singer - The Fluid Mosaic Model of the Structure of...

Info iconThis preview shows pages 1–12. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 12
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: The Fluid Mosaic Model of the Structure of CellMemb’ranes Cell membranes are viewed as two-dimensional solutions - of oriented globular proteins and lipids. S. 1. Singer and Earth L. Nicolson Biological membranes play a crucial role in almost all cellular phenomena. yet our understanding of the molecular organization of membranes is still nail— mentary. Experience has taught as, how— ever, that in order to achieve a satisfac— tory Understanding of how any biologi- cal system functions, the detailed molecular composition and structure of that system must be known. While we are still a long way from such knowl- edge about membranes in general, prog- ress at both the theoretical and experi— mental levels in recent years has brought new astagc where at Ieastthc gross aspects of the organization of e pro— teins and lipids of membranes can he discerned. There are some investigators, however, who, impressed with the great diversity of membrane compositions and functions, do not think there are any useful generalizations to be made even about the gross structure of cell mem— branes. We do not share that view. We suggest that an analogy exists between the problems of the structure of mem— branes nd the structure of proteins. The is r are tremendoust diverse in composition, function, and detailed structure. Each kind- of protein mole— cule is structurally unique. Nevertheless. . generalizations about protein structure have been very useful in understanding the properties and functions oi protein‘ molecules. Similarly, valid generaliza- lions may exist about the ways in which the proteins and lipids are organized in an intact membrane. The ultimate test of such generalizations, or models, is whether they are useful to explain old experiments and suggest new ones. Singer {I} has recently examined in Dr. Singer 1: I protest-or cl biology at the Uni- vets-hr cl' CIIEEoan. at San Diem, La .Tclla. EIr. Nit—clam is a research associate at the Armand Hint-flier Cancer Center at the Salk Institute [or Elolosical Studies, La Jolie. California. 1'10 considerable detail several models of the gross structural organization of membranes. in terms of the thermody— namics of macromolecular systems and in the light of the then available ex— perimental evidence. From this analysis, it was concluded that a mosaic struev ture of alternating globular proteins -and phospholipid hilayer was the only ' membrane mode] among those analyzed that was simultaneously consistent with metrnodynaruie restrictions and with all the experimental data available. Since that article was writtem much new evi~ deuce has been published that strongly supports and extends this mosaic model. In particular, the mosaic appears to be a fluid or dynamic one and, for many purposes, is best thought of as a twin- dhnensional oriented viscous solution. In this article, 'we therefore present and discuss a. fluid mosaic model of mem— brane structure, and propose that it is applicable to most biological mem- branes, such as plasmaiemrnal and in- tracellular membranes, including the membranes of different cell organelles, such as mitochondria and chloroplasts. These membranes are henceforth re- ferred to as functional membranes. There may be some other membrane- likc systems, such as myelin. or the iipoprotein membranes of small animal viruses, which we suggest may be rigid. rather than fluid, mosaic structures, but such membrane systems are not a pri- mary concern of this article. Our objectives are (i) to review briefly some of the thermodynamics oi macro~ molecular, altd particularly membrane, system's in an aqueous environment: {ii} to discuss some of the properties of the proteins and lipids of functional membranes: {iii} to describe the fluid mosaic model in detail; {iv}I to analyze some of the recent and more direct experimental evidence in terms of [hp model; and (v) to show that the fluid mosaic model suggests new wars o1 thinking about membrane functions and membrane phenomena, Thermodynamics and l'vieubrane Structure The uid mosaic model has evolved by a series of stages from earlier vctv sions (1—4}. Thermodynamic considera- tions about membranes and membrane componenh initiated, and are sti-l cen- tral to, these developments. ThESc con- siderations derived from two decades of intensive studies of protein and nu- cleic acid structures; the thermod ynamic principles involved, however, are per- fectly general and apply to any macro- molecular system in an aqueous en; vironment. These principles and their application to piembiane systemahave been examined in detail'elscwhere to and are only summarized here. For our present purposes, two kinds of non- covslcnt interactions are most impor- tant, hydrophobic [5} and hydropitt'l'ic {1}. By hydrophobic interactions it .meant a set of thermodynamic factors that are responsible for the sequester- ing of hydrophobic or nonpolar groups away from water, as, for example, the immiscibility of hydrocarbons and Water. To be specific, it requires the expenditure of 2.6 ltilocalories of free energy to transfer a mole of methane from a nonpolar medium to water at 25°C {5}. Free energy contributions of this magnitude, summed over the many noupolar amino acid residues of soluble proteins,‘are no doubt of primary im- portance in determining the coniomta- tions that protein motoculcs adopt in aqueous solution [a], in which ti -: non polar residues are predominantly st:- qucstcred in the interior or the mole- cules away from contact with water. By hydrophilie interactions is r'eantt set of thermodynamic factors that are responsible for the preference c." ionic and polar groups for an aqueous rather than a nonpolar environment. For ex- ample, the tree energy required to trans fer a mole of zwitterionic glycine from water to acetone is about 6.0 Real at 25°C, showing that ion pairs strongly prefer to he in water than in -'t non- polar medium {1'}. These and related free energy tenns no doubt provide the reasons why essentially all th: ionic residues of protein molecules crech- served to be in contact with water. 5CIENCE. 1L"Ell... 1?! ule. according to x—ray crystallo- phic studies. Similar thermodynamic tuncnts apply to saccharidc residues i}. it requires the expenditure of sub . nlial free energy to transfer a simple 'ateharide from water to a nonpolar heat, and such residues will therefore in a lower free energy state in con— - giant with water than in a less polar ' 'environment. There are other noncoyalent inter— ions, snelr as hydrogen bonding and inf: en- . m__ lructure. However, with respect to gross - meters, with which we are now Lies . 1u_ noenred, these are very likely of sec- }!ic tidal-y magnitude compared to hydro— itot-ie and hydrophilic interactions._ The familiar phosplrolipid bilayer Icture illustrates the combined effects hydrophobic and hydrophilic inter- ions. In this structure (Fig. l} the apolar fatty acid chains of the phos— holipids are sequestered together away at contact with water, thereby maxi- ing hydrophobic interactions. Fur— crnrore, the ionic and zwittetionic ups are in direct contact with the cons phase at the exterior surfaces hc bilayer, thereby maximizing hy— hdic interactions. In the case of terionic phospholipids such as phos- . idylcholine, dipole-dipole interac— _ns between ion pairs at the surface the bilayer may also contribute to stabiliZation of the bilayer structure. applying these thermodynamic ciples to membranes, we recognize :_t that of the three-major classes of ‘tnbrane components—proteins, lip~ and oligosaochat‘ides-n-the proteins i- predominant. The ratio by weight __ir proteins to lipids ranges from about _5_-. mt for these functional membranes 'eh have been well characterized mpare (Flt. A. substantial frac~ of this protein most probably plays ,a important role in dclennirtirtg the :I eture of membranes, and the struo- al'properties of these proteins are reiore of first-order importance. j: nrbrane proteins are considered in _i e detail in the Following section. At juncture, the significant point is ;-_-: ifhyrhophohic and hydrophiiic in- : clients are to be maximized and the _‘1ist'f.ree energy state is to he at- t'redfor the intact membrane in art - cons environment, the nonpolar ; acid residues of the proteins—— gfityith the fatty acid chains of the -|-I E F. pt. I er a E D. u‘ re E n n;- W .. E rashes“ Iss2 Fig. I. A phospho- lipid hilayer: sche- matic cross-sectional yiew. The filled eit- cles represent the ionic and polar heart groups of the phos- phoIipid molecules, which make contact with water; the easy lines represrmt the fatty acid chains. (to the maximum extent feasible} from contact with water, while the ionic and polar groups of the proteins—along with those of the lipids and the oligosac- charides—shorrltt be in contact with the aqueous solvent. These requirements place restrictions on models of mem~ brane structure; in particular, they ren- der highly ttnliltely the classical model of a trilaminar arrangement of a com linuons lipid bilayer sandwiched tre— tween two monolayers of protein. The latter model is thermodynamically un- stable becaUse not only are the non— polar amino acid residues of the mem- brane proteins in this model perforce largely exposed to water but the ionic and polar groups of the: lipid are se- questered by a layer of protein front contact with water. Therefore, neither hydrophobic nor hydrophilic interaca tions are maximized in the classical model. Entire Properties of Metnhmne Components Peripheral and integral proteins. It seems both reasonable and important to discriminate between two categories of proteins bound to membranes, which we have termed peripheral and integral proteins ll}. Peripheral proteins may be characterized lay the following