第三章 蛋白质的三维结构

The possible conformations of a protein or protein segment include any structural state it can achieve without breaking covalent bonds. A change in conformation could occur, for example, by rotation about single bonds. However, of the many conformations that are theoretically possible in a protein containing hundreds of single bonds, one or a few generally predominate under biological conditions. The need for multiple stable conformations reflects the changes that must take place in most proteins as they bind to other molecules or catalyze reactions. The conformations existing under a given set of conditions are usually the ones that are thermodynamically the most stable — that is, having the lowest free energy (G).

For the vast majority of proteins, a particular structure or small set of structures is critical to function. However, in many cases, parts of proteins lack discernible structure. These protein segments are intrinsically disordered. In some cases, entire proteins are intrinsically disordered, yet are fully functional.

Stability is the tendency of a protein to maintain a native conformation. Native proteins are only marginally stable; the ΔG separating the folded and unfolded states in typical proteins under physiological conditions is in the range of only 5 to 65 kJ/mol. A given polypeptide chain can theoretically assume countless conformations, and as a result, the unfolded state of a protein is characterized by a high degree of conformational entropy. This entropy, along with the hydrogen-bonding interactions of many groups in the polypeptide chain with the solvent (water), tends to maintain the unfolded state.

Outside the cell, the environment is often more oxidizing, and disulfide formation is more likely to occur.

In eukaryotes, disulfide bonds are found primarily in secreted, extracellular proteins (for example, the hormone insulin). Disulfide bonds are also uncommon in bacterial proteins. However, thermophilic bacteria, as well as the archaea, typically have many proteins with stabilizing disulfide bonds; this is presumably an adaptation to life at high temperatures.

For all proteins of all organisms, weak interactions are especially important in the folding of polypeptide chains into their secondary and tertiary structures. The association of multiple polypeptides to form quaternary structures also relies on these weak interactions. About 200 to 460 kJ/mol are required to break a single covalent bond, whereas weak interactions can be disrupted by a mere 0.4 to 30 kJ/mol. Individual covalent bonds, such as disulfide bonds linking separate parts of a single polypeptide chain, are clearly much stronger than individual weak interactions. Yet, because they are so numerous, the weak interactions predominate as a stabilizing force in protein structure. In general, the protein conformation with the lowest free energy (that is, the most stable conformation) is the one with the maximum number of weak interactions.

The stability of a protein is not simply the sum of the free energies of formation of the many weak interactions within it. For every hydrogen bond formed in a protein during folding, a hydrogen bond (of similar strength) between the same group and water was broken. The net stability contributed by a given hydrogen bond, or the difference in free energies of the folded and unfolded states, may be close to zero. Ionic interactions may be either stabilizing or destabilizing. We must therefore look elsewhere to understand why a particular native conformation is favored.

On carefully examining the contribution of weak interactions to protein stability, we find that the hydrophobic effect generally predominates. Pure water contains a network of hydrogen-bonded H2O molecules. No other molecule has the hydrogen-bonding potential of water, and the presence of other molecules in an aqueous solution disrupts the hydrogen bonding of water. When water surrounds a hydrophobic molecule, the optimal arrangement of hydrogen bonds results in a highly structured shell, or solvation layer, of water around the molecule (see Fig. 2-7). The increased order of the water molecules in the solvation layer correlates with an unfavorable decrease in the entropy of the water.

The result is a favorable increase in entropy. As described in Chapter 2, this increase in entropy is the major thermodynamic driving force for the association of hydrophobic groups in aqueous solution. Hydrophobic amino acid side chains therefore tend to cluster in a protein’s interior, away from water (think of an oil droplet in water). The amino acid sequences of most proteins thus include a significant content of hydrophobic amino acid side chains (especially Leu, Ile, Val, Phe, and Trp). These are positioned so that they are clustered when the protein is folded, forming a hydrophobic protein core.

Under physiological conditions, the formation of hydrogen bonds in a protein is driven largely by this same entropic effect. Polar groups can generally form hydrogen bonds with water and hence are soluble in water. However, the number of hydrogen bonds per unit mass is generally greater for pure water than for any other liquid or solution, and there are limits to the solubility of even the most polar molecules as their presence causes a net decrease in hydrogen bonding per unit mass. Therefore, a solvation layer forms to some extent even around polar molecules.

Although the energy of formation of an intramolecular hydrogen bond between two polar groups in a macromolecule is largely canceled by the elimination of such interactions between these polar groups and water, the release of structured water as intramolecular associations form provides an entropic driving force for folding.

The hydrophobic effect is clearly important in stabilizing conformation; the interior of a structured protein is generally a densely packed core of hydrophobic amino acid side chains. It is also important that any polar or charged groups in the protein interior have suitable partners for hydrogen bonding or ionic interactions.

One hydrogen bond seems to contribute little to the stability of a native structure, but the presence of hydrogenbonding groups without partners in the hydrophobic core of a protein can be so destabilizing that conformations containing these groups are often thermodynamically untenable.

The favorable free-energy change resulting from the combination of several such groups with partners in the surrounding solution can be greater than the free-energy difference between the folded and unfolded states. In addition, hydrogen bonds between groups in a protein form cooperatively (formation of one makes formation of the next one more likely) in repeating secondary structures that optimize hydrogen bonding, as described below.

The interaction of oppositely charged groups that form an ion pair, or salt bridge, can have either a stabilizing or destabilizing effect on protein structure. As in the case of hydrogen bonds, charged amino acid side chains interact with water and salts when the protein is unfolded, and the loss of those interactions must be considered when researchers evaluate the effect of a salt bridge on the overall stability of a folded protein.

However, the strength of a salt bridge increases as it moves to an environment of lower dielectric constant, ε (p. 46): from the polar aqueous solvent (ε near 80) to the nonpolar protein interior (ε near 4). Salt bridges, especially those that are partly or entirely buried, can thus provide significant stabilization to a protein structure. This trend explains the increased occurrence of buried salt bridges in the proteins of thermophilic organisms. Ionic interactions also limit structural flexibility and confer a uniqueness to a particular protein structure that the clustering of nonpolar groups via the hydrophobic effect cannot provide.

Van der Waals interactions are dipoledipole interactions involving the permanent electric dipoles in groups such as carbonyls, transient dipoles derived from fluctuations of the electron cloud surrounding any atom, and dipoles induced by interaction of one atom with another that has a permanent or transient dipole. As atoms approach each other, these dipole-dipole interactions provide an attractive intermolecular force that operates over only a limited intermolecular distance (0.3 to 0.6 nm).

Individually, van der Waals interactions contribute little to overall protein stability. However, in a well-packed protein, or in an interaction between a protein and another protein or other molecule at a complementary surface, the number of such interactions can be substantial.

Most of the structural patterns outlined in this chapter reflect two simple rules: (1) hydrophobic residues are largely buried in the protein interior, away from water, and (2) the number of hydrogen bonds and ionic interactions within the protein is maximized, thus reducing the number of unpaired hydrogenbonding and ionic groups. Proteins within membranes (which we examine in Chapter 11) and proteins that are intrinsically disordered or have intrinsically disordered segments follow different rules. This reflects their particular function or environment, but weak interactions are still critical structural elements.

Covalent bonds, too, place important constraints on the conformation of a polypeptide. In the late 1930s, Linus Pauling and Robert Corey embarked on a series of studies that laid the foundation for our current understanding of protein structure. They began with a careful analysis of the peptide bond.

The α carbons of adjacent amino acid residues are separated by three covalent bonds, arranged as Cα—C—N—Cα. X-ray diffraction studies of crystals of amino acids and of simple dipeptides and tripeptides showed that the peptide C—N bond is somewhat shorter than the C—N bond in a simple amine and that the atoms associated with the peptide bond are coplanar. This indicated a resonance or partial sharing of two pairs of electrons between the carbonyl oxygen and the amide nitrogen (Fig. 4-2a).

The oxygen has a partial negative charge and the hydrogen bonded to the nitrogen has a net partial positive charge, setting up a small electric dipole. The six atoms of the peptide group lie in a single plane, with the oxygen atom of the carbonyl group trans to the hydrogen atom of the amide nitrogen. From these findings Pauling and Corey concluded that the peptide C—N bonds, because of their partial double-bond character, cannot rotate freely.

The backbone of a polypeptide chain can thus be pictured as a series of rigid planes, with consecutive planes sharing a common point of rotation at Cα (Fig. 4-2b). The rigid peptide bonds limit the range of conformations possible for a polypeptide chain.

Peptide conformation is defined by three dihedral angles (also known as torsion angles) called ϕ (phi), ψ (psi), and ω (omega), reflecting rotation about each of the three repeating bonds in the peptide backbone. A dihedral angle is the angle at the intersection of two planes. In the case of peptides, the planes are defined by bond vectors in the peptide backbone. Two successive bond vectors describe a plane. Three successive bond vectors describe two planes (the central bond vector is common to both; Fig. 4-2c), and the angle between these two planes is what we measure to describe peptide conformation.

The important dihedral angles in a peptide are defined by the three bond vectors connecting four consecutive main-chain (peptide backbone) atoms (Fig. 4-2c): ϕ involves the C—N—Cα—C bonds (with the rotation occurring about the N—Cα bond), and ψ involves the N—Cα—C—N bonds. Both ϕ and ψ are defined as ±180° when the polypeptide is fully extended and all peptide groups are in the same plane (Fig. 4-2d).

As one looks down the central bond vector in the direction of the vector arrow (as depicted in Fig. 4-2c for ψ), the dihedral angles increase as the distal (fourth) atom is rotated clockwise (Fig. 4-2d). From the ±180° position, the dihedral angle increases from −180° to 0°, at which point the first and fourth atoms are eclipsed.

It involves the Cα—C—N—Cα bonds. The central bond in this case is the peptide bond, where rotation is constrained. The peptide bond is almost always (99.6% of the time) in the trans configuration, constraining ω to a value of ±180°. For a rare cis peptide bond, ω = 0°.

In principle, ϕ and ψ can have any value between −180° and +180°, but many values are prohibited by steric interference between atoms in the polypeptide backbone and amino acid side chains. The conformation in which both ϕ and ψ are 0° (Fig. 42d) is prohibited for this reason; this conformation is merely a reference point for describing the dihedral angles. Backbone angle preferences in a polypeptide represent yet another constraint on the overall folded structure of a protein.

Protein segments can adopt regular secondary structures such as the α helix and the β conformation. These structures are defined by particular values of ϕ and ψ and their formation is impacted by the amino acid composition of the segment. All of the ϕ and ψ values for a given protein structure can be visualized using a Ramachandran plot.

A typical protein usually has one or more stable threedimensional conformations that reflect its function. Some proteins have segments that are intrinsically disordered but are nonetheless essential for function. Whereas nonpeptide covalent bonds, particularly disulfide bonds, can play a role in stabilization of some structures, proteins are stabilized largely by multiple weak, noncovalent interactions and forces.

