Chapter 10

Membrane Structure
THE LIPID BILAYER
DEFINITIONS
101 102 103 104 105 106 107 108 Black membrane Liposome Hydrophobic Lipid raft Ganglioside Amphiphilic Phosphoglyceride Cholesterol

10
In This Chapter
THE LIPID BILAYER MEMBRANE PROTEINS A221 A227

TRUE/FALSE
109 True. The hydrophobic interior of the lipid bilayer acts as a barrier to the passage of the hydrophilic lipid head groups that must occur during flipflop. The energetic cost of this movement effectively prevents spontaneous flip-flop of lipids, so that it occurs extremely rarely in the absence of specific catalysts known as phospholipid translocators. True. The positively charged moieties in all cases are balanced by the negative charge on the phosphate group; thus, none of the common phospholipids carries a net positive charge. True. Glycolipids are synthesized in the lumen of the Golgi apparatus, which is topologically equivalent to the outside of the cell, and cannot flip-flop across the bilayer.

1010

1011

THOUGHT PROBLEMS
1012 Water is a liquid, and thus hydrogen bonds between water molecules are not static; they are continually formed and broken again by thermal motion. When a water molecule happens to be next to a hydrophobic solute, it is more restricted in motion and has fewer neighbors with which it can interact because it cannot form any hydrogen bonds in the direction of the hydrophobic solute. It will therefore form hydrogen bonds to the more limited number of water molecules in its proximity. Bonding to fewer partners results in a more ordered water structure, which constitutes the icelike cage in Figure 101. The true cage of water molecules is substantially different from that represented in the figure; the real cage exists in three dimensions, forming a

A221

A222

Chapter 10: Membrane Structure

pentagonal dodecahedron (like a soccer ball) or clusters of them that enclose the hydrophobic solute. The structure is similar to ice, although it is a more transient, less organized, and less extensive network than even a tiny ice crystal. The formation of any ordered structure decreases the entropy of the system, which is energetically unfavorable.

1013

The same forces that dictate that certain lipids will form a bilayer, as opposed to micelles, operate in the repair of a tear in the bilayer. The tear will heal spontaneously because a bilayer is the most energetically favorable arrangement. The lipids that make up a bilayer are cylindrical in shape and therefore do not readily form a micelle (or a hemi-micelle), which would require cone-shaped lipids. Lipid bilayers assemble because the surrounding water molecules exclude the component lipids; thus, analogy (2) is the correct one. If bilayers formed because of attractive forces among the lipidsanalogy (1)the properties of the bilayer would likely be quite different. Molecules attract one another by forming specific bonds that hold them together. Such bonding among lipids would make the bilayer less fluid, perhaps even rigid, depending on the strength of the interaction.

1014

1015 A. With one tail removed, the diameter of the head group would be much larger than that of the remaining hydrocarbon chain; thus, the shape of the molecule would resemble a cone more than a cylinder and would tend to form micelles rather than bilayers. B. Bilayers formed by lipids with short hydrocarbon tails would be much more fluid. The bilayers would also be less stable, since the shorter tails would be less hydrophobic and, thus, the forces that drive formation of the bilayer would be reduced. C. Bilayers formed by lipids with saturated hydrocarbon tails would be much less fluid. Whereas a normal lipid bilayer has the viscosity of olive oil, a bilayer made of lipids with saturated hydrocarbon tails would have the consistency of bacon fat. D. Bilayers formed by lipids with unsaturated hydrocarbon tails would be much more fluid. Also, because the lipids would pack together less well, there would be more gaps and the bilayer would be more permeable to small water-soluble molecules. E. If lipids with two saturated hydrocarbon tails were completely intermixed with lipids carrying two unsaturated hydrocarbon tails, the fluidity of the membrane would be about normal. In such bilayers, the saturated lipid molecules would tend to aggregate with one another because they can pack so much more tightly and would therefore form patches with much reduced fluidity. The bilayer would not, therefore, have uniform properties over its surface. Most phospholipids in normal membranes have one saturated and one unsaturated hydrocarbon chain so they do not tend to segregate; however, sphingolipids often do have long, saturated hydrocarbon chains and do tend to form microdomains, termed lipid rafts. F. If the hydrocarbon tails of the lipids in the two monolayers were covalently linked, the bilayer would have virtually unchanged properties. Each lipid molecule would now span the entire membrane, with one of its two head groups exposed at each surface. Such lipid molecules are actually found in the membranes of thermophilic bacteria, which can live at temperatures approaching that of boiling water. Their bilayers do not come apart at elevated temperatures, as usual bilayers do, because the two monolayers are covalently linked into a single membrane. 1016 1017 In a two-dimensional fluid the molecules are constrained to move in a plane; the molecules in a normal fluid can move in three dimensions. Vegetable oil is converted to margarine by reduction of double bonds (by hydrogenation), which converts unsaturated fatty acids to saturated ones.

