Voet D., Voet Ju.G. Biochemistry

Подождите немного. Документ загружается.

a major influence in determining their conformations. Lon-

don forces also provide much of the binding energy in the

sterically complementary interactions between proteins

and the molecules that they specifically bind.

B. Hydrogen Bonding Forces

Hydrogen bonds , as we discussed in Section

2-1Aa, are predominantly electrostatic interactions (but

with ⬃10% covalent character) between a weakly acidic

donor group (D¬H) and an acceptor (A) that bears a lone

pair of electrons. In biological systems, D and A can both

be the highly electronegative N and O atoms and occasion-

(D¬H

ally S atoms. In addition, a relatively acidic C¬H group

(e.g., a C

␣

¬H group) can act as a weak hydrogen bond

donor, and the polarizable ␲ electron system of an aro-

matic ring (e.g., that of Trp) can act as a weak acceptor.

Hydrogen bonds have association energies that are nor-

mally in the range ⫺12 to ⫺40 kJ ⴢ mol

⫺1

(but only around

⫺8 to ⫺16 kJ ⴢ mol

⫺1

for and hydro-

gen bonds and ⫺2 to ⫺4 kJ ⴢ mol

⫺1

for hydrogen

bonds), values which are between those for covalent bonds

and van der Waals forces. Hydrogen bonds (H bonds) are

much more directional than are van der Waals forces but

less so than are covalent bonds. The distance is nor-

mally in the range 2.7 to 3.1 Å, although since H atoms are

unseen in all but the very highest resolution macromolecu-

lar X-ray structures, a possible interaction

(where D and A are either N or O) is assumed to be a H

bond if its distance is significantly less than the 3.7

Å sum of a D¬H bond length (⬃1.0 Å) and the van der

Waals contact distance between H and A (⬃2.7 Å). Keep in

mind, however, that there is no rigid cutoff distance beyond

which H bonds cease to exist because the energy of an H

bond, which is mainly electrostatic in character, varies in-

versely with the distance between the negative and positive

centers (Eq. [8.1]).

H bonds tend to be linear, with the D¬H bond pointing

along the acceptor’s lone pair orbital (or, in

hydrogen bonds, roughly perpendicular to the aromatic

ring and pointing at its center with the distance from the D

atom to the center of the aromatic ring normally in the

range 3.2–3.8 Å). Large deviations from this ideal geome-

try are not unusual, however. For example, in the H bonds

of both ␣ helices (Fig. 8-11) and antiparallel ␤ pleated

sheets (Fig. 8-16a), the N¬H bonds point approximately

along the bonds rather than along an O lone pair

orbital, and in parallel ␤ pleated sheets (Fig. 8-16b), the H

bonds depart significantly from linearity. Indeed, many of

the H bonds in proteins are members of networks in which

each donor is H bonded to two acceptors (a bifurcated hy-

drogen bond) and each acceptor is H bonded to two donors.

For example, although the H bonds in ideal ␣ helices form

between the N¬H group at residue n and the group

at residue n ⫺ 4 (n S n ⫺ 4 H bonds; Fig. 8-11), many of the

N¬H groups in real ␣ helices associate via bifurcated H

bonds with two adjacent groups to form both

n S n ⫺ 4 and n S n ⫺ 3 H bonds.

a. Hydrogen Bonds Only Weakly Stabilize Proteins

A protein’s internal H bonding groups are arranged

such that most possible H bonds are formed (Section 8-3B).

Clearly, H bonding has a major influence on the structures

of proteins. However, an unfolded protein makes most of

its H bonds with the water molecules of the aqueous sol-

vent (water, it will be recalled, is a strong H bonding donor

and acceptor).The free energy of stabilization that internal

H bonds confer on a native protein is therefore equal to the

difference in the free energy of H bonding between the

native protein and the unfolded protein. Consequently, it

might be expected that H bonds do not stabilize (and per-

haps even slightly destabilize) the structure of a native

C “ O

D¬H

␲

D¬H

C¬H

␲

D¬H

␲C¬H

Section 8-4. Protein Stability 261

Figure 8-58 Dipole–dipole interactions. The strength of each

dipole is represented by the thickness of the accompanying

arrow. (a) Interactions between permanent dipoles.These

interactions, here represented by carbonyl groups lined up head

to tail, may be attractive, as shown, or repulsive, depending on

the relative orientations of the dipoles. (b) Dipole–induced

dipole interactions.A permanent dipole (here shown as a

carbonyl group) induces a dipole in a nearby group (here

represented by a methyl group) by electrostatically distorting its

electron distribution (shading).This always results in an attractive

interaction. (c) London dispersion forces. The instantaneous

charge imbalance (shading) resulting from the motions of the

electrons in a molecule (left) induces a dipole in a nearby group

(right); that is, the motions of the electrons in neighboring groups

are correlated.This always results in an attractive interaction.