criteria. til They require only mild treatments, such as an increase in the ionic strength of the tttedium or the addition of a cheiating agent, to dissociate them mo- leculariy intact from the membrane: (iii they dissociate tree of lipids; and {iii} in the dissociated state they are relatively soluble in neutral aqueous butters. These criteria suggest that a peripheral protein is held to the mem— brane only by rather weak nonooyalent {perhaps mainly electrostatic) interac— tions and is not strongly associated with membrane lipid. The cytochrome c of mitochondrial membranes, which can be dissociated free of lipids by high salt concentrations, and the protein spectrin {it} of erythrocyte membranes, which can be removed by chelating agents under mild conditions, are eit- amples of membrane proteins that sat— isfy the criteria for peripheral proteins. [in the other hand, the major portion II} it] percent} of the proteins of most membranes haye dltferent characteris- tics, which may be assigned to integral proteins: fl]: they require much more drastic treatments, with reagents such as detergents, bile acids, protein dena- turants, or organic solvents, to dissociate them from membranes; {ii} in many in~ stances, they remain associated with lipids when isolated; {iii} if completely freed of lipids, they are usually highly insoluble or aggregated in neutral aque- ous binders {P}. The distinction between peripheral and integral proteins may be useful in several regards. It is assumed that only the integral proteins are critical to the structural integrity of membranes. Therefore, the properties.._and.7ipterae— tions of peripheral proteins, while in— teresting in their own right, may not be directly relevant to the central prob- lems of membrane structure. The prop erties of cytochrome c, for example. may not be typical of mitochondrial membrane proteins. Furthermore, the biosynthesis of peripheral and integral proteins and their attachment to the membrane may be very different proc— esses. This is not the appropriate ocv casion to discuss membrane biogenesis in any detail, but it may be significant that, although cytochrome c is a mito- chottdriai protein, it is synthesized on cytoplasmic rather than mitochondria] ribosomes; in fact only a small fraction of the total mitochondrial protein {per— haps only the integral proteins of the inner mitochondrial membrane?) ap— pears to be synthesized on mitochon— drial rihosonres {Id}. In any event, because of the relatlycly unimportant membrane structural role assigned to the peripheral proteins, they are not a primary concern of this article. Properties of integral proteins. Since the proteins we have classified as in- tegral, according to the criteria speci— fied, constitute the major fraction of membrane proteins, we asseme that the properties to be discussed apply to the integral proteins. 1} For Seyeral well-characterized membrane systems, including erythro- cyte and other plasma membranes, and mitochondrial membranes, the proteins have been shown to he grosst hetero- geneoUs with respect to moleeular 1'1] weights (fill. There is no convincing evidence that there exists one predom- inant type of membrane protein that is specifically a structural protein; recent reports to the contrary have been with— drawn. We consider this heterogeneity to be more significant for a general model of membrane structure than the fact that in a few specialized instanees, as in the case of disk membranes of retinal rod outer segments (£2. 1.3}, a single protein species predominates. A satisfactory membrane model must be capable of explaining the heterogeneity of the integral membrane proteins. 2} The proteins of a variety of intact membranes, on the average. show ap— preciable amounts of the tic-helical con— formation, as was first shown by Ke" {M}, Wallach and Zahier t4}, and Lenard and Singer (3]. For exatttple, circular dichroism measurements of aqueous suspensions of intact and me- chanically fragmented human erythro— cy-te membranes {provided that we take into account certain optical anomalies of these measurements} reveal that about 4:} pereent of the protein is in the right-handed a—helical conformation tijl. Most soluble globular proteins whose circular dichrnism spectra have been obtained exhibit a smaller fraction of whalix in their native structures. This suggests that the integral proteins in intact membranes are largely globu- lar in shape rather than spread out as monolayers. On the other hand, a membrane model in which such globu— __ —+ + rill lar proteins are attached to the outer surfaces of a lipid bilayer {id} would not be satisfactory because. among other reasons, it would require mem— brane thicknesses much larger titan the TS to 90 angstroms generally observed. A model in 'Urltich globular protein molecules are intercalated within the membrane would, however. meet fitesc restrictions. The phospholipids of rttenthrattes. There is now substantial evidence that the major portion of the phospholipids is in bilayer form in a variety of intact membranes. For example, differential calorimetry of intact rnycoplasnta'_rn_em— hranes Sl'lCIWS that they undergo _a phase transition in a temperature range very similar to that of aqueous dispersions of the phospholipids extracted. fron‘l the membranes ti o. 1?}. Thus the structures of the lipid in the membrane and of the lipid in isolated aqueous dispersion are closely similar; presumably the latter is the bilayer fornt. This conclusion is sup- ported by x-ray diffraction {i8} and spir-label studies (£91 on similar mem- brane preparations. The bilayer character of membrane lipids rules opt models such as that of Benson {25'} in which the proteins and lipids form a single-phase lipoprotein subunit that is repeated indefinitely in two dimensions to constitute the mem- brane. In such a model. most of the lipids would he expected to have dis- tinctly different properties from those of a bilayer. 4. -——+ Fig. 2. The lipidaglobular protein mosaic model of membrane structure: schematic cross-sectional view. The phospholipids are depicted as in Fig. l. and are arranged as a discontinuous bilayer with their ionic and polar heads in contact with water. Some lipid may he structurally ditTcrcntiated from the hull: of the lipid {see text}, but this is not explicitly shown in the figure. The integral proteins, witlt the heavy lines repre senting the folded polypeptide chains. are shown as globular molecules partially etu- bcdded in, and partially protruding from, the membrane. The protruding parts have on their surfaces the ionic residues {— and +21 of the protein, while the nonpolar residues are largely in the embedded parts; accordingly, the protein molecules are arnv phipathic. The degree to which the integral proteins are embedded and,_ in particular. whether they span the entire membrane thickness depend on the sire and structure of the molecules. The arrow marks the plane of cleavage to be expected in freeze-etching experiments (see text). [From Lenard and Singer {3} and Singer Ell] 122 Two qualifications should 5 c stresse however, concerning the bi iyer for of membrane lipids. [it Time of ti evidence so far obtained for 'he bilayt form permits us to say it"tcrhcr Ii bilayer is continuous or inter rrptacf [i The calorimetrically obser ed pha- transitions, for example, oe or over broad temperature interval, a owing Ll' possibility that the coopet'ari c unit ir volved in the phase transitii-n is quit small. consisting perhaps of only ill"I lipid moldeulcs on the averagr. (ii) Nos of the experiments mentionel above sufficiently sensitive and qua titative t prove whether 1040 percent o' the pho! pholipid is in the bilayer l1rrn. Iti therefore not excluded that s ue signif cant fraction of the phosphr 'ipid {per haps as much as 30 perceru- is physi cally in a different state fro a theres of the lipid. Protein-lipid interactions in il‘iEl'll 's . drones. Seueral .li'tnds of c.perimen1 indicate that protein—lipid interaction play a direct role in a '.'ariely a membrane functions. Many | ternbrane bound enzymes and antigE=ts requin lipids. often specific phospht‘ilpids, to the expression of their acr' rities [5th table 2 in [It‘ll]. Further wore, lhi nature of the fatty acids in orporatei into phospholipids affects th' funetior of certain membrane-hound g~rotelns it bacterial membranes {22}. On the other hand, the r Iorirnelris data discussed above give no tignificart indication that the associatii 1 of pro teius with the phospholipids of intact membranes afiiects the phase ransirion of the phospholipids thenu lves. Et- perimcnts with phDSpholip‘e e (3 anti membranes have shown th: the co zytnic release of 10 percent of the phosphorylated amines fr: 'n intaa erythrocyte membranes :rofoundlt pcrturbs the physical state ot the resid- ual fatty acid chains. but has we detect able effect [as measured 1." circular diehroism spectra} on the aria-rage con formation of the membran-. proteins [2}. Such results therefore sI sgest that the phospholipids and pr items all membranes do ttot interact s' 'oneg: in fact, they appear to he lar-er inde- pendent. , This paradox, that difierel types all experiments suggest strong p: ttein—lipiti' interactions on the one hand. and weak or no interactions on the other, can in resolved in a manner consi~tent ailh all the data. if it is proposed that, whth the largest portion of the pl' tspholipid is in bilayer form and ne strongly coupled 'to proteins in the I-entbranr, SCIENI.’ . Vt'tl" [Ti ' . a small fraction of the lipid is more tightly ouptcd to protein. 1iiitfith any i i one I'll rnhranc protein, the tightly ' _ coupled lipid might be specific; that is, i.- 1he interaction might require that the If phosphclipid contain specific fatty acid .