The hydrophobic effect, derived from the increase in entropy of the surrounding water when nonpolar molecules or groups are clustered together, makes the major contribution to stabilizing the globular form of most soluble proteins. Hydrogen bonds and ionic interactions are optimized in the thermodynamically most stable structures. Van der Waals interactions involve attractive forces between molecular dipoles that occur over short distances. Individually these interactions are weak, but they combine in well-packed protein structures to provide significant effects and stabilization.

The nature of the covalent bonds in the polypeptide backbone places constraints on structure. The peptide bond has a partial double-bond character that keeps the entire six-atom peptide group in a rigid planar configuration. The N—Cα and Cα—C bonds can rotate to define the dihedral angles ϕ and ψ, respectively, although permitted values of ϕ and ψ are limited by steric clashes and other constraints.

The term secondary structure refers to any chosen segment of a polypeptide chain and describes the local spatial arrangement of its main-chain atoms, without regard to the positioning of its side chains or its relationship to other segments. A regular secondary structure occurs when each dihedral angle, ϕ and ψ, remains the same or nearly the same throughout the segment.

The most prominent are the α helix and β conformation; another common type is the β turn. Secondary structures without a regular pattern are sometimes referred to as undefined or as random coils. Random coil, however, does not properly describe the structure of these segments.

The path of most of the polypeptide backbone in a typical protein is not random; rather, it is highly specific to the structure and function of that particular protein. Our discussion here focuses on the regular structures that are most common.

Pauling and Corey were aware of the importance of hydrogen bonds in orienting polar chemical groups such as the C=O and N—H groups of the peptide bond. They also had the experimental results of William Astbury, who in the 1930s had conducted pioneering x-ray studies of proteins.

Although not an SI unit, it is used universally by structural biologists to describe atomic distances —it is approximately the length of a typical C—H bond.) With this information and their data on the peptide bond, and with the help of precisely constructed models, Pauling and Corey set out to determine the likely conformations of protein molecules.

The first breakthrough came in 1948. Pauling, at that time a visiting lecturer at Oxford University, became ill and retired to his apartment for several days of rest. Bored with the reading available, Pauling grabbed some paper and pencils to work out a plausible stable structure that could be taken up by a polypeptide chain.

The model he developed, and later confirmed in work with Corey and coworker Herman Branson, was the simplest arrangement the polypeptide chain can assume that maximizes the use of internal hydrogen bonding. It is a helical structure, and Pauling and Corey called it the α helix (Fig. 4-3).

In this structure, the polypeptide backbone is tightly wound around an imaginary axis drawn longitudinally through the middle of the helix, and the R groups of the amino acid residues protrude outward from the helical backbone (Fig. 4-3b, c). The repeating unit is a single turn of the helix, which extends about 5.4 Å along the long axis, slightly greater than the periodicity that Astbury observed on x-ray analysis of hair keratin.

Pauling and Corey considered both right-handed and left-handed variants of the α helix. The subsequent elucidation of the threedimensional structure of myoglobin and other proteins showed that the right-handed α helix is the common form (Box 4-1).

have not been observed in proteins. The α helix proved to be the predominant structure in α-keratins. More generally, about onefourth of all amino acid residues in proteins are found in α helices, the exact fraction varying greatly from one protein to another.

Make fists of your two hands with thumbs outstretched and pointing away from you. Looking at your right hand, think of a helix spiraling up your right thumb in the direction in which the other four fingers are curled as shown (clockwise). The resulting helix is right-handed.

Why does the α helix form more readily than many other possible conformations? The answer lies, in part, in its optimal use of intrahelical hydrogen bonds. The structure is stabilized by a hydrogen bond between the hydrogen atom attached to the electronegative nitrogen atom of a peptide linkage and the electronegative carbonyl oxygen atom of the fourth amino acid on the amino-terminal side of that peptide bond (Fig. 4-3a).

Within the α helix, every peptide bond (except those close to each end of the helix) participates in such hydrogen bonding. Each successive turn of the α helix is held to adjacent turns by three to four hydrogen bonds, conferring significant stability on the overall

At the ends of an α-helical segment, there are always three or four amide carbonyl or amino groups that cannot participate in this helical pattern of hydrogen bonding. These may be exposed to the surrounding solvent, where they hydrogenbond with water, or other parts of the protein may cap the helix to provide the needed hydrogen-bonding partners.

A final factor affecting the stability of an α helix is the identity of the amino acid residues near the ends of the α-helical segment of the polypeptide. A small electric dipole exists in each peptide bond (Fig. 4-2a). These dipoles are aligned through the hydrogen bonds of the helix, resulting in a net dipole along the helical axis that increases with helix length (Fig. 4-4). The partial positive and negative charges of the helix dipole reside on the peptide amino and carbonyl groups near the amino-terminal and carboxylterminal ends, respectively.

For this reason, negatively charged amino acids are often found near the amino terminus of the helical segment, where they have a stabilizing interaction with the positive charge of the helix dipole; a positively charged amino acid at the amino-terminal end is destabilizing. The opposite is true at the carboxyl-terminal end of the helical segment.

What is the length, in both Å and nm, of a polypeptide with 80 amino acid residues in a single, continuous α helix? ## SOLUTION: An idealized α helix has 3.6 residues per turn, and the rise along the helical axis is 5.4 Å. Thus, the rise along the axis for each amino acid residue is 1.5 Å.

helix, reflecting the properties of the R group and how they affect the capacity of the adjoining main-chain atoms to take up the characteristic ϕ and ψ angles. Alanine shows the greatest tendency to form α helices in most experimental model systems.

The position of an amino acid residue relative to its neighbors is also important. Interactions between amino acid side chains can stabilize or destabilize the α-helical structure. For example, if a polypeptide chain has a long block of Glu residues, this segment of the chain will not form an α helix at pH 7.0. The negatively charged carboxyl groups of adjacent Glu residues repel each other so strongly that they prevent formation of the α helix.

For the same reason, if there are many adjacent Lys and/or Arg residues, with positively charged R groups at pH 7.0, they also repel each other and prevent formation of the α helix. The size and shape of Asn, Ser, Thr, and Cys residues can also destabilize an α helix if they are close together in the chain.

The twist of an α helix ensures that critical interactions occur between an amino acid side chain and the side chain three (and sometimes four) residues away on either side of it. This is made clear when the α helix is depicted as a helical wheel (Fig. 4-3d).

A constraint on the formation of the α helix is the presence of Pro or Gly residues, which have the least likelihood of forming α helices. In proline, the nitrogen atom is part of a rigid ring (see Fig. 4-7), and rotation about the N—Cα bond is not possible.

Thus, a Pro residue introduces a destabilizing kink in an α helix. In addition, the nitrogen atom of a Pro residue in a peptide linkage has no substituent hydrogen to participate in hydrogen bonds with other residues. For these reasons, proline is found only rarely in an α helix. Glycine occurs infrequently in α helices for a different reason: it has more conformational flexibility than the other amino acid residues. Polymers of glycine tend to take up coiled structures quite different from an α helix.

In summary, five types of constraints affect the stability of an α helix: (1) the intrinsic propensity of an amino acid residue to form an α helix; (2) the interactions between R groups, particularly those spaced three (or four) residues apart; (3) the bulkiness of adjacent R groups; (4) the occurrence of Pro and Gly residues; and (5) interactions between amino acid residues at the ends of the helical segment and the electric dipole inherent to the α helix.

In 1951, Pauling and Corey predicted a second type of repetitive structure, the β conformation. This is a more extended conformation of polypeptide chains, and its structure is again defined by backbone atoms arranged according to a characteristic set of dihedral angles (Table 4-1). In the β conformation, the backbone of the polypeptide chain is extended into a zigzag rather than helical structure (Fig. 4-5).

A single protein segment in the β conformation is often called a β strand.

The arrangement of several strands side by side, all in the β conformation, is called a β sheet. The zigzag structure of the individual polypeptide segments gives rise to a pleated appearance of the overall sheet. Hydrogen bonds form between backbone atoms of adjacent segments of polypeptide chain within the sheet.

The individual segments that form a β sheet are usually nearby on the polypeptide chain but can also be quite distant from each other in the linear sequence of the polypeptide; they may even be in different polypeptide chains. The R groups of adjacent amino acids protrude from the zigzag structure in opposite directions, creating the alternating pattern seen in the side view in Figure 4-5.

The adjacent polypeptide chains in a β sheet can be either parallel or antiparallel (having the same or opposite amino-tocarboxyl orientations, respectively). The structures are somewhat similar, although the repeat period is shorter for the parallel conformation (6.5 vs. 7.0 Å for antiparallel) and the hydrogenbonding patterns are different.

The interstrand hydrogen bonds are essentially in-line (see Fig. 2-5) in the antiparallel β sheet, whereas they are distorted or not in-line for the parallel variant. In natural proteins, antiparallel β sheets are found twice as frequently as parallel β sheets. The idealized structures exhibit the bond angles given in Table 4-1; these values vary somewhat in real proteins, resulting in structural variation, as seen above for α helices.

In globular proteins, which have a compact folded structure, some amino acid residues are in turns or loops where the polypeptide chain reverses direction (Fig. 4-6). These are the connecting elements that link successive runs of α helix or β conformation.

Particularly common are β turns that connect the ends of two adjacent segments of an antiparallel β sheet. The structure is a 180° turn involving four amino acid residues, with the carbonyl oxygen of the first residue forming a hydrogen bond with the amino-group hydrogen of the fourth.

Several types of β turns have been described, each defined by the ϕ and ψ angles of the bonds that link the four amino acid residues that make up the particular turn (Table 4-1). Gly and Pro residues often occur in β turns, the former because it is small and flexible, the latter because peptide bonds involving the imino nitrogen of proline readily assume the cis configuration (Fig. 4-7), a form that is particularly amenable to a tight turn.

Beta turns are often found near the surface of a protein, where the peptide groups of the central two amino acid residues in the turn can hydrogen-bond with water. Considerably less common is the γ turn, a three-residue turn with a hydrogen bond between the first and third residues.

The α helix and the β conformation are the major repetitive secondary structures in a wide variety of proteins, although other repetitive structures exist in some specialized proteins (an example is collagen; see Fig. 4-12). Every type of secondary structure can be completely described by the dihedral angles ϕ and ψ associated with each residue.

In a Ramachandran plot, the dihedral angles that define the α helix and the β conformation fall within a relatively restricted range of sterically allowed structures (Fig. 4-8a). Most values of ϕ and ψ taken from known protein structures fall into the expected regions, with high concentrations near the α helix and β conformation values, as predicted (Fig. 4-8b).

The only amino acid residue often found in a conformation outside these regions is glycine. Because its side chain is small, a Gly residue can take part in many conformations that are sterically forbidden for other amino acids.

Any form of structural asymmetry in a molecule gives rise to differences in absorption of left-handed versus right-handed circularly polarized light.

Measurement of this difference is called circular dichroism (CD) spectroscopy. An ordered structure, such as a folded protein, gives rise to an absorption spectrum that can have peaks or regions with both positive and negative values. For proteins, spectra are obtained in the far UV region (190 to 250 nm).

the peptide bond; a signal is obtained when the peptide bond is in a folded environment. The difference in molar extinction coefficients (see Box 3-1) for left-handed and right-handed, circularly polarized light (Δε) is plotted as a function of wavelength.

The α helix and β conformations have characteristic CD spectra (Fig. 4-9). Using CD spectra, biochemists can determine whether proteins are properly folded, estimate the fraction of the protein that is folded in either of the common secondary structures, and monitor transitions between the folded and unfolded states.

Secondary structure is the local spatial arrangement of the main-chain atoms in a selected segment of a polypeptide chain; it can be completely defined by the ϕ and ψ angles of all the amino acids in that segment. In the α helix, the repeating unit is a single helical turn of ∼5.4 Å or 3.6 amino acids.

The common form found in proteins is right-handed with the amino acid R groups protruding away from the helical backbone. The propensity of a protein segment to form an α helix depends on the composition of its amino acids and amino acid positions relative to one another and relative to the helical dipole.

In the β conformation, amino acids are extended in a zigzag fashion. When several β strands are arranged adjacent to one another, they can form either parallel or antiparallel β sheets. Turns or loops connect segments of α helix or β strands.

β turns, which often contain Gly or Pro residues, tend to connect segments of antiparallel β sheets.

The Ramachandran plot is a visual description of the combinations of ϕ and ψ dihedral angles that are permitted in a peptide backbone and those that are not permitted due to steric constraints. Dihedral angles that define the α helix and the β conformation are found only within certain regions of the plot.

Circular dichroism spectroscopy is a method for assessing common secondary structure and monitoring folding in proteins based on absorption of circularly polarized UV light.

The overall three-dimensional arrangement of all atoms in a protein is referred to as the protein’s tertiary structure. Whereas the term “secondary structure” refers to the spatial arrangement of amino acid residues that are adjacent in a segment of a polypeptide, tertiary structure includes longer-range aspects of amino acid sequence. Amino acids that are far apart in the polypeptide sequence and are in different types of secondary structure may interact within the completely folded structure of a protein.

Interacting segments of polypeptide chains are held in their characteristic tertiary positions by several kinds of weak interactions (and sometimes by covalent bonds such as disulfide cross-links) between the segments. Some proteins contain two or more separate polypeptide chains, or subunits, which may be identical or different. The arrangement of these protein subunits in three-dimensional complexes constitutes quaternary structure.

In considering these higher levels of structure, it is useful to designate the major groups into which many proteins can be classified: fibrous proteins, with polypeptide chains arranged in long strands or sheets; globular proteins, with polypeptide chains folded into a spherical or globular shape; membrane proteins, with polypeptide chains embedded in hydrophobic lipid membranes; and intrinsically disordered proteins, with polypeptide chains lacking stable tertiary structures.

The groups also differ functionally: the structures that provide support, shape, and external protection to vertebrates are made of fibrous proteins. Most enzymes are globular proteins, whereas regulatory proteins can be globular, disordered, or contain both globular and disordered segments. α-Keratin, collagen, and silk fibroin nicely illustrate the relationship between protein structure and biological function (Table 4-2).

Fibrous proteins share properties that give strength and/or flexibility to the structures in which they occur. In each case, the fundamental structural unit is a simple repeating element of secondary structure. All fibrous proteins are insoluble in water, a property conferred by a high concentration of hydrophobic amino acid residues both in the interior of the protein and on its surface. These hydrophobic surfaces are largely buried, as many similar polypeptide chains are packed together to form elaborate supramolecular complexes.

The α-keratins have evolved for strength. Found only in mammals, these proteins constitute almost the entire dry weight of hair, wool, nails, claws, quills, horns, and hooves and much of the outer layer of skin. The α-keratins are part of a broader family of proteins called intermediate filament (IF) proteins. Other IF proteins are found in the cytoskeletons of animal cells. All IF proteins have a structural function and share the structural features exemplified by the α-keratins.

Two strands of α-keratin, oriented in parallel (with their amino termini at the same end), are wrapped about each other to form a supertwisted coiled coil. The supertwisting amplifies the strength of the overall structure, just as strands are twisted to make a strong rope (Fig. 4-10).

The twisting of the axis of an α helix to form a coiled coil explains the discrepancy between the 5.4 Å per turn predicted for an α helix by Pauling and Corey and the 5.15 to 5.2 Å repeating structure observed in the x-ray diffraction of hair (see end-of-chapter problem 2).

The helical path of the supertwists is left-handed, opposite in sense to the α helix. The surfaces where the two α helices touch are made up of hydrophobic amino acid residues, their R groups meshed together in a regular interlocking pattern. This permits a close packing of the polypeptide chains within the left-handed supertwist. Not surprisingly, α-keratin is rich in the hydrophobic residues Ala, Val, Leu, Ile, Met, and Phe.

These then combine in higher-order structures called protofilaments and protofibrils. About four protofibrils—32 strands of α-keratin in all—combine to form an intermediate filament. The individual two-chain coiled coils in the various substructures also seem to be interwound, but the handedness of the interwinding and other structural details are unknown. (b) A hair is an array of many α-keratin filaments, made up of the substructures shown in (a).

The intertwining of the two α-helical polypeptides is an example of quaternary structure. Coiled coils of this type are common structural elements in filamentous proteins and in the muscle protein myosin (see Fig. 5-26). The quaternary structure of α-keratin can be quite complex. Many coiled coils can be assembled into large supramolecular complexes, such as the arrangement of α-keratin that forms the intermediate filament of hair (Fig. 4-10b).

The strength of fibrous proteins is enhanced by covalent crosslinks between polypeptide chains in the multihelical “ropes” and between adjacent chains in a supramolecular assembly. In αkeratins, the cross-links stabilizing quaternary structure are disulfide bonds. In the hardest and toughest α-keratins, such as those of rhinoceros horn, up to 18% of the residues are cysteines involved in disulfide bonds.

Like the α-keratins, collagen has evolved to provide strength. It is found in connective tissue such as tendons, cartilage, the organic matrix of bone, and the cornea of the eye. In fact, collagen is the most abundant protein in mammals, usually comprising 25% to 35% of total protein content. The collagen helix is a unique secondary structure, quite distinct from the α helix.

It is left-handed and has three amino acid residues per turn (Fig. 4-11 and Table 4-1). Collagen is also a coiled coil, but one with distinct tertiary and quaternary structures: three separate polypeptides, called α chains (not to be confused with α helices), are twisted about each other. The superhelical twisting is right-handed in collagen, opposite in sense to the lefthanded helix of the α chains.

There are many types of vertebrate collagen. Typically, they contain about 35% Gly, 11% Ala, and 21% Pro and 4-Hyp (4hydroxyproline, an uncommon amino acid; see Fig. 3-8a).

The food product gelatin is derived from collagen. It has little nutritional value as a protein, because collagen is extremely low in many amino acids that are essential in the human diet. The unusual amino acid content of collagen is related to structural constraints unique to the collagen helix.

The amino acid sequence in collagen is generally a repeating tripeptide unit, Gly–X–Y, where X is often Pro and Y is often 4-Hyp.

Only Gly residues can be accommodated at the very tight junctions between the individual α chains (Fig. 4-11b). The Pro and 4-Hyp residues permit the sharp twisting of the collagen helix. The amino acid sequence and the supertwisted quaternary structure of collagen allow a very close packing of its three polypeptides.

The three-stranded collagen superhelix shown from one end, in a ball-and-stick representation. Gly residues are shown in red. Glycine, because of its small size, is required at the tight junction where the three chains are in contact. The balls in this illustration do not represent the van der Waals radii of the individual atoms. The center of the three-stranded superhelix is not hollow, as it appears here, but very tightly packed.

Scurvy is caused by lack of vitamin C, or ascorbic acid (ascorbate). Vitamin C is required for, among other things, the hydroxylation of proline and lysine in collagen; scurvy is a deficiency disease characterized by general degeneration of connective tissue. Manifestations of advanced scurvy include numerous small hemorrhages caused by fragile blood vessels; tooth loss, poor wound healing, and the reopening of old wounds; bone pain and degeneration; and eventually heart failure.

Most animals make large amounts of vitamin C, converting glucose to ascorbate in four enzymatic steps. But in the course of evolution, humans and some other animals—gorillas, guinea pigs, and fruit bats—have lost the last enzyme in this pathway and must obtain ascorbate in their diet. Vitamin C is available in a wide range of fruits and vegetables.

In 1747, James Lind, a Scottish surgeon in the Royal Navy, carried out the first controlled clinical study in recorded history. During an extended voyage on the 50-gun warship HMS Salisbury, Lind selected 12 sailors suffering from scurvy and separated them into groups of two. All 12 received the same diet, except that each group was given a different remedy for scurvy from among those recommended at the time. The sailors given lemons and oranges recovered and returned to duty.

The proline ring is normally found as a mixture of two puckered conformations, called Cγ-endo and Cγ-exo (Fig. 1). The collagen helix structure requires the Pro/4-Hyp residue in the Y positions to be in the Cγ-exo conformation, and it is this conformation that is enforced by the hydroxyl substitution at C-4 in 4-Hyp. In the absence of vitamin C, cells cannot hydroxylate the Pro at the Y positions. This leads to collagen instability and the connective tissue problems seen in scurvy.

No ascorbate is needed in this reaction. However, prolyl 4-hydroxylase also catalyzes an oxidative decarboxylation of α-ketoglutarate that is not coupled to proline hydroxylation. During this reaction, the Fe2+ becomes oxidized, inactivating the enzyme and preventing the proline hydroxylation. Ascorbate is needed to reduce the iron and restore enzyme activity so that proline hydroxylation of procollagen can continue.

Scurvy remains a problem today, not only in remote regions where nutritious food is scarce but, surprisingly, also among young adults with poor eating habits in large cities. A 2009 study of more than 1,100 men and women between the ages of 20 and 29 in Toronto, Canada, found that 1 in 7 young adults had vitamin C deficiency due to unmet dietary needs.

The tight wrapping of the α chains in the collagen triple helix provides tensile strength greater than that of a steel wire of equal cross section. Collagen fibrils (Fig. 4-12) are supramolecular assemblies consisting of triple-helical collagen molecules (sometimes referred to as tropocollagen molecules) associated in a variety of ways to provide different degrees of tensile strength. The α chains of collagen molecules and the collagen molecules of fibrils are cross-linked by unusual types of covalent bonds involving Lys, HyLys (5-hydroxylysine), or His residues that are present at a few of the X and Y positions.

Its three helically intertwined α chains may have different sequences; each chain has about 1,000 amino acid residues. Collagen fibrils are made up of collagen molecules aligned in a staggered fashion and cross-linked for strength. The specific alignment and degree of cross-linking vary with the tissue and produce characteristic cross-striations in an electron micrograph.

A typical mammal has more than 30 structural variants of collagen, particular to certain tissues and each somewhat different in sequence and function. Some human genetic defects in collagen structure illustrate the close relationship between amino acid sequence and three-dimensional structure in this protein. Osteogenesis imperfecta is characterized by abnormal bone formation in babies; at least eight variants of this condition, with different degrees of severity, occur in the human population. Ehlers-Danlos syndrome is characterized by loose joints, and at least six variants occur in humans.

These single-residue substitutions have a catastrophic effect on collagen function because they disrupt the Gly–X–Y repeat that gives collagen its unique helical structure. Given its role in the collagen triple helix (Fig. 4-11), Gly cannot be replaced by another amino acid residue without substantial deleterious effects on collagen structure.

The protein of silk, fibroin, is produced by insects and spiders. Its polypeptide chains are predominantly in the β conformation. Fibroin is rich in Ala and Gly residues, permitting a close packing of β sheets and an interlocking arrangement of R groups (Fig. 4-13). The overall structure is stabilized by extensive hydrogen bonding between all peptide linkages in the polypeptides of each β sheet and by the optimization of van der Waals interactions between sheets.

Silk does not stretch, because the β conformation is already highly extended (Fig. 4-5). However, the structure is flexible, because the sheets are held together by numerous weak interactions rather than by covalent bonds such as the disulfide bonds in α-keratins.

generating a more compact shape than is seen in the fibrous proteins (Fig. 4-14). The folding also provides the structural diversity necessary for proteins to carry out a wide array of biological functions. Globular proteins include enzymes, transport proteins, motor proteins, regulatory proteins, immunoglobulins, and proteins with many other functions.

Our discussion of globular proteins begins with the principles gleaned from the first protein structures to be elucidated. This is followed by a detailed description of protein substructure and comparative categorization. Such discussions are possible only because of the vast amount of information available online from publicly accessible databases, particularly the Protein Data Bank, or PDB (Box 4-3).

The number of known three-dimensional protein structures is now more than 100,000 and doubles every couple of years. This wealth of information is revolutionizing our understanding of protein structure, the relationship of structure to function, and the evolutionary paths by which proteins arrived at their present state,

The PDB is an archive of experimentally determined three-dimensional structures of biological macromolecules, containing virtually all of the macromolecular structures (such as proteins, RNAs, and DNAs) elucidated to date. Each structure is assigned an identifying label (a four-character identifier called the PDB ID).

The data files in the PDB describe the spatial coordinates of each atom for which the position has been determined (many of the cataloged structures are not complete). Additional data files provide information on how the structure was determined and its accuracy.

Myoglobin is particularly abundant in the muscles of diving mammals such as whales, seals, and porpoises — so abundant that the muscles of these animals are brown. Storage and distribution of oxygen by muscle myoglobin permits diving mammals to remain submerged for long periods. The activities of myoglobin and other globin molecules are investigated in greater detail in Chapter 5.

Figure 4-15 shows several structural representations of myoglobin, illustrating how the polypeptide chain is folded in three dimensions — its tertiary structure. The red group surrounded by protein is heme. The backbone of the myoglobin molecule consists of eight relatively straight segments of α helix interrupted by bends, some of which are β turns. The longest α helix has 23 amino acid residues and the shortest has only 7; all helices are right-handed.

Many important conclusions were drawn from the structure of myoglobin. The positioning of amino acid side chains reflects a structure that is largely stabilized by the hydrophobic effect. Most of the hydrophobic R groups are in the interior of the molecule, hidden from exposure to water. All but two of the polar R groups are located on the outer surface of the molecule, and all are hydrated. The myoglobin molecule is so compact that its interior has room for only four molecules of water.

The fraction of space occupied by atoms in an organic liquid is 0.4 to 0.6. In a globular protein the fraction is about 0.75, comparable to that in a crystal (in a typical crystal the fraction is 0.70 to 0.78, near the theoretical maximum). In this packed environment, weak interactions strengthen and reinforce each other. For example, the nonpolar side chains in the core are so close together that short-range van der Waals interactions make a significant contribution to stabilizing interactions.

Deduction of the structure of myoglobin confirmed some expectations and introduced some new elements of secondary structure. As predicted by Pauling and Corey, all the peptide bonds are in the planar trans configuration. The α helices in myoglobin provided the first direct experimental evidence for the existence of this type of secondary structure. Three of the four Pro residues are found at bends. The fourth Pro residue occurs within an α helix, where it creates a kink necessary for tight helix packing.

The flat heme group rests in a crevice, or pocket, in the myoglobin molecule. Within this pocket, the accessibility of the heme group to solvent is highly restricted. This is important for function, because free heme groups in an oxygenated solution are rapidly oxidized from the ferrous (Fe2+) form, which is active in the reversible binding of O2, to the ferric (Fe3+) form, which does not bind O2.

Myoglobin illustrates just one of many ways in which a polypeptide chain can fold. Table 4-3 shows the proportions of α helix and β conformation (expressed as percentage of residues in each type) in several small, single-chain, globular proteins. Each of these proteins has a distinct structure, adapted for its particular biological function, but together they share several important properties with myoglobin.

Each is folded compactly, and in each case the hydrophobic amino acid side chains are oriented toward the interior (away from water) and the hydrophilic side chains are on the surface. The structures are also stabilized by a multitude of hydrogen bonds and some ionic interactions.

To understand a complete three-dimensional structure, we need to analyze its folding patterns. We begin by defining two important terms that describe protein structural patterns or elements in a polypeptide chain; then we turn to the folding rules. The first term is motif, also called a fold.

A motif or fold is a recognizable folding pattern involving two or more elements of secondary structure and the connection(s) between them. A motif can be very simple, such as two elements of secondary structure folded against each other, and may represent only a small part of a protein. An example is a β-α-β loop (Fig. 4-16a). A motif can also be a very elaborate structure involving scores of protein segments folded together, such as the β barrel (Fig. 4-16b). In some cases, a single large motif may comprise the entire protein.

We have already encountered a well-studied motif, the coiled coil of α-keratin, which is also found in some other proteins. The distinctive arrangement of eight α helices in myoglobin is replicated in all globins and is called the globin fold. Note that a motif is not a hierarchical structural element falling between secondary and tertiary structure. It is simply a folding pattern.

The second term for describing structural patterns is domain. A domain, as defined by Jane Richardson in 1981, is a part of a polypeptide chain that is independently stable or could undergo movements as a single entity with respect to the entire protein.

In many cases, a domain from a large protein will retain its native three-dimensional structure even when separated (for example, by proteolytic cleavage) from the remainder of the polypeptide chain. In a protein with multiple domains, each domain may appear as a distinct globular lobe (Fig. 4-17); more commonly, extensive contacts between domains make individual domains hard to discern.

1. The hydrophobic effect makes a large contribution to the stability of protein structures. Burial of hydrophobic amino acid R groups so as to exclude water requires at least two layers of secondary structure. Simple motifs such as the βα-β loop (Fig. 4-16a) create two such layers. 2. Where they occur together in a protein, α helices and β sheets generally are found in different structural layers. This is because the backbone of a polypeptide segment in the β conformation (Fig. 4-5) cannot readily hydrogen-bond to an α helix that is adjacent to it.

3. Segments adjacent to each other in the amino acid sequence are usually stacked adjacent to each other in the folded structure. Distant segments of a polypeptide may come together in the tertiary structure, but this is not the norm. 4. The β conformation is most stable when the individual segments are twisted slightly in a right-handed sense. This influences both the arrangement of β sheets derived from the twisted segments and the path of the polypeptide connections between them.

Two parallel β strands, for example, must be connected by a crossover strand (Fig. 418a). In principle, this crossover could have a right-handed or left-handed conformation, but in proteins it is almost always right-handed. Right-handed connections tend to be shorter than left-handed connections and tend to bend through smaller angles, making them easier to form.

The twisting of β sheets also leads to a characteristic twisting of the structure formed by many such segments together, as seen in the β barrel (Fig. 4-16b) and the twisted β sheet (Fig. 4-18c), which form the core of many larger structures.

Following these rules, complex motifs can be built up from simple ones. For example, a series of β-α-β loops arranged so that the β strands form a barrel creates a particularly stable and common motif, the α/β barrel (Fig. 4-19).

Although many proteins contain well-folded and stable structures, this is not necessary for the biological function of all proteins. Many proteins or protein segments lack ordered structures in solution. The concept that some proteins function in the absence of a definable three-dimensional structure comes from reassessment of data from many different proteins. As many as a third of all human proteins may be unstructured or may have significant unstructured segments. All organisms have some proteins that fall into this category. Intrinsically disordered proteins have properties that are distinct from those of classical, structured proteins.

They often lack a hydrophobic core and instead are characterized by high densities of charged amino acid residues such as Lys, Arg, and Glu. Pro residues are also prominent, as they tend to disrupt ordered structures.

Structural disorder and high charge density can facilitate the function of some proteins as spacers, insulators, or linkers in larger structures. Other disordered proteins are scavengers, binding up ions and small molecules in solution and serving as reservoirs or garbage dumps. However, many intrinsically disordered proteins are at the heart of important protein interaction networks. The lack of an ordered structure can facilitate a kind of functional promiscuity, allowing one protein to interact with multiple or even dozens of partners.

In solution, p27 lacks definable structure. However, it wraps around and inhibits the action of several enzymes called protein kinases (see Chapter 6) that facilitate cell division. The flexible structure of p27 allows it to accommodate itself to its different target proteins. Human tumor cells, which are cells that have lost the capacity to control cell division normally, generally have reduced levels of p27; the lower the levels of p27, the poorer the prognosis for the cancer patient.

The mammalian protein p53 is also critical in the control of cell division. It contains both structured and unstructured segments, and the different segments interact with dozens of other proteins. An unstructured region of p53 at the carboxyl terminus interacts with at least four different binding partners and assumes a different structure in each of the complexes (Fig. 4-20).

More than 150,000 structures are now archived in the Protein Data Bank (PDB; for a deeper explanation, see Box 4-3). An enormous amount of information about protein structural principles, protein function, and protein evolution is contained in these data. Other databases have organized this information and made it more readily accessible. In the Structural Classification of Proteins database, or SCOP2 (http://scop2.mrc-lmb.cam.ac.uk), all of the protein information in the PDB can be searched within four different categories: (1) protein relationships, (2) structural classes, (3) protein types, and (4) evolutionary events.

Figure 4-21 presents examples of protein motifs taken from SCOP2 to illustrate the potential of searching within each category. The figure also introduces another way to represent elements of secondary structure and the relationships among segments of secondary structure in a protein — the topology diagram.

The number of folding patterns is not infinite. Among the tens of thousands of distinct protein structures archived in the PDB, only about 1,400 different folds or motifs are classified by the SCOP2 database. Given the many years of progress in structural biology, new motifs are now discovered only rarely. Many examples of recurring domain or motif structures are available, and these reveal that protein tertiary structure is more reliably conserved than amino acid sequence. The comparison of protein structures can thus provide much information about evolution.

Proteins with significant similarity in primary structure and/or with similar tertiary structure and function are said to be in the same protein family. The protein structures in the PDB belong to about 4,000 different protein families. A strong evolutionary relationship is usually evident within a protein family. For example, the globin family has many different proteins with both structural and sequence similarities to myoglobin (as seen in the proteins used as examples in Figures 4-30 and 4-31 and in Chapter 5).

Two or more families that have little similarity in amino acid sequence but make use of the same major structural motif and have functional similarities are grouped into superfamilies. An evolutionary relationship among families in a superfamily is considered probable, even though time and functional distinctions — that is, different adaptive pressures — may have erased many of the telltale sequence relationships.

A protein family may be widespread in all three domains of cellular life — the Bacteria, Archaea, and Eukarya — suggesting an ancient origin. Many proteins involved in intermediary metabolism and the metabolism of nucleic acids and proteins fall into this category. Other families may be present in only a small group of organisms, indicating that the structure arose more recently. Tracing the natural history of structural motifs through the use of structural classifications in databases such as SCOP2 provides a powerful complement to sequence analyses in tracing evolutionary relationships. The SCOP2 database is curated manually, with the objective of placing proteins in the correct evolutionary framework based on conserved structural features.

Structural motifs become especially important in defining protein families and superfamilies. Improved protein classification and comparison systems lead inevitably to the elucidation of new functional relationships. Given the central role of proteins in living systems, these structural comparisons can help illuminate every aspect of biochemistry, from the evolution of individual proteins to the evolutionary history of complete metabolic pathways.

Many proteins have multiple polypeptide subunits (from two to hundreds). The association of polypeptide chains can serve a variety of functions. Many multisubunit proteins have regulatory roles; the binding of small molecules may affect the interaction between subunits, causing large changes in the protein’s activity in response to small changes in the concentration of substrate or regulatory molecules (Chapter 6).

In other cases, separate subunits take on separate but related functions, such as catalysis and regulation. Some associations, such as those seen in the fibrous proteins considered earlier in this chapter and the coat proteins of viruses, serve primarily structural roles. Some very large protein assemblies are the site of complex, multistep reactions. For example, each ribosome, the site of protein synthesis, incorporates dozens of protein subunits along with RNA molecules.

A multisubunit protein can also be referred to as an oligomer or multimer. If an oligomer has nonidentical subunits, the overall structure of the protein can be asymmetric and quite complicated. However, many oligomers have identical subunits or repeating groups of nonidentical subunits, usually in symmetric arrangements. As noted in Chapter 3, the repeating structural unit in such an oligomeric protein, whether a single subunit or a group of subunits, is called a protomer.

The protein portion, the globin, consists of two α chains (141 residues each) and two β chains (146 residues each). Note that in this case, α and β do not refer to secondary structures. In a practice that can be confusing to the beginning student, the Greek letters α and β (and γ, δ, and others) are often used to distinguish two different kinds of subunits within a multisubunit protein, regardless of what kinds of secondary structure may predominate in the subunits.

Because hemoglobin is four times as large as myoglobin, much more time and effort were required to solve its three-dimensional structure by x-ray analysis, finally achieved by Max Perutz, John Kendrew, and their colleagues in 1959. The subunits of hemoglobin are arranged in symmetric pairs (Fig. 4-22), each pair having one α subunit and one β subunit. Hemoglobin can therefore be described either as a tetramer or as a dimer of αβ protomers. The role these distinct subunits play in hemoglobin function is discussed extensively in Chapter 5.

Insoluble fibrous proteins, such as those that make up keratin, collagen, and silk, have simple repeating elements of secondary structure. In some fibrous proteins, the individual polypeptide chains interact to form complex quaternary structures like coiled coils for strength and flexibility.

The first globular protein structure to be determined, by x-ray diffraction methods, was that of the O2-binding protein myoglobin. The myoglobin structure revealed for the first time how protein structure and function are connected. The complex structures of globular proteins can be analyzed by examination of folding patterns, called motifs or folds. The many thousands of known protein structures are generally assembled from a repertoire of only a few hundred motifs. Domains are regions of a polypeptide chain that can fold stably and independently. Some proteins or protein segments are intrinsically disordered, lacking definable three-dimensional structure.

Cellular proteostasis requires the coordinated function of pathways for protein synthesis and folding, the refolding of proteins that are partially unfolded, and the sequestration and degradation of proteins that have been irreversibly unfolded or are no longer needed. In all cells, these networks involve hundreds of enzymes and specialized proteins.

Protein structures have evolved to function in particular cellular or extracellular environments. Conditions different from these environments can result in protein structural changes, large and small. A loss of three-dimensional structure sufficient to cause loss of function is called denaturation. The denatured state does not necessarily equate with complete unfolding of the protein and randomization of conformation. Under most conditions, denatured proteins exist in a set of partially folded states.

The abruptness of the change suggests that unfolding is a cooperative process: loss of structure in one part of the protein destabilizes other parts. The effects of heat on proteins can be mitigated by structure. The very heat-stable proteins of thermophilic bacteria and archaea have evolved to function at the temperature of hot springs (∼100 °C).

The folded structures of these proteins are often similar to those of proteins in other organisms, but take some of the principles outlined here to extremes. They often feature high densities of charged residues on their surfaces, even tighter hydrophobic packing in their interiors, and folds rendered less flexible by networks of ion pairs. Each of these features makes these proteins less susceptible to unfolding at high temperatures.

Results are shown for proteins denatured by two different environmental changes. In each case, the transition from the folded state to the unfolded state is abrupt, suggesting cooperativity in the unfolding process. (a) Thermal denaturation of horse apomyoglobin (myoglobin without the heme prosthetic group) and ribonuclease A (with its disulfide bonds intact; see Fig. 4-26). The midpoint of the temperature range over which denaturation occurs is called the melting temperature, or Tm.

Proteins can also be denatured by extremes of pH, by certain miscible organic solvents such as alcohol or acetone, by certain solutes such as urea and guanidine hydrochloride, or by detergents. Each of these denaturing agents represents a relatively mild treatment in the sense that no covalent bonds in the polypeptide chain are broken. Organic solvents, urea, and detergents act primarily by disrupting the hydrophobic aggregation of nonpolar amino acid side chains that produces the stable core of globular proteins; urea also disrupts hydrogen bonds; and extremes of pH alter the net charge on a protein, causing electrostatic repulsion and the disruption of some hydrogen bonding. The denatured structures resulting from these various treatments are not necessarily the same.

Denaturation often leads to protein precipitation, a consequence of protein aggregate formation as exposed hydrophobic surfaces associate. The aggregates are often highly disordered. One example is the protein precipitate that can be seen after an egg white is boiled. More-ordered aggregates are also observed in some proteins, as we shall see.

The tertiary structure of a globular protein is determined by its amino acid sequence. The most important proof of this came from experiments showing that denaturation of some proteins is reversible. Certain globular proteins denatured by heat, extremes of pH, or denaturing reagents will regain their native structure and their biological activity if they are returned to conditions in which the native conformation is stable. This process is called renaturation.

A classic example is the denaturation and renaturation of ribonuclease A, demonstrated by Christian Anfinsen in the 1950s. Purified ribonuclease A denatures completely in a concentrated urea solution in the presence of a reducing agent. The reducing agent cleaves the four disulfide bonds to yield eight Cys residues, and the urea disrupts the stabilizing hydrophobic effect, thus freeing the entire polypeptide from its folded conformation. Denaturation of ribonuclease is accompanied by a complete loss of catalytic activity.

When the urea and the reducing agent are removed, the denatured ribonuclease spontaneously refolds into its correct tertiary structure, with full restoration of its catalytic activity (Fig. 4-25). The refolding of ribonuclease is so accurate that the four intrachain disulfide bonds are re-formed in the same positions in the renatured molecule as in the native ribonuclease. Later, similar results were obtained using chemically synthesized, catalytically active ribonuclease A.

The Anfinsen experiment provided the first evidence that the amino acid sequence of a polypeptide chain contains all the information required to fold the chain into its native, threedimensional structure.

Subsequent work has shown that only a minority of proteins, many of them small and inherently stable, will fold spontaneously into their native form. Even though all proteins have the potential to fold into their native structure, many require some assistance.

The maintenance of the steady-state collection of active cellular proteins required under a particular set of conditions —called proteostasis — involves an elaborate set of pathways and processes that fold, refold, and degrade polypeptide chains. The three-dimensional structure and the function of most proteins can be destroyed by denaturation, demonstrating a relationship between structure and function. Heat, extremes of pH, organic solvents, solutes, and detergents can all be used to denature proteins. Some denatured proteins can renature spontaneously to form biologically active protein, showing that tertiary structure is determined by amino acid sequence.

In living cells, proteins are assembled from amino acids at a very high rate. For example, E. coli cells can make a complete, biologically active protein molecule containing 100 amino acid residues in about 5 seconds at 37 °C. However, the synthesis of peptide bonds on the ribosome is not enough; the protein must fold.

If each conformation were sampled in the shortest possible time (∼10−13 second, or the time required for a single molecular vibration), it would take about 1077 years to sample all possible conformations! Clearly, protein folding is not a completely random, trial-and-error process. There must be shortcuts. This problem was first pointed out by Cyrus Levinthal in 1968 and is sometimes called Levinthal’s paradox.

The folding pathway of a large polypeptide chain is unquestionably complicated. However, robust algorithms can often predict the structure of smaller proteins on the basis of their amino acid sequences. The major folding pathways are hierarchical. Local secondary structures form first. Certain amino acid sequences fold readily into α helices or β sheets, guided by constraints such as those reviewed in our discussion of secondary structure. Ionic interactions, involving charged groups that are often near one another in the linear sequence of the polypeptide chain, can play an important role in guiding these early folding steps.

Assembly of local structures is followed by longer-range interactions between, say, two elements of secondary structure that come together to form stable folded structures. The hydrophobic effect plays a significant role throughout the process, as the aggregation of nonpolar amino acid side chains provides an entropic stabilization to intermediates and, eventually, to the final folded structure.

Notably, proteins dominated by close-range interactions (between pairs of residues generally located near each other in the polypeptide sequence) tend to fold faster than proteins with more complex folding patterns and with many longrange interactions between different segments. As larger proteins with multiple domains are synthesized, domains near the amino terminus (which are synthesized first) may fold before the entire polypeptide has been assembled.

Thermodynamically, the folding process can be viewed as a kind of free-energy funnel (Fig. 4-27). The unfolded states are characterized by a high degree of conformational entropy and relatively high free energy. As folding proceeds, the narrowing of the funnel reflects the decrease in the conformational space that must be searched as the protein approaches its native state. Small depressions along the sides of the free-energy funnel represent semistable intermediates that can briefly slow the folding process. At the bottom of the funnel, an ensemble of folding intermediates has been reduced to a single native conformation (or one of a small set of native conformations). The funnels can have a variety of shapes, depending on the complexity of the folding pathway, the existence of semistable intermediates, and the potential for particular intermediates to assemble into aggregates of misfolded proteins.

(a) A simple but relatively wide and smooth funnel represents a protein that has multiple folding pathways (that is, the order in which different parts of the protein fold is somewhat random), but it assumes its three-dimensional structure with no folding intermediates that have significant stability. (b) This funnel represents a more typical protein that has multiple possible folding intermediates with significant stability on the multiple pathways leading to the native structure.

Thermodynamic stability is not evenly distributed over the structure of a protein — the molecule has regions of relatively high stability and others of low or negligible stability. For example, a protein may have two stable domains joined by a segment that is entirely disordered.

allow a protein to alter its conformation between two or more states. As we shall see in the next two chapters, variations in the stability of regions within a protein are often essential to protein function. Intrinsically disordered proteins or protein segments do not fold at all.

Not all proteins fold spontaneously as they are synthesized in the cell. Folding for many proteins requires chaperones, proteins that interact with partially folded or improperly folded polypeptides, facilitating correct folding pathways or providing microenvironments in which folding can occur. Several types of molecular chaperones are found in organisms ranging from bacteria to humans. Two major families of chaperones, both well studied, are the Hsp70 family and the chaperonins.

Proteins in the Hsp70 family generally have a molecular weight near 70,000 and are more abundant in cells stressed by elevated temperatures (hence, heat shock proteins of Mr 70,000, or Hsp70). Hsp70 proteins bind to regions of unfolded polypeptides that are rich in hydrophobic residues. These chaperones thus “protect” both proteins subject to denaturation by heat and new peptide molecules being synthesized (and not yet folded).

Some chaperones also facilitate the quaternary assembly of oligomeric proteins. The Hsp70 proteins bind to and release polypeptides in a cycle that uses energy from ATP hydrolysis and involves several other proteins (including a class called Hsp40). Figure 4-28 illustrates chaperone-assisted folding as elucidated for the eukaryotic Hsp70 and Hsp40 chaperones. The binding of an unfolded polypeptide by an Hsp70 chaperone may break up a protein aggregate or prevent the formation of a new one. When the bound polypeptide is released, it has a chance to resume folding to its native structure. If folding does not occur rapidly enough, the polypeptide may be bound again and the process repeated. Alternatively, the Hsp70-bound polypeptide may be delivered to a chaperonin.

Chaperonins are elaborate protein complexes required for the folding of some cellular proteins that do not fold spontaneously. In E. coli, an estimated 10% to 15% of cellular proteins require the resident chaperonin system, called GroEL/GroES, for folding under normal conditions (up to 30% require this assistance when the cells are heat stressed). The analogous chaperonin system in eukaryotes is called Hsp60. The chaperonins first became known when they were found to be necessary for the growth of certain bacterial viruses (hence the designation “Gro”).

These chaperone proteins are structured as a series of multisubunit rings, forming two chambers oriented back to back. Inside one of the chambers, a protein is given about 10 seconds to fold. Constraining a protein within the chamber prevents inappropriate protein aggregation and also restricts the conformational space that a polypeptide chain can explore as it folds. The GroEL/GroES folding pathway is discussed in Chapter 27.

Finally, the folding pathways of some proteins require two enzymes that catalyze isomerization reactions. Protein disulfide isomerase (PDI) is a widely distributed enzyme that catalyzes the interchange, or shuffling, of disulfide bonds until the bonds of the native conformation are formed. Among its functions, PDI catalyzes the elimination of folding intermediates with inappropriate disulfide cross-links.

Protein misfolding is a substantial problem in all cells. Despite the many processes that assist in protein folding, a quarter or more of all polypeptides synthesized may be destroyed because they do not fold correctly. In some cases, the misfolding causes or contributes to the development of serious disease.

Many conditions, including type 2 diabetes, Alzheimer disease, Huntington disease, and Parkinson disease, are associated with a misfolding mechanism: a soluble protein that is normally secreted from the cell is secreted in a misfolded state and converted into an insoluble extracellular amyloid fiber. The diseases are collectively referred to as amyloidoses. The fibers are highly ordered and unbranched, with a diameter of 7 to 10 nm and a high degree of β-sheet structure. The β segments are oriented perpendicular to the axis of the fiber. In some amyloid fibers the overall structure includes two layers of β sheet, such as that shown for amyloid-β peptide in Figure 4-29.

Many proteins can take on the amyloid fibril structure as an alternative to their normal folded conformations, and most of these proteins have a concentration of aromatic amino acid residues in a core region of β sheet or α helix. The proteins are secreted in an incompletely folded conformation. The core (or some part of it) folds into a β sheet before the rest of the protein folds correctly, and the β sheets of two or more incompletely folded protein molecules associate to begin forming an amyloid fibril. The fibril grows in the extracellular space. Other parts of the protein then fold differently, remaining on the outside of the β-sheet core in the growing fibril. The effect of aromatic residues in stabilizing the structure is shown in Figure 4-29c.

The amyloid deposition diseases that trigger neurodegeneration, particularly in older adults, are a special class of localized amyloidoses. Alzheimer disease is associated with extracellular amyloid deposition by neurons, involving the amyloid-β peptide (Fig. 4-29b), derived from a larger transmembrane protein (amyloid-β precursor protein) found in most human tissues. When it is part of the larger protein, the peptide is composed of two α-helical segments spanning the membrane. When the external and internal domains are cleaved off by specific proteases, the relatively unstable amyloid-β peptide leaves the membrane and loses its α-helical structure. It can then take the form of two layers of extended parallel β sheet, which can slowly assemble into amyloid fibrils (Fig. 4-29c).

Huntington disease involves the protein huntingtin, which has a long polyglutamine repeat. In some individuals, the polyglutamine repeat is longer than normal, and a more subtle type of intracellular aggregation occurs. Notably, when the mutant human proteins involved in Parkinson disease and Huntington disease are expressed in Drosophila melanogaster, the flies display neurodegeneration expressed as eye deterioration, tremors, and early death. All of these symptoms are highly suppressed if expression of the Hsp70 chaperone is also increased.

Protein misfolding need not lead to amyloid formation to cause serious disease. For example, cystic fibrosis is caused by defects in a membrane-bound protein called cystic fibrosis transmembrane conductance regulator (CFTR), which acts as a channel for chloride ions. The most common cystic fibrosis–causing mutation is the deletion of a Phe residue at position 508 in CFTR, which causes improper protein folding. Most of this protein is then degraded and its normal function is lost (see Box 11-2). Drugs that can correct certain CFTR misfolding events have been developed. Such drugs are called “correctors” or “pharmacological chaperones.” Many of the disease-related mutations in collagen (p. 118) also cause defective folding. A particularly remarkable type of protein misfolding is seen in the prion diseases (Box 4-4).

A misfolded brain protein seems to be the causative agent of several rare degenerative brain diseases in mammals. Perhaps the best known of these is bovine spongiform encephalopathy (BSE; also known as mad cow disease). Related diseases include kuru and Creutzfeldt-Jakob disease in humans, scrapie in sheep, and chronic wasting disease in deer and elk. These diseases are also referred to as spongiform encephalopathies, because the diseased brain frequently becomes riddled with holes (Fig. 1). Progressive deterioration of the brain leads to a spectrum of neurological symptoms, including weight loss; erratic behavior; problems with posture, balance, and coordination; and loss of cognitive function. The diseases are fatal.

In the 1960s, investigators found that preparations of the disease-causing agents seemed to lack nucleic acids. At this time, Tikvah Alper suggested that the agent was a protein. Initially, the idea seemed heretical. All disease-causing agents known up to that time—viruses, bacteria, fungi, and so on—contained nucleic acids, and their virulence was related to genetic reproduction and propagation. However, four decades of investigations, pursued most notably by Stanley Prusiner, have provided evidence that spongiform encephalopathies are different.

Prion protein is a normal constituent of brain tissue in all mammals. Its role is not known in detail, but it may have a molecular signaling function. Strains of mice lacking the gene for PrP (and thus the protein itself) suffer no obvious ill effects. Illness occurs only when the normal cellular PrP, or PrPC, occurs in an altered conformation called PrPSc (Sc denotes scrapie). The structure of PrPC has two α helices. The structure of PrPSc is very different, with much of the structure converted to amyloidlike β sheets (Fig. 2). The interaction of PrPSc with PrPC converts the latter to PrPSc, initiating a domino effect in which more and more of the brain protein converts to the disease-causing form. The mechanism by which the presence of PrPSc leads to spongiform encephalopathy is not understood.

In inherited forms of prion diseases, a mutation in the gene encoding PrP produces a change in one amino acid residue that is believed to make the conversion of PrPC to PrPSc more likely. A complete understanding of prion diseases awaits new information on how prion protein affects brain function. Structural information about PrP is beginning to provide insights into the molecular process that allows the prion proteins to interact so as to alter their conformation (Fig. 2). The significance of prions may extend well beyond spongiform encephalopathies. Evidence is building that prionlike proteins may be responsible for additional neurodegenerative diseases such as multiple system atrophy (MSA), a disease that resembles Parkinson disease.

Protein folding occurs too fast for it to be a completely random process. Instead, protein folding is generally hierarchical. Initially, regions of secondary structure may form, followed by folding into motifs and domains. Large ensembles of folding intermediates are rapidly brought to a single native conformation. For many proteins, folding is facilitated by Hsp70 chaperones and by chaperonins. Disulfide-bond formation and the cis-trans isomerization of Pro peptide bonds can also be catalyzed by specific enzymes during folding. Protein misfolding is the molecular basis for many human diseases, including cystic fibrosis and amyloidoses such as Alzheimer disease.

Structural biology is the study of the three-dimensional structures of biomolecules, including proteins, nucleic acids, lipid membranes, and oligosaccharides. Structural biologists combine biochemical approaches with physical tools and computational methods to obtain these structures. Structural biology is extraordinarily powerful for elucidating the relationships between the structure and function of proteins, the molecular basis for enzymatic catalysis and ligand binding, and evolutionary relationships between proteins.

Here we focus primarily on three commonly used methods in structural biology: x-ray crystallography, nuclear magnetic resonance (NMR), and cryoelectron microscopy (cryo-EM). Which method a structural biologist uses depends on the system being studied and what information is to be learned. Often structural biologists combine multiple methods to provide a more complete view of function. Increasingly, computational methods such as molecular dynamics simulations and in silico protein folding are proving to be essential for structural biologists, and these are discussed in Box 4-5.

Computational tools are now indispensable to biochemistry. While advances in computers and software have revolutionized how protein structures can be solved by x-ray crystallography, NMR, and cryo-EM, advances in computational chemistry have now allowed a number of studies of protein structure to be carried out entirely in silico using powerful molecular modeling and dynamics software. Key developments in this field were made by Martin Karplus, Michael Levitt, and Arieh Warshel, who received the Nobel Prize in Chemistry in 2013 for their work on the theoretical and computational tools needed to carry out computer simulations of molecules as large as proteins.

Protein folding is a problem particularly well-suited for computational biochemists. Protein-folding landscapes can be complex, and nearly an infinite number of possible peptide conformations are theoretically available for the protein to sample as it folds. However, to understand how folding happens efficiently and by only sampling a small portion of all possible conformations requires computers that can quickly test tens of thousands of possible folding trajectories in a brief time. Sometimes these calculations are carried out on powerful supercomputers, but in other instances research groups have taken to crowdsourcing these problems to thousands of citizen scientists.

David Baker has pioneered crowdsourcing protein-folding problems to engage millions of computer users directly in biochemistry. The Rosetta@home project distributes complex protein-folding problems to citizen scientists who run the software in the background on their home computers. Each home computer is able to complete a portion of a protein-folding experiment, and these results can be combined with those from other users to predict protein structures.

Foldit is a video game in which users compete to solve protein-folding puzzles (Fig. 1). Users are rewarded for making stabilizing contacts in the structure, like hydrogen bonds or favorable van der Waals interactions. Virtual protein designs (sometimes called theozymes) from the video game can then be tested in the real world. DNA genes coding for the virtual proteins can be created in the laboratory and be used to recombinantly produce the proteins in bacteria, using techniques described in Chapter 9. The purified proteins can then be structurally characterized by x-ray crystallography or NMR to see how closely the real-world structures and computationally designed structures match.

One exciting application of computational biochemistry is the creation of “designer proteins.” These are proteins with completely novel folds or functions not yet identified in nature. Designer proteins have potential applications in bioengineering, medicine, materials science, and chemistry. In one recent example, the Baker Lab designed a series of Foldit puzzles (such as “Cover the Ligand,” in which users had to redesign a ligand-binding site by rebuilding the protein backbone and changing amino acid side chains) to improve the catalytic efficiency of a previously engineered Diels-Alderase (Fig. 2).

The spacing of atoms in a crystal lattice can be determined by measuring the locations and intensities of spots produced on a detector by a beam of x-rays of given wavelength, after the beam has been diffracted by the electrons of the atoms.

The spacing of the different kinds of atoms in complex organic molecules, even very large ones such as proteins, can also be analyzed by x-ray diffraction methods. However, the technique for analyzing crystals of complex molecules is far more laborious than the technique for analyzing simple salt crystals. When the repeating pattern of the crystal is a molecule as large as, say, a protein, the numerous atoms in the molecule yield thousands of diffraction spots that must be analyzed by computer.

Consider how images are generated in a light microscope. Light from a point source is focused on an object. The object scatters the light waves, and these scattered waves are recombined by a series of lenses to generate an enlarged image of the object. The smallest object whose structure can be determined by such a system — that is, the resolving power of the microscope — is determined by the wavelength of the light, in this case visible light, with wavelengths in the range of 400 to 700 nm. Objects smaller than half the wavelength of the incident light cannot be resolved.

To resolve objects as small as proteins we must use xrays, with wavelengths in the range of 0.7 to 1.5 Å (0.07 to 0.15 nm). However, there are no lenses that can recombine x-rays to form an image; instead, the pattern of diffracted x-rays is collected directly and an image is reconstructed by mathematical techniques.

However, the orderly bundles formed by fibrous proteins are not crystals — the molecules are aligned side by side, but not all are oriented in the same direction. More-detailed three-dimensional structural information about proteins requires a highly ordered protein crystal. The structures of many proteins are not yet known, simply because they have proved difficult to crystallize. Practitioners have compared making protein crystals to holding together a stack of bowling balls with cellophane tape.

An electron-density map of the protein is reconstructed from the overall diffraction pattern of spots by a mathematical technique called a Fourier transform. In effect, the computer acts as a “computational lens.” A model for the structure is then built that is consistent with the electron-density map.

John Kendrew found that the x-ray diffraction pattern of crystalline myoglobin (isolated from muscles of the sperm whale) is highly complex, with nearly 25,000 reflections. Computer analysis of these reflections took place in stages. The resolution improved at each stage until, in 1959, the positions of virtually all the nonhydrogen atoms in the protein had been determined. The amino acid sequence of the protein, obtained by chemical analysis, was consistent with the molecular model. Over 100,000 protein structures, many of them much more complex than myoglobin, have since been determined to a similar level of resolution by x-ray crystallography.

The physical environment in a crystal, of course, is not identical to that in solution or in a living cell. A crystal imposes a space and time average on the structure deduced from its analysis, and x-ray diffraction studies provide little information about molecular motion within the protein. The conformation of proteins in a crystal can also be affected by nonphysiological factors such as incidental protein-protein contacts within the crystal.

An advantage of nuclear magnetic resonance (NMR) studies is that they are carried out on macromolecules in solution, whereas x-ray crystallography is limited to molecules that can be crystallized. NMR can also illuminate the dynamic side of protein structure, including conformational changes, protein folding, and interactions with other molecules.

NMR is a manifestation of nuclear spin angular momentum, a quantum mechanical property of atomic nuclei. Only certain atoms, including 1H, 13C, 15N, 19F, and 31P, have the kind of nuclear spin that gives rise to an NMR signal. Nuclear spin generates a magnetic dipole. When a strong, static magnetic field is applied to a solution containing a single type of macromolecule, the magnetic dipoles are aligned in the field in one of two orientations: parallel (low energy) or antiparallel (high energy).

the magnetic field. Some energy is absorbed as nuclei switch to the high-energy state, and the absorption spectrum that results contains information about the identity of the nuclei and their immediate chemical environment. The data from many such experiments on a sample are averaged, increasing the signal-tonoise ratio, and an NMR spectrum such as that in Figure 4-31 is generated.

These methods allow measurement of distance-dependent coupling of nuclear spins in nearby atoms through space (the nuclear Overhauser effect (NOE), in a method dubbed NOESY) or the coupling of nuclear spins in atoms connected by covalent bonds (total correlation spectroscopy, or TOCSY).

Complementary TOCSY experiments can help identify which NOE signals reflect atoms that are linked by covalent bonds. Certain patterns of NOE signals have been associated with secondary structures such as α helices. Genetic engineering (Chapter 9) can be used to prepare proteins that contain the rare isotopes 13C or 15N. The new NMR signals produced by these atoms, and the coupling with 1H signals resulting from these substitutions, help in the assignment of individual 1H NOE signals. The process is also aided by a knowledge of the amino acid sequence of the polypeptide.

To generate a three-dimensional structure, researchers feed the distance constraints into a computer along with known geometric constraints such as chirality, van der Waals radii, and bond lengths and angles. The computer generates a family of closely related structures that represent the range of conformations consistent with the NOE distance constraints (Fig. 4-31d). The uncertainty in structures generated by NMR is in part a reflection of the molecular vibrations (known as breathing) within a protein structure in solution, discussed in more detail in Chapter 5. Normal experimental uncertainty can also play a role.

Our understanding of highly complex processes such as gene expression, mitochondrial respiration, or viral infection is aided immensely by knowing the detailed molecular structures of the proteins that participate in these processes. However, it is often difficult to determine the molecular structure of large, dynamic, macromolecular complexes that contain dozens of individual protein subunits. Moreover, integral membrane proteins often resist crystallization once they are removed from their lipid environment, making their structures difficult to solve by x-ray diffraction, and many are too large for NMR. In principle, discrete objects in the diameter range 100 to 300 Å can be visualized by electron microscopy (EM). In practice, the high intensity of the EM beam often damages the specimen before a high-resolution image can be obtained.

In cryo-electron microscopy (cryo-EM), a sample containing many individual copies of the structure of interest is quick-frozen in vitreous (or noncrystalline) ice and kept frozen while being observed in two dimensions with the electron microscope, greatly reducing damage to the specimen by the electron beam.

When cryo-EM is combined with powerful algorithms for transforming the two-dimensional structures of tens of thousands of individual, randomly oriented complexes into a three-dimensional composite, it is sometimes possible to determine molecular structures at a level comparable to that obtained by x-ray crystallography (Fig. 4-32).

Many novel structures have now been obtained by cryo-EM without models based on prior x-ray or NMR structures. Since cryo-EM relies on imaging of single molecules of a complex, this technique can also be used to computationally sort the imaged particles and simultaneously determine structures of multiple conformational states. Cryo-EM has now been used to solve the structures of some of the most dynamic and largest molecular complexes in the cell, such as the human telomerase enzyme (Fig. 4-33). Telomerase is an essential enzyme for maintaining chromosome integrity in humans (see Chapter 26) and is the target of significant medical research due to its roles in aging and cancer. Cryo-EM was critical for the laboratories of Eva Nogales and Kathleen Collins in determining the architecture of telomerase due to the heterogeneity of the complex and because only minute quantities could be purified from human cells — far too little for crystallization, but enough to observe single molecules by cryo-EM.

In x-ray crystallography, protein molecules are crystallized in well-ordered orientations that diffract x-rays. The patterns and intensities of the diffracted x-rays depend on the structure of the protein and its crystalline properties. Mathematical methods can then reconstruct the protein structure that produces a particular diffraction pattern. NMR is often carried out on molecules in solution and yields information about atomic nuclei and their chemical environment. Protein structures can be computed from NMR data using hundreds of distance and geometric constraints obtained from multi-dimensional NMR experiments. Biomolecules are frozen in vitreous ice for imaging by cryo- EM. The individual molecules are then identified and computationally sorted. The sorted two-dimensional images are then combined using computers to produce a three-dimensional structure.

In biological membranes we see this basic lipid bilayer combined with a wide array of membrane proteins specialized for one of the many roles that membranes play in cells. Each intracellular compartment has membranes with specific protein and lipid content, uniquely disposed across the two leaflets of the bilayer. The intracellular membranes together make up a dynamic endomembrane system that synthesizes and distributes membrane components to create functionally specialized cell compartments. Depending on its function, a protein may form one or several transmembrane segments, or it may adhere to the membrane bilayer through lipids attached to the protein, or it may alternate between membrane-associated and free forms.

In the generalized membrane in Figure 11-3, phospholipids form a bilayer in which proteins are embedded, their hydrophobic domains in contact with the fatty acyl chains of membrane lipids. Some proteins protrude from only one side of the membrane; others have domains exposed on both sides. The orientation of both lipids and proteins in the bilayer is asymmetric, giving the membrane structural and functional “sidedness.”