THE LIPID BILAYER

This change allows the fatty acid chains in the lipid molecules to pack more tightly against one another, increasing the viscosity, turning oil into margarine. 1018 C. Phosphatidylinositol is a minor component of the phospholipids in the plasma membrane, yet it plays a very important role in cell signaling. 1019 A. Antarctic fish, which are cold-blooded, live in freezing waters. In order to maintain an appropriate fluidity of their membranes under such extreme conditions, they require a higher proportion of unsaturated fatty acid chains in their membranes. Polar bears also live in extreme cold, but they are warm-blooded and maintain a high internal temperature; thus, they have no special requirement for unsaturated fatty acids in their membranes. 1020 The large head groups of most sphingolipids prevent the close packing of their fatty acid tails in the membrane. The planar cholesterol molecules are postulated to fill the voids that form underneath the large head groups of the sphingolipids, thereby packing tightly against fatty acid chains that could not otherwise approach closely enough to bind. Reference: Harder T & Simons K (1997) Caveolae, DIGs, and the dynamics of sphingolipid-cholesterol microdomains. Curr. Opin. Cell Biol. 9, 534542. 1021 The size of a lipid raft depends on the affinity of the sphingolipids and cholesterol molecules for one another. If they bound one another sufficiently tightly, they would aggregate into a single domain in the membrane. If they bound one another with the same affinity as they bind to other species of lipid molecules, they would remain dispersed. The small size of the lipid rafts indicates that sphingolipids and cholesterol molecules have only a slightly higher affinity for one another than for other lipids. Presumably, at this size the aggregated sphingolipids and cholesterol molecules are in equilibrium with their free forms, so that lipids are added to and leave a raft at equal rates. Lipid rafts tend to be thicker than other parts of the bilayer because the fatty acid chains attached to sphingolipids are longer (20 to 26 carbon atoms versus 16 to 22 for other lipids) and have fewer double bonds (0.1 to 0.4 per sphingolipid versus 1.1 to 1.5 for other lipids). The combination of longer and straighter fatty acid tails give rise to a thicker bilayer in the region of the raft. It is not a paradox. The fluidity of the bilayer is strictly confined to one plane. The lipid molecules can diffuse laterally, but do not readily flip from one monolayer to the other. Specific types of lipid molecule remain in the monolayer they are inserted into, unless they are actively transferred by an enzymea phospholipid translocator (a flippase). The redistribution of phosphatidylserine from the cytoplasmic to the outer monolayer of the plasma membrane occurs by two mechanisms: (1) the phospholipid translocators that normally transport this lipid from the noncytoplasmic monolayer to the cytoplasmic monolayer are inactivated in apoptotic cells; and (2) a scramblase that transfers phospholipid nonspecifically in both directions between the two monolayers is activated.

A223

1022

1023

1024

CALCULATIONS
1025 A. When lined up, it would take 4000 lipid molecules to reach from one end of a bacterium to the other (2 mm/0.5 nm = 4000). If a lipid molecule at one end moved directionally by exchanging places with its immediate neighbor

A224

Chapter 10: Membrane Structure

down the line every 107 seconds, it would reach the other end in 4 104 seconds (4000 107 seconds). This rate is some 2500 times faster than the measured rate of about 1 second. These numbers do not agree because lipid molecules do not diffuse in a straight line; they move along random paths so that it takes much longer to travel from one end of the cell to the other. B. If a 4-cm ping-pong ball exchanged places with a neighbor every 107 seconds, it would travel at a speed of 1,440,000 km/hr [(4 cm/107 sec) (km/105 cm) (3600 sec/hr)]. If its movement were only in one direction, it would reach the other wall in 1.5 105 seconds [6 m (107 sec/0.04 m)]; in a random walk it would take considerably longer.

1026

A raft 70 nm in diameter would have an area of 3.8 103 nm2 (3.14 352), and a lipid molecule 0.5 nm in diameter would have an area of 0.20 nm2 (3.14 0.252). Thus, there would be about 19,000 lipid molecules per monolayer of raft (3.8 103/0.20 = 19,000), and about 38,000 molecules in the raft bilayer. At a ratio of 50 lipids per protein, a raft would accommodate about 760 protein molecules. The true ratio of lipids to proteins in a raft is unknown.