- chains t r particular polar head groups. ,_ 1 There is at preseut, however, no satis— I i' factory .lirect evidence for such a dis- t finctive lipid fraction. This problem is _ E Hunt}- sn-nrtrtre of the proteins and _ lipids rt uterrti'n-trnes. The thcrtnodye -' _{.Itt.rnic considerations and experimental jitesults so Far discussed lit in with the _ Fidea of a mosaic structure for mem— " __ilt‘altlt§. t.-'—.i'. 2c“) in which globular mol- ecules o: the integral proteins {perhaps in particular instances attached to oli- f=goeaecb rides to form glyeoproteins, inter.._'ting strongly with specific lip- to it rm Iipoproteins} alternate with tiotts of phospholipid bilayer in the tau sc.tion ot- the membrane (Fig. 2}. din globular protein molectlles are pos- - ttttlated to be antpl‘tipathic {3, 4] as are :llttt phtupholipids. That is, they are .slntctur.:l|y asymmetric, with one highly pillar ed and one nonpolar end. The 'itigltly polar region is one in which the ' r.ittttic amino acid residues and any co-,Qvalcttllt hound saccharlde residues are tlttslcrcc. and which is in contact with :;t_he aqttt-ous phase in the intact memv :ghrane; t::c nonpolar region is devoid of iienic and saccharitie residues, contains phony ot' the nonpolar residues, and is _ .tntltcddeti in the hydrophobic interior “at the membrane. The amphipathic Eslturture adopted by a particular ins "jttgral protein {or lipoproteinl molecule, and therefore the extent to Which it is .- lhcddetl in the membrane, are under umet-t-nan-tie control; that is, they 're determined by the amino acid se- 'uence .tud covalent structure of the mtein. and by its interactions with its nnleeuler environment, so that the tree of the system as a whole is at a iairnutn. An integral protein molecule f ' the appropriate size and structure, _'r a suitable aggregate of integral pro— := ins [b 'lowl mayr transverSc the entire .It embrace {.i}; that is, they have re- t'ens it. contact with the aqueous sol- .f at on both sides of the membrane. _' It is t'lear from these considerations 'at tlill'erent proteins, it' they have the proprate amino acid seqitence to Ell-RI 'HR‘;' 191'? adopt an amphipatbic structure, can be protein or in the association of two or integral proteins of membranes: in this manner, the heterogeneity of the pro- teins of most functional membranes can be rationalized. The same considerations may also ex— plain why some proteins are membrane- bound and others are freely soluble in the cytoplasm. The dilterence may be that either the amino acid sequence of the particular protein allows it to adopt an ampltipathic structure or, on the contrary, to adopt a structure in which the distribution of ionic groups is nearly spherically symmetrical, in the lowest free energy state of the system. l1E the ionic distribution on the protein sur- face were symmetrical, the protein would he capable of interacting strongly with water all over its exterior surface, that is, it would be a monodispcrsc sol- uble protein. The mosaic slroctttre can be readily diversified in several ways. Although the nature of this diversification is a matter of speculation, it is important to recognize that the mosaic structure need not be restricted by the schematic rep- resentation in Fig. 2. Protein-protein interactions that art.- not explicitly con- sidered in Fig. 2 may he important in deterntining the properties of the mem- brane. Such interactions_ may result either in the specific binding of a peripheral protein to the exterior ecs- pnsett surface of a particular integral at r. ’ 0': I'r . \\ . fiwa an u,"- . Ilin'i was; .- J -"., ,l \ "' c . 1'l-glt KEN ye more integral protein subunits to form a Specific aggregate within the mem— brane. These features can be accom- modated in Fig. 2 without any changes in the basic structure. The phospholipids of the mosaic structure are predominantly arranged as an interrupted bilayer, with their ionic and polar head groups in contact with the aqueous phase. As has been dis- cussed, -however, a small portion of the lipid may he more intimately associated with the integral proteins. This feature is not explicitly indicated in Fig. 2. The thickness of a mosaic membrane would vary along the surface from that across a phospholipid bilayer region to that across a protein region, with an average value that could be expected to corre— spond reasonably well to experimentally measured membrane thicknesses. Matrix of the moon's.- tight or pro- tet'n? In the cross section of the mosaic structure represented in Pig. 2, it ignot indicated whether it is the protein oEEthe phospholipid that provides the matrix of the mosaic. In other words, which com- ponent is the mortar, which the bricks? This question must be answered when the third dimension of the mosaic struc— ture is specified. These two types of mosaic structure may be expected to have very diHerent structural and func— tional properties, and the question is therefore a critical one. It is out by- Fig. 3. The lipid—globular protein mosaic model with a lipid matrix {the fluid mosaic model}; schematic three-dimensional and cross-sectional views. Tlte solid bodies with stipplcd surfaces represent the globular integral proteins, which at long range are randomiy distributed in the plane of the membrane. J‘st short range, sonte may form specific aggregates. as shown, In cross section Fig. 2 appties. and in other details, the legend of TI] pethesis that functional cell membranes have a long—range mosaic structure with the lipids constitutingthe matrix. as is shown in Fig. 3. Supporting evidence is discussed later. At this point, let us consider some of the consequences of this hypothesis. I} There. should generally be no long— rangc order in a mosaic membrane with a lipid matrix. By long range, we mean ever distances of the order of a few tenths of a micrometer and greater. Suppose we have a membrane prepara~ tien containing many different protein species, and suppose further that littluli molecules of protein A are present in the membrane of a single cell or or— ganelle. How is protein A distributed over the membrane surface?I If the membrane proteins formed the matrix of the mosaic, defined by specific con- tacts between the molecules of different integral proteins, protein A might be distributed in a highly ordered, two- dimensional array on the surface. On the other hand,I if lipid formed the matrix of the mosaic, there would be no long—range interactions intrinsic to the membrane influencing the distribution of .4 molecules, and they should there- fore be distributed in an aperiodic ran- dom arrangement on the membrane surface. The absence of long-range order should not be talten to imply an ab— sence of short-range order in the mem- brane. It is very lilter that such short- range order does exist, as, for example. among at least some components of the electron transport chain in the mito- chondrial inner membrane. Such short— range order is probably mediated by specific protein (and perhaps protein—ipid} interactions leading to the forma- tion of steichiometrically defined ag- gregates within the membrane. How- ever, in a mosaic membrane with a lipid tnatrix, the long-range distribu— tion of such aggregates would be ex— pected to he random over the entire surface of the membrane. The objection may immediately be raised that long-range order clearly exists in certain cases where differen- tiated structures {for example, synapses} are found within a mentbrane. We sug— gest, in such special cases. either that short-range specific interactions among - integral proteins result in the formation of an unusually large two-dimensional aggregate or that some agent extrinsic to the membrane {either inside or out- side the cell] interacm multiply with specific integral proteins to produce a clustering of these proteins in a limited T14 area of the membrane surface. In other words, we suggest that long-range random arrangements in membranes are the norm; wherever nonrandom distri- butions are found, mechanisms must exist which are responsible for them. 2} It has been shown that, under physiological conditions, the lipids of functional cell membranes are in a fluid rather than a crystalline state. {This is not true of myelin, however.) This evidence comes from a variety of sources, such as spin~1abeling expel-iv menu (25), x—ray diffraction studies {id}, and differential calorimetry (id. H}. If a membrane consisted of integral proteins dispersed in a fluid lipid matrix, the membrane would in effect be a two— dimensiunal liquid-like solution of men— omeric or aggregated integral proteins {or lipoprotcins} dissolved in the lipid bilayer. The mosaic structure would be a dynamic rather than a static one. The integral proteins would be expected to undergo translational diffusion within the ntembrane, at rates determined in part by the effective viscosity of the lipid, unless they were tied down by some specific interactions intrinsic or extrinsic to the membrane. However, because of their amphipathic structures, the integral proteins would maintain their molecular orientation and their degree of intercalation in the membrane while undergoing translational diffusion in tile plane of the membrane {as dis— cussed below}. in contrast, if the matrix of the mo— saic were constituted of integral pro- teins, the long-range structure of the membrane would be essentially static. Large energies of activation would be required for a protein component to diffuse in the plane of the membrane from one region to a distant one be— cause ef the many noncnvalent bonds between the proteins that would have to be simultaneously broken for ex- change to take place. Therefore. a mosaic membrane with a protein ma— trix should make for a relatively rigid structure with essentially no transla- tional diffusion of its protein compo- ncnts within the membrane. From the discussion in this and the previous section, it is clear that the fluid mosaic model suggests a set of structural properties for functional membranes at least some of which can be tested experimentally. In an earlier article [1}, a large body of experimen- tal evidence was examined for its rele- vance to models of membrane structure. It was concluded that a mosaic strUc- ture was most consistent with the avail— able evidence. Some more recent suits, however, bear even m-Irc-dirct on the problem, and only this evidEI is discussed below. Some Recent Experimental I;videnct Evidence JFor proteins embedded uremic-ratios. Cine proposal c-i the fit mosaic model is that an integral p: tein is a globular molecule having significant fraction of its 1. tlttme e- bedded in the membrane. "he rose of recent freeze-etching IllirJlfl'il'l'ifll with membranes strongly stt_..