The protein composition of membranes reflects each membrane’s functional specialization. One large group of membrane proteins are the receptors for extracellular signals such as hormones and changes in membrane potential. Hundreds of membrane proteins are transporters that carry specific polar or charged compounds — sugars, amino acids, vitamins, a wide variety of metabolic intermediates, minerals (Mg2+, Ca2+, Na+, K+), and trace metal ions (Mn2+, Ni2+, Co2+) — across the plasma membrane or between organelles. All transporters and many receptors must span the membrane at least once to play their roles.

The synthesis and delivery to the site of function are more complicated for membrane proteins than for typical cytosolic proteins. Many of the proteins destined for the plasma membrane undergo extensive posttranslational modification as they pass through the ER and the Golgi apparatus. One covalent modification is the glycosylation (attachment of oligosaccharides) of proteins bound for the plasma membrane. In glycophorin, a glycoprotein of the erythrocyte plasma membrane, 60% of the mass consists of complex oligosaccharides covalently attached to specific amino acid residues (Fig. 11-8). For such glycoproteins, Ser, Thr, and Asn residues are the most common points of carbohydrate attachment (see Fig. 7-27), and the carbohydrates are invariably on the outer face of the plasma membrane.

Integral membrane proteins are firmly embedded within the lipid bilayer and are removable only by agents such as detergents or organic solvents that overcome hydrophobic interactions (Fig. 11-9). Peripheral membrane proteins associate with the membrane through electrostatic interactions and hydrogen bonding with hydrophilic domains of integral proteins and with the charged head groups of membrane lipids. In the laboratory, they can be released from their membrane association by relatively mild treatments that interfere with electrostatic interactions or that break hydrogen bonds.

The firm attachment of integral proteins to membranes is the result of the hydrophobic interaction between the nonpolar core of the lipid bilayer and nonpolar side chains of the protein. Monotopic integral proteins have small hydrophobic domains that interact with only a single leaflet of the membrane (Fig. 1110).

Bitopic proteins span the bilayer once, extending on either surface. They have a single hydrophobic sequence somewhere in the molecule. Glycophorin (Fig. 11-8) is bitopic. Polytopic proteins cross the membrane several times. They have multiple hydrophobic sequences, each of which, when in the α-helical conformation, is long enough (about 20 residues) to span the lipid bilayer. (Recall from Worked Example 4-1 that each residue in an α helix adds 1.5 Å to its length.)

Integral proteins are embedded within membranes, their nonpolar amino acid side chains stabilized by contact with the lipid bilayer. Peripheral proteins associate with membranes through electrostatic interactions and hydrogen bonding with membrane phospholipids and integral proteins. Amphitropic proteins associate reversibly with the membrane. Many integral membrane proteins span the lipid bilayer several times, with hydrophobic sequences of about 20 amino acid residues forming transmembrane α helices. Multistranded β barrels are also common in integral proteins of bacterial and mitochondrial membranes. A hydropathy plot of the amino acid sequence identifies segments that are likely to cross the lipid bilayer as helices or barrels. Covalent attachment of a hydrophobic molecule such as a fatty acid serves to anchor some membrane proteins to the bilayer.

Some membrane proteins are covalently linked to one or more lipids, which may be of several types: long-chain fatty acids, isoprenoids, sterols, or glycosylated derivatives of phosphatidylinositol (GPIs; Fig. 11-16). The attached lipid provides a hydrophobic anchor that inserts into the lipid bilayer

The strength of the hydrophobic interaction between a bilayer and a single hydrocarbon chain linked to a protein is barely enough to anchor the protein securely, but many proteins have more than one attached lipid moiety. Furthermore, other interactions, such as ionic attractions between positively charged Lys residues in the protein and negatively charged lipid head groups, can add to the anchoring effect of a covalently bound lipid.

Three clusters of positively charged Lys and Arg residues (screened blue) interact with the negatively charged head group of phosphatidylinositol 4,5-bisphosphate (PIP2) on the cytoplasmic face of the plasma membrane; five aromatic residues (screened yellow) insert into the lipid bilayer. When the head-group phosphates of PIP2 are enzymatically removed, MARCKS loses its hold on the plasma membrane and dissociates. The reversible nature of the association of MARCKS with the membrane makes it an amphitropic protein.

In polarized epithelial cells (such as intestinal epithelial cells), in which apical and basal surfaces have different roles, GPI-anchored proteins are directed specifically to the apical surface. Attachment of a specific lipid to a newly synthesized membrane protein therefore has a targeting function, directing the protein to its correct cellular location.

Studies of lipid bilayers in vitro (Fig. 11-2) show that the hydrocarbon core of the bilayer, made up of the —CH2— and —CH3 of the fatty acyl groups, is about as nonpolar as decane, and about 3 nm (30 Å) thick, roughly the width of two extended fatty acyl chains. Biological membranes, to which we turn next, are 50–80 Å thick, when proteins protruding on either side are included.

One of the best-studied polytopic proteins, bacteriorhodopsin, has seven very hydrophobic internal sequences and spans the lipid bilayer seven times. Bacteriorhodopsin is a light-driven proton pump densely packed in regular arrays in the purple membrane of the bacterium Halobacterium salinarum. X-ray crystallography and cryo-electron microscopy reveal a structure in which each of the seven α-helical segments crosses the lipid bilayer, and all are connected by nonhelical loops at the inner and outer faces of the membrane (Fig. 11-11). In the amino acid sequence of bacteriorhodopsin, seven segments of about 20 hydrophobic residues can be identified, each forming an α helix that spans the bilayer. The seven helices are clustered together and oriented not quite perpendicular to the bilayer plane, a motif that (as we shall see in Chapter 12) is very common in membrane proteins involved in signal reception. The hydrophobic interaction keeps the nonpolar amino acid residues firmly anchored among the fatty acyl groups of the membrane lipids in the membrane’s lipid core.

The presence of unbroken sequences of more than 20 hydrophobic residues in a membrane protein is commonly taken as evidence that these sequences traverse the lipid bilayer, acting as hydrophobic anchors or forming transmembrane channels. All known integral proteins have at least one such sequence. Application of this logic to entire genomic sequences leads to the conclusion that in many organisms, 20% to 30% of all proteins are integral proteins.

What can we predict about the secondary structure of the membrane-spanning portions of integral proteins? At 1.5 Å (0.15 nm) per amino acid residue, an α-helical sequence of 20 to 25 residues is just long enough to span the thickness (30 Å) of the lipid bilayer. A polypeptide chain surrounded by lipids, having no water molecules with which to hydrogen-bond, will tend to form α helices or β sheets in which intrachain hydrogen bonding is maximized. If the side chains of all amino acids in a helix are nonpolar, the helix is further stabilized in the surrounding lipids by the hydrophobic effect.

Proteins with relatively short hydrophobic helical sections (19 to 20 residues) cannot span the thicker bilayer in rafts, and thus they tend to be excluded. Proteins with longer hydrophobic helices (24 to 25 residues) segregate into the thicker bilayer regions of rafts, where the entire length of the helix is stabilized by the hydrophobic effect.

This free energy of transfer, which can be expressed as a hydropathy index (see Table 3-1), ranges from highly exergonic for charged or polar residues to highly endergonic for amino acids with aromatic or aliphatic hydrocarbon side chains. The overall hydropathy index (hydrophobicity) of a sequence of amino acids is estimated by summing the free energies of transfer for the residues in the sequence. To scan a polypeptide sequence for potential membrane-spanning segments, an investigator calculates the hydropathy index for successive segments (called windows) of a given size, from 7 to 20 residues.

For a window of seven residues, for example, the average indices for residues 1 to 7, 2 to 8, 3 to 9, and so on, are plotted as in Figure 11-13 (plotted for the middle residue in each window — residue 4 for residues 1 to 7, for example). A region with more than 20 residues of high hydropathy index is presumed to be a transmembrane segment. When the sequences of membrane proteins of known three-dimensional structure are scanned using simple online bioinformatics tools, we find a reasonably good correspondence between predicted and known membrane-spanning segments. Hydropathy analysis predicts a single hydrophobic helix for glycophorin (Fig. 11-13a) and seven transmembrane segments for bacteriorhodopsin (Fig. 11-13b) — in agreement with structures known from x-ray crystallography.

The same factors that favor α helix formation in the hydrophobic interior of a lipid bilayer also stabilize β barrels: when no water molecules are available to hydrogen-bond with the carbonyl oxygen and nitrogen of the peptide bond, maximal intrachain hydrogen bonding gives the most stable conformation. Planar β sheets do not maximize these interactions and are generally not found in the membrane interior; β barrels allow all possible hydrogen bonds and are common among membrane proteins.

A polypeptide is more extended in the β conformation than in an α helix; just seven to nine residues of β conformation are needed to span a membrane. Recall that in the β conformation, alternating side chains project above and below the sheet (see Fig. 4-5). In β strands of membrane proteins, every second residue in the membrane-spanning segment is hydrophobic and interacts with the lipid bilayer; aromatic side chains are commonly found at the lipid-protein interface. The other residues may or may not be hydrophilic.

A further remarkable feature of many transmembrane proteins of known structure is the presence of Tyr and Trp residues at the interface between lipid and water (Fig. 11-15). The side chains of these residues seem to serve as membrane interface anchors, able to interact simultaneously with the central lipid phase and the aqueous phases on either side of the membrane. Another generalization about amino acid location relative to the bilayer is described by the positive-inside rule: the positively charged Lys and Arg residues in the extramembrane loop of membrane

Depending on the precise conditions and the nature of the lipids, several types of lipid aggregate can form when amphipathic lipids are mixed with water (Fig. 11-1). Micelles are spherical structures that contain anywhere from a few dozen to a few thousand amphipathic molecules. These molecules are arranged with their hydrophobic regions aggregated in the interior, where water is excluded, and their hydrophilic head groups at the outer surface,

Micelle formation is favored when the cross-sectional area of the head group is greater than that of the acyl side chain(s), as in free fatty acids, lysophospholipids (phospholipids lacking one fatty acid), and many detergents, such as sodium dodecyl sulfate (SDS; p. 88).

A second type of lipid aggregate in water is the bilayer, in which two lipid monolayers (leaflets) form a two-dimensional sheet. Bilayer formation is favored if the cross-sectional areas of the head group and acyl side chain(s) are similar, as in glycerophospholipids and sphingolipids. The hydrophobic portions in each monolayer, excluded from water, make contact with each other. The hydrophilic head groups interact with water at one or the other surface of the bilayer.

Many membrane proteins co-crystalize with phospholipid molecules, which are presumed to be positioned in the native membranes as they are in the protein crystals. Many of these phospholipid molecules lie on the protein surface, their head groups interacting with polar amino acid residues at the inner and outer membrane–water interfaces and their side chains associated with nonpolar residues.