ANNUAL REVIEWS
Further
Quick links to online content Ann. Rev. Biochem. 1978. 47:359- and the values of T2, are affected by the structural properties of the sample on a molecular level and by molecular motions. The most important magnetic nuclei for membrane studies are IH, 2H, I3C, and 3 1p. The predominant isotopes for hydrogen and phosphorus are IH and 3 1p, thus these isotopes appear in great abundance in normal biologi cal samples. Only about 1 % of the naturally occurring carbon nuclei are I3C. Nevertheless, NMR is sensitive enough to make the technique useful for this isotope. Deuterium can be introduced by synthetic techniques into specific parts of molecules as a substitute for hydrogen. Also molecules containing the magnetic nucleus 19p have been used as probes of membrane structure. [For reviews of NMR applied to membranes, see (38, 39).] Factors that affect NMR absorption spectra The various magnetic nuclei have very different precession frequencies for any particular magnetic field strength. Thus the absorption lines for different nuclei are well separated from each other. There are three important interactions that determine NMR line shapes for any one type of nucleus. The first is the anisotropic chemical shift. The electronic cloud in the vicinity of a nucleus acts to shield the nucleus somewhat from the external magnetic field. The extent of the shielding depends on the orientation of the molecule with respect to the external magnetic field, and thus the precession frequency of the nucleus depends on this orientation. This orientation-dependent precession frequency is called the anisotropic chemical shift. The second is the magnetic dipolar interac tion between spins, which broadens the absorption lines. The third, which is important only for deuterium atoms bonded to carbon atoms, is the interaction between a deuterium nucleus and the local electric field gradient in the CD bond. This interaction changes the absorption frequency of the deuterium nucleus by an amount which depends on the orientation of the CD bond with respect to the external magnetic field. The way in which these interactions affect the spectrum depends not only upon their strength but also upon their time dependence. Suppose a nucleus in a molecule has a range of precession frequencies, aw, which is caused by the fact that the molecule can be in a variety of orientations or surround ings. If the molecules cannot change their orientations or surroundings, the spectrum will contain the entire range of frequencies and a line of width aw will result. If the orientations or surroundings of the molecules can change on a time scale of the order of 1/aw, then a much narrower line will result. In effect, each nucleus precesses at some average frequency which is in the middle of the range of width aw.
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
369
This sensitivity of the spectrum to the time dependence of the interactions is the basis for two important principles in magnetic resonance, namely proton decoupling and motional narrowing. The dipolar coupling between nuclei is usually of much less interest in learning about membrane structure than are the other interactions. The dipolar effect of protons on the NMR spectrum of 2H, 31p, and 13C can be decreased by irradiating the protons at their resonant frequencies with a strong radio frequency noise field. This causes the dipolar fields of the protons to fluctuate rapidly enough that their effect is reduced by averaging process. This technique, called proton decoupling, is of great usefulness in the NMR spectroscopy of all nuclei other than protons themselves. Motional narrowing is the name given to the process by which NMR lines are narrowed by the motion of the molecules. The line broadening effect of the anisotropic chemical shift and the quadrupole coupling (for 2H) can be reduced by molecular rotation, provided the rotation takes place on a rapid enough time scale (i.e. on times of the order of the reciprocal of the unnar rowed line width). The effect of dipolar coupling between spins is also reduced by molecular rotation and diffusion. In addition, overall rotation of membrane vesicles can contribute to the narrowing of lines. Unsonicated lamellar dispersions and membrane preparations These sys tems have broad NMR spectra with line widths of the order of several KHz. The lines are sharper above the phase transition temperature than they are below it. Partially narrowed deuterium and phosphorus spectra, obtained with the aid of proton decoupling, are useful for obtaining information about the freedom of molecular rotation above the transition temperature. The theory of the motional narrowing (40, 41) shows that the extent of narrowing can be related to a quantity called the "order parameter" in the deuterium case and to a quantity called the "chemical shielding anisotropy" in the phos phorus case. These quantities contain information about certain averages of the distribution of molecular orientation with regard to the membrane surface. (This result holds only when the bilayer surface itself is not free to rotate on a timescale of 10-5 sec or faster.) For deuterium resonance, the molecular orientation of interest is the direction of the CD bond axis, and for 31p it is the orientation of the phosphate group. The numerical value of the order parameter or chemical shielding aniso tropy can be extracted from the partially narrowed spectrum. The precise value of an order parameter or chemical shielding anisotropy ordinarily conveys little direct physical meaning. They are primarily of use for testing theories or models for molecular conformation and motion in membranes. If a model predicts values that are in disagreement with experiment, the
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
370
ANDERSEN
model must be wrong. If the model gives correct predictions, it may be right, but it could still be wrong. It is possible for very different physical models to give identical order parameters and chemical shielding anisotro pies. (For this reason, the term "order parameter" is an unfortunate one, since it conveys the false impression that the order parameter is a measure of some oversimplified notion of molecular order.) In principle, proton spectra should contain some information about molecular orientations and motion. Because of the importance of the dipo lar interaction between the two protons of a methylene group, the partially narrowed spectrum should be sensitive to motions that reorient the line joining the two protons. However, the spectrum is complicated by other dipolar couplings and there is some disagreement (42) about how the order parameters should be extracted from the spectrum. Sonicated dispersions Sonicated dispersions have much narrower lines than unsonicated preparations. The types of motions that can cause such a narrowing are overall rotation of the vesicles, diffusion of molecules around the curved surface of a vesicle, and possibly an enhanced freedom of molecular motion in the curved vesicles as compared to the flatter bilay ers of unsonicated dispersions. There has been some disagreement about the relative importance of these factors in determining the line widths (38, 43-45) and consequently the interpretation of the widths in terms of molecular structure is not certain. The lines are narrow enough that small isotropic chemical split splittings between nuclei in different chemical surroundings can be observed. For example, methyl and methylene protons cause different lines in the spec trum. Measurements of T 1 and T2, which contain information about molecular motion, can be obtained separately for each observed line. Examples of NMR studies of membranes 1 . 31p and 2H studies of polar head groups. Recent work done by three research groups, have provided much detailed information for testing mod els of the head group conformation of phospholipids in membranes. Seelig & GaIly (46) studied bilayers of dipalmitoylphosphatidylethanolamine above and below the phase tradition, using both 3 1p and 2H NMR. In the deuterium work, they used molecules that were selectively deuterated at the ethanolamine carbon atoms. Their data were consistent with a model in which the head group rotates flat on the surface of the bilayer and makes rapid transitions between just two conformations. Kohler & Klein (47) measured the 31P spectrum of bilayers of dipalmitoylphosphatidylethanola mine, DPPC, egg PC, and brain PC. They interpreted their findings in terms of two simple models of head group motion. Seelig, GaIly & Wohlgemuth (48) and Griffin, Powers & Pershan (49) studied DPPC head group orienta-
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
371
tion. The first group worked above the phase transition temperature using unoriented samples and did both 31p and 2H studies using selectively deute rated samples. The second group worked below the phase transition temper ature using oriented samples and performed 31p experiments for a variety of angles of the magnetic field relative to the bilayer normal. Both studies were consistent with a model in which the choline head group is aligned parallel to the bilayer plane. 2. 2H order parameter studies of hydrocarbon chains. Deuterium order parameter studies have been performed for a variety of bilayers in the fluid phase, using specifically deuterated molecules so that order parameters as a function of position along the chain could be measured. Seelig & Seelig studied deuterated DPPC (50) and deuterated I-palmitoyl-2-0Ieoylphos phatidylcholine (51). Stockton et al (45) studied egg PC and egg PC cholesterol bilayers that contained deuterated stearic acid. Stockton et al (52) incorporated specifically deuterated fatty acids biosynthetically into the membranes of Acholeplasma laidlawii. The results in all cases are qualitatively similar; the order parameters are nearly constant near the head group and middle of the chains and then decrease toward the methyl ends, and the values vary with temperature, degree of unsaturation, and mole fraction of cholesterol. There have been numerous attempts (44, 50, 53, 54) to understand the physical meaning of the numerical results, to use them to obtain information about hydrocarbon chain conformation and other types of molecular motion, and to understand the relationship of these deuterium order parameters and the ESR order parameters (see below). As mentioned above, it is relatively easy to disprove models using order parameters but very difficult or impossible to prove the correctness of a model. All these analyses involve so many explicit and implicit assumptions and oversimplifications that it is difficult to know how much credence to give to the conclusions. Of special note is a recent paper by Petersen & Chan (44) which emphasizes the importance of reorientation of the chains as well as rotational isomerization in accounting for the order parameters. 3. Other studies of unsonicated systems. The study of proton spectra in unsonicated systems has led to fewer detailed and conclusive results than has the study of 2H and 3 1p. This is because many different types of protons contribute to a line and because of the complicating effects of dipolar coupling among all the spins. Chan and co-workers (55, 56, 44) have developed a theory for interpreting proton spectra and relaxation times. The effect of dipolar coupling on line shapes has been described differently by Ulmius et al (42) and Bloom et al (57). The study of I3C spectra in unsonicated systems has been hampered by the ineffectiveness of the usual proton decoupling techniques for removing dipolar broadening. Urbina & Waugh (58) have applied a double resonance
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
372
ANDERSEN
technique to the study of DPPC dispersions. This method not only elimi nates dipolar coupling, giving spectra in which the width is determined by the anisotropic chemical shift, but also gives increased sensitivity. Opella, Yesinowski & Waugh (59) applied this technique to the study of cholesterol in DPPC dispersions, using cholesterols specifically enriched with \3C at two positions. 4. Sonicated vesicles. Kroon, Kainosho & Chan (60) measured protein NMR line widths and Tl values for DPPC sonicated vesicles. They used deuterated molecules to perform isotopic dilution experiments to estimate the relative influence of intermolecular and intramolecular interactions on the results. Horwitz, Horsley & Klein (61) measured Tl and T2 values for sonicated egg PC vesicles. They were able to obtain individual relaxation times for several different types of protons: N-methyl protons on the choline head group and a-carbonyl, allyl, vinyl, methylene, and methyl protons on the fatty acid chains. Lee et al (62) measured proton Tl for DPPC and egg PC vesicles and studied the effect of cholesterol. Darke et al (63) also studied the effect of cholesterol on DPPC and egg PC vesicles by measuring the changes in line shapes with composition. l3C T relaxation times give information about the correlation time for 1 reorientation of the CH bonds in a molecule. Individual values can usually be measured for many of the carbon atoms in a molecule. Such studies have been performed, for example, by Levine et al (64) on DPPC and dioleoyl phosphatidylcholine vesicles, by Godici & Landsberger (65) on egg PC vesicles, and by Barton & Gunstone (66) on a series of unsaturated PCs. Fatty acids with magnetic nuclei have been used as probes in phos pholipid vesicles. Stockton et al (45) studied 2H line widths of deuterated fatty acids, and Birdsall et al (67) studied 19F line widths of fluorinated fatty acids. De Kruijff et al (68) studied the high resolution spectrum Of31p in DPPC, distearoylphosphatidylcholine, and dimyristoylphosphatidylcholine vesi cles as a function of temperature. Separate resonances for molecules on the inside and outside layers of the vesicle were seen. ELECTRON SPIN RESONANCE OF SPIN LABELS One of the most fruit ful methods for studying membranes is to introduce stable free radical molecules into them and study the electron spin resonance (ESR) spectrum of the radical in the presence of an external magnetic field. The term "spin label" is often used to describe such molecules. A recent monograph (69) discusses many aspects of spin labeling. In a spin label ESR experiment, the spin label molecule is incorporated into the membrane sample, and the sample is placed in a magnetic field. The absorption spectrum for microwave radiation is measured. (Usually, as in
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
373