DATA HANDLING
1027 A. The phospholipase cleaves phosphatidylcholine into phosphorylcholine and diacylglycerol by breaking the ester bond that links the phosphate group to glycerol. This phospholipase is like phospholipase C in its specificity (Figure 1013). Several other types of phospholipase are indicated in the figure: phospholipase A1 and A2 remove fatty acid chains from specific positions in glycerol, and phospholipase D removes the head group but leaves the phosphate group still attached to glycerol. These enzymes are extremely useful tools for lipid analysis. Several come from the venoms of snakes. B. It may strike you as odd that removing the hydrophilic head group from phosphatidylcholine causes complete breakdown of the red cell membrane, whereas removing the exterior sialic acid groups or the external protein domains does not. After all, about half of the membrane is protein, and about a third of the outer lipid monolayer is phosphatidylcholine (but almost none is present in the inner monolayer). Removing the choline head groups destabilizes the lipid bilayer by converting a large number of the constituent phospholipids, which are amphiphilic, to diacylglycerols, which are almost entirely nonpolar. This result emphasizes that the lipid bilayer is crucial to the integrity of the membrane and further stresses that a hydrophilic head group is essential if a lipid is to form a stable bilayer. (Not all phospholipase C enzymes cause hemolysis, by the way. Some cannot act on the phospholipids in intact membranes, presumably because they cannot reach the sensitive bonds.) 1028 A. The difference in rate of loss of the ESR signals is due to the location of the nitroxide radical on the two phospholipids. The nitroxide radical in phospholipid 1 is on the head group and is therefore in direct contact with the external medium. Thus, it can react quickly with ascorbate. The nitroxide radical in phospholipid 2 is attached to a fatty acid chain and is therefore
A1 H O R2 C H O H A2 C C C H C O H O O C O P O D O choline R1 C A1 D A2

Figure 1013 Cleavage specificity of several phospholipases (Answer 1027). Susceptible bonds are indicated by arrows. Letters indicate specific phospholipases: A1 is phospholipase A1; A2 is phospholipase A2; C is phospholipase C; and D is phospholipase D.

THE LIPID BILAYER

partially buried in the interior of the membrane. As a consequence, it is less accessible to ascorbate and is reduced more slowly. B. The key observation is that the extent of loss of ESR signal in the presence and absence of ascorbate is the same in resealed red cell ghosts, but different in red cells. These results suggest that there is an undefined reducing agent in the cytoplasm of red cells (which is absent from red cell ghosts). Like ascorbate, this cytoplasmic agent can reduce the more exposed phospholipid 1, but not the less exposed phospholipid 2. Thus, in red cells, phospholipid 2 is stable in the absence of ascorbate; in the presence of ascorbate, the spin-labeled phospholipids in the outer monolayer are reduced, causing loss of half the ESR signal. Phospholipid 1, on the other hand, is not stable in red cells in the absence of ascorbate because the phospholipids in the cytoplasmic monolayer are exposed to the cytoplasmic reducing agent, which destroys half the ESR signal. When ascorbate is added, labeled phospholipids in the outer monolayer are also reduced, causing loss of the remaining ESR signal. C. The results in Figure 104 indicate that the labeled phospholipids were introduced equally into the two monolayers of the red cell plasma membrane. Phospholipid 2 was 50% sensitive to ascorbate, indicating that half the label was present in the outer monolayer, and 50% insensitive to ascorbate, indicating that half was present in the cytoplasmic monolayer. Similarly, phospholipid 1 was 50% sensitive to the cytoplasmic reducing agent and 50% sensitive to ascorbate, indicating an even distribution between the cytoplasmic and outer monolayers. Reference: Rousselet A, Guthmann C, Matricon J, Bienvenue A & Devaux PF (1976) Study of the transverse diffusion of spin labeled phospholipids in biological membranes: 1. Human red blood cells. Biochim. Biophys. Acta 426, 357371. 1029 A. Only phosphatidylserine and phosphatidylethanolamine have primary amino groups that can react with SITS. Since these phospholipids are labeled only when the red cells are made permeable (ghosts), they presumably reside in the cytoplasmic monolayer. This conclusion is supported by the results from experiments with sea snake venom, which degrades phosphatidylserine and phosphatidylethanolamine only in the permeable ghosts. These results, taken together, indicate that the phosphatidylserine and phosphatidylethanolamine are localized almost exclusively in the cytoplasmic monolayer of red cell membranes. Phospholipase degradation of phosphatidylcholine and sphingomyelin in intact red cells indicates that they are present in the outer monolayer. This conclusion depends on the red cells remaining intact during the treatment. In the case of sea snake venom, the absence of degradation of phosphatidylserine and phosphatidylethanolamine in intact red cells provides an internal control. In the case of sphingomyelinase, there is no internal control, but the absence of lysis indicates that the membrane is intact. The results in Table 102 do not exclude the possibility that phosphatidylcholine and sphingomyelin are also located in the cytoplasmic monolayer. Since most of the sphingomyelin is degraded by sphingomyelinase, most of it must be localized in the outer monolayer of the membrane. No such data are provided for phosphatidylcholine; thus, it would be incorrect to conclude that phosphatidylcholine is located exclusively in the outer monolayer. Other experiments not reported here do suggest, however, that phosphatidylcholine is found almost entirely in the outer monolayer. B. You chose red cells for these experiments because they contain no internal membranes. If the same experiments were performed on cells with internal membranes, it would have been impossible to measure the phospholipid composition of the cytoplasmic monolayer directly, since phospholipids from the cytoplasmic monolayer would have been hopelessly confused with those from internal membranes.

