Desmosomes and hemidesmosomes: structure and

non c.c.soIQc /flfllfli1071 1(111 cn IF1 CACCO 071

function of molecular componentsKATHLEEN J. GREEN* AND JONATHAN C. R. JONESt’#{149}Departmentsof Pathology and Dermatology, and the R. H. Lurie Cancer Center, tDepartment of Cell and

Molecular Biology, Northwestern University Medical School, Chicago, Illinois 60611

ABSTRACT Desmosomes and hemidesmosomesare the major cell surface attachment sites for inter-

mediate filaments at cell-cell and cell-substrate con-tacts, respectively. The transmembrane molecules ofthe desmosome belong to the cadherin family ofcalcium-dependent adhesion molecules, whereasthose in the hemidesmosome include the integrinclass of cell matrix receptors. In each junction, thecytoplasmic domains of certain tranamembranejunc-tion components contain unusually long carboxy-ter-minal tails not found in those family membersinvolved in linkage of actin to the cell surface. Thesedomains are thought to be important for the regula-tion of junction assembly and specific attachment ofintermediate filaments via associated adapter pro-teins. Recent developments have suggested the excit-ing possibility that these junctions, in addition toplaying an unportant structural function in tissueintegrity, are both acceptors and affectors of cellsignaling pathways. Many desmosomal and hemides-

mosomal constituents are phosphoproteins and incertain cases the function of specific phosphorylationsites in regulating protein-protein interactions is be-ing uncovered. In addition, a more active role intransmitting signals that control morphogenesis dur-ing development and possibly even regulate cellgrowth and differentiation are being defined forcytoplasmic and membrane components of thesejunctions.-Green, K. J., Jones, J. C. R. Desmosomesand hemidesmosomes: structure and function of mo-lecular components. FASEBJ. 10,871-881 (1996)

Key Word.,: cell junction ma! rix connector cytoskeleton

STRUCTURE AND DISTRIBUTION OFDESMOSOMES AND HEMIDESMOSOMES

The most prominent cell-surface attachment sites for in-termediate filaments (IF)2 in epithelial cells are des-mosomes and hemidesmosomes, which mediate IFanchorage at sites of cell-cell and cell-substrate contact,respectively. By anchoring IF at sites of strong intercellu-lar adhesion, desmosomes create a transcellular networkthroughout a tissue that is thought to resist forces of me-

chanical stress. This network in turn is attached to thebasal aspect of the cell by molecularly distinct junctionalstructures called hemidesmosomes, which confer addi-

tional mechanical integrity to the tissue. Although provid-ing mechanical integrity is thought to be a criticalfunction of these junctions, it is clear that they are ex-tremely dynamic structures that respond with exquisitesensitivity to environmental cues, allowing for tissue re-

modeling during development, differentiation, woundhealing, and invasion. In addition to being modulated inresponse to their environment, cell junction moleculesthemselves play active roles in signal cascades initiatedby extracellular matrix ligands and growth factors duringdevelopment and in the adult.

As their names suggest, desmosomes and hemides-mosomes exhibit similar structural characteristics (Fig.1). Each is composed of a tripartite electron-dense

plaque structure specialized for IF anchorage. In the caseof the desmosome, mirror image plaques sandwich amembrane core region, whereas a single plaque located at

the basement membrane serves this function in thehemidesmosome (insets in Fig. 1). With one known ex-

ception, the molecules comprising these junctions arecompletely distinct, although certain components are evo-lutionarily related. Extracellularly, desmosomes are sepa-rated by a 30 nm space filled with material that

represents in large part the extracellular domains of thesingle span transmembrane desmosomal cadherin mole-cules. Hemidesmosomes, on the other hand, are attachedthrough an integrin-based mechanism to the underlying

basement membrane and stroma.Although desmosomes and hemidesmosomes are both

found in epithelia where they associate with keratin-con-taming IF, desmosomes are thought to exhibit a morewidespread tissue distribution. These intercellular junc-

tions are also present in cardiac muscle where they an-

‘To whom correspondence and reprint requests should be addressed,

at: Department of Cell and Molecular Biology, Northwestern University

Medical School, 303 E. Chicago Ave., Chicago, IL 60611, USA.

2Abbreviations: IF, intermediate filaments; BP18O, 180 kDa buhlous

pemphigoid antigen; GABEB, generalized atrophic benign epidermolysis

bullosa; CP, cicatricial pemphigoid; JEB, junctional epidermolysis bul-

losa; IFAP. IF-associated protein; BP, bullous pemphigoid; APC, ade-

nomatous polyposis coli.

____ .

l L___......J’ ..

Figure 1. Electron micrographs of the basal layer of human epidermis (E, epidermal cell; ECM, extracellular matrix/dermis), a);

A region of interaction between two epidermal cells as well as epidermal cell-dermis association is shown. The upper box in panel

a has been printed at higher magnification in panel b. Note the desmosome, with its characteristic electron-dense cytoplasmic

plaques, lying either side of the contacting membranes of the epidermal cells. The lower boxed area in panel a is printed at higher

magnification in panel c and shows a typical hemidesmosome. Like the desmosome in panel b, the hemidesmosome has acytoplasmic plaque, but unlike the desmosome, each hemidesmosome abuts the dermis via the basement membrane. Panels b andc are at the same magnification, a) Bar, 500 nm; b) bar, 250 nm.

a,-, In I...... 100C Ti.....CACCD I._.._.__l

chor desmin-containing IF, and in the arachnoid and piaof meninges and follicular dendritic cells of the lymphoidsystem where they associate with vimentin-containing IF(1). In addition to skin and cornea, hemidesmosomes arealso present in transitional epithelial cells (e.g., in blad-der) and certain glandular epithelia (e.g., mammary glandepithelia and myoepithelial cells) (2, 3). The hemides-mosomes present in all these tissues are related not only

by their ultrastructural character, but also by their com-position. However, certain hemidesmosomal componentsalso occur in simple epithelial cells such as those liningthe gut that lack ultrastructurally defined hemides-mosomes (4-7). in these cases, it has been suggested thathemidesmosomal components are assembled into less or-ganized multiprotein complexes for which some authors

have now coined the term type II hemidesmosomes todistinguish them from the “classical” or type I

hemidesmsome of basal epidermal cells (7).Here we present the most recent developments ad-

dressing the molecular composition of these two junctiontypes (shown schematically in Fig. 2), as well as thestructure, function, and regulation of their constituents.This review will not be comprehensive, and we refer thereader to recent review articles for details of other corn-

ponents and a historical perspective of the subject (2, 3,8).

The desmosome

The membrane molecules

Neighboring cells are thought to be adherent at des-mosomes through interactions mediated by a relativelynew division of the cadherin family of cell adhesionmolecules known as the “desmosomal cadherins.” This di-vision includes the subclasses known as desmogleins and

desmocollins (reviewed in refs 2, 9). Like the classic

cadherins, desmosomal cadherins are single-pass, trans-membrane-spanning glycoproteins with conserved regionsof homology in the extracellular domain, thought to be in-

volved in calcium binding and adhesion, and a majorconserved region in the cytoplasmic domain required for

binding to cytoplasmic adapter proteins. In the case ofdesmosomal cadherins, a protein called plakoglobin asso-ciates with this conserved region (reviewed in ref 10).

The cytoplasmic domain of the desmogleins also harborsvariable numbers of a 29 residue repeating motif of un-

known function, unique to this cadherin subclass. Eachdesmocollin gene gives rise to two alternatively spliced

Ci.ACiCC5A A Mfl c.AIncAr,c,kAcc 071

Figure 2. Schematic showing the major components of desmosomes and hemidesmosoines in two epithelial cells. One desmosome

links the lateral domains of two epithelial cells while each is tethered to the underlying connective tissue via hemidesmosomes. Both

the desmosome and hemidesmosome are connected to the intermediate filament (IF) cytoskeleton system which shows interaction withthe surface of the nucleus (central shaded circle). We have taken some liberties in the diagram by indicating our ideas of how proteins

may interact within these complex morphological entities. Dsg and Dsc a denote the cadherin-like desmogleins and desmocollins of

the desmosome respectively. We only indicate the ‘long” or a isoform of desmocollin in the diagram. Pg. plakoglobin. desmoplakin,

desmoplakin. Lam-5, laminin-5.

mRNA transcripts resulting in an “a” and ‘b” form, whichdiffer only in the cytoplasmic domain, with the shorter“b” form containing 11 amino acids not in the “a” form.Although the functional significance of these two forms isunknown, the shorter desmocollin tail lacks the plakoglo-bin binding site present in the longer “a” form and in alldesmogleins (11).