(a)

++––

Interactions between permanent dipoles

(b) Dipole–induced dipole interactions

++––

C O

δ + δ – δ + δ –

δ + δ –

δ + δ – δ + δ –

C O

δ + δ –

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 261

protein relative to its unfolded state. However,since H bond-

ing interactions are largely electrostatic in nature, they are

likely to be stronger in the low polarity interior of a protein

than they are in the high polarity aqueous medium. More-

over, there may be an entropic effect that destabilizes the

H bonds between water and an unfolded polypeptide rela-

tive to intraprotein H bonds: The water molecules that are

H bonded to a polypeptide are likely to be more position-

ally and orientationally constrained (ordered) than those

that are H bonded to only other water molecules, thus fa-

voring the formation of intraprotein H bonds.These effects

may very well account for the observation that the muta-

genic removal of an H bond from a protein generally re-

duces the protein’s stability by ⫺2 to 8 kJ ⴢ mol

⫺1

Despite their low stability, a protein’s hydrogen bonds

provide a structural basis for its native folding pattern: If a

protein folded in a way that prevented some of its internal

H bonds from forming, their free energy would be lost and

such conformations would be less stable than those that are

fully H bonded. In fact, the formation of ␣ helices and

␤ sheets efficiently satisfies the polypeptide backbone’s

H bonding requirements.This argument also applies to the

van der Waals forces discussed in the previous section.

b. Most Hydrogen Bonds in Proteins Are Local

How can as complex a molecule as a protein fold so as to

make nearly all of its potential H bonds? The answer to this

question was revealed by a survey of the H bonds in high

resolution protein X-ray structures by Ken Dill and George

Rose: Most of the H bonds in a protein are local, that is, they

involve donors and acceptors that are close together in se-

quence and hence can readily find their H bonding mates.

1. On average, 68% of the H bonds in proteins are be-

tween backbone atoms. Of these, ⬃1/3 form n S n ⫺ 4 H

bonds (as in ideal ␣ helices), ⬃1/3 form n S n ⫺ 3 H bonds

(as in reverse turns and ideal 3

helices), and ⬃1/3 are be-

tween paired strands in ␤ sheets. In fact, only ⬃5% of the

H bonds between backbone atoms are not wholly within a

helix, sheet, or turn.

2. Hydrogen bonds between side chains and backbones

are clustered at helix-capping positions. In an ␣ helix, the

first four N¬H groups and the last four groups can-

not form H bonds within the helix (which accounts for half

the potential H bonds involving backbone atoms in an ␣ he-

lix of 12 residues, the average length of ␣ helices).These po-

tential H bonds are often made with nearby side chains. In

particular,⬃1/2 of the N-terminal N¬H groups of ␣ helices

form H bonds with polar side chains that are 1 to 3 residues

distant, and ⬃1/3 of their C-terminal groups form H

bonds with polar side chains that are 2 to 5 residues distant.

3. Over half the H bonds between side chains are be-

tween charged residues (i.e., they form salt bridges) and

are therefore located on protein surfaces between and

within surface loops (e.g., Fig. 8-57). However, ⬃85% of the

remaining side chain–side chain H bonds are between side

chains that are 1 to 5 residues apart. Hence with the excep-

tion of those in salt bridges, side chain–side chain H bonds

also tend to be local.

C “ O

C. Hydrophobic Forces

The hydrophobic effect is the name given to those influences

that cause nonpolar substances to minimize their contacts

with water and amphipathic molecules, such as soaps and de-

tergents, to form micelles in aqueous solutions (Section 2-

1Ba). Since native proteins form a sort of intramolecular

micelle in which their nonpolar side chains are largely out of

contact with the aqueous solvent, hydrophobic interactions

must be an important determinant of protein structures.

The hydrophobic effect derives from the special proper-

ties of water as a solvent,only one of which is its high dielec-

tric constant. In fact, other polar solvents, such as dimethyl

sulfoxide (DMSO) and N,N-dimethylformamide (DMF),

tend to denature proteins. The thermodynamic data of

Table 8-5 provide considerable insight as to the origin of the

hydrophobic effect because the transfer of a hydrocarbon

from water to a nonpolar solvent resembles the transfer of

a nonpolar side chain from the exterior of a protein in aque-

ous solution to its interior. The isothermal Gibbs free en-

ergy changes (⌬G ⫽⌬H ⫺ T ⌬S) for the transfer of a hydro-

carbon from an aqueous solution to a nonpolar solvent is

negative in all cases, which indicates, as we know to be the

case, that such transfers are spontaneous processes (oil and

water don’t mix). What is perhaps unexpected is that these

transfer processes are endothermic (positive ⌬H) for

aliphatic compounds and athermic (⌬H ⫽ 0) for aro-

matic compounds; that is, it is enthalpically more or equally

favorable for nonpolar molecules to dissolve in water than in

nonpolar media. In contrast, the entropy component of the

unitary free energy change, ⫺T ⌬S

(see footnote a to Table

8-5), is large and negative in all cases. Evidently, the transfer

of a hydrocarbon from an aqueous medium to a nonpolar

medium is entropically driven. The same is true of the trans-

fer of a nonpolar protein group from an aqueous environ-

ment to the protein’s nonpolar interior.

What is the physical mechanism whereby nonpolar enti-

ties are excluded from aqueous solutions? Recall that en-

tropy is a measure of the order of a system; it decreases

with increasing order (Section 3-2). Thus the decrease in

entropy when a nonpolar molecule or side chain is solvated

by water (the reverse of the foregoing process) must be

due to an ordering process. This is an experimental obser-

vation, not a theoretical conclusion.The magnitudes of the

entropy changes are too large to be attributed only to

changes in the conformations of the hydrocarbons; rather,

as Henry Frank and Marjorie Evans pointed out in 1945,

these entropy changes mainly arise from some sort of order-

ing of the water structure.