A225

A226

Chapter 10: Membrane Structure

References: Bretscher M (1972) Asymmetrical lipid bilayer structure for biological membranes. Nat. New Biol. 236, 1112. Deenen LLM & DeGier J (1974) Lipids of the red cell membrane. In The Red Blood Cell (D. MacN. Surgenor, ed), pp 147211. New York: Academic Press.

1030 A. PLAP is soluble in ice-cold Triton X-100 until 20 to 40 minutes after its synthesis. This time delay, coupled with the shift in mobility by SDS-gel electrophoresis, suggests that at 20 to 40 minutes PLAP is converted to its GPIanchored form, which migrates differently from the unmodified form. This is consistent with what is known about synthesis of GPI-anchored proteins: in the ER they are cleaved from a transmembrane anchor and attached to GPI. B. When fully solubilized in octyl glucoside, PLAP bands at the bottom of the sucrose density gradient, which is at the density characteristic of proteins. In its insoluble Triton X-100 form, PLAP bands at a less dense position, indicating it is complexed with molecules such as lipids, which are much less dense than proteins. C. The lipid composition of the low-density, insoluble form of PLAP is just what you would expect for a lipid raft, which is enriched in sphingolipids and cholesterol and depleted for phospholipids (other than sphingomyelin). Indeed, it was experiments such as this one that established the composition of lipid rafts. It remains controversial whether these experiments give information about microdomains in membranes or are artifactual results due to the method of extraction (ice-cold Triton X-100). D. The surface area of the smallest vesicles (100 nm diameter) is a little over 30,000 nm2 [4 3.14 (502)], whereas that of a lipid raft (70 nm diameter) is a little under 4000 nm2 [3.14 (352)]. Although the presence of PLAP in the vesicles would be expected if they were derived from lipid rafts, the difference in surface area is troubling because it suggests that multiple lipid rafts may have coalesced during treatment with ice-cold Triton X-100. That possibility is hard to distinguish from the coalescence of well-dispersed lipids and PLAP during treatment with Triton X-100. It is this possibility of artifact that has spurred attempts to demonstrate lipid rafts directly in the plasma membranes of living cells. Reference: Brown DA & Rose JK (1992) Sorting of GPI-anchored proteins to glycolipid-enriched membrane subdomains during transport to the apical cell surface. Cell 68, 533544. 1031 A. For randomly dispersed receptors (see Figure 106A), the polarization of the fluorescent light will depend critically on the concentration of the receptors in the membrane. At high density (pixels with high-intensity fluorescence) there will be efficient FRET (high-intensity fluorescence detected by the perpendicular filter), giving rise to a low value for polarization of the fluorescence [(Ipar Iperp)/(Ipar + Iperp)]. By contrast, at low density the overall intensity will be lower, but FRET will be much less efficient since molecules are on average farther away from one another. As a result, what fluorescence there is will be more polarized. For receptors that are confined to microdomains such as lipid rafts (see Figure 106B), the overall fluorescence intensity will decrease with decreasing density of the rafts, which is determined just by chance distribution of rafts relative to the very small window (a pixel) being examined. The polarization of the fluorescence, however, will be independent of concentration. At high density and at low density of rafts, the receptors in microdomains will always be equally close to their neighbors; thus, a constant proportion of the absorbed light energy will be transferred by FRET. As a result, receptors in microdomains will give the same low value for polarization of fluorescence, regardless of the total fluorescence intensity in a pixel. B. The results suggest that transmembrane-anchored folate receptors are randomly dispersed in the membrane, while GPI-anchored receptors are