The recent identification of three desmocollin andthree desmoglein genes has confirmed previous immu-nological evidence that the desmosomal glycoproteins areheterogeneous and expressed in tissue- and stratification-specific patterns (12-14). The desmosomal cadherinshave been mapped to a small cluster on human chromo-some lBq2l; in the case of the desmogleins, they are tan-demly linked in the order DSG1-DSG3-DSG2 from 5’ to3’ mirroring their expression pattern from suprabasal tobasal in stratified human epidermis (15).

Although the classic cadherins are typically thought to

mediate calcium-dependent homophilic adhesion, themechanism by which desmosomal cadherins function is

largely unknown. Early work demonstrated that Fab’ anti-

body fragments against desmocollin inhibit desmosomeassembly in MDBK cells (2). In addition, autoantibodiesto desmoglein family members circulating in patients withthe class of autoimmune epidermal blistering diseasescalled pemphigus have been demonstrated to be causa-tive in mouse models of the disease, consistent with a

role in disrupting intercellular adhesion (e.g., ref 16).

Convincing experimental data supporting a role for indi-vidual desmosomal cadherins in calcium-dependent ho-mophilic adhesion are lacking, however. A chimericmolecule with the Dsg3 extracellular domain fused to theE-cadherin cytoplasmic domain was shown to engage inweak homophilic adhesion that was not comparable to ad-

hesion mediated by E-cadherin (17). Likewise, full-lengthdesmoglein 1 and desmocollin 2, expressed with or with-out the associated plakoglobin molecule, are unable tosupport the level of adhesion mediated by E-cadherin (A.Kowalczyk and Green, unpublished results). The exist-ence of multiple desmosomal cadherins within a singledesmosome, as well as multiple tissue specific isoforms,suggests a functional complexity for these molecules not

exhibited by their classic cadherin relatives. For in-stance, it is possible that the active form of desmosomalcadherin is a cell type-specific heterodimer that associ-ates laterally within the junction. Such tissue-specificpairing may endow desmosomes with distinct adhesiveand/or cytoskeletal linking roles.

Although the adhesive function of all possible des-mosomal cadherin combinations has not yet been tested,the lack of demonstrable adhesion using traditional as-says nevertheless highlights the importance of consider-ing the role of other membrane molecules in desmosomefunction. Recently, a 22 kDa phospholipid-anchoredmolecule, called the E48 antigen, has been desmon-

strated to confer adhesive properties on MOP8 cells (18).

07A ‘/r.I in 1001. TF., IAcIR In,,rn,l C.PFVN,J ANIfl inijFc

The expression of E48 is restricted to specific tissues, soit is unlikely that this molecule plays a constitutive adhe-sive function in all tissues. However, the existence of as

yet unidentified cell type-specific forms of E48 required

for adhesion in other tissues should be considered.

The plaque molecules

The cytoplasmic plaque of desmosomes is complex andexhibits tissue-specific differences in both structure and

composition. The constitutive components are plakoglobinand the most abundant component, desmoplakin. Al-

though more minor, sometimes tissue-specific, compo-nents surely also play important roles in modulatingadhesive or cytoskeletal interactions, emphasis will beplaced on recent work dealing with the function andregulation of the major desmosomal components.

Plakoglobin

Plakoglobin belongs to an emerging gene family that alsoincludes -catenin, the cadherin-associated protein pl2O,and the tumor suppressor adenomatous polyposis coli

(APC). Members of this family share at the core of their

structure a series of repeating motifs first found in arma-dillo, a downstream effector in the wingless signaltransduction pathway responsible for the establishment ofsegmentation polarity in Drosophila (19, 20).

As plakoglobin binds tightly to the cytoplasmic do-mains of both desmosomal cadherins, desmocollin (thelarger “a” form) and desmogleins (11, 21-23), it mayserve as a molecular link between the outer and inner

portions of the desmosomal plaque. Consistent with thisidea, deletion of the plakoglobin binding site in desmoso-mal cadherins abrogates the ability of these molecules to

anchor IF at the plaque (11, 24).Plakoglobin is not restricted to desmosomes but is a

common component of adhesive junctions including mi-crofilament associated cell-cell adherens junctions in epi-thelial and nonepithelial cells (1, 2, 10). This distributionlikely reflects plakoblobin’s ability to associate not only

with the desmosomal cadherins, but also in separate com-plexes with the classic cadherins, albeit more weakly(25). A potential role for plakoglobin in adherens junc-tions is not clear. In fact, analysis of cross-linked junc-

tional complexes from MDCK cells suggests thatplakoglobin may not even be effectively recruited into

Triton-insoluble adherens junctions in cells that haveboth desmosomes and adherens junctions (26).

In addition to their structural roles in intercellularjunctions, members of the armadillo gene family act assignal transducers (19, 20). Armadillo is the most distalcomponent of the signaling pathway mediated by a se-creted protein in Drosophila, called wingless, which ishomologous to the vertebrate Wnt growth factor family.Although, like its vertebrate relatives, armadillo binds to

cadherins in cell-surface, adherens-type junctions, evi-dence suggests that a cytoplasmic rather than junction-associated form of armadillo proteins is active in

signaling. Wingless results in the metabolic stabilizationand accumulation of cytoplasmic armadillo, which is cor-related with a change in its phosphorylation state due to

the inactivation of the upstream serine/threonine zestewhite kinase. Evidence that such a pool exerts a signal-

ing effect in vertebrates comes from studies desmonstrat-ing that overexpression by microinjection of -catenin orplakoglobin mRNA into Xenopus embryos leads to the du-plication of the embryonic axis, resulting in embryos withtwo heads, notochords, and neural tubes (27, 28). In the

case of plakoglobin, this effect was abrogated by coex-pression with the desmoglein cytoplasmic domain (28).

The latter result suggests that the proper balance between

cadherin-bound and unbound pools of plakoglobin/f-catenin is likely to be crucial for proper signaling duringdevelopment. Along these lines, the extremely rapid deg-radation of noncadherin associated plakoglobin recentlyreported in fibroblasts ectopically expressing this protein

may represent a general mechanism for controlling theaccumulation of armadillo family members (22).

The mechanism by which the armadillo family mem-

bers actually affect downstream changes in gene expres-sion is unknown. Intriguingly, however, both -cateninand plakoglobin have been shown to accumulate in thenucleus in overexpression experiments, and some havespeculated that interaction of armadillo family memberswith a nuclear target may be involved in regulation ofgene expression (27, 28). The tumor suppressor geneproduct APC, mutated in patients with the dominantly in-herited disease familial adenomatous polyposis, also

binds in a cytoplasmic complex with 3-catenin or plako-

globin. The functional implications of this association forregulating cytoplasmic levels or signaling activity of theseproteins are not known. However, this observation sug-gests that f-catenin and plakoglobin may be involved inregulating cell growth control in addition to developmentand differentiation (29, 30).

Plakophilin/band 6

What was previously called “band 6” in enriched prepa-rations of desmosomes isolated from bovine tissues has

now been identified as a plakoglobin-like molecule. Un-like plakoglobin, in vitro evidence indicates that thismolecule may bind directly to IF polypeptides (31, 32).Band 6/plakophilin exhibits a broader tissue/cell distri-bution than previously recognized because it is found in

the cytoplasm of several cultured lines, including thosefrom simple epithelia. However, it is not a constitutivecomponent of desmosomes and thus is unlikely to be ab-solutely required for IF anchorage. If plakophilin provesto be a signaling molecule like its armadillo family rela-tives, this could provide a potential mechanism for differ-entiation-specific signaling.

Desmoplakin

Although plakoglobin appears to play an important role inestablishing contact with the IF cytoskeleton of des-

DFSMOSOMFS AND HFMIDFSMOSOMES R7’

mosomes, more likely candidates exist for direct associa-tion with IF polypeptides. The most abundant and well-studied of these is desmoplakin. Desmoplakin is a largedumbbell-shaped molecule with a central a-helical coiledrod domain flanked by two globular end domains withdistinct functions (33). So far, two alternatively spliced

forms derived from a single desmoplakin gene have beenreported. The smaller desmoplakin II product is morevariably expressed, found at lower levels in nonstratifiedtissues and absent in certain tissues such as the heart.