Liquid water has a highly ordered and extensively H

bonded structure (Section 2-1A). The insinuation of a non-

polar group into this structure disrupts it:A nonpolar group

can neither accept nor donate H bonds, so the water mole-

cules at the surface of the cavity occupied by the nonpolar

group cannot H bond to other molecules in their usual fash-

ion. In order to recover the lost H bonding energy, these

surface waters must orient themselves so as to form an H

bonded network enclosing the cavity (Fig. 8-59).This orien-

tation constitutes an ordering of the water structure, since

the number of ways that water molecules can form H bonds

262 Chapter 8. Three-Dimensional Structures of Proteins

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 262

about the surface of a nonpolar group is less than the num-

ber of ways that they can H bond in bulk water.

Unfortunately, the complexity of liquid water’s basic

structure (Section 2-1Ac) has not yet allowed a detailed

structural description of this ordering process. One model

that has been proposed is that water forms quasi-

crystalline H bonded cages about the nonpolar groups sim-

ilar to those of clathrates (Fig. 8-60).The magnitudes of the

entropy changes that result when nonpolar substances are

dissolved in water, however, indicate that the resulting water

structures can only be slightly more ordered than bulk water.

They also must be quite different from that of ordinary ice,

Section 8-4. Protein Stability 263

Table 8-5 Thermodynamic Changes for Transferring Hydrocarbons from Water to Nonpolar Solvents at 25°C

Process ⌬H (kJ ⴢ mol

⫺1

) ⫺T ⌬S

(kJ ⴢ mol

⫺1

) ⌬G

(kJ ⴢ mol

⫺1

)

11.7 ⫺22.6 ⫺10.9

10.5 ⫺22.6 ⫺12.1

9.2 ⫺25.1 ⫺15.9

6.7 ⫺18.8 ⫺12.1

0.8 ⫺8.8 ⫺8.0

0.0 ⫺17.2 ⫺17.2

0.0 ⫺20.0 ⫺20.0Toluene in H

O Δ liquid toluene

Benzene in H

O Δ liquid benzene

in H

O Δ C

in benzene

in H

O Δ C

in benzene

in H

O Δ C

in benzene

in H

O Δ CH

in CCl

in H

O Δ CH

in C

⌬G

, the unitary Gibbs free energy change, is the Gibbs free energy change, ⌬G, corrected for its concentration dependence so that it reflects only the

inherent properties of the substance in question and its interaction with solvent.This relationship, according to Equation [3.13], is

where [A

] and [A

] are the initial and final concentrations of the substance under consideration, respectively, and n is the number of moles of that

substance. Since the second term in this equation is a purely entropic term (concentrating a substance increases its order), ⌬S

, the unitary entropy

change, is expressed

Data measured at 18°C.

Source: Kauzmann,W., Adv. Protein Chem. 14, 39 (1959).

¢S

⫽ ¢S ⫹ nR ln

]

¢G

⫽ ¢G ⫺ nRT ln

]

Figure 8-59 The orientational preference of water molecules

next to a nonpolar solute. In order to maximize their H bonding

energy, these water molecules tend to straddle the inert solute

such that, for relatively small solutes, two or three of their

tetrahedral directions are tangential to its surface. This permits

them to form H bonds (black) with neighboring water molecules

lining the nonpolar surface. However, for larger (flatter) nonpolar

solutes, the adjacent water molecules are each geometrically

limited to participating in no more than three H bonds. In either

case, the ordering of water molecules extends several layers of

water molecules beyond the first hydration shell of the nonpolar

solute. [Illustration, Irving Geis. Image from the Irving Geis

Collection, Howard Hughes Medical Institute. Reprinted with

permission.]

Figure 8-60 Structure of the clathrate (n-C

)

ⴙ

ⴚ

⭈ 23H

Clathrates are crystalline complexes of nonpolar compounds

with water (usually formed at low temperatures and high pressures)

in which the nonpolar molecules are enclosed, as shown, by a

polyhedral cage of tetrahedrally H bonded water molecules

(here represented by only their oxygen atoms).The H bonding

interactions of one such water molecule (arrow) are shown in

detail. [Illustration, Irving Geis. Image from the Irving Geis

Collection, Howard Hughes Medical Institute. Reprinted with

permission.]

Nonpolar

solute

JWCL281_c08_221-277.qxd 8/10/10 11:48 AM Page 263

because, for instance, the solvation of nonpolar groups by

water causes a large decrease in water volume (e.g., the

transfer of CH

from hexane to water shrinks the water so-

lution by 22.7 mL ⴢ mol

⫺1

of CH

), whereas the freezing of

water results in a 1.6-mL ⴢ mol

⫺1

expansion.