MEMBRANE PROTEINS
clustered in microdomains. Although such microdomains are likely to be lipid rafts, these experiments do not prove that point. C. Both types of receptor behave as if randomly dispersed in cells grown in the presence of compactin, which reduces the amount of cholesterol in the membrane by blocking its synthesis. These observations suggest more strongly that the microdomains identified in part B were indeed lipid rafts, which are known to require cholesterol for their formation. Reference: Varma R & Mayor S (1998) GPI-anchored proteins are organized in submicron domains at the cell surface. Nature 394, 798801. 1032 A. The half-time for flip-flop in these experiments is the point at which 50% of the ESR signal is lost. For cells labeled in the cytoplasmic monolayer, the curve in Figure 107 suggests a half-time for flip-flop of about 7 hours. For cells labeled in the outer monolayer, the half-time of flip-flop is much longer but cannot be estimated reliably. These data indicate that the rate of flip-flop of phospholipids between the two monolayers of the plasma membrane in red cells is extremely low. Similar experiments using synthetic bilayers have given even longer times; in fact, in the best experiments, when great care was taken not to allow oxidation or other damage to the lipids, the rate of flip-flop was immeasurably low (less than once per month). B. Phospholipid 2 was used to label the cytoplasmic monolayer, and phospholipid 1 was used to label the outer monolayer. As shown by the experiments in Figure 104B, phospholipid 2 in the cytoplasmic monolayer is not reduced by the cytoplasm of red cells; when it is present in the outer monolayer, it can be reduced by ascorbate. Thus, phospholipid 2 is appropriate for measuring the rate of flip-flop from the cytoplasmic to the outer monolayer. As shown by the experiments in Figure 104A, phospholipid 1 in the cytoplasmic monolayer is reduced by red cell cytoplasm, but it is stable in the outer monolayer in the absence of ascorbate. Thus, phospholipid 1 is appropriate for measuring the rate of flip-flop from the outer to the cytoplasmic monolayer. C. One can make intact red cells with spin-labeled phospholipids exclusively in the cytoplasmic monolayer by introducing phospholipid 2 into the membrane and then incubating the red cells for 1 hour in the presence of ascorbate. Ascorbate reduces the lipids in the outer monolayer, leaving red cells that are labeled only in the cytoplasmic monolayer. Similarly, one can make intact red cells with spin-labeled phospholipids exclusively in the outer monolayer by introducing phospholipid 1 into the membrane and then incubating the red cells for 15 minutes in the absence of ascorbate. In this case the spin-labeled lipids in the cytoplasmic monolayer are reduced by agents in the cytoplasm, leaving red cells that are labeled only in the outer monolayer. Reference: Rousselet A, Guthmann C, Matricon J, Bienvenue A & Devaux PF (1976) Study of the transverse diffusion of spin labeled phospholipids in biological membranes: 1. Human red blood cells. Biochim. Biophys. Acta 426, 357371.

A227

MEMBRANE PROTEINS
DEFINITIONS
1033 1034 1035 Lectin Carbohydrate layer Spectrin

A228
1036 1037 1038 1039 1040

Chapter 10: Membrane Structure

Multipass transmembrane protein Bacteriorhodopsin Cortex Peripheral membrane protein Glycosylphosphatidylinositol (GPI) anchor

TRUE/FALSE
1041 True. The lipid bilayer defines the structure of the membrane and provides a permeability barrier that separates the inside from the outside of the cell. Specific membrane proteins allow particular solutes to enter and leave the cell, bind signaling molecules, and mediate attachment to the extracellular matrix. False. While it is true that hydropathy plots are useful for identifying a-helical transmembrane segments in proteins, they cannot identify transmembrane b strands. a-Helical transmembrane segments can be reliably identified because they involve 2030 amino acids; transmembrane b strands are more difficult to identify because they involve only about 10 amino acids, which is insufficient to distinguish b strands in the interior of proteins from those that cross a membrane. False. The carbohydrate on internal membranes is directed away from the cytosol toward the lumen of an internal membrane-enclosed compartment. Remember that the lumen of an internal compartment is topologically equivalent to the outside of the cell. False. Human red blood cells contain no internal membranes at all; at an early stage in their development they extrude their nuclei. The lack of any internal membranes is the principal reason they have been used so extensively to investigate the structure of the plasma membrane. True. The conformational changes that cause protons to be pumped are initiated by a change in shape in the retinal chromophore upon absorption of a photon of light. False. In addition to lipid rafts, which are microdomains with distinct lipid compositions, the apical and basolateral surfaces of epithelial cells, which are separated by intercellular tight junctions, also have different lipid compositions.