Molecular mapping studies using transient transfectionof constructs encoding specific domains of desmoplakinindicated for the first time that the carboxyl terminus ofthis molecule contains sequences that govern its associa-

tion with IF networks (34). These initial observationswere borne out by in vitro studies suggesting that this in-teraction is direct in the case of type II epidermal kera-tins, which interact with desmoplakin via amino-terminalsequences (35). Transient transfection studies have alsomapped sequences required for association with the des-

mosomal plaque to the amino terminus of desmoplakin(36). Together, these studies suggest that desmoplakin is

a functionally modular protein that acts as a molecularlinker to anchor IFs at the desmosome; however, thesedomain mapping studies did not directly test this hy-pothesis. To address this idea directly, we recently used a

dominant negative approach whereby a region of theamino terminus sufficient for localization and, presum-ably, binding to components of the desmosomal core wasmoderately overexpressed in stable A431 cell lines. Theresult was displacement of endogenous desmoplakin fromthe plaque and loss of IF anchorage, suggesting that des-moplakin is indeed required for this attachment (E. A.Bornslaeger and K. J. Green, unpublished observations).

Like plakoglobin, desmoplakin is a phosphoprotein.Recent evidence suggests that the interaction between thecarboxyl terminus of desmoplakin and IF networks isregulated by phosphorylation of a serine residue locatedin a cAMP-dependent kinase consensus site 23 amino ac-ids from the carboxy-terrninal end of desmoplakin (37).

This serine is within a region that had been shown to be

required for interaction with keratin (but not vimentin) IFnetworks, and may represent a regulatory site for interac-

tion with specific filament polypeptides (36). One possi-ble function for such a phosphorylation event might be toprevent desmoplakin from becoming sequestered allalong IF in the cytoplasm during its recruitment into des-mosomes.

Other members of the desinoplakin gene family

BP230/plectin/IFAP300Desmoplakin belongs to another emerging gene familywhose members are involved in the organization or an-chorage of IF networks. The first similarity identified waswith BP230, a plaque component and candidate IF linkerspecifically found in hemidesmosomes that will be de-scribed in more detail (33). The third member is plectin,

a known IF-associated protein (IFAP) with broad tissuedistribution reported to be present in desmosomes andhemidesmosomes. Like desmoplakin, plectin’s domain

functions have been mapped using transient expressionexperiments, and the carboxy-terminal repeats have beendemonstrated to associate with IF networks in cells (38).However, plectin has been shown to bind in vitro to manyIF types, including nuclear lamin B as well as micro-

tubule-associated proteins, a-spectrin and fodrin. Thus,this molecule may function as a universal linking protein.

Another potentially closely related protein calledIFAP300 has also been demonstrated to be in des-mosomes and hemidesmosomes (39). Similar to plectin,

IFAP300 binds to cytoplasmic IF networks in fibroblasts

in addition to being localized at both junction types inepithelial cells. Data supporting a central role forIFAP300 in IF anchorage will be discussed below. How-ever, the specific roles each individual family memberplays within a particular junction remain unknown.IFAP300 and desmoplakin are both IFAPs located in the

desmosome, although desmoplakin is present at higherlevels. One possibility is that IFAP300 augments interac-

tions mediated by desmoplakin, contributing to the stabil-ity of IF interactions in desmosomes. The possible celltype specificity of family members may also be important

for mediating interactions with different types of IF net-works.

THE HEMIDESMOSOME

The membrane molecules

Hemidesmosome integrins

Integrins are heterodimeric matrix receptors each com-posed of an a and a f subunit. These receptors not onlyform part of the link that integrates the extracellular ma-

trix and the cytoskeleton of cells, but also act totransduce signals (40). Until 1990, it was generally be-lieved that cytoskeleton interactions of integrins werelimited to the microfilament system of cells. However, in1990/1991, several groups showed that the epithelial cellintegrin a64 was concentrated in the hemidesmosome

and therefore was spatially associated not with the actincytoskeleton, but with keratin containing tonofilaments(reviewed in ref 3).

The Wi subunit

The a6 subunit can bind either the 1 or J4 subunit, butwhen given a choice it preferentially associates with 34(41). In many epithelial cells, therefore, despite the pres-ence of the 1 subunit, a6 is found exclusively associ-ated with 4 integrin. There are several isoforms of thea6 integrin subunit (4, 6). Each isoform is synthesized asa 150 kDa polypeptide, which is then cleaved into“heavy” and “light” chains that associate via disulfidebonding. The two best-studied a6 isoforms, a6A anda6B, differ in their cytoplasmic domains, thus providing

a7E. VnI 11) Inn iqqr. Tho tAcra Ir.,,rnI c.PrrM AMrl IClNJ

a possible opportunity for interaction with differenct cyto-

plasmic components (6). The a6A isoform has been lo-calized to tissues that possess typical hemidesmosomes(6); however, a6A is also found in gut epithelial cellslacking bona Ilde hemidesmosomes but that may assem-ble type II hemidesmosomes (6, 7). Likewise, a6B isfound primarily in the kidney and certain epithelial

glands where it may occur in a type II hemidesmosome-like structure (6, 7).

The 4 integrin subunit

The 4 integrin subunit is unique among the I integrins

so far characterized because of the presence of an ex-

tended carboxy-terminal cytoplasmic tail of more than1000 amino acids (4, 42). Alternative splicing of the 4message gives rise to two different forms, each containingtwo type III fibronectin repeat motifs connected by thevariable domain, which is also a site of proteolytic cleav-age (4, 41). Most investigators in the field assume that

the unusual structure of the subunit cytoplasmic tailexplains why a6f4 integrin is the only integrin heterodi-

mer so far identified that is found associated with the IFcytoskeleton. This assumption is based on studies demon-strating that the cytoplasmic tails of subunits of other in-

tegrin heterodimers are often involved in anchorage of theactin cytoskeleton via one or more actin cytoskeleton-as-sociated proteins. Indeed, there is now biochemical evi-dence to support this possibility, because anIF-associated protein IFAP300 that is a component of the

hemidesmosome (39) binds 4 integrin in overlay assays(S. E. Baker and J. C. R. Jones, unpublished observa-tions). Thus, IFAP300 may directly link IF to the 4 in-

tegrin cytoplasmic tail in much the same way that talin

links actin filaments to the cytoplasmic domain of the 1integrin subunit (40).

Antibodies directed against the external domains of thef4 integrin subunit inhibit hemidesmosome assembly andperturb the structural integrity of formed hemides-mosomes (43). The importance of the f4 integrin in

hemidesmosome formation and stability has been con-firmed by recent molecular genetic studies by Spinardi

and co-workers (44, 45). These authors have made use of804G cells, one of only a few cell lines that assemble

hemidesmnosomes in vitro (46). In their initial study they

presented evidence that a region of 303 amino acids in

the cytoplasmic domain of 4 is necessary for 4 subunit

incorporation into hemidesmosomes, whereas the 4 ex-

tracellular domain is essential for 4 interaction with thea6 integrin subunit (44). These same workers haveshowed that overexpression of a tailless I4 integrin, lack-ing most of the cytoplasmic domain of the wild-type

molecule, has a dominant negative effect that leads toperturbation of hemidesmosome organization (45). A mu-tation in the 4 integrin gene leading to premature termi-

nation of message transcription has now been discovered

in one patient afflicted with the blistering skin diseasejunctional epidermolysis bullosa (47). If 4 integrin plays

a role in nucleation of hemidesmosome assembly, its ab-sence could explain a key histological feature of this dis-ease, i.e., a decrease in the frequency of

hemidesmosomes.

Recent data provide circumstantial support for the pos-sibility that a64 integrin is involved in signal transduc-tion. 4 Integrin is physically associated with one ormore protein kinases; upon interaction of the a6f4 withits extracellular ligand, 4 becomes phosphorylated ontyrosine (48). Furthermore, a tyrosine phosphorylation

site in the cytoplasmic domain of 4 has been shown to

be required for its association with other hemidesmosomecomponents (48). This site lies in a tyrosine-based acti-vation motif or TAM consisting of two possible phospho-rylatable tyrosine residues, followed by a leucine atposition +3 (48). In addition, a separate tyrosine phos-

phorylation event in f4 appears to trigger binding of thesignaling adaptor molecule Shc, which upon phosphoryla-tion recruits the adaptor Grb2 (48). This study suggeststhat hemidesmosome integrins may mediate signalingevents from the matrix to an epithelial cell in a manner

similar to other integrin receptors such as the fibronectinreceptor a51 (reviewed in ref 40).