The unfavorable free energy of hydration of a nonpolar

substance caused by its ordering of the surrounding water

molecules has the net result that the nonpolar substance is

excluded from the aqueous phase. This is because the sur-

face area of a cavity containing an aggregate of nonpolar

molecules is less than the sum of the surface areas of the

cavities that each of these molecules would individually oc-

cupy.The aggregation of the nonpolar groups thereby min-

imizes the surface area of the cavity and therefore the en-

tropy loss of the entire system. In a sense, the nonpolar

groups are squeezed out of the aqueous phase by the hy-

drophobic interactions. Thermodynamic measurements in-

dicate that the free energy change of removing a ¬CH

group from an aqueous solution is about ⫺3 kJ ⴢ mol

⫺1

.Al-

though this is a relatively small amount of free energy, in

molecular assemblies involving large numbers of nonpolar

contacts, hydrophobic interactions are a potent force.

Walter Kauzmann pointed out in 1958 that hydrophobic

forces are a major influence in causing proteins to fold into

their native conformations. Figure 8-61 indicates that the

amino acid side chain hydropathies (indexes of combined

hydrophobic and hydrophilic tendencies; Table 8-6) are, in

fact, good predictors of which portions of a polypeptide

chain are inside a protein, out of contact with the aqueous

solvent, and which portions are outside, in contact with the

aqueous solvent. In proteins, the effects of hydrophobic

forces are often termed hydrophobic bonding, presumably

to indicate the specific nature of protein folding under the

influence of the hydrophobic effect. You should keep in

mind, however, that hydrophobic bonding does not gener-

ate the directionally specific interactions usually associated

with the term “bond.”

D. Disulfide Bonds

Since disulfide bonds form as a protein folds to its native

conformation (Section 9-1A), they function to stabilize its

three-dimensional structure.The relatively reducing chemi-

cal character of the cytoplasm, however, greatly diminishes

the stability of intracellular disulfide bonds. In fact, almost

all proteins with disulfide bonds are secreted to more oxi-

dized extracellular destinations, where their disulfide bonds

264 Chapter 8. Three-Dimensional Structures of Proteins

Figure 8-61 Hydropathic index plot for bovine

chymotrypsinogen. The sum of the hydropathies of nine

consecutive residues (see Table 8-6) are plotted versus the residue

sequence number.A large positive hydropathic index is indicative

of a hydrophobic region of the polypeptide chain, whereas a

–20

–40

20 40 60 80 100 120 140

Residue number

Hydrophobic

Hydrophilic

160 180 200 220 240

Hydropathic index

Table 8-6 Hydropathy Scale for Amino Acid Side Chains

Side Chain Hydropathy

Ile 4.5

Val 4.2

Leu 3.8

Phe 2.8

Cys 2.5

Met 1.9

Ala 1.8

Gly ⫺0.4

Thr ⫺0.7

Ser ⫺0.8

Tr p ⫺0.9

Ty r ⫺1.3

Pro ⫺1.6

His ⫺3.2

Glu ⫺3.5

Gln ⫺3.5

Asp ⫺3.5

Asn ⫺3.5

Lys ⫺3.9

Arg ⫺4.5

Source: Kyte, J. and Doolitle, R.F., J. Mol. Biol. 157, 110 (1982).

large negative value is indicative of a hydrophilic region.The

bars above the midpoint line denote the protein’s interior

regions, as determined by X-ray crystallography, and the bars

below the midpoint line indicate the protein’s exterior regions.

[After Kyte, J. and Doolittle, R.F., J. Mol. Biol. 157, 111 (1982).]

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 264

are effective in stabilizing protein structures [secreted pro-

teins fold to their native conformations—and hence form

their disulfide bonds—in the endoplasmic reticulum (Sec-

tion 12-4Ba), which, unlike other cellular compartments,

has an oxidizing environment]. Apparently, the relative

“hostility” of extracellular environments toward proteins

(e.g., uncontrolled temperatures and pH’s) requires the ad-

ditional structural stability conferred by disulfide bonds.

E. Protein Denaturation

The low conformational stabilities of native proteins make

them easily susceptible to denaturation by altering the bal-

ance of the weak nonbonding forces that maintain the na-

tive conformation.When a protein in solution is heated, its

conformationally sensitive properties, such as optical rota-

tion (Section 4-2A), viscosity, and UV absorption, change

abruptly over a narrow temperature range (e.g., Fig. 8-62).

Such a nearly discontinuous change indicates that the native

protein structure unfolds in a cooperative manner:Any par-

tial unfolding of the structure destabilizes the remaining

structure,which must simultaneously collapse to the random

coil. The temperature at the midpoint of this process is

known as the protein’s melting temperature, T

, in analogy

with the melting of a solid. Most proteins have T

values

well below 100°C. Recall that nucleic acids likewise have

characteristic T

’s (Section 5-3Ca).

In addition to high temperatures, proteins are dena-

tured by a variety of other conditions and substances:

1. pH variations alter the ionization states of amino

acid side chains (Table 4-1), which changes protein charge

distributions and H bonding requirements.

2. Detergents, some of which significantly perturb pro-

tein structures at concentrations as low as 10

⫺6

M,hy-

drophobically associate with the nonpolar residues of a

protein, thereby interfering with the hydrophobic interac-

tions responsible for the protein’s native structure.

3. High concentrations of water-soluble organic sub-

stances, such as aliphatic alcohols, interfere with the hy-

drophobic forces stabilizing protein structures through

their own hydrophobic interactions with water. Organic

substances with several hydroxyl groups, such as ethylene

glycol or sucrose,

however, are relatively poor denaturants because their H

bonding ability renders them less disruptive of water structure.