1042

1043

1044

1045

1046

THOUGHT PROBLEMS
1047 Thus far, arrangements A, B, D, E, F, and I have been found in biological membranes. Arrangement C, which has carbohydrate on the cytoplasmic side of the membrane, does not seem to exist. Arrangements G and H, which show proteins completely buried or with only their tips embedded in the membrane, have not been found and are thought unlikely to occur on theoretical grounds. An arrangement similar to H, but with an amphiphilic a helix inserted horizontally into the cytoplasmic monolayer, has been found. The principles are the same for both. Exposure of hydrophobic amino acid side chains to water is energetically unfavorable. There are two ways that such side chains can be sequestered away from water to achieve an energetically more favorable state. One, they can form a-helical transmembrane segments that span a lipid bilayer, so that the hydrophobic amino acid side chains can interact with the hydrophobic lipids in the membrane. Two, the hydrophobic amino acid side chains can be sequestered in the interior of the folded polypeptide chain, where they can interact with each other.

1048

MEMBRANE PROTEINS
1049 D. The mass ratio depends on the membrane. In the myelin membrane around nerve-cell axons, proteins account for less than 25% of the mass. In mitochondrial cytoplasmic membranes, proteins account for about 75% of the total mass. In typical plasma membranes the masses of proteins and lipids are about the same. 1050 Fatty acid chains, prenyl groups, and glycosylphosphatidylinositol (GPI) anchors are the three common lipid anchors for membrane proteins.

A229

1051 A. Sequence A is the actual membrane-spanning a-helical segment of glycophorin, a transmembrane protein from red blood cells. It is composed predominantly of hydrophobic amino acids, although it does contain the uncharged polar amino acids threonine (T) and serine (S), which are not uncommon in membrane-spanning a helices. Sequence B is unlikely to be a membrane-spanning segment because it contains three prolines (P), which would disrupt an a helix and thereby expose polar groups to the hydrophobic environment of the lipid bilayer. Sequence C is also unlikely to be a transmembrane segment because it contains three charged amino acids, glutamic acid (E), arginine (R), and aspartic acid (D), whose presence in the hydrophobic lipid bilayer would be energetically unfavorable. 1052 The hydrophilic faces of the five membrane-spanning a helices, each contributed by a different subunit, can come together to form a hydrophilic pore across the lipid bilayer that is lined with the hydrophilic amino acid side chains (Figure 1014). The hydrophobic side chains on the opposite sides of the a helices can then interact with the hydrophobic lipid tails in the bilayer.

1053 A. In a b strand, adjacent amino acid side chains protrude from opposite sides of the strand; thus, every other amino acid side chain will face the same side of the strand. If a b strand is part of a b-barrel pore, its amino acid side chains will alternate between hydrophobic and hydrophilic, so that one side of the strand will be hydrophobic and the other side will be hydrophilic. Only choice A has alternating hydrophobic and hydrophilic amino acids. 1054 In both an a helix and a b barrel the polar hydrogen-bonding groups in the peptide bond are fully satisfied by internal hydrogen bonds with groups in other peptide bonds. These internal hydrogen bonds dictate the secondary structures known as a helices and b sheets (or b barrels when the edges of a sheet pair to complete the cylinder). By contrast, in a disordered chain the polar groups in the peptide bonds are not involved in bonding to one another. Such disordered segments can exist in proteins because hydrogen bonds can be made with water molecules or to other polar groups in the protein. In a membrane, however, the hydrophobic hydrocarbon chains of the bilayer provide no hydrogen-bonding partners. As a result, a disordered peptide chain in a membrane is energetically very unfavorable. Transmembrane a helices are thought to be more common than transmembrane b barrels because they provide a more flexible arrangement of transmembrane segments. Because a helices can slide against one another, they allow the protein to undergo conformational changes that can be exploited to gate ion channels, transport solutes, or transmit information across the bilayer. By contrast, the b strands in a b barrel are rigidly fixed to their neighbors by hydrogen bonds that lock the protein into a single conformation. Your friends suggestion is based on an important difference between insideout and right-side-out vesicles. The contaminating right-side-out vesicles will carry carbohydrate on their exposed surface and should therefore be

hydrophilic pore hydrophilic face lipid bilayer

1055

hydrophobic face

1056

Figure 1014 Proposed structure for a hydrophilic pore formed by five membrane-spanning a helices (Answer 1052).

A230

Chapter 10: Membrane Structure

retained on a lectin affinity column. Inside-out vesicles, by contrast, will lack carbohydrate on their exposed surface and should therefore pass through the column.

1057

The environment in the cytosol is sufficiently reducing that disulfide bonds rarely form, whereas outside the cell (and in internal lumens) the environment is sufficiently oxidizing that disulfide bonds form readily. The sulfate group in SDS is charged and therefore hydrophilic. The OH group and the COC groups in Triton X-100 are polar; they can form hydrogen bonds with water and are therefore hydrophilic. The gray portions of these detergents are either hydrocarbon chains or aromatic rings, neither of which have polar groups that can hydrogen bond to water molecules; they are therefore hydrophobic. Membrane proteins anchor the lipid bilayer to the cytoskeleton, which strengthens the plasma membrane so that it can withstand the shear forces red blood cells are subjected to when they are pumped through small blood vessels. Membrane proteins also transport nutrients and ions across the plasma membrane. Normally the cytosol is sufficiently reducing that it contains no disulfide bonds, even in G6PD-deficient individuals. However, in G6PD-deficient individuals who eat fava beans, the cytosol of red cells can become sufficiently oxidizing that disulfide bonds form. Since all such bonds are inappropriate in the cytosol, they link proteins in ways that were never intended, leading to clumps and aggregates that stick to the cell membrane. The resulting distorted shape of the cell serves as a signal to the spleen to remove the damaged cells from circulation, leading to the severe anemia. Although this question was framed in an ancient context, the problem is ongoing. For centuries, schoolteachers on the Mediterranean island of Sardinia have witnessed a curious phenomenon. Every February as Spring arrives, some of their students (mostly boysthe G6PD gene is on the X chromosome) suddenly seem drained of energy. For the next three months, their schoolwork suffers, and they complain of dizziness and nausea and fall asleep at their desks. Then, just as suddenly, they return to normal and remain healthy and active until the next February rolls around. In children it might be brushed off as spring fever, but Sardinian adults (mostly males) suffer the same symptoms. It was during the Korean War that the connection was made between the Mediterranean form of G6PD deficiency and the hemolytic effects of antimalarial drugs, which, like the substance in fava beans, are oxidizing agents. The collapse of some soldiers, who were given such prophylactic drugs, led to a detailed investigation of the problem. Transmembrane domains that are composed entirely of hydrophobic amino acid side chains obviously cannot interact with one another via hydrogen bonds or electrostatic attractions, two of the more important ways to link proteins together noncovalently. Nevertheless, they can interact specifically via van der Waals attractions. If their surfaces were complementary, they could fit together well enough to make a large number of van der Waals contacts, which would hold them together. It should be noted that the transmembrane segment of glycophorin contains a few polar amino acids that may participate in the dimerization process. Proteins can be restricted to specific regions of the plasma membrane in several ways: by attachment to extracellular or intracellular proteins, by attachment to proteins in other cells, and by molecular fences that in some way corral proteins in specific membrane domains. The fluidity of the lipid bilayer is not significantly affected by the anchoring of membrane proteins; the lipid molecules flow around anchored proteins like water around rocks in a stream.

1058

1059

1060

1061

1062

MEMBRANE PROTEINS
1063 Most of the carbohydrate in the carbohydrate layer is attached to intrinsic plasma membrane molecules. The carbohydrate layer typically contains glycoproteins and proteoglycans that have been secreted into the extracellular space and then adsorbed on the cell surface. Many of the adsorbed macromolecules are components of the extracellular matrix. So, are these bridging molecules part of the plasma membrane, or part of the extracellular matrix? This is the ambiguity that makes where the plasma membrane ends and the extracellular matrix begins largely a matter of semantics.

A231

CALCULATIONS
1064 Using these average molecular weights, there are 96 lipid molecules (phospholipid + cholesterol) for every protein molecule {[(50,000/800) + (50,000/386)]/2 = 96}. A similar lipid-to-protein ratio is present in many cell membranes.

1065 A. The calculation for the number of spectrin molecules per red blood cell is shown in detail below. In essence, one first calculates the fraction of total protein that is spectrin and then converts that number into the number of spectrin molecules using the molecular weight of spectrin and Avogadros number: spectrin 5 mg protein 0.25 spectrin mmol spectrin 6 1020 molecules = total protein 250,000 mg mmol spectrin cell 1010 cells = 3 105 spectrin molecules/cell The analogous calculation gives values of 9 105 molecules of band 3 and 2.3 105 molecules of glycophorin per red cell. The calculated number of glycophorin molecules per cell is too low by a factor of 2.5 because about 60% of the molecular weight of glycophorin is carbohydrate, which is not stained by Coomassie Blue. B. The fraction of the plasma membrane that is occupied by band 3 is the area of the face of a single band 3 molecule (pr2), times the total number of band 3 molecules per cell (9 105), divided by the total area of the red blood cell (108 nm2). Note that the height of the molecule is irrelevant to the calculation. band 3 3.14 (3 nm)2 9 105 molecules 1 cell = plasma membrane molecule cell 108 nm2 = 0.25 Thus band 3 occupies about 25% of the surface area of a red blood cell. This result is consistent with freeze-fracture electron micrographs of red blood cells, which show a high density of intramembrane particles that are thought to be dimers of band 3.

DATA HANDLING
1066 A. Elimination of sialic acid staining by sialidase treatment indicates that carbohydrate is exposed on the external surface. Because the carbohydrate is attached to glycophorin, it follows that glycophorin is also exposed on the external surface. This conclusion is supported by the results with pronase digestion, which eliminates sialic acid staining, presumably by clipping the peptide backbone. The results with pronase digestion indicate that band 3 is exposed to the external surface as well. In this case the appearance of the new protein band at about 70,000 daltons allows you to estimate that at least 30,000 daltons of band 3 are exposed on the external surface. In neither

A232

Chapter 10: Membrane Structure

digestion were the two spectrin bands affected, suggesting that spectrin is not exposed on the external surface. B. One direct experimental approach for testing your colleagues objection is to break open the red cell ghosts before digesting them with pronase. If spectrin were resistant to pronase, its mobility on SDS polyacrylamide gels would be unaltered. If spectrin were located on the cytoplasmic surface, its mobility would be altered dramatically. These control experiments have been done; they show that spectrin is sensitive to pronase. Another approach is to make inside-out ghosts and see if it is possible to dissociate spectrin from the membrane by treatments that do not disrupt the membrane. This approach also has been successful, confirming that spectrin is on the cytoplasmic side and is not embedded in the membrane. C. To determine which of the red cell proteins span the membrane using this basic experimental approach, it is necessary to prepare inside-out vesicles. Such vesicles can be prepared by disrupting red cell ghosts and allowing them to reseal under defined ionic conditions. When inside-out vesicles are treated with pronase, the mobilities of band 3 and glycophorin are altered. These results, along with the results above, indicate that band 3 and glycophorin are exposed on both surfaces of the red cell membrane and must therefore be transmembrane proteins. References: Bennett V & Stenbuck PJ (1979) The membrane attachment protein for spectrin is associated with band 3 in human erythrocyte membranes. Nature 280, 468473. Bennett V & Stenbuck PJ (1980) Association between ankyrin and the cytoplasmic domain of band 3 isolated from the human erythrocyte membrane. J. Biol. Chem. 255, 64246432.
spectrin band 3 actin

ankyrin band 3 100 nm

Figure 1015 Meshwork of protein interactions on cytoplasmic surface of red cells (Answer 1067).

1067

The presence of both proteins in the pellet, as in mixture 3, 4, and 6 in Table 104, indicates an interaction between the proteins. The results with pairwise mixtures suggest that the four proteins are arranged as shown below: band 3ankyrinspectrinactin An artists conception of how these molecules are linked together to form the supporting meshwork on the cytoplasmic surface of red cells is shown in Figure 1015. In reality, the interaction between actin and spectrin is too weak to be detected by this method, unless a third protein, band 4.1, is added.

1068

There are two populations of band 3 proteins in the red cell membrane: one population is immobilized by attachment to the spectrin-based cytoskeleton; the other is freely mobile. Only the freely mobile population will be able to diffuse into the bleached spot and contribute to recovery of fluorescence. Thus, the curve for recovery of fluorescence will reach a plateau below the original level of fluorescence (Figure 1016). The extent of recovery will correspond to the proportion of band 3 proteins that are freely mobile.

BLEACH

fluorescence

RECOVERY

time

Figure 1016 Recovery of fluorescence after photobleaching of band 3 (Answer 1068).