. 17
( 66 .)


Basal lamina Hemidesmosome
Integral membrane

Fig. 4.1 Schematic representation of various cell“cell and cell“substratum adhesion
complexes. On the left, a tight junction seals neighboring cells together in an epithelial
sheet to prevent the leakage of molecules between them. An adhesion belt, also known as
an adherens junction, joins an actin bundle (green strands) in one cell to a similar bundle in
a neighboring cell via integral membrane proteins called cadherins and accessory
proteins on the cytoplasmic faces. A desmosome anchors intermediate ¬laments (blue
strands) in one cell to those in its neighbor via cadherins and accessory proteins. A gap
junction provides tubular channels made up of connexin proteins, which allow the passage
between cells of small water-soluble ions and molecules. A hemidesmosome anchors
intermediate ¬laments at the basal surface of the cell to the basal lamina, via integral
membrane proteins called integrins and cytoplasmic accessory proteins. A focal adhesion
complex or focal contact anchors actin bundles to the extracellular matrix via integrins
and cytoplasmic accessory proteins.
On the right, epithelioid cells in early embryos, which typically do not contain mature
junctional complexes, and certain cells in the adult organism such as lymphocytes and
platelets, utilize cadherins, immunoglobulin(IgG)-like CAMs (e.g., PECAM), or selectins,
adhesive interactions that are homophilic (pairs of the same molecule) or heterophilic
(pairs of different molecules). These are termed non-junctional adhesions. Integrins and
integral membrane proteoglycans mediate attachment to the ECM. (After Alberts et al.,

of the originally nonpolarized cells may lose their CAMs along part of
their surfaces, thus preventing them from adhering to one another
at those sites (Tsarfaty et al., 1992, 1994). If the cells move around ran-
domly, maximizing their contacts with one another in accordance
with this constraint (see below), a hollow region or lumen will natu-
rally arise. Cells polarized in their expression of CAMs arise during
the formation of the blastula; they are necessary to this process, as

Fig. 4.2 Schematic illustration of lumen formation when polarized cells express cell
adhesion molecules only on restricted parts of their surface. The shading in eight cells in
the middle panel represents the lack of adhesion molecules on corresponding regions of
the cells. As a consequence, if the minimization of con¬gurational energy is the driving
force in cell rearrangement, then a lumen, shown on the right, is bound to appear. (Based
on Newman, 1998a.)

we saw in Chapter 2. This is true even if the blastula is not hollow,
since it must maintain a free outer surface.
Differential adhesion also plays a role in a developmental phe-
nomenon known as ˜˜compartment formation.” Compartments are
regions of embryonic tissue, which, though similar in other respects,
do not mix (exchange cells) across a common boundary (Crick and
Lawrence, 1975; Garcia-Bellido et al., 1976). In certain cases (particu-
larly as tissues mature) such interfaces of immiscibility are not depen-
dent on adhesive differentials alone but are reinforced by specialized
structures (Heyman et al., 1995). Nonetheless, the initial establishment
of such boundaries during embryogenesis typically depends on dif-
ferential adhesion (Guthrie and Lumsden, 1991; Heyman et al., 1995;
Godt and Tepass, 1998; Gonzalez-Reyes and St Johnston, 1998; Hayashi
and Carthew, 2004).
In order to understand the physical basis of the role of differen-
tial adhesion in developmental morphogenesis it is necessary to have
some knowledge of the structure and composition of the cell surface,
which we present in the following section.

The cell surface
The majority of CAMs and SAMs are transmembrane proteins, which
attach to the actin cytoskeleton at their intracellular ends. Cadherins,
for example, accomplish this through a protein complex involving
alpha and beta catenin (Gumbiner, 1996). Moreover, if the cytoskele-
tal attachment is damaged or missing, cadherins™ ability to establish
an adhesive junction (called the adherens junction) is severely compro-
mised (Nagafuchi and Takeichi, 1988). Similarly, if integrins are un-
able to attach to the cytoskeleton, the mechanical integrity of the
focal adhesion complex (by which epithelial cells attach to the basal
lamina, or moving cells assemble on the substrate along which they
translocate) is strongly reduced (Wang et al., 1993).

The extracellular portion of most membrane proteins (and of
lipids in the outer layer of the plasma membrane) is ˜˜decorated” by
either short chains of sugars (oligosaccharides) or longer polysaccha-
rides containing amino sugars (glycosaminoglycans). Proteins linked
to oligosaccharides are referred to as glycoproteins, whereas those with
glycosaminoglycan attachments are called proteoglycans. These sugar-
containing molecules collectively form the glycocalyx (Fig. 4.1), a coat-
ing that extends several nanometers from the plasma membrane on
its noncytoplasmic face. The characteristic glycocalyx of each special-
ized cell represents the basis by which it is recognized by cells like it
and by other cell types. For the basal portions of epithelial cells, and
of mesenchymal and connective tissues (which will be discussed in
Chapter 6), the microenvironment is even more elaborate (see Fig. 4.1),
constituting an ECM that extends beyond the glycocalyx and contains
additional proteins, glycoproteins, proteoglycans, and in some cases
minerals. In this chapter we will con¬ne our discussion to epithelial
and epithelioid tissues, in which cells contact one another more or
less directly via their glycocalyces.
As a result of the complex nature of the glycocalyx, cell recog-
nition and subsequent adhesion involve a variety of chemical bonds
and forces: covalent bonds, e.g., those between protein ligands and
oligosaccharide side chains; ionic bonds or electrostatic interactions,
between charged regions of membranes; van der Waals forces between
induced and permanent dipole moments due to polar molecules; and
hydrogen bonds between membrane proteins. Each of these bonds
and the associated forces have a characteristic distance and strength.
In addition, undulations in the membrane cause steric hindrance to
adhesion, an effect that may reduce the magnitude of the overall
adhesive force.
From these considerations it is clear that membrane interactions
and cell adhesion have many facets. While we will not attempt to
present an exhaustive discussion of either the biology or the physics
of membrane interactions, an enormous ¬eld of research that extends
well beyond developmental biology, we will provide enough details in
both areas to demonstrate how adhesion and its modulation control
the changes in tissue form that occur during key developmental pro-
cesses. For more on the physical basis of membrane interactions, see
Israelachvili (1991) and Boal (2002).

Cell adhesion: speci¬c and nonspeci¬c aspects
The preceding discussion indicates that cell--cell and cell--substratum
adhesion have both nonspeci¬c and speci¬c aspects. As with many
of the other phenomena considered in this book, the nonspeci¬c,
or generic, interactions represent the physical basis that is likely to
have been built upon and embellished over the course of evolution
to generate biological speci¬city (see Chapter 10).
Before we attempt to construct a physical model of cell adhesion
we will brie¬‚y review the composition of a typical membrane and its

Fig. 4.3 Top: Schematic representation of the lipid-bilayer cell membrane, with some
of its characteristic components. Bottom: The more detailed structure of the
amphipathic (partly hydrophobic, partly hydrophilic) phospholipid molecules, the main
building blocks of the cell membrane. The lower ¬gure also shows the relationship
between the phospholipid membrane scaffold and the lipid cholesterol, an important
determinant of the viscosity of animal cell membranes.

microenvironment, shown schematically in Fig. 4.3. The membranes
surrounding and within cells consist of a double layer of a class
of molecules with charged and nonpolar regions known as phos-
pholipids. Since these elongated molecules have both water-repellent
(hydrophobic) and water-attracting (hydrophilic) domains they are

referred to as ˜˜amphipathic.” Phospholipids consist of two long in-
equivalent hydrocarbon chains, each linked to a carbon in the three-
carbon glycerol backbone. The third carbon in the glycerol backbone
is linked to a polar hydrophilic head group, which consists of a
negatively charged phosphate group (PO4 ) and a nitrogen-containing
˜˜base,” i.e., a molecule that is positively charged at neutral pH. In-
terspersed among the phospholipids in the bilayer are other lipid
molecules such as glycolipids, in which the head group contains a
sugar, and cholesterol, which disrupts the van der Waals interactions
among the phospholipids and increases the viscosity of the bilayer.
In addition, integral membrane proteins are inserted into the phospho-
lipid bilayer. These can act as channels, receptors, or mechanical links
between the inside and outside of the cell (see below).
The plasma membrane, which encloses the entire cell, forms a semi-
permeable seal selectively allowing or preventing the passage of
molecules. Water molecules, by virtue of their size, can pass freely
through the membrane. Molecules that cannot diffuse through the
membrane (because they either are too large or are charged or po-
lar and thus associate with the aqueous media on either side of the
membrane in preference to the lipid bilayer) can be speci¬cally trans-
ported from one side to the other via channels composed of integral
membrane proteins or protein complexes (e.g., the Na+ -channel and
the K+ -channel).
The lipid bilayer forms a two-dimensional ¬‚uid: its component
molecules can diffuse relatively easily in the plane of the membrane,
but because of the surrounding water cannot escape from it. This
molecular arrangement plays an important role in many cell func-
tions and, in particular, in cell adhesion. The membrane is decorated
by a myriad of receptors. These molecular antennas, most of which are
transmembrane proteins, provide communication channels between
the cell and its environment. When a receptor is activated by its lig-
and (the speci¬c molecule to which it binds), a cascade of signaling
events is typically evoked. CAMs and SAMs are important classes of
receptors. They not only establish physical links between cells or the
ECM but also are important components of signaling pathways. Their
critical location at the cell surface makes them mediators of ˜˜inside-
out” and ˜˜outside-in” signal transduction (Giancotti and Ruoslahti,
How an adhesive bond is formed when two cells approach each
other is still not fully understood. Most models suggest that bond
formation is not instantaneous: molecules diffuse within the mem-
brane and accumulate at special points (e.g., cadherins at adherens
junctions, integrins at focal adhesion plaques). Particularly detailed
studies have been performed in the case of cadherins (Shapiro et al.,
1995; Nagar et al., 1996; Angres et al., 1996; Adams et al., 1998; Bog-
gon et al., 2002). Studies on single molecules using atomic force mi-
croscopy (Baumgartner et al., 2000), or a surface force apparatus using
optical interferometry (Sivasankar et al., 1999, 2001), have clari¬ed
some aspects of the physical nature of cadherin--cadherin interactions

but were not designed to take into account important aspects such
as the cytoskeletal attachment of the proteins, which is essential to
the generation of stable cell--cell bonds (Adams et al., 1996, 1998;
Gumbiner 1996). Other studies that utilize the ability of cells with
differing adhesive properties to sort themselves out within a cell mix-
ture (Beysens et al., 2000) estimate the strength of binding indirectly,
but in a more physiologically authentic context.
In what follows, we ¬rst describe the classical Bell model of cell
adhesion (Bell, 1978), which serves as a basis for more sophisticated
models. We then apply the relevant aspects of this model, along with
the physical notions introduced earlier, to cell sorting and to certain
related morphogenetic processes underlying organ formation.

The kinetics of cell adhesion
There are many situations in development in which a cell establishes
new contacts with its surroundings. At one extreme this can happen
by default: after cytokinesis two daughter cells may remain in contact
(at least for some time) and the cells do not need to move in order to
¬nd new attachments. This typically happens in the tightly adhering
cells of epithelial tissues. At the other extreme cells may detach from
their progenitors (this happens with the neural crest, see Chapter 6),
translocate to new sites, and form connections with other cells in
an environment that is itself changing with time. For generality we
will develop a model in which cells are rather mobile. The starting
point is a mixture of cells (as in a sorting assay) in which the cells
move around until a con¬guration corresponding to a minimum (free)
energy is established. In the course of this process, cells must break
and reform adhesive bonds with one another. We thus consider two
cells as they approach each other and eventually attach.
To describe this process in a relatively simple physical way, we will
make use of the ¬‚uid mosaic model of the cell membrane (Singer and
Nicholson, 1972; Jacobson et al., 1995; see Fig. 4.3). This model assumes
that the membrane™s structural components such as phospholipids
and proteins (in particular CAMs) can undergo two-dimensional dif-
fusion in the plane of the lipid bilayer. Such diffusion has been stud-
ied extensively and diffusion coef¬cients have been measured (Saxton
and Jacobson, 1997; Jacobson et al., 1997; Fujiwara et al., 2002). Their
magnitudes depend on the permeability of the lipid bilayer, on the
mass of the diffusing molecules, and, in the case of proteins, on pos-
sible cytoskeletal attachments. Typical values (for receptors on the
surface of T lymphocytes, for example) are from 10’10 to 10’12 cm2 /s.
(Compare this with the much larger diffusion coef¬cient of a G actin
molecule in an aqueous medium (see Chapter 1), which is around
10’6 cm2 /s).
If there are no CAMs on the surfaces of the two cells, nonspeci¬c
binding may take place. Given the types of force involved this can only

be transient. More typically, the two cells would not adhere at all since
the surfaces of most cells are negatively charged and this would lead
to repulsion. In the presence of CAMs, however, more long-lived bonds
can be formed and these will be energetically more favorable than the
nonspeci¬c bonds. There are several reasons for this. First, owing to
mobile ions in the vicinity of the membrane the net negative charge
is strongly screened. As a result, beyond about a nanometer from the
cell surface the electrostatic repulsion is ineffective. Second, once
the two cells are suf¬ciently close that speci¬c bonds between them
can form, diffusion-dependent clustering of the CAMs is induced
and highly cooperative multimolecular binding interactions can take

Formation of an adhesive bond


. 17
( 66 .)