The influence of salts is more variable. Figure 8-63

shows the effects of a number of salts on the T

of bovine

SucroseEthylene glycol

H OH

HOH

CCH

HO OH

Section 8-4. Protein Stability 265

Figure 8-62 Protein denaturation. The heat-induced

denaturation of bovine pancreatic ribonuclease A (RNase A) in an

HCl–KCl solvent at pH 2.1 and 0.019 ionic strength was monitored

by several conformationally sensitive techniques.The curve is drawn

only through the points ⌬.The melting temperature, T

, is defined

as the temperature at the midpoint of the transition. Compare the

shape of this melting curve with that of duplex DNA (Fig. 5-16).

[After Ginsburg,A. and Carroll,W.R., Biochemistry 4, 2169 (1965).]

0 102030405060

0.25

0.50

0.75

1.00

Fractional change

T (°C)

Intrinsic viscosity

Specific rotation at 365 n

Molar absorbance at 287

Repeat of after cooling

for ~16 hours

01234 5

(pH6.6)

(NH

)

KCl

NaCl

LiCl

NaBr

LiBr

CaCl

KSCN

Concentration (M)

(°C)

Figure 8-63 Melting temperature of RNase A as a function of

the concentrations of various salts. All solutions also contained

0.15M KCl and 0.013M sodium cacodylate buffer, pH 7. [After

von Hippel, P.J. and Wong, K.Y., J. Biol. Chem. 10, 3913 (1965).]

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 265

pancreatic ribonuclease A (RNase A). Some salts, such as

(NH

)

and KH

, stabilize the native protein struc-

ture (raise its T

); others, such as KCl and NaCl, have little

effect; and yet others, such as KSCN and LiBr, destabilize

it.The order of effectiveness of the various ions in stabiliz-

ing a protein, which is largely independent of the identity

of the protein, parallels their capacity to salt out proteins

(Section 6-2A). This order is known as the Hofmeister

series:

Anions:

Cations:

The ions in the Hofmeister series that tend to denature

proteins, I

⫺

, , SCN

⫺

,Li

⫹

,Mg

2⫹

,Ca

2⫹

, and Ba

2⫹

,are

said to be chaotropic. This list should also include the

guanidinium ion (Gu

⫹

) and the nonionic urea, which, in

concentrations in the range 5 to 10 M, are the most com-

monly used protein denaturants. The effect of the various

ions on proteins is largely cumulative: GuSCN is a much

more potent denaturant than the often used GuCl, whereas

stabilizes protein structures.

Chaotropic agents increase the solubility of nonpolar

substances in water. Consequently, their effectiveness as

denaturing agents stems from their ability to disrupt hy-

drophobic interactions, although the manner in which they

do so is not well understood. Conversely, those substances

listed that stabilize proteins strengthen hydrophobic forces,

thus increasing the tendency of water to expel proteins.

This accounts for the correlation between the abilities of an

ion to stabilize proteins and to salt them out.

F. Explaining the Stability of Thermostable Proteins

Certain species of bacteria known as hyperthermophiles

grow at temperatures near 100°C (they live in such places

as hot springs and submarine hydrothermal vents, with the

most extreme, an Fe(III)-reducing archaeon, growing at

121°C and remaining viable as high as 130°C).These organ-

isms have many of the same metabolic pathways as do

mesophiles (organisms that grow at “normal” tempera-

tures). Yet, most mesophilic proteins denature at the tem-

peratures at which hyperthermophiles thrive. What is the

structural basis for the thermostability of hyperther-

mophilic proteins?

The difference in the thermal stabilities of the corre-

sponding (hyper)thermophilic and mesophilic proteins

does not exceed ⬃100 kJ ⴢ mol

⫺1

, the equivalent of a few

noncovalent interactions. This is probably why compar-

isons of the X-ray structures of hyperthermophilic en-

zymes with their mesophilic counterparts have failed to re-

veal any striking differences between them.These proteins

exhibit some variations in secondary structure but no more

so than is often the case for homologous proteins from dis-

tantly related mesophiles. However, several of these ther-

mostable enzymes have a superabundance of salt bridges

on their surfaces, many of which are arranged in extensive

ClO

⫺

⬎ Ca

2⫹

⬎ Ba

2⫹

⫹

, Cs

⫹

, K

⫹

, Na

⫹

7 Li

⫹

7 Mg

2⫹

⬎ Br

⫺

⬎ I

⫺

⬎ ClO

⫺

⬎ SCN

⫺

2⫺

⬎ H

⫺

⬎ CH

COO

⫺

⬎ Cl

⫺

networks. Indeed, one such network from Pyrococcus fu-

riosus glutamate dehydrogenase consists of 18 side chains.

The idea that salt bridges can stabilize a protein struc-

ture appears to contradict the conclusion of Section 8-4Aa

that ion pairs are, at best, marginally stable. The key to this

apparent paradox is that the salt bridges in thermostable

proteins form networks. Thus, the gain in charge–charge

free energy on associating a third charged group with an

ion pair is comparable to that between the members of this

ion pair, whereas the free energy lost on desolvating and

immobilizing the third side chain is only about half that lost

in bringing together the first two side chains. The same, of

course, is true for the addition of a fourth, fifth, etc., side

chain to a salt bridge network.