the case of NMR, the frequency is held fixed and the magnetic field is varied.) If oriented membrane samples are used, the spectrum can be mea sured as a function of the angle between the magnetic field and the direction normal to the membrane surfaces. The most frequently used spin labels for membrane studies are nitroxides, which contain the :::N-O group. The N atom is usually part of a 5- or 6-membered ring. This group has an unpaired electron spin located primar ily in an atomic p orbital on the nitrogen atom. The axis of the orbital is normal to the plane defined by the nitrogen and oxygen atoms and the other two atoms bonded to the nitrogen. Synthetic techniques have been developed for introducing stable nitrox ide groups into many types of compounds (70). The most common types are 1. molecules that are smaller than organic molecules typically found in membranes and that are soluble in water; 2. derivatives of fatty acids, in which the nitroxide group is attached at various places; 3. derivatives of phospholipids, in which the nitroxide group is attached at various places; 4. sterol derivatives with a structure similar to cholesterol. An example of the first type is 2,2,6,6-tetramethylpiperadine-I-oxyl (TEMPO), which is soluble in both aqueous and nonaqueous phases and which partitions between aqueous and membrane regions when added to membrane preparations. Examples of the second type are doxyI derivatives of fatty acids. (The doxyI group is a five-membered ring containing an NO group.) Examples of the third type are phosphatidylcholine molecules, one of whose acyl chains is a molecule of the second type. The 3-doxyl deriva tives of cholestane-3-one and androstane-3-one- 1 7-01 are examples of the fourth type. Factors affecting the spectrum of a nitroxide radical An electron in a magnetic field can absorb radiation at one particular frequency, much like a proton in a magnetic field. This would ordinarily give a single line in the absorption spectrum. For a nitroxide radical, however, the spectrum is affected by magnetic (hyperfine) interactions between the electron and the magnetic 14N nucleus, which has a spin of one. The result is that there are three equally spaced lines, rather than one, corresponding to the three possible quantum mechanical states of the spin- l nitrogen. For a stationary molecule, the positions of the three lines depend on the orientation of the nitroxide group relative to the external field. For mole cules which can rotate, the same principles of motional averaging apply to
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
374
ANDERSEN
ESR as to NMR. The shape of the spectrum is sensitive to the rate and anisotropy of molecular reorientation. In particular, the spacing between the high field and low field lines are sensitive to the molecular motion which can reorient the direction of the p orbital on the nitrogen nucleus. An important difference between NMR and ESR is that in the latter case the widths of the unbroadened lines is much larger (of the order of 108 Hz rather than 105 Hz), and hence ESR narrowing is sensitive only to motions that take place on a much shorter time scale. The various theories of line shapes for the ESR of nitroxide spin labels are reviewed in several articles in the recent monograph on spin labeling (4, 71-74). One very common way of analyzing an ESR spectrum involves extracting from it a number called an ESR "order parameter" which bears some conceptual similarity to the deuterium and proton NMR order parameters discussed above. When this method of analysis is valid, the order parameter is a number between -1/2 and +1 which has a well-defined meaning in terms of the average ability of the spin label p orbital to rotate during times of the order of 10-8 sec. (Unlike the NMR case, the ESR order parameter is not a measure of the average orientation of the spin label with respect to
the bilayer normal.) If a spin label has an order parameter close to unity, the physical meaning of this result is very clear, namely, that in a time of 10-8 sec the molecular motion does not appreciably change the direction of the p orbital containing the unpaired spin. However, a numerical value of the order parameter significantly less than unity is consistent with a wide variety of possible motions for the spin label. Molecular motion, discussed in the previous paragraph, is the dominant factor in determining ESR spectra. There are, however, three other effects that are of importance for certain applications of spin labels. The first two are the spin-exchange interaction and the dipole-dipole interaction, which can take place if two spin labels are close enough to each other. The third is an effect of the polarity of the surroundings of a spin label upon the spectrum.
Types ofanalysis of nitroxide spin label spectra In order to obtain informa tion about the molecular structure of a membrane from ESR spin label experiments, it is first usually necessary to reduce the data to a form that is more easily interpreted. There are only a small number of ways in which this is done. 1. The spectra can be used to estimate the relative amounts of probe mole cules in different environments. 2. The spectra can be used to derive order parameters or other numerical measures of probe mobility or anisotropic motion.
PROBES OF MEMBRANE STRUCTURE
375
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
3. The spectra can be compared to theoretical spectra calculated from detailed models of the membrane structure and dynamics. 4. The spectra can be used to derive numerical measures of the polarity of the immediate surroundings of the probe molecule. 5. When the concentration of probe molecules is high enough, the effects of spin-exchange interaction and dipole-dipole interaction between probes on the spectra can be used to obtain information about the frequency of collisions between probe molecules and the distance be tween adjacent probe molecules in a membrane. Examples of the use of ESR spin label studies The following are some of the many types of spin label studies that have been performed. [For more comprehensive and detailed reviews see (69, 75, and 76).] 1. Partitioning studies. The partitioning of TEMPO between the nonpo lar interior of artificial bilayer systems and the aqueous surroundings has been used to investigate phase transitions and lateral phase separations. The technique is based on the fact that the solubility of TEMPO in a membrane interior is greater above the phase transition temperature than it is below that temperature and that the spectrum of TEMPO in water is different from the spectrum of TEMPO in a bilayer. A graph of the fraction of label which is dissolved in the membrane shows distinct changes at temperatures at which a one-component system changes its phase or a two-component system enters or leaves a two phase region of its phase diagram. TEMPO partitioning has been used to measure phase transition temperatures for bilayers composed of a single type of lipid and phase diagrams for bilayers containing two types of lipids (77, 78). One technique for studying the effect of proteins on membrane structure is to use TEMPO partitioning to study the effect of proteins on the lipid phase transition of a model membrane. This was one method used by Hesketh et al (79) in their study of a calcium transport protein. The partitioning of phospholipid spin labels between fluid and solid phases is one of the methods used to study calcium-induced lateral phase separations in phospholipid bilayers (SO). Biological membranes may have regions of varying fluidity due not only to lateral phase separation but also because protein molecules can reduce the mobility of adjacent lipids. TEMPO partitioning between the aqueous region and the fluid region of the membrane is the basis for a method for estimating the fraction of lipids in a biological membrane which are in a fluid state (SI). Partitioning methods have been used to develop spin label techniques for measuring some electrical properties of membranes. Castle & Hubbell (S2) have shown that the partitioning of a charged spin label molecule, N,N dimethyl-N-nonyl-N-tempoylammonium ion between the lipid region of
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
376
ANDERSEN
phospholipid vesicles and the surrounding aqueous medium is sensitive to the electrical potential at the surface of the vesicle. They devised a proce dure for estimating this surface potential from an analysis of the ESR spectrum of the label. Cafiso & Hubbell (83) have shown that the partitioning of certain posi tively charged hydrophobic spin labels between a phospholipid vesicle and the surrounding aqueous medium is sensitive to the electrical potential difference between the aqueous solutions on either side of the bilayer, and have devised a method for estimating transmembrane potentials using this effect. 2. Order parameter studies and hyperfine splitting studies. A common way of presenting and analyzing ESR probe data is to derive a numerical value of the order parameter from the experimental spectrum. There are other numerical measures of probe motion which are sometimes used as alternatives to order parameters. The splittings between various hyperfine extrema (maxima and minima) in the measured spectrum or the heights of the extrema are examples of this. They are easier to obtain than the order parameters, which may require more extensive analysis of the spectrum, and for many purposes they are equally useful. Using fatty acid spin labels and phospholipid spin labels with the doxyl group attached to various positions on the hydrocarbon chains, it is possible to insert nitroxide free radicals into a membrane at various distances from the polar headgroup region. These studies are usually performed with doxyl derivatives of fatty acids and phospholipids. For these molecules, the axis of the p orbital containing the unpaired electron is always parallel to the "backbone" of the hydrocarbon chain at the point where the doxyl group is attached. The backbone of a hydrocarbon chain is most easily defined by drawing straight lines from the midpoint of each carbon-carbon bond to the midpoint of each adjacent carbon-carbon bond. Studies using such homolo gous series of spin labels have yielded the intriguing result that the ESR order parameter of the spin label decreases as the spin label is moved farther from the head group of the molecule (84, 85). This variation of the order parameter has been called a "flexibility gradient" and a "fluidity gradient," has been observed in a variety of model and biological membranes (84-86), and appears to be a general property of spin labels in bilayers. (See the section above on 2H order parameter studies of hydrocarbon chains for a brief discussion of attempts to understand the physical meaning of this effect.) Hubbell & McConnell (84) have also used the temperature dependence of the order parameter for a spin labeled phospholipid in a DPPC bilayer as a diagnostic for measuring the phase transition temperature. The transi tion from a solid to a fluid bilayer leads to an increase in molecular mobility
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
377
and hence to a measurable sharp decrease in the order parameter over a very short range of temperature. Van & Griffith (87) have used order parameters and anisotropy parameters of fatty acid spin labels to study the effect of an extrinsic protein, cytochrome c, on bilayers composed of a mixture of cardiolipin and lecithin. Changes in the splittings in spectra of spin labeled phosphatidylserine and phosphatidylcholine have been used to study CaH induced phase transitions and lateral phase separations of bilayers contain ing anionic lipids (80). Schreier-Muccillo et al (88) have used stearic acid spin labels, sterol spin labels, and a stearamide spin label with a nitroxide group in the polar region of the molecule to study bilayers of DPPC and egg PC with various percentages of cholesterol added. Keith, Snipes & Chapman (89) have used two empirical motional parameters, derived from the heights of extrema in ESR spectra, to study the motion of small spin label molecules in various aqueous regions near the surface of dimyristoyl phosphatidylcholine bilayers. The two examples mentioned above are concerned with model mem branes. An example of the use of spin labels to study biological membranes is the work of Butterfield et al (90). They used a spin label that is thought to embed itself in membranes in such a way that the nitroxide group is located near the membrane surface. They measured and compared the order parameters for this spin label embedded in erythrocytes taken from normal subjects and from patients with myotonic muscular dystrophy, Duchenne muscular dystrophy, and congenital myotonia. Relative values of the order parameters are presumably related to the relative freedom of rotational motion of molecules near the membrane surface. 3. Detailed analysis of spectral shapes. Another way to interpret ESR data is to postulate a detailed model of the dynamics of the spin labels, compute a theoretical spectrum on the basis of the model, and compare the calculated results with the experimental spectrum. The agreement between the two provides a test of whether the model is consistent or inconsistent with experiment. Many models contain some adjustable parameters that have physical meaning within the context of the model, and the parameters are varied to obtain a best fit to experiment. Examples of such models are discussed in (74). The major problem with this approach is that one must beware of interpreting the results of this type of analysis too literally, because there may be several similar, or even very different, models that agree equally well with experiment. One example of such a study is the work of Birrell and Griffith (91). They studied oriented multilayers of DPPC containing phospholipids that had doxyl spin labels attached to the hydrocarbon chains near the head group and near the hydrocarbon tail. This work was done at room temperature, which is well below the chain melting transition temperature. They found
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
378
ANDERSEN
that the spin labeled chains of both types were tilted with respect to the normal to the bilayer surface, and they were able to fit their data to a "Gaussian distribution model" for the chain backbone directions. Another example of this type of study is the work of Gaffney & McCon nell (54) who studied oriented bilayers of egg PC at room temperature. They used three different spin labeled phospholipids, with the doxyl groups at various positions along the hydrocarbon chains. A detailed fit of the spectra to a Gaussian distribution model was performed for each spin label. They concluded that the chain backbones of the spin labeled molecules were tilted with respect to the normal to the bilayer. The tilt was found to be greater when the spin label is near the head group than when it is near the methyl end of the hydrocarbon chain. 4. Other types of spin label studies of membrane structure. Certain fea tures of the spectrum of a nitroxide spin label are affected by the polarity of its immediate surroundings. This effect is the basis of several techniques for estimating a polarity index as a function of position inside a membrane, which provides some information about the distribution of water (74, 87). The electron-electron dipole-dipole interaction between spin labels can have a measurable effect on the spectrum if two spin label molecules are close together for a long enough period of time. The spectrum is sensitive to the distance between the spin labels. This principle was used by Marsh & Smith (92) to measure the change in the distance between nearest neigh bor cholestane probes in PC-cholesterol bilayers as the composition of the bilayer was changed.
