Cell Junctions
Joachim Wegener, Westfälische Wilhelms-Universität, Münster, Germany
For any multicellular organism to function, it is a prerequisite that its individual cells interact
with each other. These interactions are realized by cell-to-cell junctions that provide
mechanical stability or information exchange.
Introduction
The human body is composed of a set of different organs,
for instance the liver or the kidney, each fulfilling very
specific and important tasks indispensable for the func-
tioning of the organism as a whole. The organs themselves
are formed by different tissues, which are organized as-
semblies of specialized cells – all of the same kind or of a
well-defined but mixed community. The individual cells
themselves are regarded as the smallest units of life capable
of independent reproduction. Separated from the environ-
ment by the semipermeable plasma membrane, the cell in-
terior is a highly controlled chemical compartment loaded
with proteins that allow the cells to store or gain energy,
perform metabolism, control their own shape and migrate
within certain constraints. Altogether there are more than
200 different and individually specialized cell types in the
human body. However, it requires a population of indi-
vidual cells to work together in a controlled fashion to
make a particular tissue functional. One example, among
many others, is the blood vessels in animal bodies that
provide the infrastructure for the delivery of oxygen and
nutrients from the central to the peripheral organs. From
geometric considerations alone, it is apparent that the walls
of long and branched blood vessels, which control the ex-
change of solutes between the circulating blood and the
tissues, cannot be formed by one individual cell but only by
continuous layers of specialized cells, called endothelial
cells.
Concerted actions between an ensemble of individual
cells require cellular interactions, which are called ho-
motypic when cells of the same kind interact with each
other (e.g. the endothelial cells lining the blood vessels) or
heterotypic when the cells in contact are not of the same
type (e.g. cells cooperating during the immune response).
The simplest cellular interaction is just an exchange of sol-
uble factors, secreted by one cell and perceived and proc-
essed by another. More often, however, direct cell-to-cell
junctions (short cell junctions) are needed. From a func-
tional viewpoint cell junctions can be divided in three cat-
egories: (1) those that tie adjacent cells together
mechanically; (2) those between adjacent cells for commu-
nication purposes or metabolic coupling; (3) those that seal
the extracellular cleft between adjacent cells so that the
diffusion of solutes through the intercellular space is re-
ENCYCLOPEDIA OF LIFE SCIENCES © 2004, John Wiley & Sons, Ltd. www.els.net
Introductory article
Article Contents
. Introduction
. Tight Junctions: Barrier-Forming Cell-to-Cell Contacts
. Adherens Junctions and Desmosomes: Junctions that
Provide Mechanical Resistance to the Tissue
. Synapses, Gap Junctions and Plasmodesmata:
Communicating Cell Junctions
doi: 10.1038/npg.els.0001275
stricted. These different kinds of cell junctions will be dis-
cussed individually, using epithelial cells as an example.
Epithelial cells form continuous monolayers that line all
inner and outer surfaces of the animal body, such as the
skin, the gut lumen or the urinary bladder. The above-
mentioned endothelial cells that line the blood vessels may
be considered as a special subtype of epithelial cell. Epi-
thelial layers serve as interfacial tissues that strictly control
the transfer of ions, metabolites, proteins and cells from
one compartment to the other. Because of this barrier
function, epithelial cells are considered to be as important
for the evolution of multicellular organisms as the plasma
membrane has been for the evolution of single cells, be-
cause both provide controlled chemical compartments
within a complex environment.
Epithelial cells commonly express all the major cell
junctions in a prominent manner and, in contrast to many
other cell types, these different junctions are localized in
well-separated regions along the cell–cell contact surface.
Figure 1a illustrates two adjacent epithelial cells that are
interconnected by four different types of cell junction: tight
junctions (TJ), adherens junctions (AJ), desmosomes (DS)
and gap junctions (GJ). The cartoon is a molecular expla-
nation for the structures visualized in electron microscopic
images of the intercellular space between two opposing
cells, as shown in Figure 1b. Electron microscopy has been,
and still is, a very important tool for clarifying the struc-
tural details of cellular junctions, which are now discussed
individually.