The 180 kDa bullous pemphigoid antigen (BP18O)

Autoantibodies circulating in some patients afflicted withbullous pemphigoid (BP) recognize a 180 kDa hemides-mosomal protein, variously termed BP18O or BPAG2(bullous pemphigoid antigen 2). The cytoplasmic domain

of this type II membrane protein (i.e., its amino terminusis located in the cytoplasm) is separated by a membranedomain from a short extracellular stretch of highlycharged amino acids leading to a region containing a se-ries of GLY-X-Y or collagen-like repeats (49, 50). Basedon its collagen-like structure, BP18O has been referred to

by some investigators as type XVII collagen. It is gener-ally assumed that the collagen extracellular domain is in-volved in interactions between BP18O and components ofthe basement membrane, although the nature of such in-teractions is yet to be defined.

The importance of BP18O for epidemal-connective tis-sue interactions has been highlighted by the identifica-tion of inherited and autoimmune skin diseases that

target the BP18O gene or protein. BP18O is missing fromthe skin of individuals suffering generalized atrophic be-

nign epidermolysis bullosa (GABEB) (51,52). In one casethis has been shown to occur because a mutation in theBP18O gene leads to premature transcription terminationof the BP18O message, and thus to a lack of BP18O pro-tein in the skin (52). At the electron microscopic level,henmidesmosomes are either missing or present in a rudi-

mentary state in the skin of GABEB patients; presumably

this weakens the attachment of epidermal cells to thebasement membrane and leads to blistering. BP18O is

also a target for pathogenic antibodies in two autoimmunediseases, bullous pemphigoid and herpes gestationis (53).In particular, the perimembrane noncollagenous extracel-

DESMOSOMES AND HEMIDESMOSOMES 877

lular domain of BP18O contains an epitope recognized by

some (but not all) BP18O autoantibodies (53). Giudice

and co-workers (54) have shown that neonatal mice in-

jected with antibodies against this same epitope develop

lesions histologically identical to those seen in bullous

pemphigoid patients.A region of 36 amino acids at the amino terminus of

BP18O is required for its polarization in the plasma mem-brane (55). On the other hand, the perimembrane 27

amino acid noncollagenous domain of BP18O, the targetfor pathogenic autoantibodies, appears to be essential for

interactions between BP18O and other hemidesmosomalcomponents. With what hemidesmosome element (or ele-ments) does BP18O interact? Using a molecular geneticapproach, Hopkinson et al. (55) have provided evidence

that BP18O may interact with a6 integrin because BP18Oassociates morphologically with the a6 integrin subunit

regardless of its partner and cc6 antibodies coprecipi-

tate BP18O. Indeed, these same workers have speculated

that pathogenesis of BP involves autoantibody induced

disruption of BP18O-a6 integrin interaction leading to

perturbation of the structural integrity of the hemides-mosomes. This would parallel the disruption of thehemidesmosomes observed in in tissue explants treated

with a function blocking a6 integrin antibody (43).

Matrix molecules

Laminin-5, also referred to as GB3 antigen, epiligrin, and

kalinin, is a newly characterized member of a growingfamily of laminin heterotrimers and is composed of threesubunits termed a3, 2, and ‘y2 (56). Immunoelectronmicroscopy has revealed that laminin-5 is concentrated in

the basement membrane zone immediately underlyingeach hemidesmosome in stratified squamous epithelial

tissues (3). However, note that some if not all of thechains of laminin-5 are expressed by lung epithelial cellsthat do not possess bona fide hemidesmosomes but that

may assemble type II hemidesmosomes (5, 7).

Laminin-5, like other matrix proteins, is promiscuouswith regard to its cell receptors (57). For example, inskin cells maintained in vitro, a31 appears to initiatecell binding to laminin-5 (58). a631 may also act as alaminin-5 receptor in some tissue cultured cells whereas

a64 integrin is involved in establishment of so-calledlong term stable anchoring contacts or hemidesmosomes

on laminin-5 rich matrices (57-59). The physiologicalrelevance of these laminin-5/integrin interactions is notyet clear because they may not all occur in vivo. How-ever, it is possible that such promiscuity reflects a func-

tional diversity in laminin-5 that in some way is

modulated by the nature of its cell-surface associations.This may help explain why a protein involved in forma-tion of stable anchoring contacts, is also found at sites ofactive cell motility such as the invasion front of coloncarcinomas (60). In this regard, the apparent receptorpromiscuity of laminin-5 may result from proteolytic

processing of its component chains and thus presentation

of previously masked receptor binding sites (61). Alterna-tively, there is now evidence for the existence of laminin-

5 chain isoforms that may exhibit distinct receptor

specifities (62).Laminin-5, like BP18O, is the target molecule for anti-

bodies in an autoimmune disease (cicatricial pemphigoid,CP) and has been shown to be deficient in the skin of pa-tients with a genetic disease termed junctional epider-molysis bullosa (JEB) (63-67). CP and JEB are

characterized by loss of cohesion between an epidermalcell and the basement membrane as well as the absence

of hemidesmosomes, suggesting a role for laminin-5 inhemidesmosome assembly. Consistent with this idea, hu-man SCC12 (squamous cell carcinoma) keratinocytes

maintained under normal culture conditions fail to form

hemidesmosomes. In contrast, SCC12 cells are induced to

assemble hemidesmosomes when maintained on a

laminin-5 rich matrix secreted by 804G cells (59). Ma-

trix-induced assembly of hemidesmosomes in SCC12

cells involves reorganization of hemidesmosome compo-

nents, including a64 integrin and BP18O, along thecell-matrix interface, resulting in a distribution patternthat overlies the laminin-5 components in 804G cell ma-

trix. In other words, 804G matrix contains a structural

cue which when transduced to overlying cells, most likely

via a634 integrin, can trigger hemidesmosome assembly.

It will be interesting to address whether the mechanismof hemidesmosome assembly in this system involves asimilar signaling pathway involving phosphorylation of 4

integrin as described by Mainiero et al. (48).

The plaque molecules

BP230 and .BP230 isoforms

Autoantibodies directed against a 230 kDa plaque com-ponent of the hemidesmosome are found in the majority

of bullous pemphigoid serum samples (68). Because itwas characterized before BP18O, it is sometimes calledBPAG1, although we shall refer to it as BP230. BP230localizes within the region of the hemidesmosome plaqueto which keratin IF attach (3), providing a first clue thatBP230 may be involved in IF cell-surface anchorage.This idea is further supported by the discovery thatBP230 belongs to a family of proteins that include des-moplakin and another IF-associated protein called

plectin, all possessing striking sequence similarities intheir carboxyl terminus, a likely site of binding to IF(33). Indeed, isolated BP230 molecules even bear a su-perficial resemblance to desmoplakin because both pos-sess a central rod domain with globular ends (69).

Experimental evidence that BP230 plays a role in or-ganizing the IF cytoskeleton comes from studies of micein which BP230 has been ablated by targeted homologousrecombination (70). Hemidesmosomes in the epidermalcells of these mice lack the innermost cytoplasmic regionof the hemidesmosomal plaque and exhibit few, if any,associated IFs (70). This is not entirely surprising given

the localization of BP230 antigen and speculation that

878 Vol. 10 lune 1996 The FASEB lournal GREEN AND IONES

this protein binds IF in a manner similar to desmoplakinand plectin. However, an unexpected feature of theBP230 knockout mice is that they develop a neuropathy

with similar pathological and behavioral features to mice

homozygous for the dystonia musculorum (dt/dt) mutation.This is not just a coincidence, because the dt allele hasbeen mapped to the same chromosomal location as thegene encoding the BP230 protein (71). Furthermore, se-quence analysis of a candidate dt gene (“dystonin”) hasrevealed that it encodes a BP230 isoform (“dystonin”) dif-fering in its amino terminus from the BP230 protein ex-pressed in epidermal cells (71). The putative

amino-terminal domain of this neural isoform shows se-quence similarity to several actin binding proteins in-cluding -spectrin and a-actinin, suggesting that theprotein product of the “dystonin” gene may bind both ac-

tin-containing microfilaments and IF (71). The possibilitythat “dystonin” is found in the nervous system equivalentof the hemidesmosome is an intriguing but untested idea.

“Dystonin” is likely to be just one of a family of BP230isoforms. Hopkinson and Jones (72) have characterized a

280 kDa isoform (“BP280”) in a pancreatic carcinomacell line that is expressed in a variety of cultured epi-thelial cells and epidermis but apparently is absent fromcells of mesenchymal origin (72). Although BP280 lacksthe carboxy-terminal IF binding domain, discussedabove, it distributes with the filamentous cytoskeleton incertain epithelial cells (72). This observation suggeststhat more than one domain may be involved in the inter-

action between the cytoskeleton and BP230 and its iso-forms. Such a possibility will need to be testedexperimentally, for example, using the type of molecular

genetic approach that has proved so successful for identi-fying functional domains in desmoplakin, plectin, andBP18O (34, 38, 55).

IFAP300/HD1/plectin

As already discussed, BP230 is a strong candidate for an

IF-hemidesmosome plaque linker. However, the IF-asso-ciated protein IFAP300 has also been identified in thehemidesmosome, in addition to the desmosome (39). Be-cause this protein binds IF in vitro, it could also functionto link IF to the cell surface. By analogy to the focal con-tact, which contains several actin binding proteins,BP230 and IFAP300 could well play complementaryroles in IF-hemidesmosome interactions. Indeed,IFAP300 and BP230 may be more than simply function-

ally related because a partial sequence of IFAP300 indi-cates that it possesses some sequence similarities toplectin, a member of the same family as BP230 and des-moplakin (39).

IFAP300 was first characterized as a 300 kDa proteinassociated with the vimentin cytoskeleton of fibroblastcells (73). It is immunologically related, if not identical,to a high molecular weight protein of bovine cornealhemidesmosomes termed HD1, which has also been pro-posed as an IF-hemidesmosome linker (74; J. C. R. Jones

and K. Owaribe, unpublished observations). HD1, likeIFAP300, is not restricted to hemidesmosomes but is ex-pressed in numerous epithelial cell types as well asneuronal cells (74). Because HD1 is often coexpressedwith a6134, it has been proposed that HD1/a64 com-plexes define a type II hemidesmosome, as we have al-ready mentioned (7).

ISSUES RISING AND PERSPECTIVES

Significant advances have been made in the identification

and characterization of desmosome and hemidesmosomecomponents. In many cases, molecular genetic analysishas allowed functional assignments for individual protein

domains. It is likely that state of the art assays such asthe yeast two hybrid approach will continue to identifyprotein-protein interactions in these junctions as well asnew components that may be present at substoichiometriclevels. In addition, high-resolution structural studies

such as that recently reported for the N-cadherin ex-

tracellular domain also promise to reveal new information

regarding the structural basis of cell adhesion, and ulti-mately of cytoplasmic interactions (75).

A major challenge will be to define signaling pathwaysthat regulate junction assembly and dissolution. Bothdesmosome and hemidesmosome assembly involve a spa-

tially and temporally regulated succession of protein-pro-tein interactions. It is generally believed that assembly ofjunctional plaques is triggered by the lateral associationor clustering of transmembrane protein complexes. In thecase of the desmosome, this idea is supported by the

studies of Troyanovsky (76) demonstrating that clustering

of connexin-desmosomal cadherin tail chimeras in theplane of the membrane recruits other plaque components

and leads to anchorage of the intermediate filament cy-toskeleton. Clustering of hemidesmosomal integrins isalso likely to play a role in nucleation of junction assem-

bly as well as recruitment of cytoplasmic components, in-

cluding IF bundles (45, 59).What actually leads to clustering of desmosomal cad-

herins during normal assembly is not well understood. Incultured cell systems, cell contact and calcium lead to a

temporal sequence that begins with homophilic adhesionvia classic cadherins and proceeds with adherens junc-

tion assembly. Desmosome assembly begins shortly there-after, and although data from different systems sometimesconflict, it appears to depend on a balance in protein ki-nase and phosphatase activity (see, for example, ref 77).Data from calcium induction experiments contrast withrecent evidence demonstrating that half desmosornes canbe assembled on their own without a counterpart on anadjacent cell and suggesting that cell contact/calciumserves not as a signal, but simply allows adjacent halfdesmosomes to interact through the desmosomal cad-berm’s extracellular domains (78).

Regulating junction dissolution may be equally impor-tant, particularly in processes such as wound healing and

ncckArcrkAec AI%Jfl kAIflkACkA a

invasion. A correlation has been made between growthfactor-dependent tyrosine phosphorylation of plakoglobinand a more invasive cell state (79); plakoglobin has also

been shown to bind to tyrosine kinase growth factor re-ceptors (80). Thus, in tumors where adhesion is compro-mised and motility is enhanced, phosphorylation of thecatenins or plakoglobin may contribute to loss of cad-hem function.

The function of individual proteins, though increasinglywell defined in cultured cells, for the most part stillneeds to be determined at a tissue level. In the case ofhemidesmosomes, recently identified inherited diseasesare providing important functional information regardingprotein components of hemidesmosomes. Althoughautoimmune diseases that target the desmoglein family of

desmosomal molecules are well recognized, so far nogenodermatoses have been attributed to mutations in des-mosomal components. However, certain genodermatoseshave been narrowed down to the desmosomal cadherin

chromosome, and within the next 5 years this frontier is

sure to be broken (81). In addition, studies in a variety ofdevelopmental model systems are beginning to elucidatethe function of individual proteins in morphogenesis anddifferentiation in complex systems and tissues in vivo,and in certain instances are yielding unpredicted andsurprising results. For example, gene ablation studies inmice have led to the discovery that a BP230 isoform has

unexpected functions in the nervous system (70), whereas

studies in flies and frogs have revealed that plakoglobinand its relatives play a central signaling role in embryo-genesis (20). Advances in in vitro and cell culture tech-niques, along with the insights provided by more complexmodel systems, promise to make the near future a time ofgreat strides in our understanding of cell junctions.

The authors would like to thank all of our colleagues who generouslyshared their recent published and unpublished work with us during the

preparation of this review. Regrettably, we were unable to include cita-tions for all original research due to space constraints, but in such cases

have attempted wherever possible to cite reviews covering all the work.Thanks go to A. Kowalczyk for critical reading of the manuscript and toS. Baker for assistance with the figures. Work from the laboratory of

K.J.G. is supported by NIH ROl AR41836, NIH ROl AR 43380,CB11OB, and a basic research grant from the March of Dimes. Work

from the laboratory of J.C.R.J. is supported by NIH ROl GM38470,American Cancer Society CB69, and the U.S. Army DAMD17-94-J-4291. K.J.G. is a recipient of an American Cancer Society Faculty

Research Award.

REFERENCES

1. Schmidt, A., Heid, H. W., Schafer, S., Nuber, U. A., Zimbelmann, K., andFranke, W. W. (1994) Desmosomes and cytoskeletal architecture in epi-thelial differentiation: cell type-specific plaque components and intermedi-ate filament anchorage. Ear.).CellBiol. 65, 229-245

2. Garrod, D. R. (1993) Desmosomes and hemidesmosomes. Curr. Opin.CellBiol.5,30-40

3. Jones, J. C. R., Asmuth, J., Baker, S. E., Langhofer, M., Roth, S. 1., andHopkinson. S. B. (1994) Hemidesmosomes: extracellular matrix/intermedi-ate filament connectors. Exp. Cell Res. 213, 1-11

4. Tamura, K. N., Rozzo, C., Starr, L, Chambers, J.. Reichardt, L. F., Cooper,H. M., and Quaranta, V. (1990) Epithelial integrin a634: complete primary

structure of a6 and variant forms of 4. J.CellBiol. 111, 1593-1604

5. Kallunki, P., Saino, K., Eddy, R., Byers, M., Kallunki, T., Sariola, H., Beck,K., Hirvonen, H., Shows, T. B., and Tiyggvason, K. (1992) A truncatedlaminin chain homologous to the B2 chain: structure, spatial expression andchromosomal assignment.). CellBiol.109,3377-3390

6. Hogervorst, F., Admiraal, L. G., Niessen, C., Kuikman, I., Janssen, H.,Daams, H., and Sonnenberg, A. (1993) Biochemical characterization andtissue distribution of the A and B variants of the integrin ct6 subunit.). CellBwl. 121, 179-191

7. Uematsu, J., Nishizawa, Y.. Sonnenberg, A., and Owaribe, K. (1994) Dem-onstration of type II hemidesmosomes in a mammry gland epithelial cell line,BMGE-H.). Biochem. 115,469-476

8. Collins, J. E., and Garrod, D. R. (1994). Molecular Biology of Desmoso,nesand Hemidesmosomes, R. C. Landes Co., Austin, Texas

9. Koch, P. J., and Franke, W. W. (1994) Desmosomal cadherins: anothergrowing multigene family of adhesion molecules. Curr.Opin.CellBiol.6,682-687

10. Cowin, P. (1994). Plakoglobin. In Molecular Biology of Desinosomes andHemidesmosomes (J. E. Collins and D. R. Garrod, eds) pp. 1-131, R. C.Landes Co., Austin, Texas

11. Troyanovsky, S. M., Troyanovsky, R. B., Eshkind, 1. C., Leube, R. E., andFranke, W. W. (1994) Identification of amino acid sequence motifs indesmocollin, a desmosomal glycoprotein, that are required for plakoglohinbinding and plaque formation. Proc. Nail. Acad. Sd. USA 91, 10790-10794

12. Koch, P. J., Goldschmidt, M. D., Zimbelmann, K., Troyanovsky, K., andFranke, W. W. (1992) Complexity and expression patterns of the desmosomalcadherins. Proc. Nail. Acad. Sci. USA 89, 353-357

13. King, I. A., Sullivan, K. H., Bennett, R., and Buxton, K. S. (1995) Thedesmocollins of human foreskin epidermis: identification and chromosomalassignment of a third gene and expression patterns of the three isofonns. J.Invest. Derm. 105, 314-321

14. Yue, K. K. M., Holton, J. 1., Clarke, J. P., Hyam, J. L. M., Hashimoto, T..Chidgey, M. A. J., and Garrod, D. R. (1995) Charactensation of a desmocollinisoform (bovine DSC3) exclusively expressed in lower layers of stratifiedepithelia.). CellSci.108,2163-2173

15. Simrak, D., Cowley, C. M. E., Buxton, K. S., and Arnemann, J. (1995)Tandem arrangement of the closely linked desmoglein genes on humanchromosome 18. Genomics 25, 591-594

16. Amagai, M., Karpati, S., Prussick, R., Klaus-Kovtun, V., and Stanley, J. K.(1992) Autoantibodies against the amino-terminal cadherin-like bindingdomain of pemphigus vulgaris antigen arc pathogenic. J. Clin. Invest. 90,919-926

17. Amagai, M., Karpati, S., Klaus-Kovtun, V., Udey, M. C., and Stanley, J. R.(1994) The extracellular domain of pemphigus vulgaris antigen (desmoglein3) mediates weak homophilic adhesion.). Invest. Derm. 102,402-408

18. Brakenhoff, R. H., Gerretsen, M., Knippels, E. M. C., Dijk, M. v., Essen, H.v., Weghuis, D. 0., Sinke. K. J., Snow, C. B., and Dongen, C. A. M. S. v.(1995) The human E48 antigen, highly homologous to the murine Ly-6antigen ThB, is a GPI-anchored molecule apparently involved in keratinocytecell-cell adhesion. J. Cell Biol. 129, 1677-1689

19. Peifer, M. (1995) Cell adhesion and signal transduction: the armadilloconnection. Trends Cell Biol. 5, 224-229

20. Klymkowsky, M. W., and Parr, B. (1995) The body language of cells: theintimate connection between cell adhesion and behavior. Cell 83, 5-8

21. Mathur, M., Goodwin, 1., and Cowin, P. (1994) Interactions of the cytoslas-mic domain of the desmosomal cadherin Dsgl with plakoglohin. J.Biol.Chem. 269, 14075-14080

22. Kowalczyk, A. P., Palka, H. L., Lou, H. H., Nilles, L. A., Anderson, J. E.,Wheelock, M. J., and Green, K. J. (1994) Posttranslational regulation ofplakoglobin expression: Influence of the desmosomal cadherins on plakoglo-bin metabolic stability. I. Biol. Chem. 269, 31214-31223

23. Rob, J.-Y., and Stanley, J. R. (1995) Plakoglobin binding by human Dsg3(pemphigus vulgaris antigen) in keratinocytes requires the cadherin-likeintracytoplasmic segment.). Invest. Derm. 104, 720-724

24. Troyanovsky, S. M., Troyanovsky, R. B., Eshkind, L. C., Krutovskikh, V. A.,Leube, R. E., and Franke, W. W. (1994) Identification of the plakoglohin-binding domain in desmoglein and its role in plaque assembly and interme-diate filament anchorage.). Cell Biol. 127, 151-160

25. Peifer, M.. McCrea, P. D., Green, K. J., Wiesehaus, E., and Gumbiner, B. M.(1992) The vertebrate adhesive junction proteins Il-catenin and plakoglobinand the Drosophila segment polarity gene armadillo form a multigene family

with similar properties. J. Cell Biol. 118, 681-69126. Nathke, I.S., Hinck, L., Swedlow, J. K., Papkoff, J., and Nelson, W. J. (1994)

Defining interactions and distributions of cadhenn and catenin complexes inpolarized epithelial cells.). CellBiol.125, 1341-1352

27. Funayama, N., Fagotto, F., McCrea, P., and Combiner, B. M. (1995) Embry-onic axis induction by the armadillo repeal domain of -catenin: evidencefor intracellular signaling.). Cell Biol. 128,959-968

28. Kamovsky, A., and Klymkowsky, M. W. (1995) Anterior axis duplication inXenopus induced by the over-expression of the cadherin-hinding proteinplakoglobin. Proc.Nail.Acad.Sci.USA 92, 4522-4526

29. Rubinfeld, B., Souza, B., Albert, I., Muller, 0.. Chamberlain, S. H., Masiarz,F. R., Munemitsu, S., and Polakis, P. (1993) Association of the APC geneproduct with -catenin. Science 262, 1731-1734

30. Su, K.-K., Vogelstein, B., and Kinzler, K. W. (1993) Association of the APCtumor suppressor protein with calenins. Science 262, 1734-1737

880 Vol. 10 June 1996 The FASEB Journal GREEN AND JONES

31. Hatifeld, M.. Kristjansson. C. I., Plessmann, U., and Weber, K. (1994) Band6 protein, a major constituent of desmosomes from stratified epithelia, is anovel memherof the armadillo multigene family.). Cell Sci. 107,2259-2270

32. Held, H. W., Schmidt, A., Zimbelmann. K., Schafer, S., Winter-Simanowski,S., Stumpp, S., Keith, M., Figge, U., Schnolzer, M., and Franke, W. W. (1994)Cell type-specific desmosomal plaque proteins of the plakoglohin family:plakophilin 1 (band 6 protein). Dtffereniiadon 58, 113-131

33. Green, K. J., Vii-ata, M. L A., Elgart, C. W., Stanley, J. R., and Parry, D. A.D. (1992) Comparative structural analysis of desmoplakin, bullous pem-phigoid antigen and plectin: members of a new gene family involved inorganization of intermediate filaments. mi. J.Biol.Macromol. 14, 145-153

34. Stappenbeck, T. S., and Green, K. J. (1992) The desmoplakin carboxylterminus coaligns with and specifically disrupts intermediate filament net-works when expressed in cultured cells.). Cell Biol. 116, 1197-1209

35. Kouklls, P. D., Hutton, E., and Fuchs, E. (1994) Making a connection: directbinding between keratin intermediate filaments and desmosomal proteins.).Cell Biol. 127, 1049-1060

36. Stappenbeck. T. S., Bornsfueger, E. A., Corcoran, C. M., Luu, H. H., Virata,M. L. A.. and Green. K. J. (1993) Functional analysis of desmoplakindomains: specification of the interaction with keratin versus vimentin inter-mediate filament networks.). Cell Biol. 123, 691-705

37. Stappenbeck, 1. S., Lamb, J. A., Corcoran, C. M., and Green, K. J. (1994)Phosphorylation of the desmoplakin COOH terminus negatively regulates itsinteraction with keratin intermediate filament networks.). Biol. Chem. 269,29351-29354

38. Wiche, C.. Cmmov, D., Donovan, A., Castanon, M. J., and Fuchs, E. (1993)Expression of plectin mutant eDNA in cultured cells indicates a role ofCOOH-terminal (lomain in intermediate filament association.). Cell Biol.121,607-619

39. Skalli, 0., Jones, J. C. R., Gagescu. R., and Goldman, K. D. (1994) IFAP300 is common to desmosonies and hernidesmosomes and is a possible linkerof intermediate filaments to these Junctions.). Cell Biol. 125, 159-170

40. Clark, E. A., and Biugge, J. S. (1995) Integrins and signal transduction

pathways: the road taketi. Science 268, 233-23941. Ciancotti, F. C., Stepp, M. A.. Suzuki, S., Engvall, E., and Ruoslahti, E.

(1992) Proteolytic processing of endogenous and recombinant 34 integrin

sulsunit. J.CellBiol.118,951-95942. Hogervorst, F., Kuikman. I., Borne, A. E. C. K. v. d., and Sonnenherg, A.

(1990) Cloning and sequence analysis of beta-4 eDNA: an integrin subunitthat contains a unique 118 kDa cytoplasmic domain. EMBO). 9,765-770

43. Kurpakus. M. A., Quaranla. V., and Jones, J. C. R. (1991) Surface relocationof alpha6 beta4 integtins and assembly of hemidesmosomes in an in vitromodel of wound healing.). Cell Biol. 115,1737-1750

44. Spinardi, L., Ben, Y. 1.., Sanders, R., an(l Ciancotti, F. C. (1993) The 34subunit cytoplasmic domain mediates the interaction of a64 integrin withthe cytoskeleton of hemidesnsosomes. Mol. Biol. Cell 4,871-884

45. Spinardi, L., Einheber, S., Cohen, T., Miltier, T. A.. and Giancotti, F. C.(1995) A recombinant tail-less integrin I4 subunit disrupts hemides-mosomes. but clues not suppress a634-mec1iated cell adhesion to laminins.). Cell Biol. 129, 473-487

46. Riddelle. K. S., Green, K. J.,and Jones, J. C. R. (1991) Formation ofhemidesmosomes in vitro by a transformed rat bladder cell line.). Cell Biol.112, 159-168

47. Vidal, F.. Aherdain, D., Miquel, C., Cliristiano. A. M., Pulkkinen, L., Uitto,J., Ortonne, i-P., and Meneguzzi. C. (1995) Integrin I4 mutations associatedwith junctional epidt’rmolysis bullosa with pyloric atresia. Nature Genet. 10,229-234

48. Mainiero, F., Pepe, A.. Wary, K. K., Spinardi, L., Mohammadi, M., Schiess.inger, J., and Giancotti, F. C. (1995) Signal transduction by the a6134integrin: distinct 34 subunit sites mediate recruitment of the Shc/Grb2 andassociation with the cytoskeleton of hemidesmosomes. EMBO ). 14.4470-4481

49. Ilopkinson, S. B., Ridclelle, K. S.. and Jones, J. C. R. (1992) Cytoplasmicdomain of the 180-kD hulfous pemphigoid antigen, a hemidesmosomalcomponent: molecular and cell biologic characterization.). Invest. Derm. 99,264-270

50. (;,udk, C. J., Squiquera, H. L., Elias, P. M., and Diaz, L. A. (1991)Identification of two collagen domains within the bullous pemphigoid autoan-tigen BP18O.). Clin. invest. 87,734-738

51. Jonkman, M. F., Joiig, M. C. J. M. d., Heeres, K., Pas, H. H., Meer, J. B. v.d., Owarihe, K., Velasco, A. M. M. d., Niessen, C. M., and Sonnenberg, A.180-kD Isullous pernphigoid antigen (BP18O) is deficient in atrophic benignepidermolysis hullosa.). Clin. lnvest. 95, 1345-1352

52. McGrath, J. A., Gatalica, B., Christiano, A. M., Li, K. H., Owaribe, K.,McMillan,J. R., Eacly, R. A. J., and Uitto, J. (1995) Mutations in the 180-kDbullous pemphigotd antigen (BPAG2), a hemidesmosomal transmembranecollagen )COL17AI), in generalized atrophic benign epidermolysis bulfosa.Nature Genet. 11,83-86

53. Ciudice, C.J., Emery, D.J., Zelickson, B. D., Anhaf t, Ci., Liu, Z., and Diaz,L. A. (1993) Bullous pemphigoid and herpes gestationis autoantibodiesrecognize a common non-collagenous site on the BP18O domain.). immunol.151,5742-5750

54. l.iu, Z., Diaz, L.A., Troy, J. L., Taylor, A. F., Emery, D. J., Faiifey, J. A., andCiudice. C. J. (1993) A passive transfer model of the organ-specific autoim-

bone disease, bullous pemphigoicl using antibodies generated against thehemidesinosomal antigen, 8P180. J.Clinlnvest. 92, 2480-2488

55. Hopkinson, S. B., Baker, S. E., and Jones, J. C. R. (1995) Molecular geneticstudies of a human epidermal autoantigen (the 180-kD hullous pemphigoidantigenlBPl8O): identification of functionally important sequences withinthe BP18O molecule and evidence for an interaction between BP18O and aointegrin.). Cell Biol. 130, 117-125

56. Tryggvason, K. (1993) The laininin family. Curr. Opin. Cell Biol. 5.877-88257. Rousselle, P., and Aumailley, M. (1994) Kahinin is more efficient than

faminin in promoting adhesion of primary keratinocytes and some otherepithehial cells and has a different requirement for integrin receptors.). CellBiol. 125, 205-214

58. Carter, W. C., Kaur, P., Gil, S. C., Gahr, P. J., and Wayner, E. A. (1990)Distinct functions for integrins alpha3betal in focal adhesions and al-pha6heta4/bullous pemphigoid antigen in a new stable anchoring contact(SAC) of keratinocytes: relation to hemidesmosomes. ). Cell Biol. 111.3141-3154

59. Langhofer, M., Hopkinson, S. B., and Jones, J. C. K. (1993) The matrixsecreted by 804C cells contains laminin related components that participatein hemidesmosome assembly in vitro. J.CellSci.105, 753-764

60. Pyke, C., Romer, J., Kallunki, P., Lund, L. R., Kalfkiaer, E., Dano, K., andTryggvason, K. (1994) The gamma 2 chain of kahinin/laminin 5 is preferen-tially expressed in invading malignant cells in human cancers. Am.). Pathol.145, 782-791

61. Marinkovich, M. P., Lunstrum, C. P., arsd Burgeson, R. E. (1992) Theanchoring filament protein kalinin is synthesized and secreted as a highmolecular weight precursor.). Biol. Chern. 267, 17900-17906

62. Ryan, M. C., Tizard, R., VanDevanter, D. K., and Carter, W. C. (1994)Cloning of the LamA3 gene encoding the alpha 3 chain of the adhesive ligandepiligrin. Expression in wound repair.). Biol. Chem. 269, 22779-22787

63. Domhoge-Huhtsch, N., Gammon, W. R., Briggaman, K. A., Gil, S. C., Carter,W. C., and Yancey, K. B. (1992) Epihigrin, the major human keratinocyteintegrin ligand, is a target in both an acquired autoimmune and an inheritedsubepidernial blistering skin disease.). Clin Invest. 90, 1628-1633

64. Aherdam. D., Calhian, M.-F., Vailley, J., Bulkkinen, L., Bonifas, I., Chris-tiano, A. M., Tryggvason, K., Uitto, J., Jr., E. H. E., Ortonne, 1.-P., andMeneguzzi, C. (1994) Herlitz’s junctional epidermolysis bullosa is linked tomutations in the gene (LAMC2) for the g2 subunit of nicein/kahinirs (laminin-5). Nature Genet. 6, 299-304

65. Kivirikko, S., McGrath, I. A., Baudoin, C., Aberdam, D., Ciatti, S., Dunnill,M. C., McMillan, I. K., Eady, B. A., Ortonne, i-P., Meneguzzi, C., Uitto. J.,and Christiano, A. M. (1995) A homozygous nonsense mutation in the a3chain gene of haminin 5 (LAMA3) in lethal (Herlitz) junctional epidermolysishuh loss. Human Mol. Genet. 4, 959-962

66. Pulkkinen, L., Christiano, A. M., Airenne, T., Haakana, H., Tryggvason, K.,and Uitto, J. (1994) Mutations in the ‘(2 chain gene (LAMC2) of ka-hinirVlaminin 5 in the junctional forms of epidermolysis hullosa. NatureGenet. 6. 293-298

67. Puhkkinen, L.. Christiano. A. M., Gerecke. B., Wagnian. D. W.. Burgeson,RE., Pittelkow, M. B., and Uitto,J. (1994) A homozygous nonsense mutationin the 33 chairs gene of lansinin S (LAMB3) in Herhitz junctional epider-molysis bullosa. Genomics 24, 357-360

68. Tanaka, T., Parry, D. A. D., Klaus-Kovtun, V., Steinert, P. M., and Stanley,I. R. (1991) Comparison of molecularly cloned buhlous pemphigoid antigento clesmopf akin I confirms that they define a new family of cell adhesionjunction plaque proteins.). Biol. C/rem. 266, 12555-12559

69. Klatte, D. H., and Jones, J. C. R. (1994) Purification of the 230-kB bullouspemphigoid antigen (BP230) from bovine tongue mucosa: structural analysisanch assessment of BP230 tissue distribution using a new monocfonal anti-body.). Invest. Derm. 102, 39-..44

70. Cuo, L.. Degenstein, L., Bowling, J., Yu, Q.-C., Wohlmann, R., Perman, B.,and Fuchs, E. (1995) Gene targeting of BPAG1: alsnormalities in mechanicalstrength and cell migration in stratified epithelia and neurologic degenera-tion. Cell 81,233-243

71. Brown. A., Bemier, C., Mathieu, M., Rossaist, i.. and Kothary, R. (1995) Themouse dystonia inusculorum gene is a neural isoform of bullous pemphigoidantigen 1. Nature Gene:. 10.301-306

72. Hopkinson, S. B., arid Jones, J. C. R. (1994) Identification of a second proteinproduct of the gene encoding a human epidermal autoantigen. Biochem. J.300,851-857

73. Yang, H.-Y.. Lieska, N., Goldman, A., and Goldman, B. (1985) A 300,000MW intermediate filament-associated protein iii baby hamster kidney (BHK-21) cells. .1. Cell Biol. 100,620-631

74. Hieda, Y., Nishizawa, Y., Uematsu, J., and Owarihe, K. (1992) Identificationof a new hemidesmosomal protein. HD1: a major, high molecular masscomponent of isolated hemidesmosomes.). CellBiol.116, 1497-1506

75. Shapiro, L., Fannon, A. M., Kwong, P. B., Thompson, A., Lehmann, M. S.,Crubel, G., Legrand,J.-F., Als-Nielson,J., Colman, D. K., and Hendrickson,W. A. (1995) Structural basis of cell-cell adhesion by cadherins. Nature(London) 374, 327-337

76. Troyarsovsky, S. M.. Eshkind, L. C., Tnsyanovsky, K. B., Leuhe, K. E., aridFranke, W. W. (1993) Contributions of cytoplasmic domains of desmosomalcadherins to desmosome assembly and intermediate filament anchorage. Cell72, 561-574

DESMOSOMES AND HEMIDESMOSOMES

77. Pasdar, M., Li. Z., and Chan, H. (1995) Desmosome assembly and disassem-bly are regulated by reversible protein phosphorylation in cultured epithelialcells. CellMotil.Cytoskel. 30, 108-121

78. Demlehner, M. P., Schafer, S., Grund. C., and Franke, W. W. (1995)Continual assembly of half-desmosomal structures in the absence of cellcontacts and their frustrated endocytosis: a coordinated Sisyphus cycle. I.CellBiol.131, 745-760

79. Shibamoto, S., Hayakawa, M., Takeuchi, K.. Hori, T.. Oku, N., Miyazawa,K., Kitamura, N., Takeichi, M., and Ito, F. (1994) Tyrosine phosphorylationof -catenin and plakoglobin enhanced by hepatocyte growth factor and

epidermal growth factor in human carcinoma cells. CellAdjrer. Common. 1,295-305

80. Ochiai, A., Akimoto, S., Kanai, Y., Shibata, T, Oyama. T., and Himhashi, S.(1994) c-erbB-2 gene product associates with catenins in human cancercells. Biochem. Biophys. Res. Commun. 205,73-78

81. Hennies, H. C., Kuster, W., Mlschke, D., and Reis, A. (1995) Lecalizationof a locua for the striated form of palmoplantar keratoderma to chromosome18q near the desmosomal cadherin gene cluster. Human Mol. Genet. 4,1015-1020

COMMUNICATIONS CAPSULES

The following are summaries of original research articles that appear in this issue.

MELATONIN EXERTS CYTOPRO-TECTIVE EFFECT

In the Harderian gland of female Syr-ian hamsters, an experimental model

for human prophyries, melatonin ad-ministration decreases the cell dam-age and subsequent cell necrosiscaused by porphyrins (Antolin etal., pages 882-890). This neuro-

hormone exerts its cytoprotec-tive effect by lowering the levelof aminolevulinate synthasemRNA (and hence porphyrinsynthesis), as well as by increasing

the mRNA levels of antioxidant en-zymes, i.e., manganese superoxidedismutase and copper-zinc superox-ide dismutase.

IN VIVO NEUROPROTECTIONBY MELATONIN

Recent in-vitro studies have sug-gested that the pineal hormone mela-tonin may be used as aneuroprotective agent. Giusti et al.(pages 891-896) induced braindamage in rats in vivo by kainate, anagonist for the receptors for the exci-tatory neurotransmitter glutamate. Asa result of kainate injection, the ratsexperienced epilepsy-like seizuresthat were followed by biochemicalchanges in the brain and ultimatelyby neurodegeneration. Intraperitonealadministration of melatonin attenu-ated the severity of the seizures andreduced brain damage. A possiblemechanism of neuroprotection

lies in the antioxidant action ofthis hormone.

UPPER AIRWAY MUSCLE IN

SLEEP APNEA

Awake and sleeping upper airway

(UA) collapsibility as well as thephysiologic (maximum isometrictwitch and tetanic tension, fatigabilitymeasurements), metabolic (enzymeactivities of the anaerobic and aerobicmetabolic pathways), and histochemi-

cal characteristics (fiber type propor-tions and areas) of the uvular muscle

were obtained in 17 sleep apnea pa-tients and 3 snorers (Series et a!.,pages 897-904). Maximum ten-sions, fatigability index, anaerobicenzymes, and the percentage of sur-face occupied by type I and type hAmuscle fibers were correlated with

awake collapsibility. It is concludedthat the differences in awake UAcollapsibility contribute to deter-mine the contractile propertiesand metabolic and histochemicalcharacteristics of the genioglos-sus muscle of the tongue.

SELECTIVE ESTROGEN RECEP-TOR MODULATORS IN VIVO

Raloxifene, tamoxifen, nafoxidine,and estrogen represent four chemi-cally dissimilar compounds known to

bind to the estrogen receptor in vivo(Sato et al., pages 905-912).These clinically important com-pounds had a different profile of ef-

fects on reproductive and non-repro-ductive tissues in ovariectomized rats,

with specific advantages of differentcompounds in the management ofbone, uteri, serum lipids, or bodyweight, postovariectomy. Thesecompounds appear to be selec-tive estrogen receptor modula-tors that differentially affect estrogenresponsive tissues, depending on theligand in vivo.

RF RADIATION ALTERS LYM-PHOCYTE PROLIFERATION

A mechanism for radiofrequency (RF)radiation effects on mammalian cellproliferation was investigated byCleary et al. (pages 913-919).Theoretical studies had indicatednonuniform RF energy absorption at

membrane surfaces. The hypothesis

that RF radiation alters membranesignal transduction was tested by ex-posing IL-2-dependent cytolytic T-lymphocytes (CTLL-2) to variousdoses of radiation, while holding thetemperature constant at 37#{176}C.Prolif-eration was assayed immediately and24 h after exposure. Radiation fre-quency-dependent and dose-depend-ent proliferative alterations occurred,but only when there were IL-2 recep-tors on the CTLL-2 cell surface dur-ing exposure. These results supportthe predicted RF radiation effect oncell signal transduction: alteredbinding of IL-2 to its membranereceptor leads to changes inCTLL-2 cell proliferation.