Calorimetry Because model membranes and some biological membranes can undergo phase transitions, calorimetry techniques that are capable of measuring the latent heat associated with these transitions have been very important for the study of membranes. Standard calorimeters, which measure the total heat capacity or specific heat of a sample, are not useful for dilute phos pholipid suspensions, because the bilayer heat capacity is overwhelmed by the heat capacity of the solvent. Differential scanning calorimetry (DSq, which measures the difference in heat capacity between a membrane suspen sion and an equal volume of the solvent, is the technique used. Recently, improvements in the design of differential scanning calorimeters have led to a new generation of instruments with high sensitivity. [For reviews of calorimetry studies of membranes, see (93, 94).] The data obtained in a DSC experiment is a graph of the heat capacity of the sample, relative to that of a nominally equal volume of solvent, plotted as a function of temperature. If the sample absorbs an especially large amount of heat over a small enough temperature interval, the observed
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
379
heat capacity will have one or more peaks that rise above the baseline. For bilayers composed of one phospholipid, a narrow peak is observed, and the quantities of particular interest are the temperature at which the heat capacity is largest and the area under the peak. They correspond to the transition temperature and latent heat of the transition, respectively. (For some PCs, two separate peaks, corresponding to two transitions, are ob served.) For bilayers composed of two lipids, peaks with a more compli cated shape are observed. The shapes do not have any simple physical meaning, and the goal of an analysis of the data is usually to construct a binary phase diagram from data taken over the entire range of composition for which the lipids have a bilayer structure. Such a phase diagram shows the ranges of temperature and composition over which the bilayers consist of one phase and of two phases. It should be noted that calorimetry is, in a sense, not really a probe of membrane structure. That is, it does not provide any information about the nature of the various different structures that bilayers can have. It tells us, for example, that aqueous dispersions of bilayers of DPPC undergo a phase transition at 41°C, and that therefore it is very likely that the bilayers have a very different structure above this temperature than they do below it. However, other methods must be used to learn what these structures are and how they differ. The following paragraph illustrates the range of membrane studies that have been performed using DSC. For more complete discussions and refer ences, the review articles cited above should be consulted. The transition temperatures and enthalpies for various one-component bilayers have been measured (95-97). For PCs with saturated hydrocarbon chains, two transitions are observed. The phase transition temperature of lipids with ionizable protons can be changed by changing the pH. Also, the presence of divalent cations Ca2+ and Mg2+ in the aqueous solution can affect the phase transition by interacting with the charged head groups or perhaps forming complexes. These effects have been investigated calorimet rically (36). The phase diagrams for binary mixtures of different saturated PCs (97) and of saturated PCs and saturated fatty acids (98) have been determined. The effect of cholesterol on the phase transition of PCs has been studied (99- 102). Also, the lipids and membranes of Acholeplasma laidlawii (103) and of E. coli (104, 105) have been studied using DSC.
Freeze Fracture Electron Microscopy Freeze fracture electron microscopy is a technique for taking a "photo graph" of the interior surface of a membrane, which separates the two halves of the bilayer. [For a recent review of this method, see (106). Also see (107) for a discussion of the technical aspects of the method.]
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
380
ANDERSEN
The first step in such an experiment is a rapid freezing or quenching of the sample that contains the membrane. This is done with Freon or with mixtures of liquid and solid nitrogen. The freezing must be done rapidly enough to prevent any structural changes from taking place. The second step is the fracturing of the sample along one plane. It is generally accepted that membranes tend to fracture along the surface, mentioned above, that separates the two halves of a bilayer. The result then is two new frozen surfaces, and a part of each surface represents the interior surface of bilay ers. In the third step, a replica of one or both surfaces is made by depositing some material, such as platinum and carbon. The replicas are stable at higher temperatures and hence they can be examined by electron micros copy. A resolution of about 30 A can be achieved in the resulting electron micrograph. Various types of physical features can be observed in the resulting pic tures. For lipid dispersions of one- and two-component phospholipids, different textures are seen, corresponding to different phases. The physical meaning of the details of these surfaces is not as important as the fact that they can be observed and distinguished from each other. When two or more such textures appear on the same bilayer, it is clear that lateral phase separation exists in the membrane. Thus the technique is very useful for obtaining or confirming phase diagrams for lipid bilayers. For biological membranes the pictures typically show particles on a relatively smooth surface. The particles may either be homogeneously distributed or concen trated in patches. The particles are interpreted as intramembrane proteins and the smooth surfaces as lipids. Thus freeze fracture electron microscopy is also useful for obtaining information about protein distributions in model and biological membranes. [For a review of freeze fracture studies prior to 1975, see (106).] The following are a few of the more recent freeze fracture studies. Van Dijck et al (108) and Papahadjopoulos et al ( 1 09) used freeze fracture methods in conjunction with calorimetry to study the effect of CaH and MgH ions upon the structure of bilayers composed of dimyristoylphos phatidylglycerol. Chapman et al ( 1 10) studied the effect of monovalent ions on the phase transitions of PC bilayers. Verkleij et al ( 1 1 1) studied the outer membrane of E coli mutants using freeze fracture microscopy.
COMMENTS ON THE INTERPRETATION OF MEMBRANE PROBE DATA Although many different methods for probing membrane structure are available and much work has been done using them, we have surprisingly little unambiguous knowledge about membrane structure on a molecular
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
381
level. An important reason for this is that the quantities measured in these experiments, with the possible exception of X-ray and neutron scattering, are only indirectly related to structure. The arguments used to infer struc tural information from, for example, an NMR line width or a fluorescence depolarization ratio, or even an X-ray scattering pattern, often contain hidden structural assumptions that may be wrong. This, combined with the possibility of artifacts and systematic errors, has led to the publication of many contradictory conclusions about membrane structure. In studying the literature of this field, it is important to be aware of the physical nature of the measurements and the nature of the logic used to interpret the results obtained. ACKNOWLEDGMENTS
This review was prepared while the author was a John Simon Guggenheim Memorial Fellow. The support of the National Science Foundation (Grant CHE 75-06634) and the National Institutes of Health (Grant GM 23085) is also gratefully acknowledged. Literature Cited
I . Singer, S. J. 1974. Ann. Rev. Biochem. 2. 3. 4.
5. 6. 7. 8.
9. 10.
I I. 1 2.
13. 14.
43:805-33 Bangham, A. D. 1972. Chem. Phys. Lip ids 8:386-92 Huang, C., Thompson, T. E. 1974. Methods Enzymol. 32B:485-89 Smith, I. C. P., Butler, K. W. 1 976. In Spin Labeling: Theory and Applications, ed. L. Berliner, New York: Academic. pp. 4 1 1-51 Powers, L., Clark, N. A. 1975. Proc. Natl. Acad. Sci. USA 72:840--43 Powers, L., Pershan, P. S. 1977. Bio phys. 1. 20: 1 37-52 Melchior, D. L., Steim, J. M. 1976. Ann. Rev. Biophys. Bioeng. 5:205-38 Ranck, J. L., Mateu, L., Sadler, D. M., Tardieu, A., Gulik-Krzywicki, T., Luz zati, V. 1974. 1. MoL Bioi. 85:249-77 Akers, C. K. 1 974. See Ref. 3, pp. 2 1 1-20 Tardieu, A., Luzzati, V., Reman, F. C. 1 973. 1. MoL Bioi. 75:71 1-33 Luzzati, V., Tardieu, A. 1974. Ann. Rev. Phys. Chem. 25:79-94 Engelman, D. M. 1970. J. Mol. Biol 47: 1 1 5- 1 7 Engelman, D. M. 1 97 1 . J. Mol. BioL 58: 1 53-65 Esfahani, M., Limbrick, A. R., Knut ton, S., Oka, T., Wakil, S. J. 1 9 7 1 . Proc. Natl. Acad. Sci. USA 63:31 80-84
1 5 . Rand, R. P., Chapman, D., Larsson, K. 1975. Biophys. 1. 1 5 : 1 1 1 7-24 1 6. Janiak, M. J., Small, D. M., Shipley, G. G. 1 976. Biochemistry 1 5 :4575-80 1 7. Brady, G. W., Fein, D. B. 1977. Bio chim. Biophys. Acta 464:249-59 1 8 . Schoenborn, B. P. 1976. Biochim. Bio phys. Acta 457:41-55 19. Neutron Scattering for the Analysis of Biological Structures. Brookhaven Symp. Bioi., Vol. 27. To be published 20. Zaccai, G., Blasie, J. K., Schoenborn, B.
P. 1 975. Proc. Natl. Acad. Set: USA 72:
21. 22.
23. 24. 25. 26. 27. 28.
376-80 B1asie, J. K., Zaccai, G., Schoenborn, B. 1975. Biophys. 1. 1 5 :99a Brown, K. G., Peticolas, W. L., Brown, E. 1973. Biochem. Biophys. Res. Com mun. 54:358-64 Gaber, B. P., Peticolas, W. L. 1977. Bio chim. Biophys. Acta 465:260-74 Lippert, 1. L., Peticolas, W. L. 197 1 . Proc. Nat!. Acad. Sci. USA 68: 1 572-76 Mendelsohn, R. 1 972. Biochim. Bio phys. Acta 290: 1 5-21 Lippert, J. L., Peticolas, W. L. 1 972. Biochim. Biophys. Acta 282:8-17 Lippert, J. L., Gorczyka, L. E., Meik lejohn, G. 1975. Biochim. Biophys. Acta 382:5 1-57 Mendelsohn, R., Sunder, S., Bernstein, H. J. 1976. Biochim. Biophys. Acta 419:563-69
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
382
ANDERSEN
29. Azzi, A. 1975. Q. Rev. Biophys. 8:237316 30. Azzi, A . 1 974. See Ref. 3, pp. 234-46 3 1 . Radda, G. K. 1975. Methods Membr. Biol 4:97-188 32. Waggoner, A. S., Stryer, L. 1 970. Proc. Natl. Acad. Sci. USA 67:579-89 33. Sklar, L. A., Hudson, B. S., Simoni, R. D. 1977. Biochemistry 1 6 : 8 1 9-28 34. Lentz, B. R., Barenholz, Y., Thompson, T. E. 1976. Biochemistry 1 5 :4521-28 35. Lentz, B. R., Barenholz, Y., Thompson, T. E. 1 976. Biochemistry 1 5 :4529-37 36. Jacobson, K., Papahadjopoulos, D. 1975. Biochemistry 14:1 52-61 37. Tecoma, E. S., Sklar, L. A., Simoni, R. D., Hudson, B. S. 1977. Biochemistry 16:829-35 38. Lee, A. G., Birdsall, N. J. M., Metcalfe, J. C. 1 974. Methods Membr. Bioi. 2 : 11 56 39. Mantsch, H. H., Saito, H., Smith, I. C. P. 1 977. In Progress in Nuclear Mag netic Resonance Spectroscopy, ed. J. W. Emsley, J. Feeney, L. H. Sutcliffe, 1 1 :21 1-72 London: Pergamon 40. Seelig, J., Niederberger, W. 1974. J. Am. Chem. Soc. 96:2069-72 4 1 . Niederberger, W., Seelig, J. 1976. J. Am. Chem. Soc. 98:3704-6 42. Ulmius, 1., Wennerstrom, H., Lind blom, G., Arvidson, G. 1975. Biochim. Biophys. Acta 389: 197-202 43. Lichtenberg, D., Petersen, N. 0., Girar det, J.-L., Kainosho, M., Kroon, P. A., Seiter, C. H. A., Feigenson, G. W., Chan, S. 1. 1975. Biochim. Biophys. Acta 382 : 10-2 1 44. Petersen, N. 0., Chan, S. I. 1977. Bio chemistry 1 6:2657-67 45. Stockton, G. W., Polnaszek, C. F., Tul loch, A. P., Hasan, F., Smith, I. C. P. 1976. Biochemistry 1 5 :954-66 46. Seelig, J., Gaily, H.-U. 1976. Biochemis try 1 5:5 199-5204 47. Kohler, S. J., Klein, M. P. 1977. Bio chemistry 1 6 : 5 1 9-26 48. Seelig, J., Gaily, H.-U., Wohlgemuth, R. 1977. Biochim. Biophys. Acta 467:109-19 49. Griffin, R. G., Powers, L., Pershan, P. S. Preprint 50. Seelig, A., Seelig, J. 1974. Biochemistry 1 3 :4839-45 5 1 . Seelig, A., Seelig, J. 1977. Biochemistry 16:45-50 52. Stockton, G. W., Johnson, K. G., But ler, K. W., Tulloch, A. P., Boulanger, Y., Smith, l. C. P., Davis, J. H., Bloom, M. 1977. Nature 269:267-68 53. Schindler, H., Seelig, J. 1975. Biochem istry 14:2283-87
54. Gaffney, B. J., McConnell, H. M. 1974. J. Magn. Reson. 16: 1-28 55. Seiter, C. H. A., Chan, S. I. 1973. J. Am. Chem. Soc. 95:7541-53 56. Feigenson, G. W., Chan, S. I. 1974. J. Am. Chem. Soc. 96: 1 3 12-19 57. Bloom, M., Burnell, E. E., Valic, M. I., Weeks, G. 1975. Chem. Phys. Lipids 14: 107-12 58. Urbina, J., Waugh, J. S. 1974. Proc. Natl. Acad. Sci. USA 7 1 :5062-67 59. Opella, S. J., Yesinowski, J. P., Waugh, J. S. 1976. Proc. Natl Acad. Sci. USA 73:3 8 1 2- 1 5 60 . Kroon, P . A . , Kainosho, M . , Chan, S . I . 1976. Biochim. Biophys. Acta 433: 282-93 6 1 . Horwitz. A. F., Horsley, W. J., Klein, M. P. 1972. Proc. Natl. Acad. Sci. USA 69:590-93 62. Lee, A. G., Birdsall, N. 1. M., Levine, Y. K., Metcalfe, J. C. 1972. Biochim. Biophys. Acta 255 :43-56 63. Darke, A., Finer, E. G., Flook, A. G . • Phillips, M. C. 1972. J. Mol. BioI. 63:265-79 64. Levine, Y. K., Birdsall, N. 1. M., Lee, A. G., Metcalfe, J. C. 1972. Biochemis try 1 1 : 1 4 1 6-2 1 65. Godici, P. E., Landsberger, F. R. 1974. Biochemistry 1 3:362-68 66. Barton, P. G., Gunstone, F. D. 1975. J. Biol Chem. 250:4470-76 67. Birdsall, N. J. M., Lee, A. G., Levine, Y. K., Metcalfe, J. C. 197 1 . Biochim. Biophys. Acta 241 :693-96 68. De Kruijff, B., Cullis, P. R., Radda, G. K. 1974. Biochim. Biophys. Acta 406;6-20 69. Berliner, L., ed. 1976. Spin Labeling; Theory and Applications. New York: Academic. 592 pp. 70. Gaffney, B. J. 1976. See Ref. 69, pp. 1 8 3-238 7 1 . Nordio, P. L. 1976. See Ref. 69, pp. 5-52 72. Freed, J. H., 1976. See Ref. 69, pp. 531 32 73. Seelig, J. 1976. See Ref. 69, pp. 373-409 74. Griffith, O. H., Jost, P. C. 1976. See Ref. 69, pp. 453-523 75. Gaffney, B. 1. 1 974. See Ref. 3, pp. 1 6 1 -98 76. Gaffney, B. J., Chen, S.-C. 1977. Meth ods Membr. BioI. 8:29 1-358 77. Shimshick, E. J., McConnell, H. M. 1973. Biochemistry 12:235 1-60 78. Wu, S. H., McConnell, H. M. 1975. Bio chemistry 14:847-54 79. Hesketh, T. R., Smith, G. A., Houslay, M. D., McGill, K. A., Birdsall, N. J.
PROBES OF MEMBRANE STRUCTURE
80. 81.
82. 83. 84.
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
85.
86.
87. 88.
89. 90.
91. 92. 93. 94. 95.
M . , Metcalfe, J. C . , Warren, G . B . 1 976. Biochemistry 1 5:4145-51 Ohnishi, S. 1975. Adv. Biophys. 8 :35-82 McConnell, H. M., Wright, K. L., McFarland, B. G. 1972. Biochem. Bio phys. Res. Commun. 47:273-8 1 Castle, J. D., Hubbell, W. L. 1 9 76. Bio chemistry 1 5:48 1 8-31 Cafiso, D. S., Hubbell, W. L. 1978. Bio chemistry 1 7 : In press Hubbell, W. L., McConnell, H. M. 1 97 1 . J. Am. Chem. Soc. 93:314-26 Jost, P., Libertini, L. J., Hebert, V. C., Griffith, O. H. 1 97 1 . J. Mol. Bioi. 59:77-98 Rottem, S., Hubbell, W. L., Hayflick, L., McConnell, H. M. 1 970. Biochim. Biophys. Acta 2 1 9 : 104-13 Van, S. P., Griffith, O. H. 1975. J. Membr. Bioi. 20: 1 5 5-70 Schreier-Muccillo, S., Marsh, D., Dugas, H., Schneider, H., Smith, I. C. P. 1973. Chem. Phys. Lipids 10: 1 1-27 Keith, A. D., Snipes, W., Chapman, D. 1 977. Biochemistry 16:634-41 Butterfield, D. A., Chestnut, D. B., Ap pel, S. H., Roses, A. D. 1976. Nature 263 : 1 59-61 Birrell, G. B., Griffith, O. H. 1976. Arch. Biochem. Biophys. 1 72:455-62 Marsh, D., Smith, I. C. P. 1973. Oio chim. Biophys. Acta 298: 1 3 3-44 Ladbrooke, B. D., Chapman, D. 1 969. Chem. Phys. Lipids 3:304-56 Mabrey, S., Sturtevant, J. M. 1978. Methods Membr. Bioi. 9:ln press Hinz, H.-J., Sturtevant, J. M. 1972. J. Bioi. Chem. 247:607 1-75
383
96. Chapman, D., Urbina, J., Keough, K. M. 1974. J. BioI. Chem. 249:25 1 2-21 97. Mabrey, S., Sturtevant, J. M. 1 9 76. Proc. Natl. Acad. Sci. USA 73:3862-66 98. Mabrey, S., Sturtevant, J. M. 1977. Bio chim. Biophys. Acta 486:444-50 99. Ladbrooke, B. D., Williams, R. M., Chapman, D. 1968. Biochim. Biophys. Acta 150:333--40 100. Hinz, H.-I., Sturtevant, I. M. 1972. J. Bioi Chem. 247:3697-3700 1 0 1 . Barton, P. G. 1976. Chem. Phys. Lipids
1 6 : 1 95-200
102. Mabrey, S., Mateo, P. L., Sturtevant, J. M. 1977. Oiophys. J. 1 7:82a 103. Melchior, D. L., Morowitz, H. J., Sturtevant, J. M., Tsong, T. Y. 1970. Biochim. Biophys. Acta 2 1 9 : 1 1 4-22 1 04. Baldassare, J. J., Rhinehart, K. B., Silbert, D. F. 1976. Biochemistry 1 5 : 2986-94 105. Jackson, M. B. 1976. Fed. Proc. 35: 1 5 3 1 106. Verkleij, A. J . , Ververgaert, P. H . J. Th. 1 97 5 . Ann. Rev. Phys. Chem. 26: 101-22 107. Zingsheim. H. P. 1 972. Biochim. Bio phys. Acta 265:339-66 108. Van Dijck, P. W. M., Ververgaert, P. H . J . Th., Verkleij, A. J . , van Deenen, L . L. M., de Gier, J. 1975. Biochim. Biophys. Acta 406:465-78 109. Papahadjopoulos, D., Vail, W. J., Pang born, W. A., Poste, G. 1976. Biochim. Biophys. Acta 448:265-83 1 10. Chapman, D., Peel, W. E., Kingston, B., Lilley, T. H. 1977. Biochim. Biophys. Acta 464:260-75 1 1 1 . Verkleij, A., van Alphen, L., Bijveit, J., Lugtenberg, B. 1977. Biochim. Biophys. Acta 466:269-82
Further
Quick links to online content Ann. Rev. Biochem. 1978. 47:359- and the values of T2, are affected by the structural properties of the sample on a molecular level and by molecular motions. The most important magnetic nuclei for membrane studies are IH, 2H, I3C, and 3 1p. The predominant isotopes for hydrogen and phosphorus are IH and 3 1p, thus these isotopes appear in great abundance in normal biologi cal samples. Only about 1 % of the naturally occurring carbon nuclei are I3C. Nevertheless, NMR is sensitive enough to make the technique useful for this isotope. Deuterium can be introduced by synthetic techniques into specific parts of molecules as a substitute for hydrogen. Also molecules containing the magnetic nucleus 19p have been used as probes of membrane structure. [For reviews of NMR applied to membranes, see (38, 39).] Factors that affect NMR absorption spectra The various magnetic nuclei have very different precession frequencies for any particular magnetic field strength. Thus the absorption lines for different nuclei are well separated from each other. There are three important interactions that determine NMR line shapes for any one type of nucleus. The first is the anisotropic chemical shift. The electronic cloud in the vicinity of a nucleus acts to shield the nucleus somewhat from the external magnetic field. The extent of the shielding depends on the orientation of the molecule with respect to the external magnetic field, and thus the precession frequency of the nucleus depends on this orientation. This orientation-dependent precession frequency is called the anisotropic chemical shift. The second is the magnetic dipolar interac tion between spins, which broadens the absorption lines. The third, which is important only for deuterium atoms bonded to carbon atoms, is the interaction between a deuterium nucleus and the local electric field gradient in the CD bond. This interaction changes the absorption frequency of the deuterium nucleus by an amount which depends on the orientation of the CD bond with respect to the external magnetic field. The way in which these interactions affect the spectrum depends not only upon their strength but also upon their time dependence. Suppose a nucleus in a molecule has a range of precession frequencies, aw, which is caused by the fact that the molecule can be in a variety of orientations or surround ings. If the molecules cannot change their orientations or surroundings, the spectrum will contain the entire range of frequencies and a line of width aw will result. If the orientations or surroundings of the molecules can change on a time scale of the order of 1/aw, then a much narrower line will result. In effect, each nucleus precesses at some average frequency which is in the middle of the range of width aw.