-gcst tha: substantial antouut of protein is dee; embedded in many functional me branes. In this technique {2d} a free specimen is fractured with a inicrotet knife: some of the frozen water is in. limed fetched} from the fractured st face if desired; the surface is th- shadow cast with metal, and the surfa replica isegtamiifigi in the electron n erescope. By this—method the tops- raphy of the cleaved surtace is I . vealed, A characteristic fee-are of t exposed surface of most ionctior membranes examined by thiL techniqt' including plasmalemmal, vasttolar, it clear. chloroplast, mitochot-drial, at bacterial membranes {25". 28), is mosaicvlilte structure eonsi--ting of smooth matrix interrupted by a lat_ number of particles, Thea.- partlcl have a fairly characterist'e unifor size for a particular men iirane, f- example, about iii—fit diametc r for cryt tocylc membranes. Such s 'rfaues r sult from the cleavage of a membrar along its interior hydrophobic fat (29}. This interior face (Fig. El cerr spends to the plane indicated by tl arrow. If cleavage were to ace: smoothly between the two layers i phospholipid in the bilayer regions, b: were to circumvent the protein rnol: cules penetrating the mfduplnrte of ti membrane, then. the alternating mom and particulate regions observed on ti. freeze—etch surfaces can be .--eadiiy et plaincd by a mosaic structt-rc for It: membrane (Fig. 2}. providri that th particles can be shown to be protei in nature. That the particles are indec protein has been suggested by recer experiments {3d}. Another consequence of he mesa: mode], suggested from its inceplie {3}, is that certain integral proteins po= scssing the appropriate site and Slim one may span the entire t'i'tckness t the membrane and be expo- ed at but membrane surfaces. Chemical cvidertr SCIENIT'l-L, VOL I'. it re- rectly :ie not: ;- the number of list a trans-membrane protein, whose molecular weight is about 100,001], is present in large amounts in the human erythrocyte membrane has been ob- tained by two independent methods— eae involving protenlysis of normal omnpared to everted membranes {3!}, and the other specific chemical labeling of the membrane proteins {32]. Distribution of components in the pit-art.- of the naenrbrnne. A prediction of the fluid mosaic mode] is that the nto~dimensional longerangc distribution of any integral protein in the plane of the membrane is essentially random. To test this prediction, we have devel— opcd and applied electron microscopic techniques to visualize the distribution of specific membrane antigens over large areas of their membrane surfaces iii} and have so far studied the dis— In'hution of the Rhath} antigen on htlman erythrocyte membranes (3d), and of H—Z histocompatibility alloantigens on mouse erythrocyte membranes {5‘5}. In the case of the Ethel-D) antigen, for example, cells of CI, Rh-positive type were reacted with a saturating mount of mI-labeled purified human antibody to Ritual}. [anti—Rhomll, and I Fig. 4 {left}. The outer memb body to Elliott!) 'and then IS FEBRUARY I??? rane surface of an Rh- _‘ ien-itin-coniugated goat antibody to human suglnbulin. The coils were Ii _ Iysed at an air-water interface. I I:_ tnw magnification} were picked up on _= bodies to human s-globulin. The fertitin appears bound to the membrane .': cluster is circumscribed by a circle of radius Edi] A. The number of such cl '“I—labtiled human antibody to RhufD] molecules ham to an individual RhttD] antigenic site. Scale is lit] am; .i {'rigi'ttl. The outer membrane surface on a mouse erythrocyte (H-Z‘l sens _ .'ty antigens and stained with ferritin-oonjugated antibodies a 1 the legend to Fig. 4. The Ferritin~antihody clusters are prose face. Scale is tl.| #111. Ian1 Nicolson, Hyman, and Singer (35)] the treated {sensitized} cells were lysed at an air—water interface, so that the cell membranes were spread out flat. The flattened membranes, after being picked up on an electron microscope grid, were treated with the specific “in— direct stain," ierritin—eonjugated goat antibodies specific for human y-globuv lin. Thus, wherever the human anti—Rhu {D} molecules were bound to the Rha- {D} antigen on tlte membrane surface, the ferritin-laheled goat antibodies he— carnc specifically attached. In other words, the human y-globulin antibody now functioned as an antigen for the goat antibodies (Fig. 4). The ferritirt 1was distributed in discrete clusters, each containing two to eight ferritin Inolo- clues within a circle of radius about EDD A. The numbers of such clusters per unit area of the membrane surface corresponded to the number of 13'51— labeled human anti-Rhgtfll molecules bound per unit area. This indicates that each ferritin cluster was bound to a single anti—Rhgfflt molecule. and a clus- ur represents the number of goat anti-body molecules bound to a single human y-globnlin molecule. Each clus- ter therefore corresponds to a single nt in randomly spaced " positive human erythrocyte sensitized with human RhafD} antigen site {345) on the mem— brane. Since the clusters were distrib- uted in a random array, we conclude that the atht antigen, which exhibits preperties of an integral protein [3?], is molecularly dispersed and is distrib— uted in a random two—dimensional array on the human erythrocyte membrane. Similar experiments were carried out wit-h the H—2 alloantigenic sites on mouse erythrocyte membranes. In this case (Fig. 5} the clusters of ferritin molecules of the indirect stain were not isolated, as in the case of the RhntD} antigen, but instead occurred in patches. Tho patchy distribution of the H—2 histocontpatibility alloantigenic sites had earlier been observed by different tech- niques (33], but the two—dimensional distribution of the patches could not be ascertained. In our experiments, the patches contained variable numbers of clusters, and were arranged in an it— t'egtdar two-dimensional array on the membrane surface. The .histocompati- bitity antigen appears to'lbc glycopro- rein in nature {so}. The lon‘gLrange dis- tribution of botit the Rheum!) and 1-1-2 histocompatibility antigens on their to- spective membrane surfaces, therefore, :" Elli-Rho fur-W "- [Dl and-stained with rat labeled to saturation with purified ‘5 -labeled human anti- The erythrocyte membrane ghosts, flattened by mrfaee forces (inset, a coated, electron microscope grid and indirectly staine d with fen'itin—coniugated goat anti- in discrete clusters of two to eight fenitin-ooniugates; each ustcrs per cell [‘33de is equal within experimental error to nd per cell (102.0(1). Each cluster therefore corresponds inset scale is l-p-Ifl. [From Nicolson, Masouredis, and Singer til-fl] Fig. [tine/d with alloantibodies against H—Z“ histoeompatibil— gainst ‘i'd' mouse y-globulin. The procedures are thit- same as listed in patches“ of variable site on the membrane sur- 125 are in accord with the prediction of the fluid mosaic model that the integral proteins of membranes are randomly arranged in two dimensions. The particles on the inner membrane faces revealed by freeze-etching experi- ments, which {as discussed aboch are probably protein in natttre, are generally also relatively randomly distributed in two dimensions. Evidence that proteins are in o J'i‘nr‘rf store in inrttcr membranes. An im~ portant series of experiments has been carried out (£2, 404d} with receptor dist-t membranes from the rcu'na of the frog. This membrane system is Unusual in that it contains as its predominant, if not only, protein component the pig— ntent rhodopsin. in electron microscopy. of the negatively stained surfaces of the dried membranes, a somewhat tightly packed and ordered array of par- ticles {about till A) was observed. These particles are the individual rhodopsin molecules. Although the earlier studies suggested that there was a long—range order in the distribution of the particles {40‘}, more recent satay diffraction data {42] on pellets of wet, receptor disk membranes claimed that only a few orders of reflection were observed cor— responding to the spacings of the rite- dopsin molecules in the plane of the membrane. This indicated that a nose crystalline, aperiodic arrangement of the particlesexisted in the plane of the membrane. Furthermore, the tempera— ture dependence of the characteristics of the x-ray diffraction maxiina were Consistent with the suggestion that the particles were in a planar liquidvlilte state in the intact membrane. Additional support for the existence of this liquid- liite state was the observation that the absorption of a foreign protein {bovine serum albumin} to the membrane could definitely alter the x-ray spacings due to the rhodopsin particles; that is, the distribution of die thodopsin molecules in the plane of the membrane was tad— ically altered by the weak binding of the albumin. This alteration would not be expected if a rigid lattice structure of the rhodopsin molecules, or aggre— gates, were present in the plane of the membrane. These studies are particularly note— worthy because they involved a ment- brane which, by conventional electron microscopic techniques, appears to show long—range periodicity over its surface. Either specialized membranes have also exhibited ordered electron micrographic images of their surfaces [compare Mil]. However, it is likely that a very concen- "1'16 trated two-dimensional tluid solution of identical protein