clustering ofdesmosomal plaque complexes at cell-cell borders. J Cell Sci 114:727–38

Burgess SA, Yu S, Walker ML, Hawkins RJ, Chalovich JM, Knight PJ (2007)Structures of smooth muscle myosin and heavy meromyosin in thefolded, shutdown state. J Mol Biol 372:1165–78

Caldelari R, de Bruin A, Baumann D, Suter MM, Bierkamp C, Balmer V et al.(2001) A central role for the armadillo protein plakoglobin in theautoimmune disease pemphigus vulgaris. J Cell Biol 153:823–34

Caldelari R, Suter MM, Baumann D, De Bruin A, Muller E (2000) Long-termculture of murine epidermal keratinocytes. J Invest Dermatol 114:1064–5

Chen CP, Posy S, Ben-Shaul A, Shapiro L, Honig BH (2005) Specificity ofcell–cell adhesion by classical cadherins: critical role for low-affinitydimerization through beta-strand swapping. Proc Natl Acad Sci USA102:8531–6

Chen X, Bonne S, Hatzfeld M, van Roy F, Green KJ (2002) Protein bindingand functional characterization of plakophilin 2. Evidence for its diverseroles in desmosomes and beta-catenin signaling. J Biol Chem 277:10512–22

Choi HJ, Park-Snyder S, Pascoe LT, Green KJ, Weis WI (2002) Structures oftwo intermediate filament-binding fragments of desmoplakin reveal aunique repeat motif structure. Nat Struct Biol 9:612–20

Choi HJ, Weis WI (2005) Structure of the armadillo repeat domain ofplakophilin 1. J Mol Biol 346:367–76

de Bruin A, Caldelari R, Williamson L, Suter MM, Hunziker T, Wyder M et al.(2007) Plakoglobin-dependent disruption of the desmosomal plaque inpemphigus vulgaris. Exp Dermatol 16:468–75

Gallicano GI, Kouklis P, Bauer C, Yin M, Vasioukhin V, Degenstein L et al.(1998) Desmoplakin is required early in development for assembly ofdesmosomes and cytoskeletal linkage. J Cell Biol 143:2009–22

Garrod DR, Merritt AJ, Nie Z (2002) Desmosomal adhesion: structural basis,molecular mechanism and regulation. Mol Membr Biol 19:81–94

Green KJ, Stappenbeck TS, Parry DA, Virata ML (1992) Structure ofdesmoplakin and its association with intermediate filaments. J Dermatol19:765–9



Hatsell S, Medina L, Merola J, Haltiwanger R, Cowin P (2003) Plakoglobin isO-glycosylated close to the N-terminal destruction box. J Biol Chem278:37745–52

Hatzfeld M (2007) Plakophilins: multifunctional proteins or just regulators ofdesmosomal adhesion? Biophys Biochym Acta 1773:69–77

Hatzfeld M, Haffner C, Schulze K, Vinzens U (2000) The function ofplakophilin 1 in desmosome assembly and actin filament organization.J Cell Biol 149:209–22

He W, Cowin P, Stokes DL (2003) Untangling desmosomal knots withelectron tomography. Science 302:109–13

Hennings H, Holbrook KA (1983) Calcium regulation of cell–cell contactand differentiation of epidermal cells in culture. An ultrastructural study.Exp Cell Res 143:127–42

Huber AH, Weis WI (2001) The structure of the beta-catenin/E-cadherincomplex and the molecular basis of diverse ligand recognition by beta-catenin. Cell 105:391–402

Kapprell HP, Owaribe K, Franke WW (1988) Identification of a basic proteinof Mr 75,000 as an accessory desmosomal plaque protein in stratifiedand complex epithelia. J Cell Biol 106:1679–91

Kouklis PD, Hutton E, Fuchs E (1994) Making a connection: direct bindingbetween keratin intermediate filaments and desmosomal proteins. J CellBiol 127:1049–60

Kowalczyk AP, Bornslaeger EA, Borgwardt JE, Palka HL, Dhaliwal AS,Corcoran CM et al. (1997) The amino-terminal domain of desmoplakinbinds to plakoglobin and clusters desmosomal cadherin–plakoglobincomplexes. J Cell Biol 139:773–84

Kowalczyk AP, Hatzfeld M, Bornslaeger EA, Kopp DS, Borgwardt JE,Corcoran CM et al. (1999) The head domain of plakophilin-1 binds todesmoplakin and enhances its recruitment to desmosomes. Implicationsfor cutaneous disease. J Biol Chem 274:18145–8

Kremer JR, Mastronarde DN, McIntosh JR (1996) Computer visualization ofthree-dimensional image data using IMOD. J Struct Biol 116:71–6

Leung CL, Green KJ, Liem RK (2002) Plakins: a family of versatile cytolinkerproteins. Trends Cell Biol 12:37–45

Lewis JE, Wahl JK III, Sass KM, Jensen PJ, Johnson KR, Wheelock MJ (1997)Cross-talk between adherens junctions and desmosomes depends onplakoglobin. J Cell Biol 136:919–34

Mastronarde DN (2005) Automated electron microscope tomography usingrobust prediction of specimen movements. J Struct Biol 152:36–51

McGrath JA, McMillan JR, Shemanko CS, Runswick SK, Leigh IM, Lane EBet al. (1997) Mutations in the plakophilin 1 gene result in ectodermaldysplasia/skin fragility syndrome. Nat Genet 17:240–4

McKoy G, Protonotarios N, Crosby A, Tsatsopoulou A, Anastasakis A, CoonarA et al. (2000) Identification of a deletion in plakoglobin inarrhythmogenic right ventricular cardiomyopathy with palmoplantarkeratoderma and woolly hair (Naxos disease). Lancet 355:2119–24

Meng JJ, Bornslaeger EA, Green KJ, Steinert PM, Ip W (1997) Two-hybridanalysis reveals fundamental differences in direct interactions betweendesmoplakin and cell type-specific intermediate filaments. J Biol Chem272:21495–503

Miravet S, Piedra J, Castano J, Raurell I, Franci C, Dunach M et al. (2003)Tyrosine phosphorylation of plakoglobin causes contrary effects on itsassociation with desmosomes and adherens junction components and

modulates beta-catenin-mediated transcription. Mol Cell Biol 23:7391–402

Muller-Tidow C, Steffen B, Cauvet T, Tickenbrock L, Ji P, Diederichs S et al.(2004) Translocation products in acute myeloid leukemia activate theWnt signaling pathway in hematopoietic cells. Mol Cell Biol24:2890–904

Nelson WJ, Nusse R (2004) Convergence of Wnt, beta-catenin, and cadherinpathways. Science 303:1483–7

North AJ, Bardsley WG, Hyam J, Bornslaeger EA, Cordingley HC, TrinnamanB et al. (1999) Molecular map of the desmosomal plaque. J Cell Sci112:4325–36

O’Keefe EJ, Erickson HP, Bennett V (1989) Desmoplakin I and desmoplakin II.Purification and characterization. J Biol Chem 264:8310–8

Peifer M, McCrea PD, Green KJ, Wieschaus E, Gumbiner BM (1992) Thevertebrate adhesive junction proteins beta-catenin and plakoglobin andthe Drosophila segment polarity gene armadillo form a multigene familywith similar properties. J Cell Biol 118:681–91

Ruiz P, Brinkmann V, Ledermann B, Behrend M, Grund C, Thalhammer Cet al. (1996) Targeted mutation of plakoglobin in mice reveals essentialfunctions of desmosomes in the embryonic heart. J Cell Biol 135:215–25

Smith EA, Fuchs E (1998) Defining the interactions between intermediatefilaments and desmosomes. J Cell Biol 141:1229–41

Stappenbeck TS, Bornslaeger EA, Corcoran CM, Luu HH, Virata ML, Green KJ(1993) Functional analysis of desmoplakin domains: specification of theinteraction with keratin versus vimentin intermediate filament networks.J Cell Biol 123:691–705

Stappenbeck TS, Lamb JA, Corcoran CM, Green KJ (1994) Phosphorylation ofthe desmoplakin COOH terminus negatively regulates its interactionwith keratin intermediate filament networks. J Biol Chem 269:29351–4

Troyanovsky RB, Chitaev NA, Troyanovsky SM (1996) Cadherin binding sitesof plakoglobin: localization, specificity and role in targeting to adheringjunctions. J Cell Sci 109(Pt 13):3069–78

Troyanovsky SM, Troyanovsky RB, Eshkind LG, Leube RE, Franke WW (1994)Identification of amino acid sequence motifs in desmocollin, adesmosomal glycoprotein, that are required for plakoglobin bindingand plaque formation. Proc Natl Acad Sci USA 91:10790–4

Williamson L, Raess NA, Caldelari R, Zakher A, de Bruin A, Posthaus H et al.(2006) Pemphigus vulgaris identifies plakoglobin as key suppressor ofc-Myc in the skin. EMBO J 25:3298–309

Witcher LL, Collins R, Puttagunta S, Mechanic SE, Munson M, Gumbiner Bet al. (1996) Desmosomal cadherin binding domains of plakoglobin.J Biol Chem 271:10904–9

Yin T, Getsios S, Caldelari R, Godsel LM, Kowalczyk AP, Muller EJ et al.(2005a) Mechanisms of plakoglobin-dependent adhesion: desmosome-specific functions in assembly and regulation by epidermal growth factorreceptor. J Biol Chem 280:40355–63

Yin T, Getsios S, Caldelari R, Kowalczyk AP, Muller EJ, Jones JC et al. (2005b)Plakoglobin suppresses keratinocyte motility through both cell–celladhesion-dependent and -independent mechanisms. Proc Natl AcadSci USA 102:5420–5

Yin T, Green KJ (2004) Regulation of desmosome assembly and adhesion.Semin Cell Dev Biol 15:665–77




plakoglobin is required for effective intermediate filament

Documents