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
369
This sensitivity of the spectrum to the time dependence of the interactions is the basis for two important principles in magnetic resonance, namely proton decoupling and motional narrowing. The dipolar coupling between nuclei is usually of much less interest in learning about membrane structure than are the other interactions. The dipolar effect of protons on the NMR spectrum of 2H, 31p, and 13C can be decreased by irradiating the protons at their resonant frequencies with a strong radio frequency noise field. This causes the dipolar fields of the protons to fluctuate rapidly enough that their effect is reduced by averaging process. This technique, called proton decoupling, is of great usefulness in the NMR spectroscopy of all nuclei other than protons themselves. Motional narrowing is the name given to the process by which NMR lines are narrowed by the motion of the molecules. The line broadening effect of the anisotropic chemical shift and the quadrupole coupling (for 2H) can be reduced by molecular rotation, provided the rotation takes place on a rapid enough time scale (i.e. on times of the order of the reciprocal of the unnar rowed line width). The effect of dipolar coupling between spins is also reduced by molecular rotation and diffusion. In addition, overall rotation of membrane vesicles can contribute to the narrowing of lines. Unsonicated lamellar dispersions and membrane preparations These sys tems have broad NMR spectra with line widths of the order of several KHz. The lines are sharper above the phase transition temperature than they are below it. Partially narrowed deuterium and phosphorus spectra, obtained with the aid of proton decoupling, are useful for obtaining information about the freedom of molecular rotation above the transition temperature. The theory of the motional narrowing (40, 41) shows that the extent of narrowing can be related to a quantity called the "order parameter" in the deuterium case and to a quantity called the "chemical shielding anisotropy" in the phos phorus case. These quantities contain information about certain averages of the distribution of molecular orientation with regard to the membrane surface. (This result holds only when the bilayer surface itself is not free to rotate on a timescale of 10-5 sec or faster.) For deuterium resonance, the molecular orientation of interest is the direction of the CD bond axis, and for 31p it is the orientation of the phosphate group. The numerical value of the order parameter or chemical shielding aniso tropy can be extracted from the partially narrowed spectrum. The precise value of an order parameter or chemical shielding anisotropy ordinarily conveys little direct physical meaning. They are primarily of use for testing theories or models for molecular conformation and motion in membranes. If a model predicts values that are in disagreement with experiment, the
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
370
ANDERSEN
model must be wrong. If the model gives correct predictions, it may be right, but it could still be wrong. It is possible for very different physical models to give identical order parameters and chemical shielding anisotro pies. (For this reason, the term "order parameter" is an unfortunate one, since it conveys the false impression that the order parameter is a measure of some oversimplified notion of molecular order.) In principle, proton spectra should contain some information about molecular orientations and motion. Because of the importance of the dipo lar interaction between the two protons of a methylene group, the partially narrowed spectrum should be sensitive to motions that reorient the line joining the two protons. However, the spectrum is complicated by other dipolar couplings and there is some disagreement (42) about how the order parameters should be extracted from the spectrum. Sonicated dispersions Sonicated dispersions have much narrower lines than unsonicated preparations. The types of motions that can cause such a narrowing are overall rotation of the vesicles, diffusion of molecules around the curved surface of a vesicle, and possibly an enhanced freedom of molecular motion in the curved vesicles as compared to the flatter bilay ers of unsonicated dispersions. There has been some disagreement about the relative importance of these factors in determining the line widths (38, 43-45) and consequently the interpretation of the widths in terms of molecular structure is not certain. The lines are narrow enough that small isotropic chemical split splittings between nuclei in different chemical surroundings can be observed. For example, methyl and methylene protons cause different lines in the spec trum. Measurements of T 1 and T2, which contain information about molecular motion, can be obtained separately for each observed line. Examples of NMR studies of membranes 1 . 31p and 2H studies of polar head groups. Recent work done by three research groups, have provided much detailed information for testing mod els of the head group conformation of phospholipids in membranes. Seelig & GaIly (46) studied bilayers of dipalmitoylphosphatidylethanolamine above and below the phase tradition, using both 3 1p and 2H NMR. In the deuterium work, they used molecules that were selectively deuterated at the ethanolamine carbon atoms. Their data were consistent with a model in which the head group rotates flat on the surface of the bilayer and makes rapid transitions between just two conformations. Kohler & Klein (47) measured the 31P spectrum of bilayers of dipalmitoylphosphatidylethanola mine, DPPC, egg PC, and brain PC. They interpreted their findings in terms of two simple models of head group motion. Seelig, GaIly & Wohlgemuth (48) and Griffin, Powers & Pershan (49) studied DPPC head group orienta-
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
371
tion. The first group worked above the phase transition temperature using unoriented samples and did both 31p and 2H studies using selectively deute rated samples. The second group worked below the phase transition temper ature using oriented samples and performed 31p experiments for a variety of angles of the magnetic field relative to the bilayer normal. Both studies were consistent with a model in which the choline head group is aligned parallel to the bilayer plane. 2. 2H order parameter studies of hydrocarbon chains. Deuterium order parameter studies have been performed for a variety of bilayers in the fluid phase, using specifically deuterated molecules so that order parameters as a function of position along the chain could be measured. Seelig & Seelig studied deuterated DPPC (50) and deuterated I-palmitoyl-2-0Ieoylphos phatidylcholine (51). Stockton et al (45) studied egg PC and egg PC cholesterol bilayers that contained deuterated stearic acid. Stockton et al (52) incorporated specifically deuterated fatty acids biosynthetically into the membranes of Acholeplasma laidlawii. The results in all cases are qualitatively similar; the order parameters are nearly constant near the head group and middle of the chains and then decrease toward the methyl ends, and the values vary with temperature, degree of unsaturation, and mole fraction of cholesterol. There have been numerous attempts (44, 50, 53, 54) to understand the physical meaning of the numerical results, to use them to obtain information about hydrocarbon chain conformation and other types of molecular motion, and to understand the relationship of these deuterium order parameters and the ESR order parameters (see below). As mentioned above, it is relatively easy to disprove models using order parameters but very difficult or impossible to prove the correctness of a model. All these analyses involve so many explicit and implicit assumptions and oversimplifications that it is difficult to know how much credence to give to the conclusions. Of special note is a recent paper by Petersen & Chan (44) which emphasizes the importance of reorientation of the chains as well as rotational isomerization in accounting for the order parameters. 3. Other studies of unsonicated systems. The study of proton spectra in unsonicated systems has led to fewer detailed and conclusive results than has the study of 2H and 3 1p. This is because many different types of protons contribute to a line and because of the complicating effects of dipolar coupling among all the spins. Chan and co-workers (55, 56, 44) have developed a theory for interpreting proton spectra and relaxation times. The effect of dipolar coupling on line shapes has been described differently by Ulmius et al (42) and Bloom et al (57). The study of I3C spectra in unsonicated systems has been hampered by the ineffectiveness of the usual proton decoupling techniques for removing dipolar broadening. Urbina & Waugh (58) have applied a double resonance
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
372
ANDERSEN
technique to the study of DPPC dispersions. This method not only elimi nates dipolar coupling, giving spectra in which the width is determined by the anisotropic chemical shift, but also gives increased sensitivity. Opella, Yesinowski & Waugh (59) applied this technique to the study of cholesterol in DPPC dispersions, using cholesterols specifically enriched with \3C at two positions. 4. Sonicated vesicles. Kroon, Kainosho & Chan (60) measured protein NMR line widths and Tl values for DPPC sonicated vesicles. They used deuterated molecules to perform isotopic dilution experiments to estimate the relative influence of intermolecular and intramolecular interactions on the results. Horwitz, Horsley & Klein (61) measured Tl and T2 values for sonicated egg PC vesicles. They were able to obtain individual relaxation times for several different types of protons: N-methyl protons on the choline head group and a-carbonyl, allyl, vinyl, methylene, and methyl protons on the fatty acid chains. Lee et al (62) measured proton Tl for DPPC and egg PC vesicles and studied the effect of cholesterol. Darke et al (63) also studied the effect of cholesterol on DPPC and egg PC vesicles by measuring the changes in line shapes with composition. l3C T relaxation times give information about the correlation time for 1 reorientation of the CH bonds in a molecule. Individual values can usually be measured for many of the carbon atoms in a molecule. Such studies have been performed, for example, by Levine et al (64) on DPPC and dioleoyl phosphatidylcholine vesicles, by Godici & Landsberger (65) on egg PC vesicles, and by Barton & Gunstone (66) on a series of unsaturated PCs. Fatty acids with magnetic nuclei have been used as probes in phos pholipid vesicles. Stockton et al (45) studied 2H line widths of deuterated fatty acids, and Birdsall et al (67) studied 19F line widths of fluorinated fatty acids. De Kruijff et al (68) studied the high resolution spectrum Of31p in DPPC, distearoylphosphatidylcholine, and dimyristoylphosphatidylcholine vesi cles as a function of temperature. Separate resonances for molecules on the inside and outside layers of the vesicle were seen. ELECTRON SPIN RESONANCE OF SPIN LABELS One of the most fruit ful methods for studying membranes is to introduce stable free radical molecules into them and study the electron spin resonance (ESR) spectrum of the radical in the presence of an external magnetic field. The term "spin label" is often used to describe such molecules. A recent monograph (69) discusses many aspects of spin labeling. In a spin label ESR experiment, the spin label molecule is incorporated into the membrane sample, and the sample is placed in a magnetic field. The absorption spectrum for microwave radiation is measured. (Usually, as in
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
373
the case of NMR, the frequency is held fixed and the magnetic field is varied.) If oriented membrane samples are used, the spectrum can be mea sured as a function of the angle between the magnetic field and the direction normal to the membrane surfaces. The most frequently used spin labels for membrane studies are nitroxides, which contain the :::N-O group. The N atom is usually part of a 5- or 6-membered ring. This group has an unpaired electron spin located primar ily in an atomic p orbital on the nitrogen atom. The axis of the orbital is normal to the plane defined by the nitrogen and oxygen atoms and the other two atoms bonded to the nitrogen. Synthetic techniques have been developed for introducing stable nitrox ide groups into many types of compounds (70). The most common types are 1. molecules that are smaller than organic molecules typically found in membranes and that are soluble in water; 2. derivatives of fatty acids, in which the nitroxide group is attached at various places; 3. derivatives of phospholipids, in which the nitroxide group is attached at various places; 4. sterol derivatives with a structure similar to cholesterol. An example of the first type is 2,2,6,6-tetramethylpiperadine-I-oxyl (TEMPO), which is soluble in both aqueous and nonaqueous phases and which partitions between aqueous and membrane regions when added to membrane preparations. Examples of the second type are doxyI derivatives of fatty acids. (The doxyI group is a five-membered ring containing an NO group.) Examples of the third type are phosphatidylcholine molecules, one of whose acyl chains is a molecule of the second type. The 3-doxyl deriva tives of cholestane-3-one and androstane-3-one- 1 7-01 are examples of the fourth type. Factors affecting the spectrum of a nitroxide radical An electron in a magnetic field can absorb radiation at one particular frequency, much like a proton in a magnetic field. This would ordinarily give a single line in the absorption spectrum. For a nitroxide radical, however, the spectrum is affected by magnetic (hyperfine) interactions between the electron and the magnetic 14N nucleus, which has a spin of one. The result is that there are three equally spaced lines, rather than one, corresponding to the three possible quantum mechanical states of the spin- l nitrogen. For a stationary molecule, the positions of the three lines depend on the orientation of the nitroxide group relative to the external field. For mole cules which can rotate, the same principles of motional averaging apply to