molecules will appear, when dried, to be arranged in an or— dered array, particularly when optical tricks are used to enhance the apparent order {43). What is really a fluid phase may therefore artifactually be made to appear as a crystalline solid. This ap- pears tn be the situation with the reti- nal receptor disk membranes. A major contribution to membrane studies has been made by Frye and Edidin [44}, who investigated the mem— brane properties of some cell fusion heterolraryons. Human and ntousc cells in culture were induced to fuse with one another, with Sendai virus as the fusing agent. The distribution of human and mouse antigenic components of the fused cell membranes was then deter- mined by immunofluorescence, with the use of rabbit antibodies directed to the whole human cells, monsc antibodies directed against the H—2 alloantigcn on the mouse cell membranes, and, as in- direct stains, goat. antiserum to rabbit y—globulin and goat antiserum to mouse y-globulin labeled with two different fluorescent dyes. Shortly after cell fu— sion, the mouse and human antigenic components were largely segregated in different halves of the fused cell mema branes', but after about 40 tninutes at 3l°C the components were essentially completely intermixed. Inhibitors of protein synthesis, of adenosine triphos- phate {ATP} formation, and of gluta— mine-dependent synthetic pathways, ap- plied before or after cell fusion, had no effect on the rate of this inlermixing process, but Iosvering the temperature below I5°C sharply decreased it. Frye and Edidin [44] suggest that the interntixing of membrane compo— nents is due to diffusion of tltese com- ponents within the membrane, rather than to their removal and reinsertion, or to the synthesis and insertion of new copies of these components, into the heteroltaryon membrane. r'tn unex- plained finding of these experintents was the fairly frequent occurrence, at early and intermediate times after cell fusion, of heterokaryen membranes in which the human antigenic components were uniformly distributed over the uteru- brane surface but the mouse compo— nents were still largely segregated to about half the membrane {MHz-H1 cells]. {in the other hand, the reverse situation, with the mouse antigenic components uniformly spread out over the membrane and the human compo— nents segregated (I'm-Hm], was only rarely observed. This result can now he .mentbrane. Thus, at appropriate inter- explaincd by a. diffusion mechanism for ' the intermixing process, as follnus. The ._ antibodies to the human cell membrane were no doubt directed to a heteroge- ncous set of antigens, whereas the anti- bodies to the mouse cell were directed ' specifically to the histocompatibility alloantigcn. HDWCVBI‘, ll'll: hlsluuumlmti- . biiity antigens occur as large aggregates ' in the membrane (Fig. 5], and might therefore be expected to diffuse more slowly titan a complex mixture of largely unaggregated human antigens in the mediate times after cell fusion. tignifi- - '- cant numbers of firing-H1]: but -not of {Ml-Hug} fused cells might appear, to be converted at longer times in cells with completely intermixed components. A rough estimate may be made of ' the average effective diffusion constanI _ '- required of the membrane components -_ to account for the kinetics of in:errnit- -: ing in the Frye-Edidin eXperruans Taking the average distance of migra- tion, 1, to have bcet‘fbnbout 5 micro meters in a time, t, of db minute. gives an apparent diffusion constant, £.'=s=l 2:, of 5 X Ill—1‘l cm2fsec. For com- parison, the diffusion constant of hemo- globin in aqueous solutions is about -_ .1 l x 10—7 cmlrscc. The apparent elfeo _ -tive viscosity of the membrane fluid U phase is therefore about [03 1o 1|]E times that of water. The Frye-Edidin experiments can in rationalized by the fluid mosaic model of membrane structure as being the re- sult of the free diffusion and inlcr.tti:tiug of the lipids and the proteins [or lipo proteins} within the fluid lipid rratt‘ix. Some experiments, however, appear _ to suggest that the lipids of membranes _-_' are not readily interchangeable within the membrane and are therefore not free to diffuse independently. For es- ' ample, Wilson and Fox {23} havt studied the induction of B-galactosirlr I' and B—glucoside transport systems it mutants of Escherichia cell that cannot synthesise unsaturated fatty acids. Such fatty acids can be incorporate-I intr- phospholipids. however, if they are sup— _ plied in the growth medium. ‘When celb ' were grown in particular fatty acid sup plements and induced for the synthesis of the transport systems, the effect at temperature on the transport ra'-.: was characteristic of that fatty acid. It then, the cells were first grown in nteditutt containing oleic acid and then shifted to growth in a medium supplemented with linoleic acid during a brief period of --' induction of either of the transport sys- ' tems, the effect of temperature on trans- ' sotENca, yet. 11‘! = port-was said to be characteristic of cells 'Igrovrn continually in the linoleic acid :Emetiium. ln other words, although most jof the pliophoiipids of the membrane Ecaataincd oteic acid chains, these did rsth appear to exchange with the newly ,gthiasaca small amounts of phospholi- 'plds con1aining linoleic acid chains. _ These e-.periments, however, do not '_nscessari contradict the thesis that pins: of t.'.e phospholipids of membranes freeh- diffuslblc and, hence, ex— changeabic. i~'or example, each of the 'rt'Ia'c transport systems might be or- .ganized in the membrane as a specific :protein .=ggregate containing intercol- ' dated and strongly bound phospholipid components. If such Iipoprotein aggre— -- gates had first to be assembled in order 151:; be incorporated into the aura lipid 'natris o." the membrane, the resttlts of .' I'Wilsun and Fox would be anticipated. _ in parlicular, the small fraction of the membrane phospholipid that was -_ strongly hound, and perhaps segregated in such aggregates from the bulk of the -_mernbrane lipidII might not exchange _ rapidly with the bull-t lipid. The 1|Wilson- _-Fei experiments therefore do not re- ._ gain: that the major part of the mem- -braue phospholipid be static, but only that a small fraction of the lipids be _ «structurally differentiated from the rest. " 'Z'Ihc structural differentiation of some :of the membrane lipid by strong bind— to integral proteins is a possibility f' that was discussed above. ‘ .1. The observations of Wilsou and Fox, that there is a significant coupling of " lipid and protein incorporation into __taembrai.cs, appear to be a special case. :'_Iite experiments of Mindich {4.5} dem— :oastrate that more generally lipid and _[:rru1ein incorporation into bacterial membranes can occur independently. _'3t]d that quite Initride variations in the ratio of lipids and proteins in the membrane - ;can he produced in vivo, as might be -.etpectetl from the fluid mosaic model i membrane structure. .- The asymmetry of membranes. A substantial amount of evidence has no umulatetl showing that the two sur- I: of membranes are not identical - - composition or structure. One aspect in this asymmetry is the distribution of _-'I igasaccharldes on the two surfaces of charides on membranes in the electron microscope [33}. For example, the fer- ritin conjugate of coneanaValin A, a protein agglutinin that binds specifically to terminal mD—gltteopyranosyl or mow mannopyranosyl residues {46], attaches specifically to the outer surface of eryth- rocyte membranes and not at all to the inner cytoplasmic surface 1:33]. A similar. completely asymmetric distri- bution of ferritin conjugates of riein {a protein agglutinin] on the membranes of rabbit erythrocytEs is shown in Fig. e. Rieirt binds specifically to terminal ,d—n-gaiactopyranosyl and sterically re— lated sugar residues {45"}. Such asym- metry has now been observed with several ferritin~eonjugatcd agglutinins and a number of different mammalian eel] plasma membranes {43}. These tind— ings extend earlier results obtained by different methods [49]. The foregoing observations bear on many problems. including cell-cell inter- actions and membrane biogenesis [5th. to thc'eontcat of this article, however, the absence of oligosaccharides on in— ner membrane surfaces indicates that rotational transitions of the glyeopro- teins of erythrocyte and other plasma membranes from the enter to the inner .o, surfaces must occur at only negligibly slow rates. This conclusion probany applies to membrane proteins other than glyeoproteirts: for example, the Na,K-depcndent and trig-dependent adenosine triphusphatase activities of erythrocyte membranes are exclusively localircd to the inner cytoplasmic sur~ faces {5}]. Individual molecules of spitt- labclcd ZWittcrionic and anionic phos- pholipids also exhibit very slow inside- outside transitions in synthetic vesicles of phospholipid bilayers {.52}. The very slow or negligible rates of such transi- tions can be explained by the mosaic moch and the thermodynamic argu- ments already discussed. If the integral proteins {including the glycoprotcinsi in intact membranes have, like the phos— pholipids, an amphipathic structure, a large free energy of activation would he required to rotate the ionic and polar regions of the proteins through the hydrophobic interior of the membrane to the other side, ' _ E To accommodate the fluid rn'dhpie model to these conclusions concernirig asymmetry, we specify that, while the two-dimensional translations] 'ditfusion of the integral proteins and the phos- pholipids of memhranes occurs freely, . dens. .. . ._ . .. .. -. Fig. 6. The inner ti] and outer to} membrane surfaces of a rabbit cqrthroeytc mem- brane that has been stained with tortilla-conjugated ricin. In preparing membrane speci- mens such as are shown in Figs. 41 and 5, occasionally a cell'lyses with membrane rupture such that both inner and outer Surfaces of the membrane are