J. Cell Set. Suppl. 4, 221-237 (1986)
Printed in Great Britain © The Company of Biologists Limited 1986
221
DESMOSOMES, CELL ADHESION MOLECULES AND
THE ADHESIVE PROPERTIES OF CELLS IN TISSUES
D .R . GARROD
Cancer Research Campaign Medical Oncology Unit, University of Southampton, Centre
Block, CF99, Southampton General Hospital, Tremona Road, Southampton S09 4XY, UK
INTRODU CTION
In this article intercellular adhesion of vertebrate tissue cells, especially epithelia,
will be discussed. The underlying theme may be expressed by the question ‘What
cellular adhesive properties are required in order to develop and maintain the
organization and structure of normal tissues?’ I shall begin by outlining a number of
principles and properties of cell adhesions that have been presented previously
(Garrod, 1981; Garrod & Nicol, 1981; Garrod, 1985) and can be updated and
extended following recent studies on the molecular basis of cell adhesion. Detailed
consideration will then be given by the principal adhesive junctions of epithelial cells,
desmosomes or maculae adhaerentes, particularly in so far as recent studies illustrate
general principles of adhesion. Advances in the study of other cell-cell adhesion
mechanisms will also be considered. Cell adhesion has been reviewed recently by
Damsky, Knudsen & Buck (1984), and Edelman (1984). It is hoped that the material
presented here will be largely complementary to those reviews.
PR INCIPLES OF TISSUE CELL AD HESION
A cell has several adhesion mechanisms
Individual cells of a given type possess a variety of different molecular adhesion
mechanisms, some junctional and some non-junctional (Garrod, 1981; Garrod &
Nicol, 1981; Garrod, 1985). The most complex situation probably arises in the socalled simple epithelial cell, which is shown in a generalized diagram in Fig. 1. This
illustrates five different mechanisms of cell-cell adhesion and six different mech­
anisms of cell-matrix adhesion. Of the intercellular adhesion mechanisms, the tight
junction and gap junction have other well-established functions, occluding the
intercellular space and intercellular communication, respectively. These functions
depend upon strong intercellular binding, so that the junctions must contribute to
the adhesiveness of the cell. Some earlier electron-microscopical observations
suggested that, under some experimental circumstances, intercellular bonding at
these junctions is more resistant to disruption than that at the other junctional types
(Sedar & Forte, 1964; Muir, 1967; Berry & Friend, 1969). The adhesive function of
the desmosome and the zonula adhaerens are widely recognized. These, together
with non-junctional adhesion molecules, probably constitute the major adhesive
222
D. R. Garrod
Fig. 1. Diagram showing multiple adhesion mechanisms of epithelial cells. (N.B. This
diagram shows a probable maximum number of mechanisms not all of which are present
in all epithelial cells.)
mechanisms in most cell types. In a previous review (Garrod, 1985) it was suggested
that so-called liver cell adhesion molecule (L-CAM) (Edelman, 1984) might be
involved in non-junctional membrane adhesion. However, it has recently been
shown that uvomorulin (probably equivalent to L-CAM) is concentrated in the
zonula adhaerens of intestinal epithelial cells (Boiler, Vestweber & Kemler, 1985). It
may therefore be necessary to postulate another molecular adhesion mechanism or
Desmosomes, cell adhesion
223
mechanisms for non-junctional membranes. In liver, this could be the molecule
known as cell CAM-105 (Ocklind & Obrink, 1982). It is also possible that adhesion
molecules that are concentrated at junctions may be present at lower density on other
parts of the cell surface, where they may mediate adhesion.
The cell-matrix adhesion mechanisms consist of receptors for matrix components
and two ‘junctional’ types, the hemidesmosome and the focal contact. They have
been discussed in a previous review where appropriate references are given (Garrod,
1985).
Epithelial cells possess a number of adhesion mechanisms because they require a
range of different adhesive properties in order to maintain their epithelial organ­
ization (Garrod, 1985). They need links (desmosomes) between their intermediate
filament cytoskeletons in order to provide structural strength and continuity
throughout the tissue (Arnn & Staehelin, 1981; Ellison & Garrod, 1984; Docherty,
Edwards, Garrod & Mattey, 1984; Garrod, 1985), they need links (zonulae adhaerentes) between their contractile microfilaments (Baker & Schroeder, 1967;
Karfunkel, 1971; Burgess, 1982; Garrod, 1985), they need to bind to and interact
with their substratum (matrix receptors and hemidesmosomes) and to form links
between their actin cytoskeleton and the matrix (focal contacts) to facilitate move­
ment during morphogenesis or wound healing (Billig et al. 1982). Each of these
functions involves a specialized adhesion mechanism that has evolved in response to
the need for its particular function. Simple epithelial cells may possess nine or ten
different adhesion mechanisms, in order to fulfil a complex set of adhesive
requirements.
This plurality of adhesion mechanisms is not restricted to epithelial cells. Nerve
cells may possess a minimum of four mechanisms, the neural cell adhesion molecule
(N-CAM) (Thiery, Brackenbury, Rutishauser & Edelman, 1977), the LI antigen
(Rathjen & Schachner, 1984) (now shown to be definitely distinct from N-CAM;
Rathjen & Rutishauser, 1984), the calcium-dependent adhesion mechanism recog­
nized by monoclonal antibody NCD-1 (Hatta, Okada & Takeichi, 1985) and the
neuron—glial cell adhesion molecule (Ng-CAM) (Thiery, Delouvee, Grumet &
Edelman, 1985). They also possess cell-matrix adhesion mechanisms (Cole &
Glaser, 1984; Schubert & LaCorbiere, 1985).
Fibroblasts are probably less complex than either epithelial cells or neurones but
certainly form adhaerens-type junctions (Heaysman & Pegrum 1973), and gap
junctions (Pitts, 1978), and possess calcium-dependent and calcium-independent
adhesion mechanisms (Takeichi, Atsumi, Yoshida & Okada, 1981) as well as various
types of cell-substratum adhesions.
The possession of multiple adhesion mechanisms is the general rule among tissue
cells. Any cells that are shown to possess only a simple mechanism will be highly
specialized for a particular adhesive function.
Finally, in this context the term cellular adhesiveness is sometimes used in a
strictly quantitative sense. For example, quantitative differences in adhesiveness
have been invoked to explain the experimentally induced phenomenon known as
sorting-out (Steinberg, 1964). The present analysis demonstrates that the total
D. R. Garrod
224
adhesiveness of a cell at a given time in a given situation receives a quantitative
contribution from those of its molecular adhesive mechanisms that are functional at
that time and in that situation. When we have identified all the molecular adhesive
mechanisms in a cell, quantified their expression on the cell surface and determined
the strength of the adhesive bonds that they form, we shall be able to compute
cellular adhesiveness for that cell type.