Not all thermostable proteins have such a high inci-

dence of salt bridges. Structural comparisons suggest that

these proteins are stabilized by a combination of small ef-

fects, the most important of which are an increased size in

the protein’s hydrophobic core, an increased size in the in-

terface between its domains and/or subunits, and a more

tightly packed core as evidenced by a reduced surface-to-

volume ratio.

The fact that the proteins of hyperthermophiles and

mesophiles are homologous and carry out much the same

functions indicates that mesophilic proteins are by no

means maximally stable.This, in turn, strongly suggests that

the marginal stability of most proteins under physiological

conditions (averaging ⬃0.4 kJ/mol of amino acid residues)

is an essential property that has arisen through evolutionary

design. Perhaps this marginal stability helps confer the

structural flexibility that many proteins require to carry

out their physiological functions (Section 9-4). Other possi-

bilities are that it may facilitate the elimination of other-

wise stable non-native conformations (Section 9-2C), it

may promote the unfolding of proteins so as to permit their

insertion into or transport through membranes (Section

12-4E), and/or it may expedite their programmed degrada-

tion (Section 32-6).

5 QUATERNARY STRUCTURE

Proteins, because of their multiple polar and nonpolar

groups, stick to almost anything; anything, that is, but other

proteins. This is because the forces of evolution have

arranged the surface groups of proteins so as to prevent

their association under physiological conditions. If this

were not the case, their resulting nonspecific aggregation

would render proteins functionally useless (recall, e.g., the

consequences of sickle-cell anemia; Section 7-3A). In his

pioneering ultracentrifugational studies on proteins, how-

ever, The Svedberg discovered that some proteins are

composed of more than one polypeptide chain. Subsequent

studies established that this is, in fact, true of most proteins,

including nearly all those with molecular masses ⬎100 kD.

Furthermore, these polypeptide subunits associate in a

geometrically specific manner. The spatial arrangement of

these subunits is known as a protein’s quaternary structure

(4° structure).

266 Chapter 8. Three-Dimensional Structures of Proteins

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 266

There are several reasons why multisubunit proteins are

so common. In large assemblies of proteins, such as colla-

gen fibrils, the advantages of subunit construction over the

synthesis of one huge polypeptide chain are analogous to

those of using prefabricated components in constructing a

building: Defects can be repaired by simply replacing the

flawed subunit rather than the entire protein, the site of

subunit manufacture can be different from the site of as-

sembly into the final product, and the only genetic infor-

mation necessary to specify the entire edifice is that speci-

fying its few different self-assembling subunits. In the case

of enzymes, increasing a protein’s size tends to better fix

the three-dimensional positions of the groups forming the

enzyme’s active site. Increasing the size of an enzyme

through the association of identical subunits is more effi-

cient, in this regard, than increasing the length of its

polypeptide chain since each subunit has an active site.

Additionally, in some multimeric enzymes, the active site oc-

curs at the interface between subunits where it is com-

prised of groups from two or more subunits. More impor-

tantly, however, the subunit construction of many enzymes

provides the structural basis for the regulation of their ac-

tivities. Mechanisms for this indispensable function are dis-

cussed in Sections 10-4 and 13-4.

In this section we discuss how the subunits of multisub-

unit proteins associate, what sorts of symmetries they have,

and how their stoichiometries may be determined.

A. Subunit Interactions

A multisubunit protein may consist of identical or non-

identical polypeptide chains. Recall that hemoglobin, for

example, has the subunit composition ␣

␤

. We shall refer

to proteins with identical subunits as oligomers and to

these identical subunits as protomers. A protomer may

therefore consist of one polypeptide chain or several unlike

polypeptide chains. In this sense, hemoglobin is a dimer

(oligomer of two protomers) of ␣␤ protomers (Fig. 8-64).

The association of two subunits typically buries 1000 to

2000 Å

of surface area (minimally ⬃600 Å

) that would

otherwise be exposed to solvent. The resulting contact re-

gions superficially resemble the interiors of single subunit

proteins: They contain closely packed nonpolar side

chains, hydrogen bonds, and in some cases, interchain disul-

fide bonds. However, protein–protein interfaces differ

from subunit interiors in several respects:

1. They tend to have hydrophobicities between those of

protein interiors and exteriors. In particular, the subunit in-

terfaces of proteins that dissociate in vivo have lesser hy-

drophobicities than do permanent interfaces.

2. An average of ⬃77% of intersubunit hydrogen

bonds are between side chains. In contrast, an average of

⬃68% of the hydrogen bonds within subunits are between

backbone atoms. This is mainly because secondary struc-

tural elements are not continued across subunit boundaries

(with the occasional exception of ␤ sheets; see below).

3. Around 56% of protein–protein interfaces contain

salt bridges.Th\ese contribute to the specificity as well as to

the stability of subunit associations.

In addition, there are negligibly few hydrogen bonds and

salt bridges at the edges of the contact regions. Not surpris-

ingly, the residues at protein–protein interfaces are evolu-

tionarily well conserved compared to other surface

residues.

B. Symmetry in Proteins

In the vast majority of oligomeric proteins, the protomers

are symmetrically arranged; that is, the protomers occupy

geometrically equivalent positions in the oligomer. This

implies that each protomer has exhausted its capacity to bind

to other protomers; otherwise, higher oligomers would

form.As a result of this limited binding capacity, protomers

pack about a single point to form a closed shell, a phenom-

enon known as point symmetry. Proteins cannot have in-

version or mirror symmetry, however, because such sym-

metry operations convert chiral

L-residues to D-residues.

Thus, proteins can only have rotational symmetry.

Section 8-5. Quaternary Structure 267

Figure 8-64 The quaternary structure of hemoglobin. The ␣

␣

, ␤

, and ␤

subunits in this stereo, space-filling drawing are

colored yellow, green, blue, and cyan, respectively. Heme groups

are red.The protein is viewed along its molecular 2-fold rotation

axis, which relates the ␣

␤

protomer to the ␣

␤

protomer.

Instructions for viewing stereo drawings are given in the

appendix to this chapter. [Based on an X-ray structure by Max

Perutz, MRC Laboratory of Molecular Biology, Cambridge, U.K.

PDBid 2DHB.]

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 267

Various types of rotational symmetry occur in proteins,

as X-ray crystal structure determinations have shown:

1. Cyclic symmetry

In the simplest type of rotational symmetry, cyclic sym-

metry, subunits are related (brought to coincidence) by a

single axis of rotation (Fig. 8-65a). Objects with 2-, 3-,...,

or n-fold rotational axes are said to have C

, C

,...,or C

symmetry, respectively. An oligomer with C

symmetry

consists of n protomers that are related by (360/n)° rota-

tions. C

symmetry is the most common symmetry in pro-

teins; higher cyclic symmetries are relatively rare.

A common mode of association between protomers

related by a twofold rotation axis is the continuation of a

␤ sheet across subunit boundaries. In such cases, the 2-fold

axis is perpendicular to the ␤ sheet so that two symmetry

equivalent ␤ strands hydrogen bond in an antiparallel fash-

ion. In this manner, the sandwich of two 4-stranded ␤ sheets

in a protomer of transthyretin (also known as prealbumin)

is extended across a 2-fold axis to form a sandwich of two

8-stranded ␤ sheets (Fig. 8-66). Hemoglobin’s two ␣␤ pro-

tomers are also related by C

symmetry (Fig. 8-64).

2. Dihedral symmetry

Dihedral symmetry (D

), a more complicated type of

rotational symmetry, is generated when an n-fold rotation

axis and a 2-fold rotation axis intersect at right angles

(Fig. 8-65b). An oligomer with (D

) symmetry consists of

2n protomers. D

symmetry is, by far, the most common

type of dihedral symmetry in proteins.

Hemoglobin’s ␣ and ␤ subunits have such similar

structures that, in the hemoglobin ␣

␤

tetramer, they are

related by pseudo-2-fold rotational axes that are perpen-

dicular to the tetramer’s exact 2-fold axis (lie in the plane

of Fig. 8-64; Section 10-2B). Hence the tetramer is said to

be pseudosymmetric and have pseudo-D

symmetry. The

X-ray structure of glutamine synthetase reveals that this

enzyme consists of 12 identical subunits that are related

by D

symmetry (Fig. 8-67).

Under the proper conditions, many oligomers with D

symmetry dissociate into two oligomers, each with C

sym-

metry (and which are related by the 2-fold rotation axes in

the D

oligomer).These, in turn, dissociate to their compo-

nent protomers under more stringent denaturing condi-

tions.

3. Other rotational symmetries

The only other types of rotationally symmetric objects

are those that have the rotational symmetries of a tetrahe-

dron (T ), a cube or octahedron (O), or a dodecahedron or

icosahedron (I ), and hence have 12, 24, and 60 equivalent

positions, respectively (Fig. 8-65c). Certain multienzyme

complexes are based on octahedral symmetry (Section

21-2A), whereas others have icosahedral symmetry (Sec-

tion 21-2Aa). The protein coats of the so-called spherical

viruses also have icosahedral symmetry (Section 33-2A).

268 Chapter 8. Three-Dimensional Structures of Proteins

(a) Cyclic symmetries

(b) Dihedral symmetries

Tetrahedral

symmetry

Octahedral (cubic)

symmetry

Icosahedral

symmetry

(c)

Figure 8-65 Some possible symmetries of

proteins with identical protomers. The lenticular

shape, the triangle, the square, and the pentagon at

the ends of the dashed lines indicate, respectively,

the unique 2-fold, 3-fold, 4-fold, and 5-fold

rotational axes of the objects shown. (a)

Assemblies with the cyclic symmetries C

, C

, and

.(b) Assemblies with the dihedral symmetries

, D

, and D

. In these objects, a 2-fold axis is

perpendicular to the vertical 2-, 4-, and 3-fold axes.

that the tetrahedron has some but not all of the

symmetry elements of the cube, and that the cube

and the octahedron have the same symmetry.

[Illustration, Irving Geis. Image from the Irving

Geis Collection, Howard Hughes Medical Insti-

tute. Reprinted with permission.]

See the

Animated Figures

JWCL281_c08_221-277.qxd 8/10/10 11:48 AM Page 268

a. Helical Symmetry

Some protein oligomers have helical symmetry (Fig.8-68).

The chemically identical subunits in a helix are not strictly

equivalent because, for instance, those at the ends of the

helix have a different environment than those in the mid-

dle. Nevertheless, the surroundings of all subunits in a long

helix, except those near its ends, are sufficiently similar that

the subunits are said to be quasi-equivalent. The subunits

of many structural proteins, for example, those of actin

(Section 35-3Ad) and tubulin (Section 35-3Gc), assemble

into fibers with helical symmetry.

Section 8-5. Quaternary Structure 269

Figure 8-66 A dimer of transthyretin as viewed down its

2-fold axis (black lenticular symbol). Each protomer, which is

colored in rainbow order from its N-terminus (blue) to its

C-terminus (red), consists of a ␤ barrel (really a ␤ sandwich)

containing two Greek keys (Fig. 8-50a). Note how both of its ␤

sheets are continued in an antiparallel fashion in the symmetry-

related protomer to form a sandwich of two 8-stranded ␤ sheets.

Two of these dimers associate back to back in the native protein

to form a tetramer with D

symmetry. [Based on an X-ray

structure by Colin Blake, Oxford University, U.K. PDBid 2PAB.]

Figure 8-67 X-ray structure of glutamine synthetase from

Salmonella typhimurium. The enzyme consists of 12 identical

subunits, here drawn in worm form, arranged with D

symmetry.

(a) View down the 6-fold axis of symmetry showing only the six

subunits of the upper ring in alternating blue and green.The

subunits of the lower ring are roughly directly below those of the

upper ring.The protein, including its side chains (not shown), has

a diameter of 143 Å.The six active sites shown are marked by

pairs of bound Mn

2⫹

ions (magenta spheres).Also drawn in one

active site are ADP (cyan) and the inhibitor phosphinothricin

(red). (b) Side view along one of the protein’s 2-fold axes

showing only its eight nearest subunits.The molecule extends 103

Å along the 6-fold axis, which is vertical in this view. [Based on

an X-ray structure by David Eisenberg, UCLA. PDBid 1FPY.]

(a)

(b)

Subunit

Helix segment

Helix

Figure 8-68 A helical structure composed of a single kind of

subunit.

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 269

b. Obtaining the Atomic Coordinates of a

Biologically Functional Quaternary Structure

Crystals consist of three-dimensional lattices of identi-

cal unit cells (the smallest portion of a crystal lattice that is

repeated by translation) that usually have internal symme-

try. The crystal’s asymmetric unit is the unique portion of

the unit cell from which the entire unit cell can be gener-

ated through the operation of its symmetry elements. In a

crystal of a symmetrical protein, if one or more of protein’s

symmetry axes are coincident with unit cell’s symmetry

axes, the asymmetric unit would contain a subset of the

protein’s protomers (often only one), which would be re-

lated to the other such subsets by crystallographic symme-

try. Alternatively, the asymmetric units in crystals of many

oligomeric proteins contain one or more entire proteins, in

which case their protomers are said to be related by non-

crystallographic symmetry.

A Protein Data Bank (PDB) coordinate file for an X-

ray crystal structure contains the atomic coordinates of the

protomers occupying an asymmetric unit. The entire crys-

tal structure can then be generated through the application

of its crystallographic symmetry. Thus, a symmetric pro-

tein’s PDB coordinate file might contain the coordinates of

only one of its several symmetry-related protomers. More-

over, in some cases, the intermolecular contacts in a crystal

may sufficiently resemble the contacts between the pro-

tomers in an oligomer so that its quaternary structure may

be ambiguous.

To alleviate these difficulties, computerized procedures

have been devised to generate the coordinates of the most

probable biologically functional molecule based on several

criteria, including maximizing the solvent-accessible sur-

face area that is buried on forming the oligomer. The

coordinates of the most probable quaternary structures

of macromolecules whose structures have been deter-

mined by X-ray crystallography are publicly available at

http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html. The bi-

ological unit may also be viewed directly from the corre-

sponding PDB Structure Summary page, although note

that the two algorithms do not always agree.

C. Determination of Subunit Composition

In the absence of an X-ray or NMR structure, the number

of different types of subunits in an oligomeric protein may

be determined by end group analysis (Section 7-1A). In

principle, the subunit composition of a protein may be de-

termined by comparing its molecular mass with those of its

component subunits. In practice, however, experimental

difficulties, such as the partial dissociation of a supposedly

intact protein and uncertainties in molecular mass determi-

nations, often provide erroneous results.

a. Cross-Linking Agents Stabilize Oligomers

A method for 4° structure analysis, which is especially

useful for oligomeric proteins that decompose easily, em-

ploys cross-linking agents such as dimethylsuberimidate or

glutaraldehyde (Fig. 8-69). If carried out at sufficiently low

protein concentrations to eliminate intermolecular reac-

tions, cross-linking reactions will covalently join only the

270 Chapter 8. Three-Dimensional Structures of Proteins

Figure 8-69 Cross-linking agents. Dimethylsuberimidate and glutaraldehyde are bifunctional

reagents that react to covalently cross-link two Lys residues.

(CH

)

2CH

C (CH

)

(CH

)

NH (CH

)

(CH

)

(CH

)

(CH

)

(CH

)

CHNN

⫹

C (CH

)

(CH

)

⫹

2 Subunit

Subunit

Subunit Subunit

Subunit

Lys Dimethylsuberimidate

Lys Glutaraldehyde

JWCL281_c08_221-277.qxd 2/23/10 1:59 PM Page 270