exposed. In this case the mounted membrane was stained with ferritin conjugated to ricitt, a plant. agglu- tinin that specifically binds to terminal ttvii-galactopyranosyl and sterieally related terminal sugar residues in ollgosaccharides The ferritia-agglutinin is found on the outer membrane surface only. The scale is equivalent to {Ll am; the insert scale is equiva- lent to: 1 pm. J ‘r i-. In. embranes. There exist plant proteins, ' ed Iecu'ns or plant agglutinins, which led to specific sugar residues, and, as resell. can cause the agglutination of ' lls bearing the sugar residues on their uitaccs. By conjugating several such ggltltinins to territin, we have been able visualise the distribution of oligosaca '_ ran at an! 1er 12: I' the rotational diffusion of these corn— poncnts is generally restricted to axes perpendicular to the plane of the mem— brane; that is. in general, molecular tumbling does not occur at. significant rates within the membrane. The asym- metry of the membrane introduces another factor into the problem of translational diffusion of membrane components. In the experiments of Frye and Edidin (44} only those membrane antigens exposed at the outer surface of the membrane were labeled by fluores- cent antibodies, and the conclusion that these particular antigens were mobile in the plane of the membranes therefore, strictly speaking, applies only to those components accessible at the outer sur— face. 1|til-"hether components confined to the inner surfaces also intermix and diffuse should be separately established. Thus, recent evidence obtained with many experimental methods and differ— ent ltinds of functional membrane sys- tems is entirely consistent with the pre- dictions of the fluid mosaic moch of membrane structure and provides strong support for the model. It seems amply justified, therefore, to speculate about Itow‘a fluid mosaic structure might carry out various membrane functions. and to suggest specific mechanisms for various functions-that can be subjected to experimental tests. ' The Fluid Mosaic Model and Membrane Functions Tilt: hypOfl'lESis that a membrane is an oriented, two-dimensional, 1uriscous solution of amphipathic proteins {or lipoproteinsl and lipids in instantaltcoUs thermodynamic equilibnim, leads to many specific predictions about the mechanisms of membrane functions. Rather than catalog a large number of these, we suggest some directions that such speculations may usefully take. Among these problems are nerve im— pulse transmission, transport through membranes, and the effects of specific drugs and hormones on membranes (J). The fluidity of the mosaic structure, which introduces a new factor into such speculations, is emphasised here. This new factor may be stated in general form as follows. The physical or chemv ical perturbation of a membrane may affect or alter a particular membrane component or set of components; a rep distribution of membrane components can then occur by translational diffu- sion through the viscous two-dimen— 7'28 sional solution, thereby allowing new merinodynamic interactions among the altered components to take effect. This general mechanism may play an im- portant role in various membrane-ruc— diatcd cellular phenomena that occur on a time scale of minutes or longer. Much more rapidly occurring phenom- ena, such as nerye impulse transmission, would find the mosaic structure to be a static one, insofar as translational diffu— sion of the membrane components is concemcd. In order to illustrate the concepts involved, we discuss two spe- cific membrane phenomena. Mniignont transformation of ceiis and the “exposure of cryptic sites." Normal mammalian cells grown in monolayer culture generally exhibit “contact in- hibition"; that is, they divide until they form a confluent monolayer and they then stop dividing. Cells that have be- come transformed to malignancy by‘ oncogenie viruses or by chemical car— cinogens lose the property of contact inhibition; that is, they overgrow the monolayer. For some time. this experi— mental finding has been thought to rev fleet the difference between the normal and the malignant states in vivo, and to be due to differences in the surface properties of normal and malignant cells. Much excitement and investiga- tive activity therefore attended the dam- onstration {53, 54} that malignant trans— formation is closely correlated with a greatly increased capacity for the trans formed cells to be agglutinated by sev- eral saccharine-binding plant aggluti— nios. Furthermore, mild treatment of normal cells with proteolytic enzymes can render them also more readily ag— glutinablc by these protein agglutinins. Burger (54} has suggested, therefore, that the agglutinin-hinding sites are pres- ent on the membrane surfaces of nor- mal cells but are "cryptic" (Fig. TA] {that is, they are shielded by some other membrane components from effectively participating in the agglutination proc- ess}, and that proteolytic digestion of normal cells or the processes of malig- uant transformation "exposes" fliese cryptic sites on the membrane surface. In some cases, quantitative binding studies have indeed indicated that no significant change. itt the numbers of agglutininvbindihg sites on the mcmv brane accompanies either mild pro— teolysis of normal cells or malignant transformation [55}. An alternative explanation of these phenomena (Fig. TB}, based on the fluid mosaic model of membrane struc— ture, may be proposed. Consi or first the proteolysis experiments with nor- mal cells. Suppose that the integral gly- coprotcins in the normal cell mem- brane are molecularly dispersed in the fluid mosaic structure. It is likely that mild proteolysis would preferentially release a small amount of glycopcptides and other polar peptides fro-t these proteins because these are its most exposed portions of the imaged pro- teins at the outer surface of th: mem- brane (Figs. 2. and 3}. The re nainlng portions of these proteins may still contain a large fraction of their original oligosaccharidc chains after the limited proteolysis, but the release of xome of the more polar structures would matte the remaining portions more hydro- phobic. As these more hydtophobic glycoproteins diffused in the membrane, diey might then aggregate in the plane of the membrane. The result n said he :1 clustering of thchagglutinin binding sites on the—“cozyrriE-treated cell sur—_ face, as compared to the normal una treated surface. Such clustering {With no increase, or perhaps even a t-Iecreasc in the total numbers of sites because . of digestion]: could enhance the agglu- j lination of the treated cells, .is com-- pared to that of normal cells, because it would increase the probability of agglutinin bridges forming between the surfaces of two cells. In malignant transformation, distinct chemical changes in the glycolipids and the glycoproteins of the cell membrane are known to occur [36}, and the en- hanCed agglutlnability of the transform; ed cells may be much more complicated than is the case in the proteolysis of normal cells- If, howcvcr, the two phe- nomena do have a basic feature in cont- rnoII, it could be a similar clus1ering of saccharidc-binding sites on the trans- formed and the enzyme-treated normal cells. In malignant transformation, such clustering could be the result of the chemical changes in the membrane mentioned above: or some virus-induced gene product [57} may be incorporated into the cell membrane and serve as a nucleus for the aggregation of the ag- glutinin—binding glycoproteins within the membrane. These suggestions can be tested are perimentally by the use of ferritin-coa- jugated agglutinins [33} as already disr cussed (Fig. til. The prediction is that With normal cells subjected to m.in proteolysis, and also with ntalignanl transformed cells, the total number of ferritinwagglutlnin particles specificallyI SCIENCE. VOL. til 1r' 'lll lI-I-Il-I- l Iain to the outer surfaces of the cells "it ight not be greatly different from those {LI normal cells, but larger clusters of a 'tin particles would be found. Cooperative phenomena in ment— ~states. By a cooperative phenomenon j_"c mean an effect which is initiated _ l one site on a complex structure and 'rismitted to another remote site by .s- structural coupling between the sites. a number of important embrane phenomena may fall into '. it category. However, before enum- -' [- ting them, we should first discrimi— tiatc between two types of cooperative disco that may occur. These can be termed irons and ct's'. Trans effects refer cooperative (allostcrici changes that an- been postulated to operate at some _: . alired region on the membrane sur- ' face. to transmit an effect from one.side the membrane to the other [53}. For _ '_ ample, an integral protein may exist in ' 51hr. membrane as an aggregate of two [or are} subunits, one of which is exposed the aqueous solution at the outer sur- face of the membrane, and the other ' is exposed to the cytoplasm at the inner .‘titttftice. Thespecific binding of a drug --;.