The Electron Microscopy of Macromolecules

The Electron Microscopy of Macromolecules

BY RALPH W. G. WYCKOFF Experimental Biology and Medicine Institute, National Institutes of Health, Bethesda, Maryland

CONTENTS ...........................

Page

...............

ining Macromolecular Visibility ....... . 1. Resolution.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ................. .................................................... ..................................................... 4. Specimen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Visualization of Macromolecules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Resulting Problems. . . . . . . . . . . . . . . . . IV. Macromolecular Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .................................................. acromolecules . . . . . . . . . . . . . . . . . . . . . . . . . 3. Filamentous Macromolecules. . . . . . . . . . . . . . . . . . . . . . . . . V. Structure of Macromolecular Solids. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. The Structure of Single Crystals.. . . . . . . . . 2. The Growth of Crystals.. . . . . . . . . . . . . . . . . 3. Crystalline Order in Muscle.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Sub-crystalline Soh ...................................... VI. Formation of Macromol ...................................... 1. “Healthy” Macromolecules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Viruses.. . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 2 2 4 4 5 5 6 6

15

22 25 29 29 31 32

I. INTRODUCTION Our developing ability t o see particles of macromolecular dimensions with the help of the electron microscope permits a study of organization within biological structures on a new and more fundamental level. Heretofore, the cell and its optically visible components and products have been the units in terms of which the organization of living matter has been discussed; but we are now beginning to see the macromolecules that are essential parts of these cells and the way they are organized to carry out its functions. The cell of course will never lose its importance as the unit of vital activity, but it is inevitable that before long conventional cytology will be supplemented by a kind of molecular cytology that describes the macromolecular architecture of cells, and the way this 1

2

RALPH W. G. WYCKOFF

architecture is built up during growth and altered by the cell’s metabolic activities.

11. FACTORS DETERMINING MACROMOLECUL-4R

VISIBILITY

1. Resolution Present day electron microscopes have a resolving power t h a t is more than adequate for the visualization of particles having the size of all but the smallest proteins. I n addition, the smallest known viruses provide spherical objects with diameters as small as 100-150 A. and weights in the range between one and four millions. Molecules of some hemocyanins, easily seen in the electron microscope, are a t least this big; others have molecular weights in the hundred thousands and sizes comparable with molecules of the plant and animal globulins, including the larger enzymes. All these are not hard t o photograph. Examining still smaller molecules is primarily a problem of incorporating them into satisfactory preparations. It is hard t o determine precisely the practical lower limit of existing electron microscopic vision because of this difficulty in getting preparations suitable for its measurement. The theoretical lower limit of resolution of the electrons used for microscopy is only a fraction of a n angstrom unit-far less than the diameter of a n atom. Existing electron lenses are, however, completely uncorrected and can only yield satisfactory images after being so closely apertured that the resulting optical system achieves a mere fraction of this higher limit of resolution. The average present electron microscope of precision, such as the RCA Type E M U with which most work is done in this country, should have a resolution of better than 100 A. under the most ordinary conditions of operation. If it is equipped with a good objective lens that has little astigmatism, either inherently or by correction, its average working resolution mill probably be in the neighborhood of 50 A. For better resolution than this the microscope must be in the hands of a n operator of more than average experience and the preparations must be of more than average thinness. Under such circumstances resolution of the order of 20 A. can be attained and resolutions of half this value have been claimed under certain conditions.

2. Contrast While resolution sufficient t o portray minute particles in their true shapes is needed, it is not the only factor essential for the photography of macromolecules. The other factors must be understood in terms of physical considerations that are different from those important in optical microscopy. We see things under the optical microscope by reason af

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

3

various amounts of light absorption in a preparation or through differences in refractive index over a transparent object. Neither of these is important in establishing what detail is seen in a satisfactory electron microscopic image. The degree of detail is determined by the different scattlering powers for electrons of different parts of a preparation. The amount of this scattering depends on the number and weight of the atoms involved. Dense objects and especially those containing heavy atoms scatter strongly and appear relatively opaque under the electron microscope. Macromolecules of biological origin consist almost exclusively of very light atoms which scatter electrons poorly. The amount of this poorly scattering material diminishes rapidly as a function of molecular size. In almost all cases they must be supported for microscopy on a substrate which also scatters electrons. When therefore we look at smaller and smaller molecules a point is soon reached where the molecule does not scatter enough more than this substrate to be clearly distinguished from it. Such a light molecule is thus practically invisible because its image does not contrast appreciably with that of its support, and experience shows that the limit of visibility through this lowered contrast is reached with objects that greatly exceed in size the limit of resolution of the microscope as outlined above. If nothing could be done to circumvent this difficulty, there would be little hope of finding out much about biological macromolecules. Fortunately, however, it is possible to enhance preferentially the scattering and visibility of macromolecular particles. In some cases this can be done by incorporating heavy atoms into the molecular particles, either through chemical reaction or by adsorption. Seemingly a more broadly useful procedure for enhancing contrast is metal shadowing. This is accomplished by evaporating a metal obliquely and in an exceedingly thin layer upon the surface of a preparation. The metal thus deposited is thickest on those aspects of an object that face the evaporating atoms and is completely absent from parts of the substrate on the far side of tall objects. The scattering power from such an unevenly distributed deposit varies so as to bring out with added clarity the micro detail of an object, including its macromolecules. The most satisfactory and informative electron micrographs yet made of macromolecules are ones that have been shadowed. All those shown in connection with this chapter have been so treated. Definite conditions must be fulfilled if metal shadowing is to reveal small macromolecules. The layer of deposited metal should be SO thin that it does not perceptibly disturb the shape of the molecular particles. This means that the metal should be of high atomic weight and consequent scattering power. It should be chemically inert and its atoms

4

RALPH W. G . WYCKOFF

must not be able to move about and recrystallize after deposition, even under the electron bombardment incident to microscopy. A number of materials meet these requirements well enough to use for shadowing the larger macromolecular particles, but only the noble metals have the chemical inertness and high melting points required for the best photography of small molecules. Platinum is especially satisfactory for this purpose, but there are serious difficulties in evaporating a substance of so high a melting point without damaging, through radiation, the preparation being shadowed-and these difficulties have not yet been fully met. 3. Substrate The ability to see small molecules is as dependent on the smoothness as on the thinness of the substrate that is supporting them. It is evident that for good visibility they can scarcely be smaller than the molecular texture of the supporting membrane. Known plastics have a texture of their own that is of the order of 100 A.; they, therefore, have a limited usefulness when studying molecules smaller than this. To examine smaller molecules we should if possible employ the smoother backgrounds that are provided either by exceedingly thin metallic films or by atomic replicas. They are both technically far more difficult to use, but for the present their roughness is not the limiting factor in seeing small particles. 4. Specimen The maximum of resolution of which a microscope is capable can only be realized from preparations of extreme thinness. Most electrons lose energy when they are scattered. Since electron lenses cannot focus electrons of different velocities in the same plane (i.e., have no “color” correction), these slowed-up electrons strike the image plane in such a way as to diffuse and damage the image produced. Some which are bent through especially large angles can be screened out by close aperturing of the objective lens, and this does greatly improve the quality of the final image. But under all circumstances the ability to perceive small detail falls rapidly with increase in specimen thickness, and this fact must be given great weight when seeking to see macromolecular particles. The most serious limiting factors now reside in the substance itself. One of these lies in the need for having clean molecules to look at. Experience shows that a relatively small amount of a low molecular weight impurity may be sufficient to obscure completely the smaller macromolecules and to smear over and disturb the appearance of the larger ones. The monomolecular films formed on the air-liquid interfaces of solutions of all but the big molecular proteins interfere very seriously

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

5

with the visualization of their molecules. In some cases we know how to evade the obscuring effects of these films, but they are one of the most serious problems we now face in the photography of protein molecules.

111. VISUALIZATION OF MACROMOLECULES Macromolecular particles of many different sorts are now accessible to observation and photography. In general they are of two kinds: (a) proteins and other substances that can be obtained as solutions, or suspensions, which can be purified and sometimes crystallized and (b) parts of a more or less insoluble macromolecular fabric. To the first group belong the viruses, the respiratory proteins of vertebrates and invertebrates, seed globulins, and proteins constituting animal sera or derived from tissue extracts. The majority of these substances seem to have spherical or short rod-like molecules. Macromolecular particles of the second group are typified by those building the cellulosic structures of plants and the muscle, connective tissue and nerve of animals. They usually are long filaments. Resulting Problems The present problems arising from the ability to see these macromolecules with the electron microscope deal mainly with (a) a description of the shapes and sizes of members of these two groups, (b) the arrangement of their particles as they occur in nature or in the solids they form under laboratory conditions, (c) the changes they undergo as a result of chemical reactions in which they participate and (d) the mechanisms whereby they are produced in nature. For many of these substances a determination of their particle dimensions must be considered largely of academic interest or as information essential to an attack on the other problems; but for viruses such knowledge has very immediate and practical value for their recognition and for the study of the way they cause disease. We are only beginning to penetrate the molecular architecture of the fixed tissues of the animal body and of the cellulose that is the principal fixed tissue of plants. But it is already obvious that the determination of this architecture will give quite a new insight into how these tissues function and a study of how this architecture is modified with time will provide a new approach to many problems of degenerative disease and of the aging process. Though many of the macromolecular particles derived from living matter serve as frameworks and fabrics to support more actively metabolizing components, others, such as the respiratory proteins and the enzymes, fulfill their purposes by being centers of intense chemical activity. It is bound to be an increasing concern of electron microscopy to

6

RALPH

W.

G. WYCKOFF

observe the changes that these molecules produce in their environment and that they themselves undergo as their chemical reactions proceed. It is impossible at this time to perceive more than the broadest outlines of this visual chemistry or even to be sure of the directions in which it will most actively develop, but that such a chemistry will inevitably arise and that it will yield exciting new prospects over the way living matter functions is certain beyond doubt. The mechanisms through which these macromolecular particles are produced in nature are intimately tied up with the way both they and the living organisms that originate them function. As yet we have no good picture of how large molecules, some of several million molecular weight, can be so constructed that each is exactly like all the others of a given substance. But we are beginning to perceive details of the mechanism whereby the viruses develop within their living hosts, and it is not beyond hope that this accumulating knowledge, besides clearing up many of the mysteries surrounding the fundamental nature of these agents of disease, mill indicate how we should proceed to find out more about the benign macromolecular entities, of similar size, that are to be found throughout the living world.

IV. MACROMOLECULAR MORPHOLOGY 1, Historical

Until a generation ago very little was known about the elementary particles of the materials dealt with here. Over the past years the ultracentrifugal studies of Svedberg supplemented by measurements of rates of diffusion, viscosity, double refraction of flow and other physicochemical characteristics have, however, given much information about the weights and probable shapes of such of these molecules as can be obtained in solution. The electron miscroscope has for the most part confirmed this information and has extended our knowledge to many substances whose particles are altered by attempts to put them into solution. There are two phases to be recognized in the electron microscopy of macromolecular particles. One is what might be termed the historical phase, the other is the phase of present-day development. During the first few years in which electron microscopes were available, beginning in the later 1930’s’ a considerable number of macromolecular objects were examined (cf. 1 and 3 for early references). The first adequate microscopes, manufactured by Siemens, were all in Germany; during these earlier years of the war, they were used to look at most of the materials we still find it most profitable to study. The same sort of exploratory work was carried out in this country (cf. 46 for many early references)

THE ELECTRON MICROSCOPY OF MACROMOLECULES

7

after the first RCA microscopes were built. The methods of specimen preparation developed during these early years were not well adapted to show small organic particles such as the biological macromolecules; therefore, though many particles of macromolecular dimensions were seen during this early work both here and abroad, the visualization was usually not clear enough to add much to the knowledge we already had of them. Subsequent development of methods for reinforcing the scattering from small and light particles, notably shadowing, made the electron microscopic images of macromolecules so much clearer that it has proved well worthwhile to repeat and greatly extend the earlier explorations. [n all preliminary investigations of this sort, designed to find out the range and limitations of a new method, the objects for study will in the main be those that are especially easy to prepare and that have already been examined by other methods. We therefore continue to look a t the same kinds of macromolecular particles that were first put under the electron microscope : for example, the hemocyanins, plant globulins, the viruses, muscle, connective tissue, fibrin, etc. As our understanding of these and similar “classical” objects of investigation is increased through improved experimental procedures and increased experience, the field of investigation will inevitably be widened. 2. “Spherical ” Macromolecules

As already stated nearly all the macromolecular particles thus far visualized are either nearly spherical, or they are filaments of great length. Where comparisons can be made they have dimensions that for the most part accord well with the ultracentrifugal data, but some are not as asymmetric as these less direct methods of measurement had led us to expect. A number of the plant virus proteins have spherical macromolecules with diameters that range between ca. 150 and 300 A. Some of these, like the tomato bushy stunt, the tobacco necrosis and the turnip yellows viruses, seem to be strictly spherical; others, like the Southern bean mosaic virus (Fig. 1) and the non-infectious Rothamsted protein, are nearly, but not strictly, spherical. Some of the smaller animal viruses such as those causing the encephalomyelitides, papillomas in rabbits and warts in man, poliomyelitis and related diseases, have macromolecular infectious units that are nearly or exactly spherical particles with diameters in the range between 500 and 100 A. Aside from these products of virus activity the hernocyanins and erythrocruorins, as respiratory proteins of invertebrates, have the largest centro-symmetrical molecular particles yet seen. Molecules of the hemocyanin of the horseshoe crab,

8

RALPH W. G . WYCKOFF

Limulus polyphemus, have appeared after ultracentrifugal purification as spherical objects ca. 160 A. in diameter (40). Careful examination of their electron microscopic images led to the rather surprising result. that they were not of identical size but had a broad gaussian distribution about a mean value. The reason for this is not altogether clear, but it could be due either to an irregular flattening of the molecules during drying or to a minimal denaturation and alteration in molecular shape brought about by the procedures used in purification.

FIG.1. A dried deposit from a purified suspension of the Southern bean mosaic virus protein showing a n orderly grouping of its elementary particles. In many regions the packing is the closest possible and yields an hexagonal net, in others it is t h e more open arrangement of a square net. Especially over the close-packed areas t h e particles are often more than one layer deep. Platinum shadowing. Rlagnification = 32,000 X.

Interesting photographs (26) have been made of very cautiously purified hemocyanin of the sea snail, Busycon canaliculatum. The molecules found in such carefully purified solutions are cube-like objects* having a well-defined inner structure. Inspection seems to show that each cube is a group of four short rods. This is one of the hemocyanins which Eriksson-Quensel and Svedberg (5) have shown by ultracentrifuga-

* Electron micrographs of this and many other macromolecular particles have been collected i n a recently published book by the writer (Electron Microscopy, Interscience, New York, 1949). These have not been reproduced here nor have all the bibliographic references given there been repeated.

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

9

tion t o possess molecular weights that split into simple fractions when the p H is raised. I n agreement with this, alkaline solutions do not show these cube-like aggregates, but instead have rod-like molecules which appear t o be the components of the cubes. If this hemocyanin is not freshly taken from the animal, or if it is subjected t o chemicals or t o repeated ultracentrifugations for purification, these cube-like particles are no longer seen. Instead the molecules seem t o be roughly spherical objects of about the same size. It is possible that they represent the first stage in denaturation of the original molecules. Plant globulins, of which edestin is the most familiar example, offer a series of middle-sized macromolecules which can easily and profitably be observed in the electron microscope. Edestin was in fact one of the first molecules t o be photographed (36) and both then and later (10) has shown itself as an approximately spherical particle ca. 70 A. in diameter. I n this case ultracentrifugal data (38) had pointed to a somewhat elongated particle, and we are not a t this moment in a position t o explain this difference of result. The molecules of the serum globulins, with half the molecular weight of the plant globulins, are not much smaller and in some sera, such as those of the horse, there are antibody molecules th a t are considerably heavier and very easy to photograph. As previously indicated, the resolution of good present-day electron microscopes is sufficient to allow us to see molecules of weights somewhat under 50,000. It thus has become feasible to try to look a t molecules of hemoglobin and of the most important enzymes. There have as yet been few serious attempts t o examine these smaller and in some respects more interesting molecules. As earlier suggested, useful knowledge about them will only follow when satisfactory methods of preparation are found; and this in its turn will depend on being able t o spread them over a sufficiently thin and smooth substrate in such a way th a t the molecules are unobscured by the monomolecular films th a t ordinarily form on the surfaces of their aqueous solutions. The more active cultivation of this field has been delayed by the numerous rewarding investigations that can more easily be made involving larger particles. Macromolecular particles of all sorts are seen within the protoplasm of living cells. Until recently it has been possible to examine them only after cellular disruption; under such circumstances i t has seldom been possible t o identify what is seen and to determine its relationship t o the grosser cellular components. This situation is being changed by our increasing ability to obtain sections thin enough for satisfactory electron microscopy. Both spherical and filamentous macromolecular particles of a wide variety of sizes are evident in sections through tissue cells.

10

RALPH W . G . WYCKOFF

Such particles also can be seen in great numbers in the protoplasm from colon bacilli (44) lysed by bacteriophage (Fig. 2).

FIG. 2. A portion of the protoplasm released from a colon bacillus lysed by bacteriophage. The spherical macromolecules it contains in great numbers can be seen both along the edge and throughout the mass. 24,000 X

3. Filamentous Macromolecules Most filamentous macromolecules are components of framework structures of plants and animals and as such are not soluble in water or saline solvents. A few plant virus proteins, notably those of the tobacco mosaic (Fig. 3) and potato X groups and the muscle proteins myosin and F-actin are, however, both filamentous and soluble. The elementary particles of cellulose, whether they come from wood, a wide variety of textile fibers, or are synthesized by microorganisms, have uniform particles a couple of hundred angstrom units in diameter and indefinite in length. Agreement does not exist in the literature as t o the diameters of the fibrils from different plants. It has been thought b y some (17) that this thickness is not always the same and can be as little as 100 A. but other work has demonstrated (20) that the microfibrils of a given plant are of the same thickness and that those from a variety of plants are a t least of very similar diameters. This is true of the microfibrils from bacterial cellulose (21), wood, cotton, ramie, sisal and flax.

THE ELECTRON MICROSCOPY O F XACROMOLECULES

11

Beyond doubt, filamentous macromolecules are an important ingredient of the protoplasm of many kinds of cells. Where they have thus far been recognized, they are most often part of the gel-type structure to be discussed later, but apparently isolated filamentous macromolecules can be found in great numbers in the protoplasm of aging colon bacilli. Their relative amount compared to the spherical particles also present seems to determine the physical properties of this protoplasm. The protoplasm of young, actively growing cells of E. coli does not contain

FIG.3. Elementary particles in an old purified suspension of the tobacco mosaic virus protein. In such a solution the particles are all rods of the same width (150 A.) but of different and often of great lengths. 45,000 X

filaments (Fig. 2) and is so fluid that it flows out in a thin layer when the cell is ruptured. Old cells commonly have their spherical macromolecules embedded in a matrix of filaments which causes the bacterial contents t o retain their shape if the membrane breaks. We must imagine that as a general thing the relative amount of such filamentous macromolecules is important, in determining the viscosity and other physicochemical properties of any protoplasmic system of which they form a part. Most filamentous macromolecules seem to be built by the end-to-end association of short units. This is true of both the soluble and insoluble materials. In some instances, the components can be seen in the filaments; in others they must be inferred from the way the filaments

12

RALPH W. G. WYCKOFF

develop or associate together, or from the products of their chemical disintegration. Fibrin is a substance within whose macromolecular filaments a repetitive structure can be seen. After staining with phosphotungstic acid, a system of cross striations has been manifested which consists of a n alternation of broad and narrow dark bands with intervening light regions (9). The interval between repetitions, for instance from one light band t o the next, has been measured as ca. 227 A. with most values lying betmeen 210 and 240 A. Within these dark bands a still finer structure has been seen. This gives them a definitely particulate appearance, but the detail is so minute that it has only been imperfectly resolved in the photographs thus far made. When fibrin (9, 14, 27, 41) has been clotted on the preparation, along with the well-defined fibrin fibrils just discussed, innumerable thinner threads reaching down in dimensions towards the limit of visibility have been present ; their relation t o the striated fibrils is not yet certain. One study of fibrinogen itself (8, 9, 27) failed t o find fibrous particles after drying, another described elemeptary particles consisting of beaded threads ca. 40 A. wide and between 200-1100 A. long. Filaments of clam muscle paramyosin (11) seem t o be examples of a degree of order within single fibrils more complex than that expressed by simple cross-striations. After staining with phosphotungstic acid, these fibrils have shown a detail which indicates t h a t they must have components arranged in a kind of two dimensional, repetitive order. The larger elementary fibrils of connective tissue have cross striations that repeat themselves on the macromolecular level. These will be discussed in a following section which deals with the structures they form. Filamentous macromolecular substances differ not only in the amount of fine structure that can be seen within their particles, but equally in the extent t o which they can be dissociated or split into components from which they may originally have been assembled. Some, such as fibrin, cannot now be split, others, like cellulose, can be broken down into more or less uniform fragments by treatment with appropriate chemicals, and a few can pass reversibly from a fibrous t o a n essentially globular condition. The fibrous form of the muscle protein myosin (16, 34) offers a n example on a very fine scale of filaments which are not dissociated into components. Its seemingly structureless molecules are as thin as any we have yet been able to distinguish (Fig. 4) and they are of indefinite lengths. The tobacco mosaic virus protein particles (Fig. 3) as seen in older purified suspensions are examples on a larger scale of particles t h a t cannot be dissociated into their constituent elements. Their long rods must have developed by the end-to-end association of the 2700 A. long rodlets

THE E L E C T R O N MICROSCOPY OF M A C R O M O L E C U L E S

13

present in freshly expressed and carefully handled virus preparations. Nevertheless, the association is so exact that the junctions between rodlets are not apparent and the long rods are devoid of periodic structure. Thus far no fine structure has been demonstrated within the elementary filaments of cellulose (Fig. 5). They can, however, by suitable treatment be broken down into uniform fragments which, under the action of some reagents, approach in size the “crystallites” that X-ray

FIG.4. A portion of a paracrystalline aggregate of the filamentous macromolecules of myosin. Separate, extremely thin myosin threads can be seen at many places throughout this approximately parallel aggregate. From unpublished work by Rozsa, Szent-Gyorgyi, and Wyckoff. 28,000 X

diffraction indicates as being present. Hydrogen peroxide yields fragments bigger than the expected crystallites, but the breakdown during conversion to viscose, nitrocelluloses (13), methyl and ethyl celluloses and similar derivatives is more complete. The X-ray crystallites are within the resolving power of existing microscopes; our failure to detect them in the intact fibrils may be due either to a lack of morphologic distinctness or to the difficulty that exists in freeing most cellulose from the lignin

14

RSLPH W. G. WYCKOFF

and hemicelluloses with which it is so intimately associated in nature. There are forms of cellulose that are especially clean, and detailed examination of them at high resolution may show internal structure that has not otherwise been apparent. Such cellulose (Fig. 5) is made by certain bacteria (21) and it occurs in simple plants such as valonia (30). The muscle protein actin is the best example to date of a system which exhibits a reversible transition from the globular to the fibrous macro-

FIG.5. A mass of micro fibrils of cellulose as produced by a culture of Acetobacter xylinum. Most of the filaments shown here are ribbons or twisted ropes composed of several microfibrils. 11,000X

molecular state. It has been possible to gain some information (31) about how this can occur and about the fine structure of the filamenk that participate in this transformation. When G-actin is transformed into F-actin in very dilute solution, the filaments formed are of indefinite length and devoid of clearly defined fine detail (Fig. 6). They have substantially the same diameter as the filaments seen in teased (Fig. 7) or sectioned muscle, especially clearly after digestion with trypsin. When the G-actin to be transformed is in greater concentration, and when the polymerization is carried out under conditions which permit electron microscopic examination of the undisturbed product, the close contact of the newly formed F-actin threads brings out their inner structure.

T H E ELECTRON MICROSCOPY OF MACRO-VOLECULES

15

Examination of photographs of such aggregates (Fig. 8) has indicated that the molecular unit of F-actin is a short cylinder about 300 A. long and 100 A. thick. At great dilution these units associate end-to-end t o yield the filaments of indefinite length; a t greater concentrations they are in side-by-side, as well as end-to end, association and this is responsible for the networks that reveal the rodlets themselves. Their dimensions are compatible with other data (35, 37) concerning the macromolecules of F-actin. These have shown that G-actin with a molecular weight of

FIG.6. Filaments of F-actin obtained by polymerizing a very dilute G-actin solution directly on the substrate. It is impossible t o tell from a n electron micrograph such as this if apparent structure seen in filaments is real or a reflection of roughnesses in the underlying substrate. 41,000 X

ca. 70,000 polymerizes t o a particle of weight CU. 1,500,000 when it becomes fibrous: the observed rodlets should have about this heavier weight. V. STRUCTURE OF MACROMOLECULAR SOLIDS

The methods of X-ray analysis have already given considerable insight into the way the molecules are arranged in many macromolecular crystals; electron microscopy permits us t o see this arrangement both in crystals and in other solids whose particle arrangement is not sufficiently ordered to give useful X-ray diffraction. The small ambunt of electron microscopy already done demonstrates that there are inhature materials

16

RALPH W. G . WYCKOFF

showing many degrees of order in molecular arrangement intermediate between crystals and amorphous solids. Included among these are certain newly-found two-component solids whose particles interlace to form what can perhaps be described as fabric-structures. The electron microscope is the most direct method of gaining insight into the fine structure of these subcrystalline types of order and their further inrestigation with its help is therefore bound to be particularly rewarding.

FIG.7. A short segment teased from striated rabbit muscle and partly dispersed. Here longitudinal filaments having the same general diameter as F-actin threads are clearly visible. Remains of a minor band can be seen in the center of the bundle. A single thicker, continuously cross-striated fiber of connective tissue crosses the top of the photograph.

29,000 X

1. The Structure of Single Crystals

It is evident that the ability to observe directly the molecules composing a crystal permits an unexpectedly direct attack on many of the most fundamental problems of crystallography. For the most part, this attack has yet t o be developed, but some progress has been made in the visualization, and immediate determination,. of structure that thus becomes ,possible. Electron micrographs have now been prepared showing the molecular arrangement on single faces of several virus protein and other crystals and in some instances they have been obtained from several different faces of one kind of crystal. Evidently, if the molecular arrangement on several faces and the crystal symmetry are

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

17

FIG.8. Groups of the filamentous macromolecules of F-actin obtained by polymerizing a solution of G-actin, more concentrated than t h a t used for Fig. 6, on the substrate. The close association of these threads with one another brings out the fact that they are composed of a succession of short cylindrical elements. 59,000 X

FIG.9. A single crystal of a tobacco necrosis virus protein showing the regular arrangement of its macromolecular particles. On the top face t h e net is square, on the side face a t t h e bottom of the photograph the hexagonal character of the prevailing net is especially clear. 50,000 X

18

RALPH W. G . W Y C K O F F

known, it becomes a simple matter to deduce the arrangement within the body of the crystal. It has been shown (43) in this way that at least two strains of the tobacco.necrosis virus have their molecules in a cubic close packing (Fig. 9). Recently photographs have been published (10) which indicate that the same arrangement prevails in crystals of the plant globulin edestin. The molecules of the lorn symmetry crystals of the Rothamsted tobacco necrosis protein are in a slight distortion of this cubic close packing. Though its structure has not been completely

FIG.10. A section through three polyhedral bodies developed in a virus-diseased caterpillar. Regularities appearing in parts of these sections may reflect the crystalLine arrangement of the macromolecules of nucleoprotein of which these bodies are composed. 19,000 X

established, the Southern bean mosaic virus protein seems to have the much greater distortion of this arrangement that would be required by a close packing of its definitely spheroidal elementary particles. At present these studies of crystalline order are restricted to macromolecular substances whose crystals do not collapse when dehydrated and whose faces are not obscured by monomolecular denatured films. How far these methods can be extended to substances of lower molecular weight can only be told through further trials, but even with thg materials at hand there are many novel and interesting experiments waiting to be done. The structures of the crystals just mentioned have been deduced from the molecular arrangement on their faces. The ability to cut sec-

THE E L E C T R O N MICROSCOPY O F M A C R O M O L E C U L E S

19

tions thin enough for electron microscopy makes it conceivable that in some instances molecular arrangement within a crystal may be seen in sections cut through single individuals. We have partially succeeded in doing this while examining the so-called polyhedral bodies formed within virus-diseased caterpillars (33). Figure 10 shows a section cut through one of these bodies, which are composed primarily of nucleoproteins having a molecular weight that is probably of the order of half a million. In places within this body there are regions of orderly particle arrangement that may reproduce the original structure undamaged by the sectioning. 2. The Growth of Crystals The electron microscope makes it possible to follow (43) the way a crystal is built up as well as to see it in its finished form. Suspensions of

FIG.11. A dried deposit of a suspension of polystyrene latex spheres. Multiple layers occur in the bottom, more opaque half of the picture. The spheres are transparent enough so that a second layer can show through the topmost one. Regularities in its arrangement result in the regular patterns seen within the top spheres. 7,000 X

uniform spherical particles of a wide variety of sizes have given ordered deposits when dried down on a smooth substrate. At great dilutions this arrangement is usually a close packing in a plane, but this hexagonal net is sometimes interspersed with areas where the packing is more open and the resulting net is square. At greater concentrations these ordered layers mill form, overlying one another in regular fashion (Fig.1). In the sense that a crystal is by definition a solid whose elementary particles

20

RALPH W. G . WYCKOFF

are in a three-dimensional orderly array, these aggregates have regions of crystallinity. It is important that such orderly stackings have been observed not only of molecules of crystallizable proteins but also of several viruses and bacteriophage and even of the much larger latex spheres of Fig. 11. Since they are as extensive and perfect when composed of the super-molecular latex as of the molecules of many crystallizable proteins, it seems evident that they are not in themselves expressions of molecular attractions operating between the particles.

FIG.12. This and the following two figures show similarly shaped single crystals of t h e Rothamsted tobacco necrosis protein i n various stages of development. I n t h e least “mature” stage shown in this photograph the “crystal” is merely a thick terraced deposit superimposed on the kind of ordered deposit shown in Fig. 1. At t h e top right hand corner, this underlying deposit is approximately in a plane; its thickness a t the bottom right is made evident by the number of narrow terraces t h a t appear there.

33,000 X

Single crystals are polyhedra bounded by more or less flat faces. Such crystals can be obtained of some proteins giving these ordered molecular stackings, but they have not yet been prepared of solids composed of large virus particles or of latex. By working with certain easily

THE ELECTRON MICROSCOPY OF MACROMOLECULES

21

crystallizable proteins, it is possible t o follow the steps whereby the orderly stacking of their molecules can develop into single crystals with well-defined faces. For such a study of the details of crystal formation the so-called Rothamsted protein has proved especially instructive, largely because it can so easily be crystallized from salt-free aqueous solution. Representative transitional steps between a stacked deposit

FIG. 13. In this more developed crystal a side face has appeared on its upper right portion. Both it and the top face are, however, still series of terraces. 44,000 X

such as Fig. 1 and a single crystal such as Fig. 9 are shown in Figs. 12, 13 and 14. In the earliest stages the top layers of a deposit are of irregular extent and outline, but they usually have less area than those below, and this makes it easy to see how exactly their positions and orientations are determined by those of the layers below. Such a deposit can properly be called a crystalline mass. Polygonal outlines later develop (Fig. 12), and systems of terraces (Figs. 12 and 13) are found in the positions of ultimate crystal faces. Such partly formed crystals seem to “mature”

22

RALPH W. G . WYCKOFF

by a filling-up and levelling-off of these terraces till faces are produced that are often molecularly flat (Fig. 14).*

FIG.14. In this nearly finished crystal the top layer is planar except for a few irregularly distributed groups of adsorbed molecules and a few terraces remaining in t h e upper left corner. The side face on the upper right is still somewhat shingled. I n t h e two preceding photographs t h e individual molecules, a few tenths of a millimeter in diameter, are clearly visible; at the low magnification of this photograph they are barely perceptible. 24,000 X

3. Crystalline Order in Muscle Truly crystalline arrangements have thus far been rare among naturally-occurring biological structures. Crystalline inclusions such as those just mentioned in Sect. V, 1 do arise through the activity of certain viruses, but most recognizably crystalline materials found within

* Figs. 12, 13, and 14 from R. W. G. Wyckoff, Electron Microscopy, Interscience, New York, 1949.

T H E E L E C T R O N MICROSCOPY OF M A C R O M O L E C U L E S

23

living organisms have been inorganic structures such as tooth and bone. It has, therefore, been rather unexpected to discover (18) that the macromolecules comprising parts of intact striated muscle are in a threedimensionally-ordered array. A fragment of striated muscle (11, 12, 32), teased from a fiber, shows a succession of bands corresponding to the isotropic and anistropic regions seen with optical microscopes (Fig. 15). These ribbons consist of bundles of the parallel filaments of Fig. 7 crossed in many places by a fine striation that repeats itself about every 400 A.

FIG.15. A teased ribbon of striated rabbit muscle chosen to show the kind of repetitive order that can sometimes be seen along the lengths of such muscle. The relatively transparent isotropic band with its bisecting Z sub-band is in the center of the photograph. 29,ooO x

Until sections through intact fibers could be examined it was impossible to determine the relation between them and the teased ribbons, and to perceive the significance of the transverse order they exhibit. From an early study of sectioned muscle (25) it had been concluded that the myofibril is a hollow tube of which the ribbons are ruptured walls. Our photographs (18) of transversely sectioned muscle on the contrary demonstrate (Fig. 16) that in the intact fiber the filaments of Figs. 7 and 15 are in a compact bundle. In parts of a fiber that can be shown to correspond to the anistropic regions, these filaments are regularly close

24

RALPH W . G . WYCKOFF

packed so that as in Fig. 16 their cross-section is a n hexagonal net of dots. Many of these sections show th at there is a pronounced tendency for this packing t o split parallel to the lines of the net (Fig. 17) ; and this indicates that the ribbons may be such cleavage sheets. Longitudinal sections show a repeated structure which appears t o correspond to the 400 A. striation seen in ribbons. This repetition takes the form of a net which,

FIG. 16. A section almost exactly a t right angles to the long axis of a striated rabbit muscle fiber. The long filamentous macromolecules, perceptible in Fig. 15 and well shown i n Fig. 7, are seen here in section as a close-packed net of points. The separate blocks may be “myofibrils” or they may arise through the shrinkage that is the consequence of desiccating so highly hydrated a structure. 29,000 X

though usually oblique, does not depart too far from rectangular. Its existence, taken in connection with the close packing of filaments revealed in transverse section, shows that some sort of three-dimensional regularity exists within the anistropic regions of these muscle fibers. It is too early t o decide about its fine details, but inasmuch as muscle is a structure that changes dimensions as it does its work, there may well be change in this crystalline order during muscular contraction. Its further

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

25

investigation is bound to contribute towards an understanding of the mechanism of this contraction. 4. Sub-Crystalline Solids

Macromolecular solids can be found whose particles are arranged in various gradations between the perfect three-dimensional order of true crystals and a state of complete molecular disorder. The cellulose fibrils

FIG.17. Another section through muscle cut not quite normal to the fiber axis. This gives the sectioned macromolecular filaments linear as compared with the circular outlines of Fig. 16. In some places the tendency to part along these lines is perceptible; i t is much more pronounced in other photographs. 30,000 X

composing the primary walls of plant cells, for example, are tangled networks devoid of symmetry. Though these are flat, other structures have been found that are irregular three-dimensional networks of anastomosing filaments. The apparent framework of the axon of myelinated nerve is of this character (Fig. 18) and there is increasing evidence that the protoplasm of some cells may be interlaced by similar systems of interconnected threads. Macromolecular filaments are found aligned in bundles and sheets to provide an example of one-dimensional order. The so-called ((crystals” of the tobacco mosaic virus protein, or of “crystalline” myosin (34) (Fig. 4) thus are rather imperfectly aligned sheaves of filaments. Analo-

26

RALPH W. G . WYCKOFF

gous structures seemingly are produced by the larger elongated particles in a vanadium pentoxide gel (39). I n the secondary wall of plants the elementary fibrils lie more or less perfectly parallel t o one another. The extensive membranes that can be formed from the tobacco mosaic virus protein are sheets of parallel molecules, and it is possible that they are models on a large scale of the molecular arrangement prevailing in many naturally-occurring membranes. A similar but more perfect one-dimen-

FIG.18. A section through a myelinated nerve fibril from the rabbit showing the kind of irregular three-dimensional network of fine filaments that is seen in the axon in many preparations. The bounding myelin appears in the diagonal upper right and lower left corners. 15,000 x

sional order has been seen on the surface of gels of the tobacco mosaic protein (43), but present observations do not show if this order is more than one molecular layer thick. Tendon is a parallel array of fibrillar elements, but the arrangement here is at once more and less symmetrical than the foregoing. The separate collagen fibrils are not of a uniform diameter, but they do have an obvious repetitive structure along their lengths; and in a carefully teased-out fragment of tendon these repetitions on neighboring fibers are often seen in a regular transverse alignment that gives the whole something approaching two-dimensional order. The structure that is responsible for these repetitions along the fibrils of tendon has been the subject

T H E ELECTRON MICROSCOPY O F MACROMOLECULES

27

of several studies. Usually a separated fibril appears as a string of discs which are ca. 650 A. from center to center. It has been difficult to decide concerning the nature and fine structure of the ridges that produce the disc-like appearance. Occasionally traces can be seen of structure within t,he discs and a better knowledge of their probable structure has been gained through the examination of collagen from other forms of connective tissue. Though agreement has not yet been reached concerning the fine structure of these collagens, and it may well be that this structure is not

FIG.19. A group of connective tissue fibrils from heart muscle showing the character of their cross striations. The distance between pairs of their cross threads is cu. 650 A. 40,000X

always the same, certain facts have already become clear. The raised discs are regions where the fibrils are crossed by transverse ridges which often seem embedded in an obscuring amorphous material. The discrepancies concern the number and nature of these ridges. In tendon there apparently are two principal ridges per disc ; in connective tissue from human skin (7), staining with phosphotungstic acid has sometimes brought out a third striation per disc and as many as three additional striations between discs. In connective tissue of the heart (29), it is common to find filaments like those of Fig. 19. Shadowing brings out a pair of ridges corresponding to each disc of a fibril of tendon; sometimes a third ridge between each pair gives the impression of a continuous cross striation. Often these continuously striated regions and regions of paired ridges occur along the same fibril. More than one intermediate ridge, yielding a total of four or more striations in a 650 A. period, have

28

RALPH W. G . WYCKOFF

not been observed in this type of collagen. The white connective tissue associated with muscle and with yellow connective tissue contains masses of thin continuously striated fibrils all of about the same diameter (Fig. 20). Their distances between ridges have been ca. 650/3 = ca. 215 A. These are probably structurally the same as the “unit fibers” found in tissue cultures of chick embryo skin (28); they are indistinguishable from the continuously striated microfibrils that are present in great numbers in the mesodermal layer of the chorioallantoic membrane of the chick. Wherever they occur, they are often seen grouped together in tight bundles and in some of these the cross striations are well-aligned

FIG.20.

A fragment of continuously cross-striated connective tissue. These cross striations are ca. 215 A . apart. 35,000 X

over large areas. It will be important t o determine what relation if any exists between these uniform fibers and the larger discontinuously striated collagen. The relation between the ridges on connective tissue fibrils and the internal structure of the fibrils as revealed by staining or by swelling remains obscure. As a result of earlier work dealing with the behavior of collagen on heating and with diseased tissue, it was thought t h a t collagen had two forms which could transform into each other either through maturation or as a consequence of disease (41). The idea sometimes propounded that the thin continuously striated fibrils are

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

29

steps towards the formation of the 650 A. fibrils fits such a picture. One hypothesis concerning the nature of the cross striae has made them reflections of periodic alterations of the internal structure of the fibrils. Some knowledge of its internal structure has been gained by swelling tendon from young animals in very dilute acid. This shows that the fibrils are made up of bundles of seemingly thread-like molecules having diameters nearly at the limit of resolution of the microscope. They could be the same as the separate “protofibrils” seen in the tissue cultures (28). There is, on the other hand, considerable evidence that the ridges are remains of a system of cross threads binding the fibrils into a sort of fabric. In support of this idea is the parallel alignment of adjacent fibrils in many specimens and the fact that in some places the ridges seem to pass unbroken over more than one fibril. If this interpretation is substantiated, such connective tissue will be an especially good example of a kind of repetitive order which, like a cloth, involves the regular interplay of more than one kind of filamentous component. To date work with connective tissue has necessarily been carried out with samples that have been disintegrated mechanically in order to have them thin enough for observation. Obviously only traces of their original arrangement can remain. When sections satisfactory for electron microscopy have been studied a far clearer picture will be gained of the structure of and relation between these various forms of tissue. Yellow, elastic, connective tissue has revealed a very interesting structure (6). Without special treatment it does not show an evident fine structure, but after tryptic digestion many spiral elements as well as the bundles of continuously striated filaments referred to above are left behind. Presumably in the intact tissue the spiral elements embedded in a seemingly amorphous menstruum play a part in determining the elastic properties this tissue shows. VI. FORMATION OF MACROMOLECULES 1. ((Healthy” Macromolecules

Though it is one of the most promising fields of investigation opened up by the electron microscope, very little has yet been done to find out how macromolecular particles are formed in nature. Such study will proceed in two directions, one designed to find out how viruses proliferate, the other concerning itself with the growth of the macromolecules of healthy tissue. The examination of the viruses will be at the outset more rewarding and in many respects easier, but there are healthy’’ macromolecular systems that call for immediate attention. