BiochemicalSociety

AnnualSym

posiumN

o.75

Cryoelectron tomography of isolated desmosomesGethin Rh. Owen*, Devrim Acehan†, K.D. Derr*, William J. Rice* and David L. Stokes*†1

*New York Structural Biology Center, 89 Convent Avenue, New York, NY 10027, U.S.A., and †Skirball Institute of Biomolecular Medicine, Department of Cell

Biology, New York University School of Medicine, New York, NY 10016, U.S.A.

AbstractDesmosomes are a complex assembly of protein molecules that form at the cell surface and mediate cell–celladhesion. Much is known about the composition of desmosomes and there is an established consensusfor the location of and interactions between constituent proteins within the assembly. Furthermore, X-raycrystallography has determined atomic structures of isolated domains from several constituent proteins.Nevertheless, there is a lack of understanding about the architecture of the intact assembly and the physicalprinciples behind the adhesive strength of desmosomes therefore remain vague. We have used electrontomography to address this problem. In previous work, we investigated the in situ structure of desmosomesfrom newborn mouse skin preserved by freeze-substitution and imaged in resin-embedded thin sections. Inour present work, we have isolated desmosomes from cow snout and imaged them in the frozen unstainedstate. Although not definitive, the resulting images provide support for the irregular groupings of cadherinmolecules seen previously in mouse skin.

IntroductionEpithelia are specialized cellular assemblies that line theinternal and external surfaces of most tissues. By adoptingthis topology, epithelia provide selective permeability andprotect surrounding tissues from mechanical and chemicalinsult. Adhesive junctions are prevalent between thecells that compose epithelia: both adherens junctions anddesmosomes serve to maintain cellular organization and toresist mechanical forces [1]. Desmosomes are particularlyimportant in tissues that experience mechanical challenge,such as skin and heart, where genetic or autoimmune defectsare manifested in disease states [2,3]. In addition to a structuralrole, other reports have also indicated that desmosomes canbe dynamic [4] and may play a role in cell–cell signalling[5].

Desmosome architectureStructurally, the desmosome is a discrete molecular assemblythat couples the membranes of two adjacent cells [6]. Thisassembly consists of a cluster of transmembrane proteinsanchored to an extended scaffold of intermediate filamentsthat propagate across the entire epithelium. The anchoringis carried out by a heterogeneous assembly of proteins(plakoglobin, plakophilin and desmoplakin) that form aplaque at the intracellular surface of the membrane. Thetransmembrane proteins (desmoglein and desmocollin)belong to the cadherin family, which generally mediatescalcium-dependent cell–cell adhesion in vertebrate tissue [7].Cadherins are composed of an extracellular portion with five

Key words: cadherin, cell adhesion, desmosome, electron tomography, three-dimensional

reconstruction.1To whom correspondence should be addressed (email Stokes@saturn.med.nyu.edu).

Biochem. Soc. Trans. (2008) 36, 173–179; doi:10.1042/BST0360173

tandem Ig-like domains, a single transmembrane helix and acytosolic domain designed to interact with various proteinsthat compose the intracellular plaque. Each extracellulardomain consists of ∼110 amino acids [8] that adopt a β-sandwich fold with the topology of a Greek key [9,10]. Threecalcium ions are bound to the loop connecting successivedomains [11,12] via conserved sequence motifs. In this way,calcium binding confers rigidity to the domain interface andhas been shown to induce the entire extracellular domain toassume an extended structure [13].

Desmosomal cadherins are closely related to the betterstudied classical cadherins, which are also known as type Icadherins and are found primarily in adherens junctions. Bothtypes of cadherins possess a conserved tryptophan residuenear their N-terminus that has been shown to be critical toadhesivity [14]. However, desmosomal and classical cadherinsare functionally distinct. The former engage in heterophilicinteractions, whereas the latter employ homophilic inter-actions; they also use different cytoplasmic proteins to coupleto unique cytoplasmic elements: intermediate filaments andactin filaments respectively [1]. Nevertheless, a compellingmodel for adhesion involves the conserved tryptophanresidue and probably applies to both desmosomal and classi-cal cadherins. In particular, the side chain of Trp2 has been ob-served binding within a hydrophobic pocket in the first extra-cellular domain of a crystallographic neighbour, which pre-sumably corresponds to a partner cadherin from the apposingcell membrane [9,12]. However, alternative models have beenproposed based on different molecular contacts observed inother crystal forms [11,13] and on surface force measurement[15–17]. None of these models provides an explanation for theability of cadherins to recognize specific binding partners.Furthermore, there is little consensus for the location ofthe cis interactions between cadherins on the same cellsurface.

C©The Authors Journal compilation C©2008 Biochemical Society 173Bio

chem

ical

So

ciet

y T

ran

sact

ion

s

ww

w.b

ioch

emso

ctra

ns.

org

174 Biochemical Society Transactions (2008) Volume 36, part 2

Electron tomographyElectron microscopy has the potential to distinguish betweenalternative models for cadherin interactions. Using electrontomography, it is possible to visualize the three-dimensionalstructures of pleomorphic cellular assemblies [18]. Thistechnique involves recording two-dimensional projectionimages of the specimen over a wide range of tilt angles,typically ±70◦, followed by alignment and back projectionto create a three-dimensional representation of the originalobject [19]. The success of this technique depends on theability to preserve the macromolecular assembly in the elec-tron microscope during the extensive imaging procedure.In previous work, we used rapid freezing and freeze-substitution to prepare thin sections of resin-embeddedepidermis from newborn mouse skin [20]. The resultingelectron tomograms revealed gently curved shapes withinthe intercellular region of the desmosomes that closelyresembled the extracellular domain of classical cadherinmolecules as determined by X-ray crystallography [12].This X-ray model was fitted to the three-dimensionalmap from electron tomography, demonstrating how theproposed Trp2 interaction could mediate both cis and transinteractions. Furthermore, flexibility in the N-terminaltail would allow molecules to form irregular groupings,where enhanced intermolecular interactions would provideadditional mechanical strength to the junctional assembly.More recently, images were presented of human skin sectionspreserved in the frozen unstained state [21]. Although thesewere two-dimensional projection images and do not benefitfrom the three-dimensional analysis of tomographic imaging,the images show cadherins that appeared to be regularlyarranged with a 5 nm periodicity, rather than the groupingobserved by He et al. [20]. Furthermore, desmosomes appearto have a 7.5 nm periodicity in historical images of the guinea-pig heart [22,23]. In order to investigate whether preparativemethods were responsible for these apparent differences, wehave prepared isolated desmosomes from cow snout in orderto image them in the frozen unstained state using cryoelectrontomography.

Structure of in situ desmosomesInitially, we compared the in situ structure of desmosomesfrom neonatal mouse and cow snout epidermis usingelectron tomography on thin sections of freeze-substitutedresin-embedded tissue. Cow snouts were obtained from alocal slaughterhouse, and epidermis was removed from thesurface using a scalpel. Thin slices that included the stratumspinosum were immediately processed for high-pressurefreezing as follows. A 1 mm diameter tissue punch was usedto cut plugs from the epidermis, which were then loadedinto specimen planchets that had been dipped in hexadecene.Type-A planchets (Bal-tec) were selected to provide aninternal cavity with 200 μm thickness. In parallel, we alsoprepared neonatal mouse skin epidermis as described byHe et al. [20]. Freeze-substitution involved incubation ofsamples in a 1 ml solution of acetone containing 1% osmium

tetroxide and 0.1% uranyl acetate as follows: −90◦C for 48 h,−60◦C for 24 h, −30◦C for 18 h, with all transitions at 5◦C/h.After warming to 0◦C, samples were washed with freshacetone and then embedded in LX112 epoxy resin, whichwas polymerized at 60◦C. Ultrathin sections of approx.50 nm were cut with a 45◦ diamond knife and post-stainedwith 2% uranyl acetate and 0.4% aqueous lead citrate.

In the resulting images (Figure 1), mouse epidermis wascharacterized by linear cell borders. Although punctuated byfrequent desmosomes, these borders nonetheless had consid-erable amounts of free membrane surface area. In contrast,cow snout epidermis was characterized by numerous circularcell boundaries, reflecting a preponderance of interdigit-ating filipodia that serve to maximize contact area betweenadjacent cells. At least 50% of the surface area surroundingthese filipodia was engaged in desmosomes, thus conferringnecessary mechanical resilience to this tissue.

Electron tomography was used to provide a more detailedview of desmosome architecture. Individual images werecollected using the SerialEM program [24] on a 200 kV, fieldemission electron microscope (Tecnai F20) using a high-tilttomography holder and a 4 megapixel × 4 megapixel CCD(charge-coupled device) camera. The tilt angle was ±70◦

with an interval of 2◦ at zero tilt that varied according tothe cosine of the tilt angle. The defocus was 8–12 μm andfiducial free image alignment was done by cross-correlationprior to calculating the final three-dimensional structureby back projection [25]. A Gaussian low-pass filter wasapplied to each tomogram to suppress information beyondthe first zero of the contrast transfer function and a 3 × 3median filter was applied to reduce noise and enhanceedges.

Slices from the resulting tomograms show that desmo-somes from cow snout and mouse skin display similarfeatures (Figure 2). In the mouse desmosomes, the intra-cellular region consists of inner and outer dense plaques thattether intermediate filament bundles to the membrane, andthe intercellular region is characterized by a line of densityrunning halfway between the apposed cell membrane termedthe midline. The inset of Figure 2(A) illustrates how thismidline arises from the overlapping of cadherin proteins,which are visible as thin strands crossing between the apposedcells [20]. These features are greatly exaggerated in cowsnout desmosomes. The density of cadherins is much greaterand, although individual densities are occasionally visiblein the intercellular space (arrows in Figure 2B), individualmolecules are generally difficult to discern even in thetomographic slices. Interestingly, the intercellular distancein cow snout desmosomes (42.7±2.63 nm; n = 8) wassignificantly larger than in mouse skin desmosomes (32.70±1.97 nm; n = 4). Finally, the density of intermediate filamentsin the cytoplasm also appeared to be much higher in cowsnout epidermis, appearing as an undifferentiated clot in thetomographic slices compared with the distinct filaments inequivalent slices from mouse skin. Indeed, the preponderanceof intermediate filaments in cow snout appeared toexclude virtually all other organelles from the nearby

C©The Authors Journal compilation C©2008 Biochemical Society

Biochemical Society Annual Symposium No. 75: Structure and Function in Cell Adhesion 175

Figure 1 Appearance of desmosomes in freeze-substituted resin-embedded thin sections

(A, B) Desmosomes from newborn mouse epidermis. Cell borders are linear and punctuated by frequent desmosomes,

although most of the cell surface is free from intercellular contacts. The desmosomes are characterized by a midline with

fine strands extending from the cell surface. A dense intracellular plaque is visible next to the membrane, and intermediate

filaments are frequently seen inserting into this plaque, although, in (B), these filaments run obliquely through the sample

and are therefore not apparent. Other organelles are visible in the nearby cytoplasm. (C, D) Desmosomes from cow snout

epidermis. The circular structures correspond to interdigitating filopodia that characterize keratinocytes from the stratum

spinosum. The surface of these filopodia is >50% covered with desmosomes. Densities of cadherins and intermediate

filaments are much higher and other organelles are excluded from the cytoplasm. Scale bars, 30 nm.

cytoplasm (e.g. ribosomes visible in Figure 1B, but not inFigure 1D).

Conditions for isolating desmosomes forelectron tomographyIn order to optimize conditions for isolating desmosomes, weinvestigated the effects of different buffers on desmosomestructure in situ. Previous methods for isolating desmo-somes from cow snout have relied on citrate to solubilize thenon-cornified layers of the epidermis [26,27]. These isolateddesmosomes were shown to have a similar morphology to

those in the tissue and were used to establish the biochemicalcomposition [28,29]. Nevertheless, citrate is a calciumchelator and therefore has the potential to interfere withcadherin structure, which relies on Ca2+ ions to mediateinteractions between the extracellular cadherin domains andmaintain the extended shape of the molecule [13].

To evaluate the effect of pH and citrate on desmosomestructure, slices of the intact stratum spinosum wereincubated at 4◦C in 0.1 M glycine/HCl buffer (pH 2.6), 0.1 MPipes buffer (pH 7.4) or 0.1 M citrate buffer (pH 2.6) foreither 1 or 7 days. All three buffer solutions also containedprotease inhibitors (5 μg/ml each of pepstatin and leupeptin)

C©The Authors Journal compilation C©2008 Biochemical Society

176 Biochemical Society Transactions (2008) Volume 36, part 2

Figure 2 Electron tomography of desmosomes from

freeze-substituted resin-embedded epidermis

(A) Tomographic slice from mouse skin desmosomes reveals individual

cadherin densities spanning the intermembrane space and individual

intermediate filaments in the cytoplasm. In the inset, the membrane is

marked by lines and cadherins with arrows. (B) In cow snout, individual

cadherins are occasionally visible (arrows in the upper left), but the

density of cadherin molecules is generally much higher. Even in regions

of undifferentiated intercellular density, the midline can be appreciated

by viewing the slice at a glancing angle. Scale bars, 6 nm.

and 0.05% Nonidet P40. The characteristic midline was seenin all samples after 1 day of incubation in the three buffers(Figures 3A–3C). However, after 7 days of incubation, thismidline had disappeared in samples incubated in citrate(Figure 3F), but was still apparent in those from glycine/HCland Pipes buffer (Figures 3D and 3E). We therefore decidedto use the glycine/HCl buffer for isolation of desmosomesfor electron tomography.

Morphology of isolated desmosomesDesmosomes were isolated at 4◦C from 20 cow snoutsaccording to a modified method of Gorbsky and Steinberg[27]. To start, 12–15 g of the tissue consisting primarily ofstratum spinosum were added to 0.1 M glycine/HCl buffer(pH 2.6) containing 0.05% Nonidet P40 and 5 μg/ml each ofpepstatin and leupeptin. After incubation in buffer for 30 min

at 4◦C, the stratum spinosum slices were finely minced witha razor blade, suspended and stirred vigorously overnight inglycine/HCl buffer at 4◦C. After filtering through a 51 μmpolyester netting, the filtrate was centrifuged at 13 000 gfor 30 min. The pellet was resuspended in the same bufferand sonicated with a microprobe. The resulting suspensionwas centrifuged at 750 g for 30 min, and a pellet containingresidual cell debris was discarded. The supernatant wasfurther centrifuged at 12 000 g for 30 min, producing atrilaminar pellet with desmosomes on the top, pigment inthe middle and cell debris and nuclei at the bottom. Thetopmost pellet was retrieved, washed and repelleted three orso times until the pellet contained only a single layer. Thiswashed pellet was resuspended in glycine/HCl and sonicatedfurther. This solution was placed on to a discontinuoussucrose gradient (40–60% w/v) followed by centrifugationat 180 000 g for 3 h. A single dense white band formed at the55–60% interface, which was collected and resuspended inglycine/HCl buffer. Finally, these purified desmosomes werewashed three times to remove the sucrose by centrifugationat 22 000 g for 30 min at 4◦C.

The morphology of isolated desmosomes was initiallyevaluated by negative staining and in thin sections preparedby freeze-substitution (Figure 4). In negative staining,the desmosomes appear to be disc-shaped with diametersbetween 300 and 600 nm. The desmosomes generally liewith their membrane plane oriented parallel to the specimensupport, which does not provide a direct view of the inter-cellular gap. However, negatively stained desmosomes areoccasionally curled up at the edges to reveal the two apposingmembranes and the intervening gap, although the corre-sponding cadherin structures were not generally well pre-served by this technique. On the other hand, thin sectioningof desmosome pellets produces frequent views parallel to themembrane plane and indicated that the overall morphologyof the desmosomes had survived the isolation procedure. Inparticular, transverse sections of resin-embedded desmosomepellets revealed the leaflets of the bilayer and the midlineproduced by the overlapping cadherin molecules (Figure 4C,inset). In addition, components of the intracellular plaquewere visible at the periphery of the structure and formeddistinct bands parallel to the membrane, similar to those seenin the desmosomes from mouse skin (Figures 1A and 2A).Intermediate filaments were not present in these preparations,because the isolation conditions are specifically tailored todepolymerize intermediate filaments and thus disconnect thedesmosomes from the cytoskeletal network.

Tomography of frozen isolateddesmosomesWe turned next to imaging isolated desmosomes in thefrozen unstained state. Our goal was to compare tomogramsof these frozen-hydrated samples with those from freeze-substituted epidermis to investigate whether the proceduresof freeze-substitution and plastic embedding had negativelyaffected the preservation and/or the cadherin organization.We collected tilt series of these samples with two different

C©The Authors Journal compilation C©2008 Biochemical Society

Biochemical Society Annual Symposium No. 75: Structure and Function in Cell Adhesion 177

Figure 3 Tomogram slices from cow snout desmosomes after incubation in various buffers at 4◦C

(A) After 1 day in glycine/HCl buffer (pH 2.6). (B) After 1 day in Pipes buffer (pH 7.4). (C) After 1 day in citrate buffer (pH 2.6).

(D) After 7 days in glycine/HCl buffer (pH 2.6). (E) After 7 days in Pipes buffer (pH 7.4). (F) After 7 days in citrate buffer

(pH 2.6). The presence of a dense midline was taken to indicate a native organization of cadherin molecules. This midline

disappeared after incubation in citrate buffer for 7 days. Despite the disappearance of the midline, the overall desmosome

structure remained intact. Scale bar, 12 nm.

electron microscopes operating at either 200 kV (FEI TecnaiF20) or 300 kV (FEI Polara). Tilt series were recorded at25000 × magnification using a total electron dose of 50electrons/A2 (1 A = 0.1 nm) and a defocus of 16 μm. The300 kV instrument was coupled to an energy filter, anda 15 μm slit was used to remove inelastically scatteredelectrons and thus produce zero-loss images.

We used Fourier shell correlation to assess the resolutionof these datasets. For this calculation, images were dividedinto two halves containing even and odd images from the tiltseries [30]. Based on a cut-off of 0.5, this analysis indicated aresolution of ∼10 nm for the cryotomograms. In comparison,tomograms from stained sections produced a resolution of3 nm, indicating that the lower electron dose required forcryotomography poses a significant limitation. Nevertheless,these values represent underestimates of the true resolutionin the final tomogram, because of statistical effects of dividingthe data into two. Furthermore, resolution is anisotropicgiven the tilt geometry used for data collection, thusproducing better resolution parallel to the specimen supportand worse resolution in the perpendicular direction. For visu-alization of the tomograms, a Gaussian low-pass filter was ap-plied to the final dataset in order to suppress information be-yond the first zero of the contrast transfer function followedby a 3 × 3 median filter to reduce noise and enhance edges.

Similarly to negatively stained samples, desmosomesimaged in the frozen unstained state tend to lie with themembrane plane parallel with the support film. This situationis inevitable due to the trapping of a disc-shaped structurewithin a thin (100–200 nm) layer of water before freezing.However, similarly to samples in negative stain, some

desmosomes curl up at the edges and thus provide views of themembrane bilayer and the cadherins within the intercellulargap (Figure 5). These tomograms revealed the basic trilaminarfeatures seen in stained desmosomes in situ: namely, apposedmembranes separated by a distinct midline and flanked byremnants of intracellular material. These features and thecorresponding resolutions were equivalent in tomogramstaken from the two microscopes, suggesting that the higheraccelerating voltage and energy filtration were not essentialfor imaging this particular specimen. Close examination ofthe midline reveals an irregular pattern of globular densitieswithin the intercellular space that is not seen along themembrane and is consistent with the presence of unresolvedgroups of cadherin molecules along this midline. We hopedto see strand-like densities, corresponding to individual cad-herins, running between this midline and the membrane. In-stead, we observed a dark band separating these two features.We believe that this band represents an artefact arising fromthe high defocus required to produce phase-contrast from un-stained samples. This dark fringe can, in fact, be seen on bothsides of the membrane and will tend to mask connectionsto the membrane, although our observations of in situstained desmosomes suggest that the cadherin density in cowsnout epidermis may be too high to distinguish individualdensities in this region even in the absence of this defocusfringe.

DiscussionIn the present paper we have compared desmosomes fromnewborn mouse epidermis with those from cow snout. The

C©The Authors Journal compilation C©2008 Biochemical Society

178 Biochemical Society Transactions (2008) Volume 36, part 2

Figure 4 Isolated desmosomes from the cow snout stratum

spinosum

(A) Negative staining of the sample indicated the presence of disc-like

structures that lay flat on the specimen support with diameters between

300 and 600 nm. Scale bar, 30 nm. (B) Thin sections of pellets

that were freeze-substituted and resin-embedded revealed isolated

desmosomes in various orientations. Scale bar, 100 nm. (C) Close-up of

a resin-embedded section reveals the expected trilaminar structure with

midline, apposing membranes and remnants of the cytoplasmic plaque.

Intermediate filaments were removed by the isolation procedure. Scale

bar, 30 nm.

number of desmosomes, the density of proteins that composethese desmosomes and the density of intermediate filamentsassociated with the desmosomes were all significantlyhigher in cow snout. Undoubtedly, this situation reflectsthe increased mechanical challenge to the epidermis in thesnout. Although we were not able to distinguish indivi-dual cadherin molecules in tomographic slices from cowsnout desmosomes, the irregular appearance of the midlineis consistent with the grouping of cadherin moleculesobserved previously in freeze-substituted mouse epidermis.Furthermore, the appearance of the intercellular region infreeze-substituted desmosomes from cow snout is rathersimilar to frozen unstained isolated desmosomes, suggestingthat the protocols used for freeze-substitution do notcause major rearrangement of cadherin packing within theintercellular region. Thus the reported differences between

Figure 5 Cryoelectron tomography of isolated desmosomes in

the frozen unstained state

(A) Slice from a tomogram recorded with a 200 kV microscope. (B)

Slice from a tomogram recorded with 300 kV using an energy filter to

exclude inelastically scattered electrons. The same features are visible

in both tomograms. These particular examples were selected because

the desmosome curled up at the edge to provide a direct view of

intermembrane space. The two bilayers (arrows) are clearly visible

as well as a build-up of material at the midline (arrowhead). This

midline appears to be composed of globular densities, which probably

correspond to the groups of cadherin molecules seen previously in

freeze-substituted samples. Scale bars, 50 nm.

cadherin packing in freeze-substituted resin-embeddedmouse epidermis [20] and frozen-hydrated sections of humanforearm epidermis [21] may be due to different conditionswithin these respective tissues. Indeed, the model presentedby He et al. [20] emphasized inherent structural flexibility inthe association of cadherin molecules. Others have suggestedthat pairwise cadherin interactions are actually rather weakand that desmosomal strength results from the sum of a greatnumber of weak interactions between molecules clustered ata discrete location on the cell membrane [31]. Thus, in tissueplaced under repeated, possibly directional, stress, irregular

C©The Authors Journal compilation C©2008 Biochemical Society

Biochemical Society Annual Symposium No. 75: Structure and Function in Cell Adhesion 179

groups of cadherin molecules might be remodelled, insome circumstances, to form a quasi-ordered array. Furthercomparisons of the architecture of desmosomes in a varietyof tissues will be required to address this possibility.

We documented changes in desmosome organization whenslices of cow snout epidermis were incubated in citrate buffer.In particular, these desmosomes lost their midline, which isconsistent with previous observations of desmosomes formedbetween cultured cells that had been incubated in the calciumchelator EGTA. These morphological effects were also asso-ciated with a weakening of cell–cell adhesion [23,32]. Sincecitrate buffer also serves as a calcium chelator, we assume thatthis may be a similar effect. Interestingly, wounding of mouseepidermis was also shown to disrupt desmosome morphologyand to weaken cell adhesion in a similar way [23]. Theseobservations indicate that desmosome structure is dynamicand thus responsive to environmental cues. From a techno-logical point of view, isolated desmosomes are relatively easyto prepare and image by electron tomography. However, toaddress native morphology, tissue-specific architecturaldifferences and structural dynamics, it will be preferable toimage desmosomes within their native cellular environment.For this reason, we are currently using cryo-ultramicrotomyto produce sections from frozen unstained epidermis.Although this approach suffers from the innately low contrastand severe radiation-sensitivity of unstained material, it isessential to establish the reliability of freeze-substitution inpreserving structural detail at a molecular level.

We thank John Berriman and Chris Petzold for technical assistance,

Kent McDonald for advice on high-pressure freezing, Sriram

Subramaniam, Rachid Saugrat and Wim Hagen for facilitating

tomographic data collection on the Polara microscope at the

National Cancer Institute. This work was supported by NIH (National

Institutes of Health) grant GM071044 to D.L.S.

References1 Gumbiner, B.M. (1996) Cell adhesion: the molecular basis of tissue

architecture and morphogenesis. Cell 84, 345–3572 Amagai, M. (1999) Autoimmunity against desmosomal cadherins in

pemphigus. J. Dermatol. Sci. 20, 92–1023 McGrath, J.A. (2005) Inherited disorders of desmosomes.

Aust. J. Dermatol. 46, 221–2294 Chidgey, M. (2002) Desmosomes and disease: an update.

Histol. Histopathol. 17, 1179–11925 Jamora, C. and Fuchs, E. (2002) Intercellular adhesion, signalling and the

cytoskeleton. Nat. Cell Biol. 4, E101–E1086 Garrod, D.R., Merritt, A.J. and Nie, Z. (2002) Desmosomal adhesion:

structural basis, molecular mechanism and regulation. Mol. Membr. Biol.19, 81–94

7 Nollet, F., Kools, P. and van Roy, F. (2000) Phylogenetic analysis of thecadherin superfamily allows identification of six major subfamiliesbesides several solitary members. J. Mol. Biol. 299, 551–572

8 Hatta, K., Nose, A., Nagafuchi, A. and Takeichi, M. (1988) Cloning andexpression of cDNA encoding a neural calcium-dependent cell adhesionmolecule: its identity in the cadherin gene family. J. Cell Biol. 106,873–881

9 Shapiro, L., Fannon, A.M., Kwong, P.D., Thompson, A., Lehmann, M.S.,Grubel, G., Legrand, J.F., Nielson, A., Colman, J. and Hendrickson, W.A.(1995) Structural basis of cell–cell adhesion by cadherins. Nature 374,327–337

10 Overduin, M., Harvey, T.S., Bagby, S., Tong, K.I., Yau, P., Takeichi, M. andIkura, M. (1995) Solution structure of the epithelial cadherin domainresponsible for selective cell adhesion. Science 267, 386–389

11 Nagar, B., Overduin, M., Ikura, M. and Rini, J.M. (1996) Structural basis ofcalcium-induced E-cadherin rigidification and dimerization. Nature 380,360–364

12 Boggon, T.J., Murray, J., Chappuis-Flament, S., Wong, E., Gumbiner, B.M.and Shapiro, L. (2002) C-cadherin ectodomain structure and implicationsfor cell adhesion mechanisms. Science 296, 1308–1313

13 Pertz, O., Bozic, D., Koch, A.W., Fauser, C., Brancaccio, A. and Engel, J.(1999) A new crystal structure, Ca2+ dependence and mutationalanalysis reveal molecular details of E-cadherin homoassociation. EMBO J.18, 1738–1747

14 Tamura, K., Shan, W.S., Hendrickson, W.A., Colman, D.R. and Shapiro, L.(1998) Structure–function analysis of cell adhesion by neural (N-)cadherin. Neuron 20, 1153–1163

15 Sivasankar, S., Brieher, W., Lavrik, N., Gumbiner, B. and Leckband, D.(1999) Direct molecular force measurements of multiple adhesiveinteractions between cadherin ectodomains. Proc. Natl. Acad. Sci. U.S.A.96, 11820–11824

16 Sivasankar, S., Gumbiner, B. and Leckband, D. (2001) Directmeasurements of multiple adhesive alignments and unbindingtrajectories between cadherin extracellular domains. Biophys. J. 80,1758–1768

17 Zhu, B., Chappuis-Flament, S., Wong, E., Jensen, I.E., Gumbiner, B.M. andLeckband, D. (2003) Functional analysis of the structural basis ofhomophilic cadherin adhesion. Biophys. J. 84, 4033–4042

18 Baumeister, W. (2002) Electron tomography: towards visualizing themolecular organization of the cytoplasm. Curr. Opin. Struct. Biol. 12,679–684

19 McEwen, B.F. and Marko, M. (2001) The emergence of electrontomography as an important tool for investigating cellular ultrastructure.J. Histochem. Cytochem. 49, 553–564

20 He, W., Cowin, P. and Stokes, D.L. (2003) Untangling desmosomal knotswith electron tomography. Science 302, 109–113

21 Al-Amoudi, A., Norlen, L.P. and Dubochet, J. (2004) Cryo-electronmicroscopy of vitreous sections of native biological cells and tissues.J. Struct. Biol. 148, 131–135

22 Rayns, D.G., Simpson, F.O. and Ledingham, J.M. (1969) Ultrastructure ofdesmosomes in mammalian intercalated disc: appearances afterlanthanum treatment. J. Cell Biol. 42, 322–326

23 Garrod, D.R., Berika, M.Y., Bardsley, W.F., Holmes, D. and Tabernero, L.(2005) Hyper-adhesion in desmosomes: its regulation in wound healingand possible relationship to cadherin crystal structure. J. Cell Sci. 118,5743–5754

24 Mastronarde, D.N. (2005) Automated electron microscope tomographyusing robust prediction of specimen movements. J. Struct. Biol. 152,36–51

25 Winkler, H. and Taylor, K.A. (2006) Accurate marker-free alignment withsimultaneous geometry determination and reconstruction of tilt series inelectron tomography. Ultramicroscopy 106, 240–254

26 Skerrow, C.J. and Matoltsy, A.G. (1974) Isolation of epidermaldesmosomes. J. Cell Biol. 63, 515–523

27 Gorbsky, G. and Steinberg, M.S. (1981) Isolation of the intercellularglycoproteins of desmosomes. J. Cell Biol. 90, 243–248

28 Kapprell, H.P., Cowin, P., Franke, W.W., Ponstingl, H. and Opferkuch, H.J.(1985) Biochemical characterization of desmosomal proteins isolatedfrom bovine muzzle epidermis: amino acid and carbohydratecomposition. Eur. J. Cell Biol. 36, 217–229

29 Cowin, P., Kapprell, H.P. and Franke, W.W. (1985) The complement ofdesmosomal plaque proteins in different cell types. J. Cell Biol. 101,1442–1454

30 Cardone, G., Grunewald, K. and Steven, A.C. (2005) A resolution criterionfor electron tomography based on cross-validation. J. Struct. Biol. 151,117–129

31 Chen, C.P., Posy, S., Ben-Shaul, A., Shapiro, L. and Honig, B.H. (2005)Specificity of cell–cell adhesion by classical cadherins: critical role forlow-affinity dimerization through β-strand swapping. Proc. Natl. Acad.Sci. U.S.A. 102, 8531–8536

32 Kimura, T.E., Merritt, A.J. and Garrod, D.R. (2007) Calcium-independentdesmosomes of keratinocytes are hyper-adhesive. J. Invest. Dermatol.127, 775–781

Received 14 November 2007doi:10.1042/BST0360173

C©The Authors Journal compilation C©2008 Biochemical Society