Tight Junctions: Barrier-Forming Cell-
to-Cell Contacts
The tight junctions (the former Latin term is zonula oc-
cludens or zo) are localized very close to the apical pole of
the cells at the entrance of the intercellular cleft (TJ in
Figure 1). These cell–cell contacts are the boundary that
separates the apical from the basolateral membrane do-
main. Two basic functions are assigned to the tight junc-
tions: the ‘gate’ and ‘fence’ functions. The gate function
1
 Cell Junctions

Figure 1 (a) Diagram, and (b) electron micrograph of cell-to-cell junctions between two adjacent epithelial cells. TJ, tight junctions; AJ, adherens
junctions; DS, desmosomes; GJ, gap junctions. On the cytoplasmic site of the membrane adherens junctions are associated with the actin cytoskeleton,
whereas desmosomes are linked to intermediate filaments, for instance keratin filaments in epithelial cells. Anchorage of the cells to the extracellular matrix
(ECM) is provided by focal contacts (FC) or hemidesmosomes (HD), which also differ with respect to their particular connections to the cytoskeleton. The
apical membrane surface shows membrane protrusions (microvilli, Mv) that are typical of transporting epithelia. Bar, 200 nm. (b) From Tsukita et al.
(2001).

describes the tight occlusion of the intercellular cleft so that
solutes cannot simply diffuse from the apical to the baso-
lateral compartment or vice versa. Instead, all hydrophilic
compounds that have a tendency to migrate across an ep-
ithelial cell layer have to cross the cellular plasma mem-
branes via corresponding transport proteins. Since these
are highly specific and strictly regulated, the uptake or re-
lease of metabolites across an epithelial diffusion barrier is
under cellular control. It is this barrier function that allows
the establishment of chemical or electrochemical gradients
across an epithelial cell layer, which is of vital importance
for all transepithelial transport processes (e.g. glucose up-
take in the gut) from one compartment to the other. But
tight junctions do not only restrict diffusion through the
intercellular cleft, they also serve as a diffusion barrier for
lipids and proteins within the cellular plasma membrane.
Once proteins have been integrated either in the apical or
the basolateral membrane domain by the protein targeting
machinery, they cannot diffuse through the tight junctions
and migrate into the other part of the membrane. This
fence function of the tight junctions helps to preserve a
polar distribution of proteins between both membrane
domains, which is a second and equally important require-
ment for transepithelial transport.
Ultrastructural morphology
In thin section electron micrographs of the intercellular
cleft between adjacent epithelial cells, tight junctions can be
identified as a 0.1–0.7-mm area in which the opposing
plasma membranes form intimate contacts that resemble
points of membrane fusion and are thus called membrane
kisses. These can be clearly seen in high-resolution electron
micrographs like the one shown in Figure 2a. At these
membrane kisses (arrowheads in Figure 2a), the intercellu-
lar diffusion pathway is effectively blocked. This efficient
occlusion of the diffusion pathway has been experimen-
tally localized to points of membrane kisses by applying
electron-dense ions like La3+ from the basolateral side of
an epithelial monolayer and preparing thin-section elec-
tron micrographs after the ions had infiltrated the inter-
cellular shunt. The flow of La3+ ions up the intercellular
channel is abruptly stopped at the point of a membrane
kiss. Applying the lanthanum ions from the apical side
stains the upper part of the intercellular cleft that is beyond
the membrane kiss but leaves the lower one bright. Figure 2b
shows the morphology of tight junctions in freeze-fracture
electron micrographs. In this technique the lipid bilayer of
the plasma membrane is broken up along the hydrophobic
core, so it is possible to examine the membrane from the

2

Figure 2 Morphology of tight junctions in electron micrographs. (a)
High-resolution thin-section electron micrograph of the tight junction
area. The image is a magnification of the highlighted area shown in Figure
1b. Arrowheads indicate points of membrane kisses. Bar, 50 nm. (b) Freeze-
fracture replicas of the tight junction area show strands on the p-face
(arrowheads) and corresponding grooves on the e-face (arrows). Ap,
apical; Bl, basolateral; Mv, microvilli, .Bar, 200 nm. From Tsukita et al.
(2001).
inside, either viewing in the direction of the cytoplasm (p-
face) or in towards the extracellular space (e-face). Freeze-
fracture micrographs show that the points of closest mem-
brane apposition are not individual spots but a network of
two-dimensional strands that surrounds the cell like a belt.
There is experimental evidence to show that the number of
strands in the apical to basolateral direction and the com-
plexity of the network are correlated with the tightness of
the junctions, which was found to vary considerably in
epithelia derived from different origins. However, the in-
itial concept that junctional tightness increases linearly
Cell Junctions
Apical
Apical
Apical
Cell 1 Cell 2
(a) Basolateral (b) Basolateral Basolateral
Figure 3 (a) Protein model of the tight junctions according to which
strands, as seen in a freeze-fracture replica, correspond to linear aggregates
of transmembrane proteins. (b) Lipid model of the tight junctions
according to which strands, as seen in freeze-fracture electron
micrographs, correspond to cylinders of lipids in inverted micellar
arrangement.
with the number of strands is now generally accepted to be
too simple.
Molecular architecture
On the p-face tight junctions look like strands, with cor-
responding furrows on the e-face. This structural appear-
ance raised the question about the molecular architecture
of these barrier-forming cell-to-cell junctions. The strands
and grooves have been interpreted as linear arrays of
transmembrane proteins that interact with their extracel-
lular domains within the intercellular shunt, thereby bring-
ing the opposing membranes so close together that the
shunt effectively becomes blocked (Figure 3a). Much ex-
perimental evidence has been gathered to prove this hy-
pothesis. Several transmembrane proteins that are
exclusively associated with the tight junctions have been
identified and could function in a manner similar to that
illustrated in Figure 3a. The names of the most prominent
tight junction-specific proteins are occludin, JAM (junc-
tional adhesion molecule) and a whole family of proteins,
the claudin family, that already has more than 20 members,
with more to come. Occludin and the claudins are very
similar with respect to their three-dimensional topology,
although they have very little sequence similarity. They
have four transmembrane domains which form two closed
extracellular loops with the N- and C-termini of the protein
both located in the cytoplasm. The two extracellular loops
of proteins from opposing cells are then thought to interact
and form a tight seal. It is, however, not yet clear whether
homophilic or heterophilic interactions are realized. Inter-
estingly, we will meet this kind of protein folding again
when we discuss the connexin proteins that form the gap
junctions. It has been shown that some of the tight junc-
tion-associated transmembrane proteins are connected to
the actin filaments via several linker proteins, so that the
total molecular architecture of tight junctions closely re-
3
 Cell Junctions
sembles that of adherens junctions and desmosomes, which
are discussed in the next paragraph.
However, as not all functional experiments can be en-
tirely explained by a tight junction architecture that is ex-
clusively based on transmembrane proteins, an alternative
hypothesis has been proposed. According to this model,
the strands and furrows, as seen in freeze-fracture micro-
graphs, are cylinders made from lipids in an inverted mi-
cellar arrangement. As shown in Figure 3b, such a lipid
arrangement implies a continuous outer leaflet of the lipid
double layer. This continuity of the outer membrane leaflet
provides the missing molecular explanation, for instance,
for the experimental finding that fluorescently labelled lip-
ids that are selectively incorporated into the outer leaflet of
the apical membrane cannot diffuse through the plane of
the membrane into the basolateral membrane domain.
When fluorescent lipid probes are incorporated into the
inner leaflet they are free to diffuse all around the cell.
These and other functional observations can be readily
explained by a membrane structure as described in the lipid
model of the tight junctions. If tight junction strands are
indeed made from lipid cylinders, it is very reasonable to
assume that these are stabilized and thus regulated by
junction-associated transmembrane proteins.
Adherens Junctions and Desmosomes:
Junctions that Provide Mechanical
Resistance to the Tissue
Moving along the intercellular cleft between two epithelial
cells in apical to basolateral direction (Figure 1), the adher-
ens junctions (the former Latin term is zonula adhaerens or
za) are located right underneath the tight junctions, fol-
lowed by the desmosomes further down. Since both of
these cell junctions were first identified structurally from
electron micrographs without detailed knowledge of their
molecular composition, the initial terms were belt desmo-
somes for adherens junctions and point desmosomes for
what are now simply called desmosomes. This initial ter-
minology already implies that adherens junctions sur-
round the cells like a belt, whereas the desmosomes are
more spot-like structures distributed along the contact ar-
ea between adjacent cells. What is actually seen in electron
micrographs (Figure 1b) is not really the junction itself but
much more the association of the two types of junction
with the underlying cytoskeleton via plaques of linker or
adapter proteins. It is this association of the junctional
proteins with the cytoskeleton that makes adherens junc-
tions and desmosomes truly distinct. Whereas adherens
junctions are connected to thick bundles of actin filaments
that form a belt-like structure around the cell close to the
apical surface, the desmosomes are linked to intermediate
4
filaments that traverse the entire cytoplasm and form a
dense network around the nucleus.
Connecting intracellular filaments
The architecture of adherens junctions and desmosomes is
individual with respect to the participating molecules,
which will be discussed below, but it follows a common
concept: transmembrane junction proteins make contact
with a corresponding transmembrane protein on the op-
posing cell surface with their extracellular domains, while
their intracellular domains are connected to the cytoskel-
eton via an individual set of adapter proteins, as indicated
in Figure 1a. Accordingly, these two cell junctions inter-
connect both the microfilaments and the intermediate fil-
aments of individual cells so that they run – in a sense
continuously – through the whole epithelial monolayer.
The mechanical bonding of the filaments that is provided
by adherens junctions and desmosomes gives epithelial cell
layers a very unique mechanical stability, which is most
obvious in skin epithelium. It is, for instance, a daily ex-
perience that the skin does not show cuts and openings
even for mechanical load forces that would be too severe
for individual cells. By bridging and bonding of the indi-
vidual cytoskeletons, mechanical stress on one particular
spot is distributed into the whole cell layer, which greatly
reduces the mechanical input on a single cell.
Molecular architecture
From the molecular viewpoint the transmembrane com-
ponents in both adherens junctions and desmosomes be-
long to the same protein family, the cadherins. Cadherins
were named according to their Ca2+-dependent adhesion
functionality. Desmosomal cadherins are subdivided into
desmogleins and desmocollins. As indicated in Figure 4,
Cell 2
Cell 1
Intercellular
cleft
Figure 4 Molecular composition of adherens junctions and desmosomes.
Transmembrane proteins of the cadherin family associate to cis-dimers with
cadherins in the same membrane, which then form homophilic
interactions (trans-dimers) with dimers on the surface of the opposing cell.
The intracellular domains of cadherins are linked to cytoskeletal filaments
through a dense plaque of individual adapter proteins.

they form single-span transmembrane proteins with five
extracellular domains and a total of four Ca2+-binding
sites at the interfaces between the individual extracellular
domains. Cadherins expressed on the same cell surface (e.g.
cell 1 in Figure 4) form so-called cis-dimers, which then
establish homophilic interactions with cis-dimers on the
opposing cell surface. These constructs are then referred to
as trans-dimers. In mature adherens junctions a large
number of cadherin dimers group together and, because of
their interdigitated and alternating association, the result-
ing structures are called cadherin zippers (Figure 4). This
model of cadherin interactions is also proposed for des-
mosomes but some very recent studies provided evidence
that desmosomal cadherins of the desmoglein type need to
interact with cadherins of the desmocollin type and thus
form a heterophilic combination. One major difference
between cadherins found in adherens junctions and those
present in desmosomes is their cytoplasmic domain. The
intracellular parts of the proteins determine which adapter
proteins can bind, and thus which kind of filament is con-
nected to the transmembrane components. The most
prominent linker proteins in adherens junctions are the
a-, b- and g-catenins, vinculin, paxilin and a-actinin,
whereas plakoglobin and desmoplakin are the molecular
adapters found in desmosomes. Plaques of these proteins
and the corresponding filaments form on the intracellular
side of adherens junctions or desmosomes, which gives
these junctions their characteristic appearance in electron
micrographs. During embryonic development, but also
when individual cells are stimulated to make junctions,
adherens junctions form first and are a prerequisite for
subsequent desmosome formation.
The present molecular model of adherens junctions, with
its manageable number of molecular constituents, may
have to be revised in the near future, as a new junctional
protein combination has been found in adherens junctions
under certain circumstances. The transmembrane protein
nectin, which is not a member of the cadherin family but
belongs to the immunoglobulin superfamily, together with
its intracellular adapter proteins afadin and ponsin, was
proposed to be mixed into the cadherin aggregates. Con-
sistent with the cadherin-based cell contacts, afadin and
ponsin provide a molecular connection to the actin cyto-
skeleton.
Cell adhesion outside mature junctions
As mentioned above, adherens junctions and desmosomes
are ensembles of aggregated cadherins that are associated
with thick filament bundles of the cytoskeleton, which
makes them easy to spot in electron micrographs. In ad-
dition to these mature cell junctions, individual spots of
cadherin-mediated cell adhesion are spread all over the
contact area between two opposing cells. However, since
these are only formed by oligomers of cadherin dimers,
Cell Junctions
they are not easily detectable by electron microscopy. It is
generally assumed that these scattered interaction sites are
the starting point of all cell adhesion that eventually cul-
minates in the formation of mature cell junctions, as de-
scribed above. But even these distributed sites of cell
adhesion have a direct molecular linkage to the cytoskel-
eton, which already provides an enormous mechanical
stability.
More about cadherins
Cadherins are categorized by the tissue in which they are
predominantly expressed. E-cadherin received its name
from its prominent occurrence on epithelial cells, whereas
VE-cadherin is assigned to vascular endothelial cells. N-
cadherin is found on nerve and heart cells, whereas P-cad-
herin is characteristic for cells derived from the placenta
and the epidermis. It is, however, important to note that
the cell types mentioned above do not exclusively express
one single cadherin type but often a mixture. The name was
only derived from the tissue of strongest expression. Cad-
herins are regarded as the most important cell adhesion
molecules in the developing embryo, and the Ca2+-de-
pendent activity seems to be an important developmental
mechanism for either immobilizing cells in a certain envi-
ronment or releasing them for migration to a different lo-
cation. There is also strong evidence that the metastatic
potential of carcinoma cells (i.e. tumour cells that have
developed from epithelial cells) is strongly correlated with
the expression of E-cadherin. Loss of E-cadherin-mediated
cell–cell interactions allows carcinoma cells to leave the
primary tumour and migrate through adjacent tissue into
the blood vessels and beyond.
Other cell adhesion molecules
In addition to cadherins, other proteins have been assigned
to the group of cell adhesion molecules (CAMs). Ca2+-
independent cell adhesion is predominantly mediated by
CAMs that belong to the immunoglobulin superfamily, as
they have at least one domain that has strong similarity to
the three-dimensional folding of antibodies. The neural cell
adhesion molecule (NCAM), highly expressed on nerve
cells, is the best-characterized molecule among these and
forms homophilic interactions with NCAMs on the other
cell surface. There are more than 20 different NCAMs,
differing mostly in their cytoplasmic domains and the way
they participate in cytoplasmic structures or signal trans-
duction pathways. The intercellular adhesion molecule
(ICAM) is expressed on stimulated endothelial cells and it
interacts with a certain subtype of surface protein on leu-
cocytes, the integrins. This heterotypic interaction allows
the leucocytes to bind to the vessel wall and extravasate
into the surrounding tissue. A very similar function is as-
cribed to a group of proteins referred to as selectins, which
5
 Cell Junctions
are also expressed on endothelial cells. These proteins rec-
ognize certain carbohydrate structures on the surface of
leucocytes and thereby form a site of heterotypic cell ad-
hesion.
Synapses, Gap Junctions and
Plasmodesmata: Communicating Cell
Junctions
Although very different in molecular composition, and
even the organisms from which they are expressed,
synapses, gap junctions and plasmodesmata share a basic
function, serving as cell junctions for communication pur-
poses. The mechanisms by which communication is
achieved are rather different for synapses on the one hand
and gap junctions and plasmodesmata on the other, so they
are discussed separately.
Synapses
Synapses are very special cell junctions between two neu-
rons (interneuronal junctions) or neurons and muscle cells
(neuromuscular junctions). In these junctions, the mem-
brane of the so-called presynaptic cell (1 in Figure 5), which
sends a signal, is locally brought into very close contact
with the membrane of the postsynaptic cell (2 in Figure 5),
which receives and processes the signal. The synaptic cleft
(3 in Figure 5) between these two membranes is only 20–
30 nm wide and it is filled with components of the extra-
cellular matrix. This protein and glycoprotein filling is one
mechanism by which the membranes of the two cells are
locally stabilized in such close apposition. Both cells on
either side of the junction express the corresponding cell
surface receptors, so the extracellular membrane serves as
an adhesion mediator or a molecular glue. A molecular
1 the hexagon (Figure 6). This functional unit of a gap junc-
5 opposing cell, a direct, water-filled connection of approx-
4
2
3
6 are, however, size-selective. Whereas molecules up to a
Figure 5 Cartoon of a chemical synapse as it is formed between neurons
or between neurons and muscle cells. 1, Presynaptic membrane; 2,
postsynaptic membrane; 3, synaptic cleft; 4, synaptic adherens junction; 5,
neurotransmitter-loaded secretory vesicles; 6, transmitter-gated ion
channels like the acetylcholine receptor.
6
assembly that is ultrastructurally very similar to adherens
junctions in epithelial cells, called a synaptic adherens
junction, also provides mechanical stability at synapses. N-
cadherin transmembrane proteins on the surface of either
cell form clustered homophilic interactions in the synaptic
cleft which are stabilized by their linkage to the microfil-
ament system via a set of intracellular adapter proteins (4 in
Figure 5). An action potential arriving at the presynaptic
membrane (1) triggers the fusion of neurotransmitter-
loaded vesicles (5 in Figure 5) with the axonal plasma mem-
brane. The neurotransmitters are released into the synaptic
cleft, diffuse very fast across this narrow extracellular space
and bind to corresponding transmitter-gated ion channels
(6 in Figure 5) on the postsynaptic membrane. Upon chan-
nel opening, the associated influx of ions depolarizes the
cell and thereby generates an action potential that can now
migrate along the membrane of the postsynaptic nerve or
muscle cell. Any excess of the neurotransmitter in the
synaptic cleft that is no longer needed is quickly removed
and the state of the synaptic junction is regenerated. Taken
together, synaptic junctions provide the molecular re-
quirements to transmit electrical impulses from one cell to
the next via a burst of a transmitter molecule secreted into a
defined volume close to the receiver membrane where the
corresponding receptors are located.
Gap Junctions and Plasmodesmata
By a very different mechanism both gap junctions and
plasmodesmata also allow the communication between
two adjacent cells, which are animal cells in the case of the
former and plant cells in the case of the latter. Both of these
junctions are water-filled channels between the interiors of
neighbouring cells that form a direct exchange pathway for
ions, signal molecules and metabolites, as long as these do
not exceed a certain molecular threshold size.
Gap junctions, also referred to as electrical synapses, are
basically formed by a group of transmembrane proteins
called connexins. Six of these connexins are arranged in
such a way that a hydrophilic pore opens up in the core of
tion is named a connexon. When this connexon hemichan-
nel binds a corresponding structure on the surface of the
imately 1.5 nm diameter is formed between the two cyto-
plasms (Figure 6). These channels can be used to share
information but also to allow diffusion of metabolites from
one cell to the other (metabolic coupling). Gap junctions
molecular weight of 1000 g mol 2 1 (1 kDa), for instance
sugars, amino acids and nucleotides, can almost freely dif-
fuse through the pore, larger molecules are rejected. Ac-
cordingly, the cells may share small metabolites and ions
but not macromolecules like proteins, DNA or RNA. The
term electrical synapses, in contrast to the chemical

Cytoplasm
Ions, amino
acids, sugars,
nucleotides
Intercellular cleft
Figure 6 Gap junction channels connecting the interiors of two adjacent
cells. Due to the size-selectivity of gap junctions, only ions, amino acids,
sugars and other small metabolites may pass the junction, whereas
macromolecules like proteins and nucleic acids are rejected.
synapses described in the preceding paragraph, arose be-
cause gap junctions allow a free flow of ions from one cell to
the next through the gap junction channels, which provides
the requirements to transmit electrical signals without
making use of chemical transmitters. This mode of action
potential propagation is faster than the transmitter-based
mechanism described above. Heart muscle cells are exten-
sively interconnected by gap junctions, which provide the
basis for highly synchronized contraction of the heart
muscle.
The participating cells can adjust the permeability of the
gap junction channels. When the connexon hemichannels
are open, the six connexin proteins are somewhat tilted
with respect to the direction of the channel. A sudden in-
crease of the intracellular Ca2+ concentration or a drop of
pH provides conformational changes within the connex-
ins, which eventually change their tilt angles and thereby
close the channel. This is a rather important safety pre-
caution because it avoids that breakdown of the osmotic
balance in one cell, for instance when the plasma mem-
brane has become permeable for any reason, propagates
into all neighbouring cells and destroys the tissue.
Plasmodesmata between plant cells have a very similar
function, although the structure is rather different. This
different structural solution for the same problem is ap-
parently necessary because the rather thick cell wall does
not allow plants cells to get into sufficiently intimate con-
tact to build gap junction-like structures in which proteins
bridge between the two cytoplasms. At plasmodesmata,
the otherwise continuous cell wall around plant cells forms
Cell Junctions
Cytoplasm Cell wall
Desmotubule
Cytoplasm Cytoplasm
Proteins,
nucleic acid
Cell wall
Plasma
membranes
Figure 7 Plasmodesma forming a water-filled channel between the
interior of two plant cells. The channel width is reduced by a vesicle
(desmotubule) that originates from the endoplasmic reticulum.
point-like openings at which the plasma membranes of
neighbouring cells fuse (Figure 7). This provides a contin-
uous channel lined by the plasma membrane, which is
roughly 30–50 nm in diameter (compared to 1.5 nm for gap
junction channels). To reduce the size of this connection
and to provide some size selectivity, a membrane vesicle
(desmotubule) that has budded off the endoplasmic re-
ticulum acts like a cork in a bottle (Figure 7). This plugging
of the intercellular connection slows down diffusional ex-
change and only allows sufficiently small molecules to mi-
grate into neighbouring cells.
Further Reading
Alberts B, Bray D and Lewis J et al. (1994) Molecular Biology of the Cell.
New York: Garland.
Cereijido M (ed.) (1991) Tight Junctions. Boca Raton, FL: CRC Press.
Evans WH and Martin PEM (2002) Gap junctions: structure and func-
tion. Molecular Membrane Biology 19(2): 121–136.
Green KJ and Gaudry CA (2000) Are desmosomes more than tethers for
intermediate filaments?. Nature Reviews: Molecular Cell Biology 1:
208–216.
Tepass U, Truong K and Godt D et al. (2000) Cadherins in embryonic
development and neural morphogenesis. Nature Reviews: Molecular
Cell Biology 1: 91–100.
Tsukita S, Furuse M and Itoh M (2001) Multifunctional strands in tight
junctions. Nature Reviews: Molecular Cell Biology 2: 285–293.