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
374
ANDERSEN
ESR as to NMR. The shape of the spectrum is sensitive to the rate and anisotropy of molecular reorientation. In particular, the spacing between the high field and low field lines are sensitive to the molecular motion which can reorient the direction of the p orbital on the nitrogen nucleus. An important difference between NMR and ESR is that in the latter case the widths of the unbroadened lines is much larger (of the order of 108 Hz rather than 105 Hz), and hence ESR narrowing is sensitive only to motions that take place on a much shorter time scale. The various theories of line shapes for the ESR of nitroxide spin labels are reviewed in several articles in the recent monograph on spin labeling (4, 71-74). One very common way of analyzing an ESR spectrum involves extracting from it a number called an ESR "order parameter" which bears some conceptual similarity to the deuterium and proton NMR order parameters discussed above. When this method of analysis is valid, the order parameter is a number between -1/2 and +1 which has a well-defined meaning in terms of the average ability of the spin label p orbital to rotate during times of the order of 10-8 sec. (Unlike the NMR case, the ESR order parameter is not a measure of the average orientation of the spin label with respect to
the bilayer normal.) If a spin label has an order parameter close to unity, the physical meaning of this result is very clear, namely, that in a time of 10-8 sec the molecular motion does not appreciably change the direction of the p orbital containing the unpaired spin. However, a numerical value of the order parameter significantly less than unity is consistent with a wide variety of possible motions for the spin label. Molecular motion, discussed in the previous paragraph, is the dominant factor in determining ESR spectra. There are, however, three other effects that are of importance for certain applications of spin labels. The first two are the spin-exchange interaction and the dipole-dipole interaction, which can take place if two spin labels are close enough to each other. The third is an effect of the polarity of the surroundings of a spin label upon the spectrum.
Types ofanalysis of nitroxide spin label spectra In order to obtain informa tion about the molecular structure of a membrane from ESR spin label experiments, it is first usually necessary to reduce the data to a form that is more easily interpreted. There are only a small number of ways in which this is done. 1. The spectra can be used to estimate the relative amounts of probe mole cules in different environments. 2. The spectra can be used to derive order parameters or other numerical measures of probe mobility or anisotropic motion.
PROBES OF MEMBRANE STRUCTURE
375
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
3. The spectra can be compared to theoretical spectra calculated from detailed models of the membrane structure and dynamics. 4. The spectra can be used to derive numerical measures of the polarity of the immediate surroundings of the probe molecule. 5. When the concentration of probe molecules is high enough, the effects of spin-exchange interaction and dipole-dipole interaction between probes on the spectra can be used to obtain information about the frequency of collisions between probe molecules and the distance be tween adjacent probe molecules in a membrane. Examples of the use of ESR spin label studies The following are some of the many types of spin label studies that have been performed. [For more comprehensive and detailed reviews see (69, 75, and 76).] 1. Partitioning studies. The partitioning of TEMPO between the nonpo lar interior of artificial bilayer systems and the aqueous surroundings has been used to investigate phase transitions and lateral phase separations. The technique is based on the fact that the solubility of TEMPO in a membrane interior is greater above the phase transition temperature than it is below that temperature and that the spectrum of TEMPO in water is different from the spectrum of TEMPO in a bilayer. A graph of the fraction of label which is dissolved in the membrane shows distinct changes at temperatures at which a one-component system changes its phase or a two-component system enters or leaves a two phase region of its phase diagram. TEMPO partitioning has been used to measure phase transition temperatures for bilayers composed of a single type of lipid and phase diagrams for bilayers containing two types of lipids (77, 78). One technique for studying the effect of proteins on membrane structure is to use TEMPO partitioning to study the effect of proteins on the lipid phase transition of a model membrane. This was one method used by Hesketh et al (79) in their study of a calcium transport protein. The partitioning of phospholipid spin labels between fluid and solid phases is one of the methods used to study calcium-induced lateral phase separations in phospholipid bilayers (SO). Biological membranes may have regions of varying fluidity due not only to lateral phase separation but also because protein molecules can reduce the mobility of adjacent lipids. TEMPO partitioning between the aqueous region and the fluid region of the membrane is the basis for a method for estimating the fraction of lipids in a biological membrane which are in a fluid state (SI). Partitioning methods have been used to develop spin label techniques for measuring some electrical properties of membranes. Castle & Hubbell (S2) have shown that the partitioning of a charged spin label molecule, N,N dimethyl-N-nonyl-N-tempoylammonium ion between the lipid region of
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
376
ANDERSEN
phospholipid vesicles and the surrounding aqueous medium is sensitive to the electrical potential at the surface of the vesicle. They devised a proce dure for estimating this surface potential from an analysis of the ESR spectrum of the label. Cafiso & Hubbell (83) have shown that the partitioning of certain posi tively charged hydrophobic spin labels between a phospholipid vesicle and the surrounding aqueous medium is sensitive to the electrical potential difference between the aqueous solutions on either side of the bilayer, and have devised a method for estimating transmembrane potentials using this effect. 2. Order parameter studies and hyperfine splitting studies. A common way of presenting and analyzing ESR probe data is to derive a numerical value of the order parameter from the experimental spectrum. There are other numerical measures of probe motion which are sometimes used as alternatives to order parameters. The splittings between various hyperfine extrema (maxima and minima) in the measured spectrum or the heights of the extrema are examples of this. They are easier to obtain than the order parameters, which may require more extensive analysis of the spectrum, and for many purposes they are equally useful. Using fatty acid spin labels and phospholipid spin labels with the doxyl group attached to various positions on the hydrocarbon chains, it is possible to insert nitroxide free radicals into a membrane at various distances from the polar headgroup region. These studies are usually performed with doxyl derivatives of fatty acids and phospholipids. For these molecules, the axis of the p orbital containing the unpaired electron is always parallel to the "backbone" of the hydrocarbon chain at the point where the doxyl group is attached. The backbone of a hydrocarbon chain is most easily defined by drawing straight lines from the midpoint of each carbon-carbon bond to the midpoint of each adjacent carbon-carbon bond. Studies using such homolo gous series of spin labels have yielded the intriguing result that the ESR order parameter of the spin label decreases as the spin label is moved farther from the head group of the molecule (84, 85). This variation of the order parameter has been called a "flexibility gradient" and a "fluidity gradient," has been observed in a variety of model and biological membranes (84-86), and appears to be a general property of spin labels in bilayers. (See the section above on 2H order parameter studies of hydrocarbon chains for a brief discussion of attempts to understand the physical meaning of this effect.) Hubbell & McConnell (84) have also used the temperature dependence of the order parameter for a spin labeled phospholipid in a DPPC bilayer as a diagnostic for measuring the phase transition temperature. The transi tion from a solid to a fluid bilayer leads to an increase in molecular mobility
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
377
and hence to a measurable sharp decrease in the order parameter over a very short range of temperature. Van & Griffith (87) have used order parameters and anisotropy parameters of fatty acid spin labels to study the effect of an extrinsic protein, cytochrome c, on bilayers composed of a mixture of cardiolipin and lecithin. Changes in the splittings in spectra of spin labeled phosphatidylserine and phosphatidylcholine have been used to study CaH induced phase transitions and lateral phase separations of bilayers contain ing anionic lipids (80). Schreier-Muccillo et al (88) have used stearic acid spin labels, sterol spin labels, and a stearamide spin label with a nitroxide group in the polar region of the molecule to study bilayers of DPPC and egg PC with various percentages of cholesterol added. Keith, Snipes & Chapman (89) have used two empirical motional parameters, derived from the heights of extrema in ESR spectra, to study the motion of small spin label molecules in various aqueous regions near the surface of dimyristoyl phosphatidylcholine bilayers. The two examples mentioned above are concerned with model mem branes. An example of the use of spin labels to study biological membranes is the work of Butterfield et al (90). They used a spin label that is thought to embed itself in membranes in such a way that the nitroxide group is located near the membrane surface. They measured and compared the order parameters for this spin label embedded in erythrocytes taken from normal subjects and from patients with myotonic muscular dystrophy, Duchenne muscular dystrophy, and congenital myotonia. Relative values of the order parameters are presumably related to the relative freedom of rotational motion of molecules near the membrane surface. 3. Detailed analysis of spectral shapes. Another way to interpret ESR data is to postulate a detailed model of the dynamics of the spin labels, compute a theoretical spectrum on the basis of the model, and compare the calculated results with the experimental spectrum. The agreement between the two provides a test of whether the model is consistent or inconsistent with experiment. Many models contain some adjustable parameters that have physical meaning within the context of the model, and the parameters are varied to obtain a best fit to experiment. Examples of such models are discussed in (74). The major problem with this approach is that one must beware of interpreting the results of this type of analysis too literally, because there may be several similar, or even very different, models that agree equally well with experiment. One example of such a study is the work of Birrell and Griffith (91). They studied oriented multilayers of DPPC containing phospholipids that had doxyl spin labels attached to the hydrocarbon chains near the head group and near the hydrocarbon tail. This work was done at room temperature, which is well below the chain melting transition temperature. They found
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
378
ANDERSEN
that the spin labeled chains of both types were tilted with respect to the normal to the bilayer surface, and they were able to fit their data to a "Gaussian distribution model" for the chain backbone directions. Another example of this type of study is the work of Gaffney & McCon nell (54) who studied oriented bilayers of egg PC at room temperature. They used three different spin labeled phospholipids, with the doxyl groups at various positions along the hydrocarbon chains. A detailed fit of the spectra to a Gaussian distribution model was performed for each spin label. They concluded that the chain backbones of the spin labeled molecules were tilted with respect to the normal to the bilayer. The tilt was found to be greater when the spin label is near the head group than when it is near the methyl end of the hydrocarbon chain. 4. Other types of spin label studies of membrane structure. Certain fea tures of the spectrum of a nitroxide spin label are affected by the polarity of its immediate surroundings. This effect is the basis of several techniques for estimating a polarity index as a function of position inside a membrane, which provides some information about the distribution of water (74, 87). The electron-electron dipole-dipole interaction between spin labels can have a measurable effect on the spectrum if two spin label molecules are close together for a long enough period of time. The spectrum is sensitive to the distance between the spin labels. This principle was used by Marsh & Smith (92) to measure the change in the distance between nearest neigh bor cholestane probes in PC-cholesterol bilayers as the composition of the bilayer was changed.
Calorimetry Because model membranes and some biological membranes can undergo phase transitions, calorimetry techniques that are capable of measuring the latent heat associated with these transitions have been very important for the study of membranes. Standard calorimeters, which measure the total heat capacity or specific heat of a sample, are not useful for dilute phos pholipid suspensions, because the bilayer heat capacity is overwhelmed by the heat capacity of the solvent. Differential scanning calorimetry (DSq, which measures the difference in heat capacity between a membrane suspen sion and an equal volume of the solvent, is the technique used. Recently, improvements in the design of differential scanning calorimeters have led to a new generation of instruments with high sensitivity. [For reviews of calorimetry studies of membranes, see (93, 94).] The data obtained in a DSC experiment is a graph of the heat capacity of the sample, relative to that of a nominally equal volume of solvent, plotted as a function of temperature. If the sample absorbs an especially large amount of heat over a small enough temperature interval, the observed
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
379
heat capacity will have one or more peaks that rise above the baseline. For bilayers composed of one phospholipid, a narrow peak is observed, and the quantities of particular interest are the temperature at which the heat capacity is largest and the area under the peak. They correspond to the transition temperature and latent heat of the transition, respectively. (For some PCs, two separate peaks, corresponding to two transitions, are ob served.) For bilayers composed of two lipids, peaks with a more compli cated shape are observed. The shapes do not have any simple physical meaning, and the goal of an analysis of the data is usually to construct a binary phase diagram from data taken over the entire range of composition for which the lipids have a bilayer structure. Such a phase diagram shows the ranges of temperature and composition over which the bilayers consist of one phase and of two phases. It should be noted that calorimetry is, in a sense, not really a probe of membrane structure. That is, it does not provide any information about the nature of the various different structures that bilayers can have. It tells us, for example, that aqueous dispersions of bilayers of DPPC undergo a phase transition at 41°C, and that therefore it is very likely that the bilayers have a very different structure above this temperature than they do below it. However, other methods must be used to learn what these structures are and how they differ. The following paragraph illustrates the range of membrane studies that have been performed using DSC. For more complete discussions and refer ences, the review articles cited above should be consulted. The transition temperatures and enthalpies for various one-component bilayers have been measured (95-97). For PCs with saturated hydrocarbon chains, two transitions are observed. The phase transition temperature of lipids with ionizable protons can be changed by changing the pH. Also, the presence of divalent cations Ca2+ and Mg2+ in the aqueous solution can affect the phase transition by interacting with the charged head groups or perhaps forming complexes. These effects have been investigated calorimet rically (36). The phase diagrams for binary mixtures of different saturated PCs (97) and of saturated PCs and saturated fatty acids (98) have been determined. The effect of cholesterol on the phase transition of PCs has been studied (99- 102). Also, the lipids and membranes of Acholeplasma laidlawii (103) and of E. coli (104, 105) have been studied using DSC.
Freeze Fracture Electron Microscopy Freeze fracture electron microscopy is a technique for taking a "photo graph" of the interior surface of a membrane, which separates the two halves of the bilayer. [For a recent review of this method, see (106). Also see (107) for a discussion of the technical aspects of the method.]
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
380
ANDERSEN
The first step in such an experiment is a rapid freezing or quenching of the sample that contains the membrane. This is done with Freon or with mixtures of liquid and solid nitrogen. The freezing must be done rapidly enough to prevent any structural changes from taking place. The second step is the fracturing of the sample along one plane. It is generally accepted that membranes tend to fracture along the surface, mentioned above, that separates the two halves of a bilayer. The result then is two new frozen surfaces, and a part of each surface represents the interior surface of bilay ers. In the third step, a replica of one or both surfaces is made by depositing some material, such as platinum and carbon. The replicas are stable at higher temperatures and hence they can be examined by electron micros copy. A resolution of about 30 A can be achieved in the resulting electron micrograph. Various types of physical features can be observed in the resulting pic tures. For lipid dispersions of one- and two-component phospholipids, different textures are seen, corresponding to different phases. The physical meaning of the details of these surfaces is not as important as the fact that they can be observed and distinguished from each other. When two or more such textures appear on the same bilayer, it is clear that lateral phase separation exists in the membrane. Thus the technique is very useful for obtaining or confirming phase diagrams for lipid bilayers. For biological membranes the pictures typically show particles on a relatively smooth surface. The particles may either be homogeneously distributed or concen trated in patches. The particles are interpreted as intramembrane proteins and the smooth surfaces as lipids. Thus freeze fracture electron microscopy is also useful for obtaining information about protein distributions in model and biological membranes. [For a review of freeze fracture studies prior to 1975, see (106).] The following are a few of the more recent freeze fracture studies. Van Dijck et al (108) and Papahadjopoulos et al ( 1 09) used freeze fracture methods in conjunction with calorimetry to study the effect of CaH and MgH ions upon the structure of bilayers composed of dimyristoylphos phatidylglycerol. Chapman et al ( 1 10) studied the effect of monovalent ions on the phase transitions of PC bilayers. Verkleij et al ( 1 1 1) studied the outer membrane of E coli mutants using freeze fracture microscopy.
COMMENTS ON THE INTERPRETATION OF MEMBRANE PROBE DATA Although many different methods for probing membrane structure are available and much work has been done using them, we have surprisingly little unambiguous knowledge about membrane structure on a molecular
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
PROBES OF MEMBRANE STRUCTURE
381
level. An important reason for this is that the quantities measured in these experiments, with the possible exception of X-ray and neutron scattering, are only indirectly related to structure. The arguments used to infer struc tural information from, for example, an NMR line width or a fluorescence depolarization ratio, or even an X-ray scattering pattern, often contain hidden structural assumptions that may be wrong. This, combined with the possibility of artifacts and systematic errors, has led to the publication of many contradictory conclusions about membrane structure. In studying the literature of this field, it is important to be aware of the physical nature of the measurements and the nature of the logic used to interpret the results obtained. ACKNOWLEDGMENTS
This review was prepared while the author was a John Simon Guggenheim Memorial Fellow. The support of the National Science Foundation (Grant CHE 75-06634) and the National Institutes of Health (Grant GM 23085) is also gratefully acknowledged. Literature Cited
I . Singer, S. J. 1974. Ann. Rev. Biochem. 2. 3. 4.
5. 6. 7. 8.
9. 10.
I I. 1 2.
13. 14.
43:805-33 Bangham, A. D. 1972. Chem. Phys. Lip ids 8:386-92 Huang, C., Thompson, T. E. 1974. Methods Enzymol. 32B:485-89 Smith, I. C. P., Butler, K. W. 1 976. In Spin Labeling: Theory and Applications, ed. L. Berliner, New York: Academic. pp. 4 1 1-51 Powers, L., Clark, N. A. 1975. Proc. Natl. Acad. Sci. USA 72:840--43 Powers, L., Pershan, P. S. 1977. Bio phys. 1. 20: 1 37-52 Melchior, D. L., Steim, J. M. 1976. Ann. Rev. Biophys. Bioeng. 5:205-38 Ranck, J. L., Mateu, L., Sadler, D. M., Tardieu, A., Gulik-Krzywicki, T., Luz zati, V. 1974. 1. MoL Bioi. 85:249-77 Akers, C. K. 1 974. See Ref. 3, pp. 2 1 1-20 Tardieu, A., Luzzati, V., Reman, F. C. 1 973. 1. MoL Bioi. 75:71 1-33 Luzzati, V., Tardieu, A. 1974. Ann. Rev. Phys. Chem. 25:79-94 Engelman, D. M. 1970. J. Mol. Biol 47: 1 1 5- 1 7 Engelman, D. M. 1 97 1 . J. Mol. BioL 58: 1 53-65 Esfahani, M., Limbrick, A. R., Knut ton, S., Oka, T., Wakil, S. J. 1 9 7 1 . Proc. Natl. Acad. Sci. USA 63:31 80-84
1 5 . Rand, R. P., Chapman, D., Larsson, K. 1975. Biophys. 1. 1 5 : 1 1 1 7-24 1 6. Janiak, M. J., Small, D. M., Shipley, G. G. 1 976. Biochemistry 1 5 :4575-80 1 7. Brady, G. W., Fein, D. B. 1977. Bio chim. Biophys. Acta 464:249-59 1 8 . Schoenborn, B. P. 1976. Biochim. Bio phys. Acta 457:41-55 19. Neutron Scattering for the Analysis of Biological Structures. Brookhaven Symp. Bioi., Vol. 27. To be published 20. Zaccai, G., Blasie, J. K., Schoenborn, B.
P. 1 975. Proc. Natl. Acad. Set: USA 72:
21. 22.
23. 24. 25. 26. 27. 28.
376-80 B1asie, J. K., Zaccai, G., Schoenborn, B. 1975. Biophys. 1. 1 5 :99a Brown, K. G., Peticolas, W. L., Brown, E. 1973. Biochem. Biophys. Res. Com mun. 54:358-64 Gaber, B. P., Peticolas, W. L. 1977. Bio chim. Biophys. Acta 465:260-74 Lippert, 1. L., Peticolas, W. L. 197 1 . Proc. Nat!. Acad. Sci. USA 68: 1 572-76 Mendelsohn, R. 1 972. Biochim. Bio phys. Acta 290: 1 5-21 Lippert, J. L., Peticolas, W. L. 1 972. Biochim. Biophys. Acta 282:8-17 Lippert, J. L., Gorczyka, L. E., Meik lejohn, G. 1975. Biochim. Biophys. Acta 382:5 1-57 Mendelsohn, R., Sunder, S., Bernstein, H. J. 1976. Biochim. Biophys. Acta 419:563-69
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
382
ANDERSEN
29. Azzi, A. 1975. Q. Rev. Biophys. 8:237316 30. Azzi, A . 1 974. See Ref. 3, pp. 234-46 3 1 . Radda, G. K. 1975. Methods Membr. Biol 4:97-188 32. Waggoner, A. S., Stryer, L. 1 970. Proc. Natl. Acad. Sci. USA 67:579-89 33. Sklar, L. A., Hudson, B. S., Simoni, R. D. 1977. Biochemistry 1 6 : 8 1 9-28 34. Lentz, B. R., Barenholz, Y., Thompson, T. E. 1976. Biochemistry 1 5 :4521-28 35. Lentz, B. R., Barenholz, Y., Thompson, T. E. 1 976. Biochemistry 1 5 :4529-37 36. Jacobson, K., Papahadjopoulos, D. 1975. Biochemistry 14:1 52-61 37. Tecoma, E. S., Sklar, L. A., Simoni, R. D., Hudson, B. S. 1977. Biochemistry 16:829-35 38. Lee, A. G., Birdsall, N. J. M., Metcalfe, J. C. 1 974. Methods Membr. Bioi. 2 : 11 56 39. Mantsch, H. H., Saito, H., Smith, I. C. P. 1 977. In Progress in Nuclear Mag netic Resonance Spectroscopy, ed. J. W. Emsley, J. Feeney, L. H. Sutcliffe, 1 1 :21 1-72 London: Pergamon 40. Seelig, J., Niederberger, W. 1974. J. Am. Chem. Soc. 96:2069-72 4 1 . Niederberger, W., Seelig, J. 1976. J. Am. Chem. Soc. 98:3704-6 42. Ulmius, 1., Wennerstrom, H., Lind blom, G., Arvidson, G. 1975. Biochim. Biophys. Acta 389: 197-202 43. Lichtenberg, D., Petersen, N. 0., Girar det, J.-L., Kainosho, M., Kroon, P. A., Seiter, C. H. A., Feigenson, G. W., Chan, S. 1. 1975. Biochim. Biophys. Acta 382 : 10-2 1 44. Petersen, N. 0., Chan, S. I. 1977. Bio chemistry 1 6:2657-67 45. Stockton, G. W., Polnaszek, C. F., Tul loch, A. P., Hasan, F., Smith, I. C. P. 1976. Biochemistry 1 5 :954-66 46. Seelig, J., Gaily, H.-U. 1976. Biochemis try 1 5:5 199-5204 47. Kohler, S. J., Klein, M. P. 1977. Bio chemistry 1 6 : 5 1 9-26 48. Seelig, J., Gaily, H.-U., Wohlgemuth, R. 1977. Biochim. Biophys. Acta 467:109-19 49. Griffin, R. G., Powers, L., Pershan, P. S. Preprint 50. Seelig, A., Seelig, J. 1974. Biochemistry 1 3 :4839-45 5 1 . Seelig, A., Seelig, J. 1977. Biochemistry 16:45-50 52. Stockton, G. W., Johnson, K. G., But ler, K. W., Tulloch, A. P., Boulanger, Y., Smith, l. C. P., Davis, J. H., Bloom, M. 1977. Nature 269:267-68 53. Schindler, H., Seelig, J. 1975. Biochem istry 14:2283-87
54. Gaffney, B. J., McConnell, H. M. 1974. J. Magn. Reson. 16: 1-28 55. Seiter, C. H. A., Chan, S. I. 1973. J. Am. Chem. Soc. 95:7541-53 56. Feigenson, G. W., Chan, S. I. 1974. J. Am. Chem. Soc. 96: 1 3 12-19 57. Bloom, M., Burnell, E. E., Valic, M. I., Weeks, G. 1975. Chem. Phys. Lipids 14: 107-12 58. Urbina, J., Waugh, J. S. 1974. Proc. Natl. Acad. Sci. USA 7 1 :5062-67 59. Opella, S. J., Yesinowski, J. P., Waugh, J. S. 1976. Proc. Natl Acad. Sci. USA 73:3 8 1 2- 1 5 60 . Kroon, P . A . , Kainosho, M . , Chan, S . I . 1976. Biochim. Biophys. Acta 433: 282-93 6 1 . Horwitz. A. F., Horsley, W. J., Klein, M. P. 1972. Proc. Natl. Acad. Sci. USA 69:590-93 62. Lee, A. G., Birdsall, N. 1. M., Levine, Y. K., Metcalfe, J. C. 1972. Biochim. Biophys. Acta 255 :43-56 63. Darke, A., Finer, E. G., Flook, A. G . • Phillips, M. C. 1972. J. Mol. BioI. 63:265-79 64. Levine, Y. K., Birdsall, N. 1. M., Lee, A. G., Metcalfe, J. C. 1972. Biochemis try 1 1 : 1 4 1 6-2 1 65. Godici, P. E., Landsberger, F. R. 1974. Biochemistry 1 3:362-68 66. Barton, P. G., Gunstone, F. D. 1975. J. Biol Chem. 250:4470-76 67. Birdsall, N. J. M., Lee, A. G., Levine, Y. K., Metcalfe, J. C. 197 1 . Biochim. Biophys. Acta 241 :693-96 68. De Kruijff, B., Cullis, P. R., Radda, G. K. 1974. Biochim. Biophys. Acta 406;6-20 69. Berliner, L., ed. 1976. Spin Labeling; Theory and Applications. New York: Academic. 592 pp. 70. Gaffney, B. J. 1976. See Ref. 69, pp. 1 8 3-238 7 1 . Nordio, P. L. 1976. See Ref. 69, pp. 5-52 72. Freed, J. H., 1976. See Ref. 69, pp. 531 32 73. Seelig, J. 1976. See Ref. 69, pp. 373-409 74. Griffith, O. H., Jost, P. C. 1976. See Ref. 69, pp. 453-523 75. Gaffney, B. 1. 1 974. See Ref. 3, pp. 1 6 1 -98 76. Gaffney, B. J., Chen, S.-C. 1977. Meth ods Membr. BioI. 8:29 1-358 77. Shimshick, E. J., McConnell, H. M. 1973. Biochemistry 12:235 1-60 78. Wu, S. H., McConnell, H. M. 1975. Bio chemistry 14:847-54 79. Hesketh, T. R., Smith, G. A., Houslay, M. D., McGill, K. A., Birdsall, N. J.
PROBES OF MEMBRANE STRUCTURE
80. 81.
82. 83. 84.
Annu. Rev. Biochem. 1978.47:359-383. Downloaded from www.annualreviews.org Access provided by University of Western Ontario on 02/07/15. For personal use only.
85.
86.
87. 88.
89. 90.
91. 92. 93. 94. 95.
M . , Metcalfe, J. C . , Warren, G . B . 1 976. Biochemistry 1 5:4145-51 Ohnishi, S. 1975. Adv. Biophys. 8 :35-82 McConnell, H. M., Wright, K. L., McFarland, B. G. 1972. Biochem. Bio phys. Res. Commun. 47:273-8 1 Castle, J. D., Hubbell, W. L. 1 9 76. Bio chemistry 1 5:48 1 8-31 Cafiso, D. S., Hubbell, W. L. 1978. Bio chemistry 1 7 : In press Hubbell, W. L., McConnell, H. M. 1 97 1 . J. Am. Chem. Soc. 93:314-26 Jost, P., Libertini, L. J., Hebert, V. C., Griffith, O. H. 1 97 1 . J. Mol. Bioi. 59:77-98 Rottem, S., Hubbell, W. L., Hayflick, L., McConnell, H. M. 1 970. Biochim. Biophys. Acta 2 1 9 : 104-13 Van, S. P., Griffith, O. H. 1975. J. Membr. Bioi. 20: 1 5 5-70 Schreier-Muccillo, S., Marsh, D., Dugas, H., Schneider, H., Smith, I. C. P. 1973. Chem. Phys. Lipids 10: 1 1-27 Keith, A. D., Snipes, W., Chapman, D. 1 977. Biochemistry 16:634-41 Butterfield, D. A., Chestnut, D. B., Ap pel, S. H., Roses, A. D. 1976. Nature 263 : 1 59-61 Birrell, G. B., Griffith, O. H. 1976. Arch. Biochem. Biophys. 1 72:455-62 Marsh, D., Smith, I. C. P. 1973. Oio chim. Biophys. Acta 298: 1 3 3-44 Ladbrooke, B. D., Chapman, D. 1 969. Chem. Phys. Lipids 3:304-56 Mabrey, S., Sturtevant, J. M. 1978. Methods Membr. Bioi. 9:ln press Hinz, H.-J., Sturtevant, J. M. 1972. J. Bioi. Chem. 247:607 1-75
383
96. Chapman, D., Urbina, J., Keough, K. M. 1974. J. BioI. Chem. 249:25 1 2-21 97. Mabrey, S., Sturtevant, J. M. 1 9 76. Proc. Natl. Acad. Sci. USA 73:3862-66 98. Mabrey, S., Sturtevant, J. M. 1977. Bio chim. Biophys. Acta 486:444-50 99. Ladbrooke, B. D., Williams, R. M., Chapman, D. 1968. Biochim. Biophys. Acta 150:333--40 100. Hinz, H.-I., Sturtevant, I. M. 1972. J. Bioi Chem. 247:3697-3700 1 0 1 . Barton, P. G. 1976. Chem. Phys. Lipids
1 6 : 1 95-200
102. Mabrey, S., Mateo, P. L., Sturtevant, J. M. 1977. Oiophys. J. 1 7:82a 103. Melchior, D. L., Morowitz, H. J., Sturtevant, J. M., Tsong, T. Y. 1970. Biochim. Biophys. Acta 2 1 9 : 1 1 4-22 1 04. Baldassare, J. J., Rhinehart, K. B., Silbert, D. F. 1976. Biochemistry 1 5 : 2986-94 105. Jackson, M. B. 1976. Fed. Proc. 35: 1 5 3 1 106. Verkleij, A. J . , Ververgaert, P. H . J. Th. 1 97 5 . Ann. Rev. Phys. Chem. 26: 101-22 107. Zingsheim. H. P. 1 972. Biochim. Bio phys. Acta 265:339-66 108. Van Dijck, P. W. M., Ververgaert, P. H . J . Th., Verkleij, A. J . , van Deenen, L . L. M., de Gier, J. 1975. Biochim. Biophys. Acta 406:465-78 109. Papahadjopoulos, D., Vail, W. J., Pang born, W. A., Poste, G. 1976. Biochim. Biophys. Acta 448:265-83 1 10. Chapman, D., Peel, W. E., Kingston, B., Lilley, T. H. 1977. Biochim. Biophys. Acta 464:260-75 1 1 1 . Verkleij, A., van Alphen, L., Bijveit, J., Lugtenberg, B. 1977. Biochim. Biophys. Acta 466:269-82
