Models of Membrane Structure: An Experimental Perspective Until electron microscopy was applied to

the study of cell structure in the early 1950s, no one had ever seen a membrane. Yet indirect evidence
led biologists to postulate the existence of membranes long before they could actually be seen. In fact,
researchers have been trying to understand the molecular organization of membranes for more than a
century. The intense research effort paid off, however, because it led eventually to the fluid mosaic
model of membrane structure. This model, which is now thought to be descriptive of all biological
membranes, envisions a membrane as two quite fluid layers of lipids, with proteins localized within and
on the lipid layers and oriented in a specific manner with respect to the inner and outer membrane
surfaces. Although the lipid layers are turning out to be much more complex than originally thought, the
basic model is almost certainly correct as presently envisioned. Before looking at the model in detail, we
will describe some of the central experiments leading to this view of membrane structure and function.
As we do so, you may also gain some insight into how such developments come about, as well as a
greater respect for the diversity of approaches and techniques that are often important in advancing our
understanding of biological phenomena. Figure 7-3 presents a chronology of membrane studies Models
of Membrane Structure: An Experimental Perspective 159 that began over a century ago and led
eventually to our current understanding of membranes as fluid mosaics. Overton and Langmuir: Lipids
Are Important Components of Membranes A good starting point for our experimental overview is the
pioneering work of German scientist Charles Ernest Overton in the 1890s. Working with cells of plant
root hairs, he observed that lipid-soluble substances penetrate readily into cells, whereas water-soluble
substances do not. From his studies, Overton concluded that lipids are present on the cell surface as
some sort of “coat” (Figure 7-3a). He even suggested that cell coats are probably mixtures of cholesterol
and lecithin, an insight that proved to be remarkably farsighted in light of what we now know about the
prominence of sterols and phospholipids as membrane components. A second important advance came
about a decade later through the work of Irving Langmuir, who studied the behavior of purified
phospholipids by dissolving them in benzene and layering samples of the benzenelipid solution onto a
water surface. As the benzene evaporated, the molecules were left as a lipid film one molecule thick—
that is, a “monolayer.” Because phospholipids are amphipathic molecules (see Figure 2-11), Langmuir
reasoned that the phospholipids orient themselves on water such that their hydrophilic heads face the
water and their hydrophobic tails protrude away from the water (Figure 7-3b). Langmuir’s lipid
monolayer became the basis for further thought about membrane structure in the early years of the
twentieth century. Gorter and Grendel: The Basis of Membrane Structure Is a Lipid Bilayer The next
major advance came in 1925 when two Dutch physiologists, Evert Gorter and F. Grendel, extracted the
lipids from a known number of erythrocytes (red blood cells) and used Langmuir’s method to spread the
lipids as a monolayer on a water surface. They found that the area of the lipid film on the water was
about twice the estimated total surface area of the erythrocyte. Therefore, they concluded that the
erythrocyte plasma membrane consists of not one but two layers of lipids. Hypothesizing a bilayer
structure, Gorter and Grendel reasoned that it would be thermodynamically favorable for the nonpolar
hydrocarbon chains of each layer to face inward, away from the aqueous milieu on either side of the
membrane. The polar hydrophilic groups of each layer would then face outward, toward the aqueous
environment on either side of the membrane (Figure 7-3c). Gorter and Grendel’s experiment and their
conclusions were momentous because this work represented the first attempt to understand
membranes at the molecular level. Moreover, the lipid bilayer they envisioned became the Polar
Nonpolar 1900 1880 Overton Langmuir Gorter and Grendel Davson and Danielli Robertson Singer and
Nicolson Unwin and Henderson 1920 1940 1960 1980 2000 (b) Lipid monolayer (c) Lipid bilayer (d) Lipid
bilayer plus protein sheets (e) Unit membrane (f) Fluid mosaic model (g) Membrane protein structure (h)
Lipid raft (a) Lipid nature of membrane Alpha helix Lipid FIGURE 7-3 Timeline for Development of the
Fluid Mosaic Model. The fluid mosaic model of membrane structure that Singer and Nicolson proposed
in 1972 was the culmination of studies dating back to the 1890s (a)–(e). This model (f) has been
significantly refined by subsequent studies (g and h). 160 Chapter 7 Membranes: Their Structure,
Function, and Chemistry basic underlying assumption for each successive refinement in our
understanding of membrane structure. Davson and Danielli: Membranes Also Contain Proteins Shortly
after Gorter and Grendel proposed their bilayer model in 1925, it became clear that a simple lipid
bilayer, could not explain all the properties of membranes— particularly those related to surface
tension, solute permeability, and electrical resistance. For example, the surface tension of a lipid film
was significantly higher than that of cellular membranes but could be lowered by adding protein to the
lipid film. Moreover, sugars, ions, and other hydrophilic solutes readily moved into and out of cells even
though pure lipid bilayers are nearly impermeable to water-soluble substances. To explain such
differences, Hugh Davson and James Danielli suggested that proteins are present in membranes. They
proposed in 1935 that biological membranes consist of lipid bilayers that are coated on both sides with
thin sheets of protein (Figure 7-3d). Their model, a protein-lipid-protein “sandwich,” was the first
detailed representation of membrane organization and dominated the thinking of cell biologists for the
next several decades. The original model was later modified to accommodate additional findings.
Particularly notable was the suggestion, made in 1954, that hydrophilic proteins might penetrate into
the membrane in places to provide polar pores through an otherwise hydrophobic bilayer. These
proteins could then allow water-soluble substances to cross the cell membrane. Specifically, the lipid
interior accounted for the hydrophobic properties of membranes, and the protein components
explained their hydrophilic properties. The real significance of the Davson–Danielli model, however, was
that it recognized the importance of proteins in membrane structure. This feature, more than any other,
made the Davson–Danielli sandwich the basis for much subsequent research on membrane structure.

Robertson: All Membranes Share a Common Underlying Structure With the advent of electron
microscopy in the 1950s, cell biologists could finally verify the presence of a plasma membrane around
each cell. They could also observe that most subcellular organelles are bounded by similar membranes.
Furthermore, when membranes were stained with osmium, a heavy metal, and then examined closely
at high magnification, they were found to have extensive regions of “railroad track” structure that
appeared as two dark lines separated by a lightly stained central zone, with an overall thickness of 6–8
nm. This pattern is seen in Figure 7-4 for the plasma membranes of two adjacent cells that are separated
from each other by a thin intercellular space. Because this same staining pattern was observed with
many different kinds of membranes, J. David Robertson suggested that all cellular membranes share a
common underlying structure, which he called the unit membrane (see Figure 7-3e). When first
proposed, the unit membrane structure seemed to agree remarkably well with the Davson–Danielli
model. Robertson suggested that the lightly stained space (between the two dark lines of the trilaminar
pattern) contains the hydrophobic region of the lipid molecules, which do not readily stain. Conversely,
the two dark lines were thought to represent phospholipid head groups and the thin sheets of protein
bound to the membrane surfaces, which appear dark because of their affinity for heavy metal stains.
This interpretation appeared to provide strong support for the Davson–Danielli view that a membrane
consists of a lipid bilayer coated on both surfaces with thin sheets of protein. Further Research Revealed
Major Shortcomings of the Davson–Danielli Model Despite its apparent confirmation by electron
microscopy and its extension to all membranes by Robertson, the Davson–Danielli model encountered
difficulties in the 1960s as more and more data emerged that could not be reconciled with their model.
Based on electron microscopy, most membranes were reported to be about 6–8 nm thick—and, of this,
the lipid bilayer accounted for about 4–5 nm. That left only about 1–2 nm of space on either surface of
the bilayer for the membrane protein, a space that could at best accommodate a thin monolayer of
protein. Yet after membrane proteins were isolated and studied, it became apparent that most of them
were globular proteins with sizes and shapes that are inconsistent with the concept of thin sheets of
protein on the two surfaces of the membrane. As a further complication, the Davson–Danielli model did
not readily account for the distinctiveness of different kinds of membranes. Depending on their source,
membranes vary considerably in chemical composition and Cell 1 Cell 2 Intercellular space FIGURE 7-4
Trilaminar Appearance of Cellular Membranes. This electron micrograph of a thin section through two
adjacent cells shows their plasma membranes separated by a small intercellular space. Each membrane
appears as two dark lines separated by a lightly stained central zone in a staining pattern that gives each
membrane a trilaminar, or “railroad track,” appearance (TEM). Models of Membrane Structure: An
Experimental Perspective 161 especially in the ratio of protein to lipid (Table 7-1), which can vary from 3
or more in some bacterial cells to only 0.23 for the myelin sheath surrounding nerve axons. Even the
two membranes of the mitochondrion differ significantly: The protein/lipid ratio is about 1.2 for the
outer membrane and about 3.5 for the inner membrane, which contains all the enzymes and proteins
related to electron transport and ATP synthesis. Yet all of these membranes look essentially the same
when visualized using electron microscopy. The Davson–Danielli model was also called into question by
studies in which membranes were exposed to phospholipases, enzymes that degrade phospholipids by
removing their head groups. According to the model, the hydrophilic head groups of membrane lipids
should be covered by a layer of protein and therefore protected from phospholipase digestion.
However, up to 75% of the membrane phospholipid can be degraded when the membrane is exposed to
phospholipases, suggesting that many of the phospholipid head groups are exposed at the membrane
surface and not covered by a layer of protein. Moreover, the surface localization of membrane proteins
specified by the Davson–Danielli model was not supported by the experience of scientists who tried to
isolate such proteins. Most membrane proteins turned out to be quite insoluble in water and could be
extracted only by using organic solvents or detergents. These observations indicated that many
membrane proteins are hydrophobic (or at least amphipathic) and suggested that they are located, at
least partially, within the hydrophobic interior of the membrane rather than on either of its surfaces.

Singer and Nicolson: A Membrane Consists of a Mosaic of Proteins in a Fluid Lipid Bilayer The preceding
problems with the Davson–Danielli model stimulated considerable interest in the development of new
ideas about membrane organization, culminating in 1972 with the fluid mosaic model proposed by S.
Jonathan Singer and Garth Nicolson. This model, which now dominates our view of membrane
organization, has two key features, both implied by its name. Simply put, the model envisions a
membrane as a mosaic of proteins embedded in, or at least attached to, a fluid lipid bilayer (Figure 7-3f).
This model retained the basic lipid bilayer structure of earlier models but viewed membrane proteins in
an entirely different way—not as thin sheets on the membrane surface but as discrete globular entities
within the lipid bilayer (Figure 7-5a). This way of thinking about membrane proteins was revolutionary
when Singer and Nicolson first proposed it, but it turned out to fit the data quite well. Three classes of
membrane proteins are now recognized based on differences in how the proteins are linked to the
bilayer. Integral membrane proteins are embedded within the lipid bilayer, where they are held in place
by the affinity of hydrophobic segments of the protein for the hydrophobic interior of the lipid bilayer.
Peripheral proteins are much more hydrophilic and are therefore located on the surface of the
membrane, where they are linked noncovalently to the polar head groups of phospholipids and/or to
the hydrophilic parts of other membrane proteins. Lipid-anchored proteins are essentially hydrophilic
proteins and therefore reside on membrane surfaces, but they are covalently attached to lipid
molecules that are embedded within the bilayer. The fluid nature of the membrane is the second critical
feature of the Singer–Nicolson model. Rather than being rigidly locked in place, most of the lipid
components of a membrane are in constant motion, capable of lateral mobility (i.e., movement parallel
to the membrane surface). Many membrane proteins are also able to move laterally within the
membrane, although some proteins are anchored to structural elements such as the cytoskeleton on
one side of the membrane or the other and are therefore restricted in their mobility.