- hormone molecule to the active site 3- the outwardsoriented subuait may induce a conformational rearrangement . I'itii'hin the aggregate, and mercby change functional property of the aggro ~ gate or of its inward-oriented subunit. By as effects, on the other-hand, we refer to cooperative changes that may produced over the entire membrane, aI least large areas of itJ as a conse- . quence of some cycnt or events occur- :ting at only one or a few localized :pdlnlfi on the membrane surface. For eminple, the killing effects of certain 'hacteriucins on bacteria {.59}, the lysis -=-:-' the cortical granules of egg cells ,tipon fertilization of eggs by sperm :idtii, attd'thc interaction of growth hormone with erythrocyte membrast 'idi] are cases which may involve '.": transmission and amplification of local— : ited events over the entire surface of membt'ttnc. These phenomena may gnot all occur by the same or related mechanisrns, hut in at least two experi- if‘ [Motel studies, that involving the inter- " action of eoliein E1 with intact Esche- richia colicclis {tilt}, and that of human growth hormone and isolated human erythrocyte membranes (on, there is 'tubslanlial evidence that long-range cine " gtype cooperative eifects intrinsic to the _ membranes are involved. " The question we now address is, How might such cis' effects work? Changed]: '- 13 F EBltUARY t9?! and his co-worlters (63] have pro posed an extension to membranes of the Monod-Wytnan-Changeux aJlosterie model of protein cooperative phenom— ena, using as a model of membrane structure an infinite two~dlmensional aggregate of identical lipoproteln sub- units [as, for example, the model de- scribed by Eenson (20)]. In this theo- retical treatment, the individual subunits are capable of existing in either of two conformational states, one- of which has a much larger binding afinity for a specific ligand than does the other. The binding of a single ligand molecule to any one subunit then triggers the cooperative conversion of many of the subunits to the ligand-bound confor— motion, in order to maximize the inter- actions among the subunits. This theory as presented relies on the membrane model used. If, however, the membrane is not a two-dimensional aggregate of lipoprotcin subunits, but is instead a fluid mosaic of proteins and lipids. the physical situation would iii? O "—F- —+ be quite different. The basic theory of Changeees et at. (dd) might still be formally applicable. but with imporv tant changes in physical significance. It is possible, for example, that a particle lar integral protein can exist in either of two conformational states, one of which is favored by ligand binding; in its normal unbound conformation the ititegra] protein is monomolecularly dispersed within the membrane, but in the conformation promoted by ligand binding, its aggregation is thermody— namically favored. The binding of a ligand molecule at one .integral protein site, followed by diffusion of the non- llganded protein molecules to it, might then lead to an aggregation and simul— taneous change in conformation of the aggregated protein within the mem- brane. This mechanism could result in a longsrangc ctr-type cooperative phe— nomenon, if the eventual aggregate size was very large and it its rescncc produced local penurbgtions'rin the properties of the membrane. However, 0 . Fig. '3‘. Two different mechanisms to explain the findings that e1thn1-.malignantiy trans- formed cells or normal cells that are subjected to mild proteolysis bemme. much more readily agglutinablc by scvcral plant agglutinins. {A} The mechanism. of Burger E54}: agglutinin—binding sites that are present on the surfaces of normal cans, bps, an: obstructed [“uryplic sites“). are exposed by proteolysis or the processes of malignant transformation, (Bl The redistribution mechanism {see text); the agglutinin sites on normal cell surfaces are largely mononIoIeeularly dispersed in the fluid mosaic strut:— '1urc, but on protcolysis or malignant transformation, they dili'use and aggregate in clusters. The probability of agglutination of two such modified cells is enhanced by the clustering of binding sites. 1'19 the transition would occur at. a rate and over a time period determined by the Tale of diffusion of the molecules, of the integral protein in the fluid mosaic membrane. This time period is lilter to be relatively long, of the order of minutes {4‘4}, as already mentioned. On the ether hand, if ctr-type cooper— ative effects occurred in a lipoprotein subunit made! according to the mecha— nism postulated by Changeux et at. {d3}, one would expect the coopera— tive change to be much faster. Con- formation changes in the soluble allo- steric protein aspartyltmnscarbamylase, for example, have halfrtimes of the order of ][l milliseconds [64}. It is therefore of some interest. that in the studies of the interaction of coliciu E1 and E. coil the fluorescence changes that marked the apparent ctr-wpe co- operative trausitions in the cell memw branc oeetn'red over intervals of one to several minutes (.52). If this sug— gested mechanism for the cuticle effect is valid, one would predict that ii] freeze—etclnng experiments on the coli- cin—treated bacteria [23) might reveal an aggregation of normally dispersed particles at the inner membrane face, or {ii} changes in membrane fluidity, such as would be produced by suitable changes in telnperature or by different compositions of membrane phospho- lipirls {65}, might markedly afiect the kinetics of the fluorescence changes that are observed on addition of the oolicin to the bacteria, In this discussion of membrane func— tions, some detailed mechanisms to account for two membrane phenomena have been_ presented. It may well turn out mat these mechanisms are incor— rout. lClur object. has been not so much to argue for these. specific. mechanisms, as to illustrate that the fluid mosaic model of membrane structure can sug— gest novel ways of thinking about membrane functions—"ways that are amenable to experimental tests. Other membrane phenomena may be influ- enced by similar diffusional mechanisms: for example, cell-eel] and cell-sub- strate interactions, where the apposition of intense local electric fields to a cell membrane may affect the distribution of electrically charged integral proteins within the membranes; or the specific binding of mnltivalent antibody to cell surface antigens, where the simultane- ous binding of one antibody molecule to several molecules of the antigen may induce rearrangements of the dis tribution of the antigen in theplanc of 1'30 the membrane, an effect that may be involved in the phenomenon of anti— genic modulation {as}. There are other specific examples as well. It may well be that a number of critical metabolic functions performed by cell membranes may require the translational mobility of some impor— tant integral proteins. This could be the ultimate significance of the long— standing observation (d?}-that the membrane lipids of poikilothormic orga- nisms contain a larger fraction of un- saturated fatty acids the lower their temperature of growth. Appropriate enzymes apparently carry out the nec- essary biochemical adjustment {63) that keeps the membrane lipids in a fluid state at the particular temperature of growth: it these enzymes are not functional, for example, because of mutations, the organism—to grow at the lower temperature [dill—must be supplied with the unsaturated fatty acid exogenousiy. Ttitlhile it has. been sug— gested before that the maintenance of lipid fluidity may be important to carrier mechanisms operating across a functional membrane, it is also possible that the real purpose of fluidity is to permit some critical integral proteins to retain their translational mobility in - the plane of the membrane, as an obligatory step in their function. Summary a fluid mosaic model is presented for the gross organization and structure of the proteins and lipids of biological membranes. The model is consistent with the restrictions imposed by ther— ntodynamics. In this model, the pro— teins that are integral to the membrane are a heterogeneous set of globular molecules, each arranged in an fltnpht—d parlric structure, that is, with the ionic and highly polar groups protruding from the membrane into the aqueous phase, and the nonpolar groups largely buried in the hydrophobic interior of the membrane. These globular molecules are partially embedded in a matrix of phospltolipid. The halls of the phospho— lipid is organized as a discontinuous, fluid bilayer, although a small fraction of the lipid may interact specifically with the membrane proteins. The fluid mosaic structure is therefore formally analogous to a two-dimensional ori- ented solution of integral proteins {or tipoproteins} in the viscous phospho- lipid bilayer solvent. Recont experi- ments with a 1wide variety of Ir :hniques and several different membrne sys- tems are described, all- of which are consistent With, and add much detail to, the fluid mosaic model. It '-1ercfore seems appropriate to suggest possible mechanisms for various m:mhratte functions and membrane-mediated phenomena in the light of the model. As examples, experimentally testable mechanisms are suggested for cell sur- face changes in malignant transforms- h tion, and for cooperative effects ex- hibited in-the interactions of membranes with some specific ligands. Note added in proof: Since this ar- ticle was written, we have obtained electron microscopic evidence r69} that the ooncanavalin A binding sites on the membranes of awe virus—trans formed mouse fibroblasts {ET} cells: are more clustered than the sit-:5 on tht membranes of normal cells, as predicted by the hypmsesishgspmscunn in Fig. 13. There has also appeared a study by Taylor et ul. {Fifi} showing the re- marltable effects produced on lympho- cytes by the addition of antit adios di- rected to their surface immur 1globulin molecules. The antibodies induce a to distribution and pinocytosis of these- surface immunoglobulins, so that withini about 3t] minutes at 3?°C the surface immunoglobnlins are completely swept out of the membrane. These effects do not occur. however, if the bivalent anti- bodies are replaced by their univaleat Fab fragments or if the antibody es- pcrimcnts are carried out at- [11°C in- stead of Slat}. Theso and related results strongly indicate that the bivalent anti- bodies produce an. aggregation of the surface immunoglobulin mo|:cu|es in the plane of the membrane, v. hich can occur only it the itrtrnunnglohulin mole rules are free to diIluSe in the meta: brane. This aggregation then appears to trigger off the pinocytosis of ma mem- brane components by some unknown mechanism. Such membrane transfer mations may be of crucial i: 1portanet in the induction of an antibody re- sponse to an antigen, as well as in other processes of cell differentiation. Internets and Notes I. 5. 1- Singer. It'l. Structure and! Function n1 Bldffllft‘trf .Hrmhrnnes, ]__ ]_ Itn1hficfd‘ Ed, (Academic Press, New York. IDTIJ, 1:. Ho. 1‘. H. Glaser. H. Sbttpktns, 5. 1. Singer, ltt Sheets, .5. 1. Chan, Pm. Nat. Ac..|l. Set. Us or, 111 new}. 3. J'. Lenard and S. J. Singer, in' J'. 56, Ill: (I995). ' 4. D. F. H. Wallaclt thrill P. H. It'll-Jet, Elli. P. I551. 5. W. KBLlZmaJLt‘I, Admit. Protein r'iierrt. Id_| “955”. SCIENE' . VflL. li'l "3.5 - c. tumors. J. Arne-r. Chem. Soc. e4. 414a Y5! 1: [194-33- '. .13. It. tCnrn. Artmt. Her. Eistehlm. as. are 5-1? _ . totes. .-o—':' I ' 1". T. Marchesl and. E. Steers, Jr.‘ Eden“ ' .159. to: mass}. - . 5. El. Richardson. H. o. Hultitt. u. E. Green. 3:3 For. n'ur. Amt. so. ens. st. 1321 tress}. -' i"- H. :“htliell and T. 5. Work. Artmt. Rev. Bio. I'te_ .- tits-s. 39.351 {1930}. Cd I. D. tlatdst. K. Frets-11:111. T. 5_ Work Nan-ire ' - 1”. fl [1966]: E. D. Kiehn and .T. I. Holland. et.‘ but. Nut. Ami. .t't-t. ti..5'. 151. Inn (1953]; :" D. E1. Green. N. F. Haard. fi_ Lena; II. I. .' Siintan. I-lJl'I'l. 150. 11‘? ([963]; S. A. Rosenberg . Bull '3. Guidottl. ill. Her! Cell! Membrane. (1 ' h. Izn'n'cson and T. J. Greenwstt. Eds. [L111- " pinull. Philadelphia. 19691. p. 93-; I. Lenard. ..' flrr- -'|errr|'§tr_|' s. 1129 tiara}. . I' 1]. ti. Blasie. C. R. Worthington. M_ M_ .es :. L'II." e'y. r. Met. Biol. 39. 40? “959). I "I D. tattoos and A. C. GHltIl-Hunuenirl. Non- .' ' I'E'J.115,. 3'70 “91"”. tr- 5'3. tie. Arch. fltorhrm. Bios-lite. III. 5.5-: 3 “IL-5}. “1' M. onset and s. .T. Singer. Flm‘he‘mbtn- it}, tat: l'lhll {1911}. -'_ l5. 1. ". Swim. M. E. Totlrtellotte. I. C. Reinert. 3“. -:.' R. \‘. MEElhaney. R. L. Rader. Pros. Not. .5. i AoJ. Set. v.3. :53. Joe (19199}. -' ' II. Til. 1. Melcholr. H. .T. Horowitz... I. M. Stung. 3} =1 " on !. T. Y. Tsong. Btoehtm. Hiapfiyn'. data he __ Il'u' 1H {19TH}. ' .. ._II. D. -l. Engeln'lan. J. Mol. Biol. I'J'. 115 [1979}: :CT' ' b1. 1!. F. Wilkins. a. E. Bieurnet. p. M. .f' Eruxhnatt. Nutter: 13¢ 12 [19h]. 3' = u. it. e. Tourteltotte. is. Emma. .L Keith, 3?“ Fmr. Nat. Anni. 5H. 3.5. 645-. 9m {1m}. ‘ 313-. A. Henson. J. Amer. 01'! E'Frem. Save. 43. ‘3' ' '- tie. Um]. I)- .' 'II. D. Tt'izule. Recent Pray. Surface Set. 3, 1H . ' _- [Ill-I'D}. ll" 11“. U. Sshairer and P. Duluth. I. Hot. Biol. m ': 1-1. 1119 t rsssi . -E. G. Wilson and C. F. FM, thief. 55. 45' [197”. E" .. KI. LEI-1311' and S. I. Eager. Science Iii 133 I a: ' '- {19531: G. Vanderkool and D. E. Green. I Pier. Net. Ara-d. Set. v.5. as. 1515 um}. -_ ts. w. L. Hubbell and H M. McConnell, Proe. .f; him. Aead. .S'et. ILLS. 61. 11'. ([9631: A. ll - . " Kinh. a. s. Waggoner, o. H. Grll’t'lflt. rant. Flt .- 1' PL “9. - H. L. S1eere. J'. Biophys. Bioel'rem. Eyre-t. 3¥ 1‘3“ 45 195?11H.Moor.K.Mlin1:tha1er.lt.wald. .1. _' ._ nL'. i1. Frey-Wyssting. taint. In. 1 {Islet}. - ' 21'. D. Brenton. Ame. Rev. Piss: ray-stat. 1o. 10-; Flt ' thus]. I. tvt. Womeswenh. L. Packer. 1'1. X— 1— ts [,3 ' 'nevvs nun ooMMENT . _ I—"—-—‘—" _ A tow and authoritative study of the ‘-. Soviet space program indicates that. while Arnet'iean Spaco efforts oontinue n ._ i: winding down toward the last Apollo i; : 1li.gh1 this year. the overall Soviet space I ' 'K' program remains “a strong and growing '.-.' enterprise.” its ambitions unhindered by 5: hudy.:tary strain and undimrned by the 1. 4.... 1 " .'Dl.'-'l'S-I -n ;: .1IEIIIE 5-3: SID-cit numb-er 5111-0353. 5 ' rI;_1:tnaauanvto}: flrrtrtton [Biochtrm Biophys. Acre 2115. 115 {1930}] find no evidence by frem—etdttng for the custom: of the stalked-leech! on mi1t:l- chondrial Inner membrane-t that are obsenreu With nettnlively s1n|nett preparations. 2.3. M. E. Bayer and C. C. Rernsen. 1. Bacterial. lot. at}: tlvtot. 29.13". Pinto da Silva and. D. :Flrat'lttzlrtI J. Cell Biol. 45. 593 [IQTO]; T. W. Tillaclt and V. T. hfnrchesi. tom. 1:. 64‘}. 1D. I'. Pinto da Silva. S. D. Douglu. 1']. Brenton. Abstracts of the J'I‘J‘J'J't Mrrtl'ttg of tire American Society to: Cell Biology. 51211 Diego, Calm. November 11339 (lino). 11. I59; T. W. Tillaek, R. E. Scott. v. T. Mai-ehen. fluid... In. :13. 3|. T. L. Strsk. G. Fairbanks. D. F. It. Wanneh. Biochemistry 1o. Edl‘l' {19'J'I‘.I. 32. M. 5. Bromine. Nature 31. 2211' {IBM}; .L Moi. Biol. 5!}. 351 twirl}. 31 G. L. Nicolson. and .5. J. Slower. J'ras'. Nat. Aunu‘. Sci. H.511 65_ 5'41 {IPTIL 34. G. L. Nicol—toll. 5. P. Masotlredls. .5. J. 5inger_ thief... 11. ms. 35. G. L. Nicolsnrt. R. Hyman. 5. 1'. Singer. J'. Cell Biol. 50. EDS “Till. 36. R. E. Lee and J. D. Feldmsn. that. 1]. 39:5 “9641. 31'. F. A. Green. immunachrmflrfl' l, '14? {1951'}; J. Biol. Chem. 343. 5515} (1963.1. 33. W. C. Davis and L.‘ Silvertaan. Inenrptmrro- tion fr. 3351 H953}; T. Avoid. U. Ititrnn-lefljng. E. or Hem-n. E. A. Horse. 1.. I. Did. J. Exp. Med. no. 91"? {I969}. 35‘. A. Shlmada and 5. G. Nathettsou, Hitachi-ml;- H't' 3.. 4343 {[969}: T. Mornrnstsu and 3. G. Natltensnn. told. 13. dull-1‘5 {1971)}. 4-0. 1. IL Blasts. M. M. Dewey. A. E. mauroek. C. R. Wetthinttton. J'. Mol. Biol. 14. 1-13 {1965]. 1|. M. M. Dew-tr. P'. K. Davis. I. K. hissie. t.. Barr. lbtd. 39. 3135 {19159}. 41. J. K. Elesie and C. R. Woeflthtgtnur. this!" D. 11?. 43. R. C. warren and R. M. Hleloz. Nature 117‘ / sex 330 Elfi'l'fi}. hid-flE. D. Frye and M. Edidin. J. Cell SH. '1', 3-13 {1911]}. 45. I.. htindirh. J'. Moi. Biol. 1-9.. AIS. 433 [19H]. do. it. D. Pot-eta and l. J. Unhimin. Biochemis- try it. IP30 {ETD}. 47. R. G. Dryerlale. P. R. Herrick. D. Franks. Var Snug. 15. 194 “91581:. 43. G. L. Nit-Olson and S. I. Singer. in prepara- “an. 4‘}. E. H. Eylar. M. A. Madofi. CI. V. Brody. .T. 1. Uncles. J. Biol. Chem. 133. 15m! {1962): E. L. Benedettl and P. Emmi-Jot. J. (2.1.! Set. 1. 499 (191511. I The Soviet Space Program: Effort .. '; Said to Surpass Peak U.S. Level deaths of three cosmonauts last year. The study.* produced for the Senate Committee on Aeronautical and Space Soiences by analysts in three divisions of the Library of Congress. concludes that the current level of Soviet space activity exceeds that of the United States at its peak in 1966. The space __ vist Spoor Prenatal-111. [goo-'10" Report at the Committee on Aeronautical 11nd Spat-e Selenees. '-".|1rttm.-.-d let the Science PGl—lc)‘ Rflenrlfll Division. Foreign Muirs Division. and the European Law . Library of Conercss: availuhle from the omens-went PrI'n1Irto oh'nee. Washington. 11C. 5“. G. L. Nioolson and S. .‘F. Sistger. in prepara- tine. 51. V. T. Man-tesi and G. E. Palette. J. Celt not. as. as: {test}. ' 51. R. K. Kornberg and H. M. MeCnnneH. Etn- chrottrtry Ill. 1111 {twirl}. 53. M. M. Burst! and A. R. Goldberg. Proc. Nat. Ant-Id. Sci. US. 51'. 359 {1961'}; M. Inhar and L. Sachs. tbid. 63. 1413 [1969}. 5-4. M. tvl. Rumor. told. 63. 99-1 {1969}. 55. B. Solo. [1. lie. N. Sharon. L. Sachs. flit:L villus. Broom-Is. Acre. in press: B. Dunne and .T.. Sambroolt. Nature 11:. 155 {1971}. 56. 5. Jinkln'lon' and W. T. Mumhami. Proc. Nat. Anni. Set. UJS‘. 53. 15-1- [I'Dfihl] P. T. Mora. H. 0. Brady. R. M. Bradley. V. M. McFar- land. laid. 63. 11911 “369]: C. A. Buck. M. L'. Gtieh. 1.. Warren. Biochemistry 9. 156? [two]; H. C. Wu. E. Meetan. P. H. Black. P. w. senses. that. a. mo tlssst. ST. T. 1.. Benjamin and M. ht. Burger. Free. Hot. Amt. set. on. 61.929 {1970). 53. T. R. I'Ddlesll and I.—]". Chanson. in Firm-Im- rnentol Conceer In Drug-Rennie: Interne- tinnr. D. J. Toasts, I. F. Dsnlelfi. t. P. Moran. Eds. {Academic Press. New Yoflt. 19159}. p. 5'3. 59. M. Nomun. Pm. Not. Ana-i. Set. U5. 51. 1514 (Ed-t). (to. D. Epel, II. C. Pressman, 5‘. Elsaesser. A. ht. 1'I'I'r'ltfl'll't". in TI'H' Ct" frrlt.‘ Gear-imam: Jr:- H'Fflfllofll'. G. N. Padilla. G. L. Whitson. l. L. Cnnterson. Eds. {Academic Press. New York. I969}. p. 215}. 6!. M. Sonenberg. Bis-chem. Beehive. Res. Com- mtm. 36. 45:1 (19619); Free. Not. steed. Sci. [123. 6!. mil (19111. "i . W. A. Creme: and 5. K. Pitt‘l‘lips. I; Bt'lcteriol'. [II-t. 1119 new. ~ . J. P. Change-1.x. .1. mm. Y. Tues. C. KitteL Frat. Nor. Amt. set. 0.5. to. 3:15 (Jae—t}. J. Eekt'eldt. G. G. Hamsters. 5. C. Mohr. C. W. Wu. Biochemistry SI. 3333 [193D]. I55. D. F. Gilbert and P. It. Vast-Ins. Prue. Nat. Arno. Sci. LES. 53,. 1519 {we'll}. 615. E. A. Horse and L. .T. {listI Arenas. Rev. Genet. 3.2159 (1969). 6'1. E. F. Tremaine. C. Hotter-er. P. Rot-mtg. Bull. 30:. Chart. Biol. :2. 651' (1930}: G. Frankel and J't. 8. Host. Bloch-Eur. I. 34. 1035 {I940}. es. M. B'menslty. r. Barret-tot. res. 4-19 rte-t1}. 6-9. G. L. Nioolwn. Nature 133, 244 flfl'l'li. 'i'fl. R. B. Tudor. w. P. H. Dulles. M. C. Rail. S. dePeu-is. tbid'... p. 135. 'it. The originaI studies reported in this article tvero supported by erant. GM: 15911 trons the National Institutes ot‘ Health [to 5.1.5.1. a s a study also indicates that the Soviet Union is almost certainly pressing ahead —caul.iously but intently—with a manned lunar program that may be ex— [tested to ptlt cosmonauts on the moon in the mid~19‘iti‘s and possibly as early as 1933. A related eonelosion. perhaps the most surprising of the Gill-page study. is that the Russians may end up speuding the equivalent of $49 billion to land men on the moon. far more than the cost of the Apollo program. Whether or not the Soviets actually earry throngh with their evident ittten- tions, the study goes on. “it. is not pos- sible to establish that the Russians have invested smaller total resources in Inner exploration than the United States" even though the Soviet efiort “has not pro— duced the visible result in this regard which the United States-has achieved." These and other findings stand in direct contradiction of assertions by Soviet T31 ...
View Full Document

This note was uploaded on 07/17/2008 for the course CHEM 763 taught by Professor Dr.hille,dr.sayre during the Spring '07 term at Ohio State.

Page1 / 12

singer - The Fluid Mosaic Model of the Structure of...

This preview shows document pages 1 - 12. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online