CELLS OF DIFFEREN T TYPES SHARE AD HE SION MECHANISMS
Ultrastructural studies have shown that intercellular junctions are similar in cells
of different tissues and different species (McNutt & Weinstein, 1973; Overton,
1974a; Staehelin, 1974). It is believed that all epithelia possess desmosomes, except
the pigmented epithelium of the eye (Middleton & Pegrum, 1976; Nicol & Garrod,
1982; Docherty et al. 1984) and lens epithelium. Moreover, the only non-epithelial
tissue that possesses desmosomes is cardiac muscle, where they occur in the inter­
calated discs. Endothelial cells do not possess desmosomes, except in fish (Cowin,
Mattey & Garrod, 1984a).
The distribution of cell adhesion molecules (CAMs) has been studied by Edelman
and his colleagues (Edelman, 1983, 1984). L-CAM is present in adult animals in the
stratum germinativum of the skin, the epithelium of the urinogenital, digestive and
respiratory tracts, the lymphoid organs and secretory glands. L-CAM is probably
similar or identical to the other calcium-dependent adhesion molecules isolated from
other sources by other workers. These include uvomorulin (Hyafil, Morello, Babinet
& Jacob, 1980), cell CAM 120/80 (Damsky et al. 1983) and cadherin (YoshidaNoro, Suzuki & Takeichi, 1984) (now called E-cadherin (Hatta et al. 1985). These
also occur in a wide distribution and in different animal species (mammals and
birds). (If these molecules are identical, the name cadherin would seem preferable to
the others, which all carry somewhat restrictive connotations with regard to function,
distribution or molecular weight.)
N-CAM is present in adult animals in nervous system, testis and cardiac muscle,
and has been found in fish, lizard, frog, chick, mouse, rat and human (Edelman,
1984). A very similar tissue distribution to that of N-CAM (which is a calcium­
independent mechanism) has been found for the calcium-dependent neural cell
adhesion mechanism, N-cadherin (Hatta et al. 1985). Ng-CAM, is found exclusively
in the nervous system (Thiery et al. 1985).
Some adhesion molecules appear to have a fairly restricted distribution. Antibody
studies showed that cell CAM-105, first identified in adult rat hepatocytes (Ocklind
& Obrink 1982), was found only in mouse liver, rat liver, rat kidney and rat small
intestine (Vestweber et al. 1985). However, the survey was rather limited and the
failure of antibodies to cross-react does not necessarily mean that the molecule is
absent from a tissue.
The general conclusion from study of the distribution of adhesion molecules is
that adhesion mechanisms are widely shared between different cell types. In his
‘modulation’ theory of cell adhesion, Edelman has asserted that, in general, tissues
Desmosomes, cell adhesion
225
will have only a few cell-cell adhesion molecules, corresponding in number perhaps
only to the number of classes of cells and tissues (Edelman, 1983). This formulation
is in direct contrast to previous theories, which require a large repertoire of cell typespecific adhesion molecules (Moscona, 1962; Sperry, 1963). We have pointed out
that such a large repertoire of molecules is not required in order to account for the
adhesive behaviour of cells (Garrod, 1981; Garrod & Nicol, 1981). I believe that
there are indeed a limited or restricted number of molecular adhesion mechanisms,
in fact far fewer than there are differentiated cell types within the vertebrate body.
However, the number of molecules appears to be greater than that suggested
by Edelman, because cells of any one tissue class clearly possess a number of
mechanisms.
One of the consequences of the sharing of adhesion mechanisms is that cells from a
wide variety of tissues and from different animal species exhibit mutual adhesion. A
cell has a much greater potential for adhesion to other cell types than it shows in its
normal tissue location (Garrod, 1981). Desmosomes (Overton, 19746; Mattey &
Garrod, 1985a; and below), adhaerens type junctions (Armstrong, 1970, 1971) and
gap junctions (Epstein & Gilula, 1977) have all been shown to be capable of forming
between some different types of cells. Like L-CAM and N-CAM (Edelman, 1984),
these junctions probably represent examples of ‘homophilic’ adhesive binding in
which the adhesion molecules bind to the same or similar adhesion sites on the other
cell. Thus molecularly specific adhesion binding can give rise to non-specific cell-cell
adhesion, and in doing so shows that the adhesion binding sites are conserved
between different tissues and different species (see later discussion on desmosomes).
Another type of adhesion between different cell types arises from ‘heterophilic’
binding, in which a CAM recognizes a different molecule on another cell. Examples
are Ng-CAM, which is involved in neuron-glial cell adhesion and is present only on
neurons (Edelman, 1984; Thiery et al. 1985), and the Ji glycoprotein, which is
involved in astrocyte-neuron adhesion and is present only on astrocytes (Kruse et al.
1985). In neither case has the complementary molecule been identified.
THE D IFFER IN G ADHESIVE PROPERTIES OF DIFFERENT CELL TYPES
The fact that cells share adhesion mechanisms, and if confronted experimentally
can use these to form mutual adhesions, makes it necessary to list the ways in which
the adhesive properties of different cell types can differ from each other.
(a) Cells can possess different combinations of intercellular junctions and CAMs.
(b) Cells can possess different numbers of intercellular junctions and CAMs.
(c) Cells can possess different surface distributions of junctions or CAMs.
(d) Cells can have adhesion mechanisms of different stability, i.e. adhesion may
be reversible or irreversible.
Differences in category (a) are qualitative and have been amply illustrated above.
Differences in categories (b) to (d) are more subtle. A single adhesion mechanism
such as the desmosome, may vary in one or all of these ways between different cell
types. I shall now review recent work on desmosomes laying stress on two aspects:
D. R. Garrod
226
how desmosomes in different tissues resemble each other and how desmosomes can
give rise to different adhesive properties between different tissues.
DESMOSOMES
Desmosome ultrastructure has been reviewed many times (McNutt & Weinstein,
1973; Overton, 1974a; Staehelin, 1974; Skerrow, 1978; Arnn & Staehelin, 1981;
Garrod & Cowin, 1985; Garrod, 1985). Here I shall simply note that desmosomes
are symmetrical about the intercellular mid-line, the two halves of a desmosome
being contributed by adjacent cells. Formation of a desmosome must therefore
depend upon mutual recognition and adhesion between molecules on the surfaces of
the two cells.
The major components of desmosomes are as follows. The plaque components
are: (1) a pair of neutral proteins of MT (by electrophoresis) 250000 and 215 000
(desmoplakins I and II); (2) a neutral protein of 83 000 Mr\ (3) a basic protein of
75 000M t (Franke et al. 1982; Mueller & Franke, 1983). The glycoproteins are: (1) a
triplet of 175000-164000; (2) two bands of Mr 130000 and 115 000 the former of
which often appears as a doublet; and (3) a single band of 22000 Mr (Gorbsky &
Steinberg, 1981).
Gorbsky & Steinberg (1981) proposed that all the desmosomal glycoproteins are
involved in desmosomal adhesion and suggested the collective name ‘desmogleins’ for
them. However, the first evidence for involvement of a molecule in desmosomal
adhesion was provided by ourselves, since univalent Fab' fragments of antibody
against the 130000 and 115 000 Mr glycoproteins inhibited desmosome formation in
MDBK (Madin-Darby Bovine Kidney) cells (Cowin, Mattey & Garrod, 19846).
Similar results were not obtained with Fab' against the 175000-164000 Mr com­
ponents and, indeed, none of our polyclonal antisera recognized the surface domain
of this triplet in MDBK cells. (Since the 175 000-164000 Mr components are
glycoproteins, the existence of a surface domain seems almost certain.) We have
therefore proposed the name desmocollins I and II for the 130000 and 115 000 Mr
components to denote their demonstrated function in desmosomal adhesion (Cowin
et al. 19846). We think that different names are desirable for the glycoprotein
components because they are clearly different molecules, both antigenically (Cohen,
Gorbsky & Steinberg, 1983; Cowin & Garrod, 1983), and in amino acid and sugar
composition (Kaprell et al. 1985). It is possible that the 175000-164000 Mr com­
ponent is also involved in cell adhesion, but the data presented by Gorbsky &
Steinberg (1981) would be equally consistent with that of a transmembrane molecule
with a small surface domain, primarily concerned with linking the adhesion mol­
ecules with the plaque (Cowin et al. 19846). No information about the location and
function of the 22000 Mr component has yet been published.
DESMOSOMAL COMPONENTS IN DIFFERENT TISSUE AND ANIMAL SPECIES
Polyclonal antisera have been raised in guinea pigs, against desmosomal proteins
and glycoproteins from bovine nasal epithelium obtained by preparative gel electro­
phoresis (Cowin & Garrod, 1983; see also Franke et al. 1981; Giudice, Cohen, Patel
Desmosomes, cell adhesion
227
& Steinberg, 1984). Fluorescent antibody studies of tissues from different animal
species (man, cow, guinea pig, rat, chicken, lizard (Lacerta viridis), frog (Rana
pipiens), axolotl and trout) revealed that the desmosomal antigens are widely
distributed (Cowin & Garrod, 1983; Cowin et al. 1984a). The results suggested that
desmosomal antigens are well conserved in the epidermis of mammals, birds, reptiles
and anuran amphibians, since equally bright staining for all antigens was obtained.
In non-epidermal epithelia, while staining for most antigens was as bright as that
in the epidermis, staining for the desmocollins was conspicuously less bright.
One possible explanation of this is that the desmocollins vary in composition
between epidermal and non-epidermal tissues (see Cowin et al. 1984a, for detailed
discussion).
Desmocollin-specific polyclonal antisera raised in mice confirm the difference
between epidermal and non-epidermal epithelia, although not quite in the way we
had expected (Parrish et al. 1986). They recognize only the suprabasal layers of
epidermis and the arachnoid layer of the meninges, but no other epithelial tissues
whatsoever. These antisera also stained meningiomas, raising the possibility of
developing monoclonal reagents for the diagnosis of these tumours. This result
demonstrates that there are indeed chemical differences between desmocollins in
different tissues. In addition, Giudice et al. (1984), using monoclonal antibodies
that recognize different epitopes of desmosomal glycoproteins, demonstrated differ­
ences between the 175 000—164000 Mr glycoproteins and the desmocollins in bovine
nasal, corneal and oesophageal epithelia.
We have carried out immunoblotting studies using polyclonal antisera on
desmosome-enriched fractions from cells from different tissues and species
(Suhrbier & Garrod, 1986). These were human foreskin keratinocytes, MadinDarby bovine kidney cells (MDBK), Madin-Darby canine kidney cells (MDCK),
chicken epidermis and frog epidermis. We found that desmosomal glycoproteins
from various sources vary in apparent molecular weight, heterogeneity and antibody
cross-reactivity. For example, bands reacting with antibody against bovine 175 GOO164 000 Mr glycoprotein have Mr values of 175 000, 169000 and 164000 in bovine
nasal epithelium, but 245 000, 230000 and 210000 in MDCK cells, while keratin­
ocytes possess a single band of 190000. In mammalian cells anti-desmocollin
antibodies do not cross-react with bands recognized by anti-175 000-164000 Mr
antibodies. However, in chickens and frogs there are glycoprotein bands that react
with both antibodies, as well as distinct desmocollin bands (for details refer to
Suhrbier & Garrod, 1986). (Conservation of desmosomal plaque constituents is
discussed by Franke et al. (1982), Mueller & Franke (1983), Giudice et al. (1985),
Suhrbier & Garrod (1985).)
DESM OS OM AL CROSS-REACTIVITY: FORMATION OF DESMOSOMES BETW EEN
DIFF E R EN T CELL TYPES
Desmosomes form between embryonic chick corneal epithelial cells and mouse
epidermal cells (Overton, 1977). Furthermore, desmosome formation between cells
of a metastatic oat cell carcinoma of the lung and liver parenchyma cells has been
228
D. R. Garrod
reported (Jesudason & Iseri, 1980). In view of the apparent conservation of desmosomal components, we wished to determine the capacity of a wider range of
tissues and species for mutual desmosome formation. We have therefore made all
binary combinations in culture between HeLa cells (human cervical carcinoma line),
MDBK cells, MDCK cells, embryonic chick corneal epithelial cells and adult frog
(R. pipiens) corneal epithelial cells (Mattey & Garrod, 1985). In every case mutual
desmosome formation was obtained. Thus, irrespective of whether the cells were
from man or the frog, from a simple (MDBK, MDCK) or a stratified (cornea)
epithelium, from a cell line or a primary culture, from a normal cell or a neoplastic
cell, the adhesive components of their desmosomes were sufficiently well conserved
to participate in recognition and mutual binding.
In contrast, Overton (19746) was unable to find evidence of mutual desmosome
formation between embryonic chick cornea and liver cells, and between cornea and
heart myocytes (see also Nicol & Garrod, 1982). Chick embryonic liver cells and
myocytes possess very few desmosomes. However, the observation of Jesudason &
Iseri (1980), described above, shows that hepatocytes can form desmosomes with
other tissues.
The conclusion from this work is that, although there may be a few exceptions,
cells from different tissues and animal species are capable of mutual desmosome
formation. How can this conclusion be rationalized in terms of the variability
of desmosomal glycoproteins demonstrated by fluorescent antibody staining and
Western blotting?
Present evidence suggests that the biological activity of many glycoproteins is due
to their protein components, while their carbohydrate moieties are thought to be
important as sorting signals in glycoprotein routing, metabolic stability and cellular
differentiation (Olden, Parent & White, 1982; Warren, Buck & Tuszyinski, 1978).
This is probably true for desmosomal glycoproteins, since desmosomes can form in
the absence of N-linked carbohydrates (King & Tabiowao, 1981; Overton, 1982 and
see below). The most likely explanation of our results therefore is that they suggest
that desmosomal adhesion is dependent upon a highly conserved protein domain in
the desmosomal adhesion molecules, and that the variability between tissues and
species arises because of differences in carbohydrate moieties, which may reflect
differences in carbohydrate control mechanisms. Alternatively, the adhesion domain
alone may be well conserved, the rest of the polypeptide showing divergence.
Why should desmosomal adhesive recognition be highly conserved? Probably
because desmosomal adhesion is a cellular property fundamental to the organization
of epithelia. Any alteration caused by mutation would tend to disrupt epithelial
organization and would therefore be selected against.
D IFFER IN G AD HESIVE PROPERTIES OF DIFFERENT CELL TYPES: NUMBER
AN D DISTRIBUTION OF DESMOSOMES
In spite of the ability of their desmosomes to ‘cross-react’, the cells of different
epithelia possess and, indeed, require different adhesive properties, as may be
illustrated by the following considerations.
Desmosomes, cell adhesion
229
Fig. 2. Tissue sections stained with monoclonal antibody against the desmosomal plaque
constituents, desmoplakins I and II, to illustrate differences in distribution and quantity
between different tissues. A. Skin, showing punctate staining between lateral borders of
basal cells and continuous all-round staining of cells in spinous and granular layers.
Cornified squames are not stained by this antibody. B. Liver, showing punctate staining
concentrated along bile canaliculi and interfaces between hepatocytes. c. Kidney tubules
(cuboidal epithelium), showing concentration of staining at apico-lateral cell interfaces
and diminishing towards the bases of the cells. Note that the luminal surfaces of the cells
are unstained (arrow). D . Uterus, showing columnar epithelium with staining con­
centrated at the apico-lateral interfaces. Note that the luminal surfaces are again
unstained (arrow). The antibody was raised and the photographs provided by Elaine
Parrish, a - d , X314.
Ultrastructural observations make it quite clear that the number and distribution of
desmosomes differs between different tissues. Thus cells of the intermediate layers of
stratified epithelia have numerous desmosomes distributed all around their surfaces,
mediating all-round adhesion with other cells. In contrast, cells from simple epithelia
have fewer desmosomes located, as is the case with other junctional types, on their
lateral surfaces. Quantitative variations in desmosomes (number and size) between
different layers of epidermis have been demonstrated, and similar studies on other
tissues have been reviewed by White & Gohari (1984). This evidence has been
greatly reinforced by recent studies of Franke and colleagues (Franke et al. 1981,
1982) and ourselves (Cowin & Garrod, 1983; Cowin et al. 1984a; Garrod, 1984), in
which widely cross-reacting anti-desmosomal antibodies have enabled surveys of
many tissues and cells, using immunofluorescence. Examples of simple and stratified
epithelia stained with fluorescent anti-desmosomal antibodies are shown in Fig. 2.
230
D. R. Garrod
The number of desmosomes possessed by a tissue may be directly related to its
adhesiveness. Thus both Overton (1977), working with embryonic chick corneal
epithelium and mouse epidermis, and Wiseman & Strickler (1981), working with
chick embryonic heart of different ages, have shown that in sorting-out experiments
the cells possessing the greater number of desmosomes adopted the internal position
in aggregates of binary cell combinations. Adoption of the internal position suggests
that cells are more adhesive than those that surround them (Steinberg, 1964).
The functional significance of variation in desmosome number may relate to the
amount of stress a tissue is subjected to : tissues that have to resist stress require more
desmosomes to link their intermediate filament cytoskeletons from cell to cell (Arnn
& Staehelin, 1981; Garrod, 1985).
M OD ULATION OF DESMOSOME DISTRIBUTION VARIES BETW EEN DIFFEREN T
TYPES OF EPITHELIA
We have shown that cells in tissue culture can modulate the distribution of
desmosomal components in order to achieve the characteristic patterns of simple or
stratified epithelium, depending upon cell type.
MDBK cells are derived from bovine kidney tubule and in culture form a simple,
polarized epithelium, each cell having a single row of desmosomes around its apicolateral border. When MDBK cells are passaged their polarity is destroyed, their
desmosomal plaques are internalized and their desmocollins can be removed by the
action of trypsin and EDTA (Cowin et al. 1984b). When they are replated, the
desmocollins initially reappear all over the upper surfaces of the cells, a process that
is independent of intercellular contact. During a culture period of several days,
however, polarity is re-acquired and the desmocollins become removed from the
upper cell surfaces and confined to the lateral cell margins. Thus MDBK cells in
culture exhibit a reorganization or modulation of adhesive molecules, generating this
polarized arrangement of adhesive properties.
A different pattern of modulation is seen in stratified epithelia. Calcium-induced
desmosome formation between keratinocytes in monolayer culture (Hennings et al.
1980; Jones et al. 1982; Hennings & Holbrook, 1983) has been followed with
fluorescent antibody staining (Watt, Mattey & Garrod 1984; Garrod, 1985). In
< 0'1 mM-calcium the desmosomal components were diffusely distributed, but
within 15 min of raising the calcium concentration to 1'8 mM all the molecules were
concentrated at cell peripheries in the region of contact. At this stage the cells were
still in monolayer, with desmosomes between their lateral surfaces. Stratification
began after about 9 h in high concentrations of calcium. This was characterized by
the appearance of desmosomes between the upper and lower surfaces of super­
imposed cells, detectable by punctate staining with anti-desmosomal antibodies. We
are not sure whether this process involves re-orientation of existing desmosomes, or
breakdown and re-formation of desmosomes, though we have some preliminary
evidence for the latter. (For details and photographs, see Watt et al. (1984).) The
antigenic difference that we have demonstrated between the desmocollins of basal
Desmosomes, cell adhesion
231
and suprabasal epidermal appears to be a consequence of stratification rather than a
cause of it (Parrish et al. 1986). It could be associated with a difference in
desmosomal stability in the different layers (see below). In this context the most
important point is that formation of desmosomes between upper and lower surfaces
of stratifying cells is a property that distinguishes stratified from simple epithelial
cells.
The occurrence of desmosomal redistribution during morphogenesis in vivo, has
been demonstrated in duct formation in the rat mammary gland (Dulbecco, Allen &
Bowman, 1984). The epithelium of the duct consists of two layers, polarized cells
with a luminal surface apically and myo-epithelial cells basally. In the early bud
stages, microvillin and p80, markers of the apical cell surface, and desmoplakins are
located together in the immature epithelial cells. In the two-layered ductal epi­
thelium, however, the apical markers are present at the luminal surface, while the
desmoplakins are concentrated at the interface between the apical cells and the myo­
epithelial cells. The acquisition of the correct pattern of adhesive properties must be
fundamental to the development and maintenance of epithelial structure. It is clear
that different types of epithelial cells are able to regulate the distribution of des­
mosomes and desmosomal components in specific ways in order to achieve patterns
appropriate to their tissues.
STABILITY OF DESMOSOMES
The capacity to modify adhesive contacts is necessary during development. In
certain sites in adult organisms, such as intestinal crypts and the germinative layer of
the epidermis, much cell rearrangement is also required. In other situations, such as
.the upper layers of the epidermis, the oesophagus and vaginal epithelium, firm
linkages between cells are necessary (Revel, 1974). Variability in the stability of
adhesive contacts, some being reversible, some more enduring, would thus seem a
necessary requirement for a functional adhesion mechanism.
This idea was first raised in relation to desmosomes by Borysenko & Revel (1973),
who showed that desmosomes of stratified squamous epithelia and glandular épi­
thélia were insensitive to EDTA, whereas desmosomes of simple columnar epithelia
were EDTA-sensitive. The former group of desmosomes were, however, disrupted
by trypsin, while the latter were not. A third group of desmosomes, found in
rat pancreas, were disrupted only by the detergent sodium deoxycholate. They
suggested that the EDTA-resistant desmosomes were functionally stable, while
those that were EDTA-sensitive were physiologically labile.
Our own data relating to desmosome stability come from calcium-switching
experiments on keratinocytes (Watt et al. 1984), and MDCK and MDBK cells. I
should first point out that MDCK cells exhibit rapid triggering of desmosome
formation by low-high switching of calcium concentration, whereas MDBK cells
show slow desmosome formation on calcium switching, requiring protein and RNA
synthesis for production of the desmoplakins and 175 000—164000 Mr glycoproteins
(Mattey & Garrod, unpublished).
232
D. R. Garrod
Once calcium-induced desmosome formation has taken place, the stability of
desmosomes to calcium reduction changes with time and varies between the three
cell types. Keratinocyte desmosomes may be disrupted by calcium depletion for up
to 2h after formation, but thereafter become resistant. MDCK cell desmosomes may
be disrupted by calcium depletion for up to 3 days after formation but then acquire
resistance. MDBK cell desmosomes do not become resistant to calcium depletion for
at least 14 days after formation.
We are aware that calcium-switching may not truly parallel the in vivo situation
but other environmental and, or, developmental triggers may be involved in con­
trolling desmosome assembly and disassembly. Differential desmosomal stability
may certainly provide a basis for adhesive differences between cell types.
THE MOLECULAR ROLE OF CALCIUM
We are beginning to unravel the molecular basis of calcium involvement in
desmosomal adhesion (Suhrbier, Mattey & Garrod, unpublished observations). The
desmosomal adhesion molecules, the desmocollins, are calcium protected: in the
presence of 0-1 mM-calcium they are resistant to removal from the cell surface by
0-1 % trypsin, but in the presence of EGTA they are rapidly removed by 0/0001 %
trypsin. Furthermore, trypsin digestion (0-1%) of desmosomal cores from bovine
nasal epithelium yields a single soluble desmocollin fragment of Mr 42000, which is
digested in the presence of EGTA. Moreover, both the desmocollins and the
175 000—164000 Mr desmosomal glycoproteins bind calcium. An essential point to
note is that the calcium requirements for calcium protection and calcium-induced
desmosome formation differ by at least an order of magnitude (approx. 0'lmM and
1-2 mM respectively). Calcium-induced desmosome assembly and stability probably
involve desmosomal carbohydrate, since cells may be induced to form stable des­
mosomes in low calcium medium by treatment with tunicamycin.
ADHESIVE DIFFERENCES WITH CELL AD HESI ON MOLECULES
Variation in the cellular distribution of cell adhesion molecules has been described.
Thus, for example, N-CAM is concentrated at the end-feet of neuroepithelial cells in
the optic nerve pathway, where it plays an important role in guidance of developing
optic nerve fibres (Silver & Rutishauser, 1984), and at the nerve-muscle junction in
developing mouse diaphragm (Rieger, Grumet & Edelman, 1985), while Ng-CAM is
present on neuronal processes rather than cell bodies (Thiery et al. 1985). In
developing optic retinal ganglian cells there is a differential distribution of the
embryonic (E) and adult (A) forms of N-CAM within the same cells. The sialic-acidrich E-form is present in the optic nerves in the 5—10 day chick embryo, while
the sialic-acid-poor A-form is present in the retina (Schlosshauer, Schwarz &
Rutishauser, 1984). The E-form of N-CAM has a lower binding affinity than the
A-form (Edelman, 1983, 1984), conversion of the former to the latter providing a
means of increasing cellular adhesiveness during development. In the case of
N-CAM, variation in adhesiveness with time also occurs via down- or up-regulation
Desmosomes, cell adhesion
233
of the molecule. Thus N-CAM disappears or decreases from skeletal muscle post­
natally but reappears when regeneration of new synapses is required following
denervation or disease (Moore & Walsh, 1985; Rieger et al. 1985). Reduction in the
amount of N-CAM on the cell surface has also been shown to occur following
viral transformation of embryonic chick retinal cells (Brackenbury, Greenberg &
Edelman, 1984). A full discussion and summary of how variation or modulation of
CAMs can generate differences in adhesive properties is given by Edelman (1984).
It has recently been reported that individual neural cell types express immuno­
logically distinct forms of N-CAM (Williams, Goridis & Akeson, 1985). However, it
is not known whether this is due to differences in glycosylation of N-CAM protein in
different cells or to differences in the protein itself. Murray et al. (1984) reported
that a N-CAM, cDNA probe detected only one N-CAM gene in the chick, but
Goridis et al. (1985) reported that there are possibly three N-CAM genes in the
mouse genome. Southern blot hybridization with an L-CAM probe suggested that
there may be three distinct L-CAM genes in chick (Gallin, Prediger, Edelman &
Cunningham 1985).
CONCLUSION
It seems that similar general principles may apply to adhesion mechanisms
involving CAMs and intercellular junctions. Adhesive differences between cells may
be qualitative, depending upon the presence of different molecular adhesion mech­
anism. A single adhesion mechanism, present in many cell types, may give rise to
adhesive differences because of variation in quantity, distribution, stability or
affinity. It is important to determine how cells control these parameters.
I thank Dr Derek Mattey for his comments on the manuscript. This work was supported by the
Cancer Research Campaign.
REFERENCES
P. B. (1970). A fine structural study of adhesive cell junctions in heterotypic cell
aggregates, J. Cell Biol. 47, 197-210.
A rm stro n g , P. B. (1971). Light and electron microscope studies of cell sorting in combination of
chick embryo neural retina and retinal pigment epithelium. Wilhelm Roux Arch. EntwMech. Org.
168, 125-141.
A rn 'n , J. & S t a e h e l in , L. A . (1981). The structure and function of spot desmosomes.
Dermatology 20, 330-339.
B a k e r , P. C. & S c h r o ed er , T. E. (1967). Cytoplasmic filaments and morphogenetic movements
in the Amphibian neural tube. Devi Biol. 15, 432-450.
B erry , M. N. & F r ie n d , D. S. (1969). High-yield preparation of rat liver parenchymal cells.
A biochemical and fine structural study. J. Cell Biol. 43, 506-520.
B il l ig , D., N ic o l , A., C o w in , P., M c G in ty , R., M o rg a n , J. & G a rr o d , D. R. (1982). The
cytoskeleton and substratum adhesion in chick embryonic corneal epithelial cells. J. Cell Sci. 57,
51-71.
B o ller , K ., V estw eb er , D. & K e m l e r , R. (1985). Cell-adhesion molecule uvomorulin is
localized in the intermediate junctions of adult intestinal epithelial cells. Jf. Cell Biol. 100,
327-332.
A rm stro n g ,
234
D. R. Garrod
J. Z. & R e v el , J. P. (1973). Experimental manipulation of desmosome structure.
J. Anat. 137, 403-422.
B ra ck en bu ry , R., G r een b erg , M. E . & E d e l m a n , G . M. (1984). Phenotypic changes and loss of
N-CAM-mediated adhesion in transformed embryonic chicken retinal cells, jf. Cell Biol. 99,
1944-1954.
B u r g e s s , D . R . (1982). R eactivation of intestinal epithelial b ru sh b o rd er m o tility : ATPd ep en d e n t co n tractio n via a term in al w eb contractile ring. J'. Cell Biol. 95 , 853—863.
C o h e n , S . M., G orbsky , G . & S tein b erg , M. S. (1983). Immunochemical characterization of
related families of glycoproteins in desmosomes. J. biol. Chem. 258, 2621-2627.
C o l e , G . J. & G l a se r , L. (1984). Inhibition of embryonic neural retina cell—substratum adhesion
with monoclonal antibody. J. biol. Chem. 2559, 4031-4034.
C o w in , P. & G a rr o d , D. R. (1983). Antibodies to epithelial desmosomes show wide tissue and
species cross-reactivity. Nature, Lond. 302, 148—150.
CowiN, P., M attey , D. L. & G a rr o d , D. R. (1984a). Distribution of desmosomal components in
the tissues of vertebrates, studied by fluorescent antibody staining. J. Cell Sci. 66, 119-132.
C o w in , P., M attey , D. L. & G a rr o d , D. R. (19846). Identification of desmosomal surface
components (desmocollins) and inhibition of desmosome formation by specific Fab'.X Cell Sci.
70, 41-60.
D am sky , C. H., K n u d s e n , K . & B u c k , C. A. (1984). Integral membrane proteins in cell-cell and
cell-substratum adhesion. In The Biology of Glycoproteins (ed. R. Ivatt), pp. 1-64. New York:
Plenum.
D am sky , C. H., R ic h a , J., S olter , D ., K n u d s e n , K . & B u ck , C. A. (1983). Identification and
purification of a cell surface glycoprotein mediating intercellular adhesion in embryonic and
adult tissues. Cell 34, 455-466.
D o ch erty , R. J., E d w a r d s , J. G ., G a rro d , D . R. & M attey , D . L. (1984). Chick embryonic
pigmented retina is one of a group of epithelial tissues that lack cytokeratins and desmosomes and
have intermediate filaments composed of vimentin. J. Cell Sci. 71, 61—74.
DULBECCO, R., A l l e n , W. R. & B o w m a n , M. (1984). Lumen formation and redistribution of
intramembranous proteins during differentiation of ducts in the rat mammary gland. Proc. natn.
Acad. Sci. U.SA. 81, 5763-5766.
E d e l m a n , G. M. (1983). Cell adhesion molecules. Science 219, 450-457.
E d e l m a n , G. M. (1984). Modulation of cell adhesion during induction, histogenesis, and perinatal
development of the nervous system. A. Rev. Neurosci. 7, 339-377.
E l l is o n , J. E . & G arrod, D. R. (1984). Anchoring filaments of the Amphibian dermal-epidermal
junction traverse the basal lamina entirely from the plasma membrane of hemidesmosomes to the
dermis. J ’. Cell Sci. 72, 163-172.
E p st e in , M. L. & G il u l a , N . B. (1977). A study of communication specificity between cells in
culture. J. Cell Biol. 75, 769-787.
F r a n k e , W. W., M o l l , R., S c h il l e r , D. L., S c h m id , E., K a rten beck , J. & M u e l l e r , H.
(1982). Desmoplakins of epithelial and myocardial desmosomes are immunologically and
biochemically related. Differentiation 23, 115-127.
F r a n k e , W. W., S c h m id , E ., G r u n d , C., M u e l l e r , H., E n g elbr ech t , I., M o l l , R., S t a d l e r ,
J . & J a ra sch , E . D . (1981). Antibodies to high molecular weight polypeptides of desmosomes:
specific localisation of a class of junctional proteins in cells and tissues. Differentiation 20,
217-241.
G a l l in , W . J ., P r e d ig e r , E . A ., E d e l m a n , G . M . & C u n n in g h a m , B. A. (1985). Isolation of a
c D N A clone for th e liver cell adhesion m olecule (L -C A M ). Proc. nat. Acad. Sci. U.S.A. 82,
2 80 9-28 1 3.
G a rr o d , D. R. (1981). Adhesive interactions of cells in development: specificity, selectivity and
cellular adhesive potential. Fortschr. Zool. 26, 184-195.
G a rr o d , D . R . (1985). The adhesions of epithelial cells. In Cellular and Molecular Control for
Direct Cell Interactions in Developing Systems (ed. H.-J. Marthy), NATO Advanced Studies
Institute, series A. Life Sci. New York: Plenum (in press).
G a r r o d , D. R. & C o w in , P. (1985). Desmosome structure and function. In Receptors in Tumour
Biology (ed. C. M. Chadwick). Cambridge University Press (in press).
G a r r o d , D. R. & N ic o l , A. (1981). Cell behaviour and molecular mechanisms of cell adhesion.
Biol. Rev. 56 , 199-242.
B o ry sen k o ,
Desmosomes, cell adhesion
235
J., C o h e n , S. M., P a tel , N. H. & S t ein b er g , M. S. (1984). Immunological
comparison of desmosomal components from several bovine tissues. J. Cell Biochem. 26, 35-45.
G o rb s k y , G . & S te in b e rg , M. S. (1981). Isolation of the intercellular glycoproteins of
desmosomes. J. Cell Biol. 90, 243 -2 48 .
G o r id is , C., H ir n , M., S a n t o n i , M.-J., G e n n a r in i , G ., D ea g o stin i -B a z in , H ., J o r d a n ,
B . R., K ie f e r , M. & S tein m e t z , M. (1985). Isolation of mouse N-CAM-related cDNA:
detection and cloning using monoclonal antibodies. EMBOjf. 4 , 631-635.
H atta , K., O k a d a , T. S. & T a k eic h i , M. (1985). A monoclonal antibody disrupting calciumdependent cell-cell adhesion of brain tissues: possible role of its target antigen in animal pattern
formation. Proc. natn. Acad. Sci. U.SA. 82, 2789-2793.
H ea y sm an , J. E. M. & P eg ru m , S. M. (1973). Early contacts between fibroblasts. An
ultrastructural study. Expl Cell Res. 78, 71-78.
H e n n in g s , H . & H o lbro o k , K. A. (1983). Calcium regulation of cell-cell contact and
differentiation of epidermal cells in culture. An ultrastructural study. Expl Cell Res. 143,
127-142.
G iu d ic e , G .
H e n n in g s , H ., M ic h a e l , D ., C h e n g , C ., S teinert , P ., H olbrook , K . & Y u spa , S. H . (1980).
Calcium regulation of growth and differentiation of mouse epidermal cells in culture. Cell 19,
245-254.
H y a fil , F., M o rello , D., B a bin et , C. & J acob , F. (1980). A cell surface glycoprotein involved
in the compaction of embryonal carcinoma cells and cleavage stage embryos. Cell 21 , 927-939.
J e s u d a s o n , M. L. & I s e r i , O. A. (1980). Host—tumor cellular junctions: an ultrastructural study
of hepatic metastases in bronchogenic oat cell carcinoma. Human Path. 11, 67-70.
J o n e s , J . C. R., G o l d m a n , A. E., S tein er t , P . M., Y u s pa , S. & G o ld m a n , R. D. (1982).
Dynamic aspects of the supramolecular organization of intermediate filament networks in
cultured epidermal cells. CellMotil. 2, 197—213.
K a ppr e ll , H .-P ., C o w in , P ., F ra n k e , W . W ., P o stin g l , H . & O pfer k u c h , H . J. (1985).
Biochemical characterization of desmosomal proteins isolated from bovine muzzle epidermis:
amino acid and carbohydrate composition. Eur.jf. Cell Biol. 36, 217-229.
K a r fu n k e l , P. (1971). The role of microtubules and microfilaments in neurulation in Xenopus.
Devi Biol. 25, 30-56.
K in g , I. A. & T abio w o , A. (1981). Effect of tunicamycin on epidermal glycoprotein and
glycosominoglycan synthesis in vitro. Biochem. J. 198, 331—338.
K r u s e , J ., K e il h a u e r , G., F a issn e r , A., T im pl , R. & S ch a c h n er , M. (1985). T h e J1
gly co protein - a novel n ervous system cell adhesion m olecule of th e L2/HNK-1 fam ily. Nature,
Land. 316, 146-148.
M attey , D . L . & G a rr o d , D . R . (1985). M u tu al desm osom e form ation betw een all b inary
co m b in ation s of h u m an , bovine, canine, avian and am p h ib ian cells: desm osom e fo rm atio n is not
tissue o r species specific. J. Cell Sci. 75, 3 77-399.
M c N u t t , M . S . & W e in s te in , R . S. (1973). M em brane u ltrastru ctu re at m am m alian in tercellular
ju n ction s. Prog. Biophys. molec. Biol. 26, 4 5-10 1.
M id d l e t o n , C. A. & P eg ru m , S. M . (1976). Contacts between pigmented retinal ephithelial cells
in culture. X Cell Sci. 22, 371—383.
Moore, S. E. & W a l s h , F. S. (1985). Specific regulation of N-CAM/D2-CAM cell adhesion
molecule during skeletal muscle development. EMBOJ. 4 , 623-630.
M o sc o n a , A. A. (1962). Analysis of cell recombinations in experimental synthesis of tissues
in vitro. J. cell. comp. Physiol. 60 (suppl. I), 65-80.
M u e l l e r , H. & F ra n k e , W. W. (1983). B iochem ical and im m unological characterization of
desm oplakins I and I I , th e m ajor polypeptides of th e desm osom al plaque. J. molec. Biol. 163,
647-671.
M u ir , A. R. (1967). The effects of divalent cations on the ultrastructure of the perfused rat heart.
J.Anat. 101, 239-261.
M u rr a y , B. A ., H em perley , J. J ., G a l l in , W. J ., M a c G reggor , J. S ., E d e l m a n , G . M . &
C u n n in g h a m , B. A . (1984). Isolation of cD N A clones for th e chicken n eu ral cell adhesion
m olecule (N -C A M ). Proc. natn. Acad. Sci. U.SA. 81, 5584-5588.
N ic o l , A. & G a rr o d , D. R. (1982). Fibronectin, intercellular junctions and the sorting-out of
embryonic chick tissue cells in monolayer. J. Cell Sci. 57, 357-372.
236
D. R. Garrod
O c k l in d , C. & Ô b r in k , B. (1982). Intercellular adhesion of rat hepatocytes. Identification of a cell
surface glycoprotein involved in the initial adhesion process. X biol. chem. 257, 6788-6795.
K., P a r en t , J. B. & W h it e , S. L. (1982). Carbohydrate moieties of glycoproteins: a re­
evaluation of their function. Biochim. biophys. Acta 650, 209-232.
O v er to n , J. (1974a). C ell junctions an d th e ir developm ent. Prog. Surface Membr. Sci. 8 ,
161 -2 08 .
O v er to n , J. (19746). Selective formation of desmosomes in chick cell reaggregates. Devi Biol. 39,
210-225.
O v erto n , J. (1977). Formation of junctions and cell sorting in aggregates of chick and mouse cells.
Devi Biol. 55 , 103-116.
O v erto n , J. (1982). Inhibition of desmosome formation with tunicamycin and with lectin in
corneal cell aggregates. Devi Biol. 92, 6 6 -7 2 .
P a r r is h , E., G a rr o d , D. R ., M attey , D. L., H a n d , L., S tea rt , P. & W e l l e r , R . O . (1986).
Mouse antisera specific for desmosomal adhesion molecules of suprabasal skin cells, meninges
and meningioma. Proc. natn. Acad. Sci. U.S.A. (in press).
PlTTS, J. D. (1978). Junctional communication and cellular growth control. In Intercellular
Junctions and Synapses, Receptors and Recognition, series B, vol. 2 (ed. J. Feldman, N. B.
Gilula & J. D. Pitts), pp. 61-79. London: Chapman and Hall.
R a t h je n , F. G. & R u t ish a u se r , U . (1984). Comparison of two cell surface molecules involved in
neural cell adhesion. EMBOjf. 3 , 461-465.
R a t h je n , F. G. & S ch a c h n er , M. (1984). Immunocytological and biochemical characterization
of a new neuronal cell surface component (LI Antigen) which is involved in cell adhesion.
EMBOJ. 3, 1-10.
R e v e l , J.-P. (1974). Some aspects of cellular interactions in development. In The Cell Surface in
Development (ed. A. A. Moscona), pp. 51-66. New York, London, Sydney, Toronto: John
Wiley and Sons.
R ie g e r , F., G ru m et , M. & E d e l m a n , G . M. (1985). N-CAM at the vertebrate neuromuscular
junction. J. Cell Biol. 101, 285-293.
S c h l o ssh a u e r , B., S c h w a r z , U. & R u t ish a u se r , U. (1984). Topological distribution of
different forms of neural cell adhesion molecule in the developing chick visual system. Nature,
Land. 310, 141-143.
S ch u b e r t , D. & L a C o rbière , M. (1985). Isolation of a cell-surface receptor for chick neural
retina adhesions. X Cell Biol. 100, 56-63.
S e d a r , A. W. & F orte , J. G. (1964). Effects of calcium depletion on the junctional complex
between oxyntic cells of gastric glands. J. Cell Biol. 22, 173—188.
S il v e r , J. & R u t ish a u se r , U. (1984). Guidance of optic axons in vivo by a preformed adhesive
pathway on neuroepithelial endfeet. Devi Biol. 106, 485-499.
S k err o w , C. J. (1978). Intercellular adhesion and its role in epidermal differentiation. Invest. Cell
Path. 1, 23-37.
S perry , R . W. (1963). Chemoaffinity in the orderly growth of nerve fibre patterns and
connections. Proc. natn. Acad. Sci. U.SA.. 50, 703—710.
S t a e h e l in , L. A. (1974). Structure and function of intercellular junctions. Int. Rev. Cytol. 39 ,
191-283.
S te in b e r g , M. S. (1964). The problem of adhesive selectivity in cellular interactions. In Cellular
Membranes in Development, 22nd Symp. Soc. Study of Development and Growth (ed. M.
Locke), pp. 321-366. New York: Academic Press.
S u h r b ie r , A. & G a rr o d , D. R . (1985). An investigation of the molecular Components of
desmosomes in epithelial cells of five vertebrates. J. Cell Sci. 81, 223-242.
T a k e ic h i , M., A tsu m i , T ., Y o sh id a , C., U n o , K. & O k a d a , T . S. (1981). Selective adhesion of
embryonal carcinoma cells and differentiated cells by Ca2+-dependent sites. Devi Biol. 87 ,
340-350.
T h ier y , J.-P., B ra ck en bu ry , R ., R u t ish a u se r , U. & E d elm a n , G. M. (1977). Adhesion among
neural cells of the chick embryo. II. Purification and characterisation of a cell adhesion molecule
from neural retina. J. biol. Chem. 252, 6841-6845.
Olden,
Desmosomes, cell adhesion
237
T h ier y , J.-P., D elouv S e , A ., G ru m et , M . & E d e l m a n , G . M . (1985). In itial ap p earan ce and
regional d istrib u tio n of th e n eu ro n -g lial cell adhesion m olecule in th e chick em bryo. J. Cell Biol.
100, 4 4 2 -4 5 6 .
V e stw eb er , D., O c k l in d , C., G o ssler , A., O d in , P., O br in k , B. & K e m ler , R. (1985).
Comparison of two cell-adhesion molecules, uvomorulin and cell-CAM 105. Expl Cell Res. 157,
451-461.
W a r r e n , L., Buck, C. A. & T u szy n sk i , G. P. (1978). Glycopeptide changes in malignant
transformation: a possible role for carbohydrate in malignant behaviour. Biochim. biophys. Acta
516, 97-127.
W a tt , F . M ., M attey , D . L . & G a rr o d , D . R . (1984). C alcium -induced reorganization of
d esm osom al com ponents in cu ltured h um an keratinocytes. Jf. Cell Biol. 99 , 2211-2215.
W h it e , F. H. & G o h a r i , K. (1984). Desmosomes in hamster cheek pouch epithelium: their
quantitative characterization during epithelial differentiation. Jf. Cell Sci. 66, 411-429.
W il l ia m s , R. K ., G o r id is , C . & A k e so n , R. (1985). Individual cell types express immuno­
logically distinct N -C A M forms. Jf. Cell Biol. 101, 36-42.
W ise m a n , L. L. & S trickler , J. (1981). Desmosome frequency: experimental alteration may
correlate with differential a d h e s io n .Cell Sci. 49, 217-223.
Y o sh id a -N o r o , C., S u z u k i , N . & T a k e ic h i , M. (1984). Molecular nature of the calciumdependent cell—cell adhesion system in mouse teratocarcinoma and embryonic cells studied with
monoclonal antibody. Devi Biol. 101, 19-27.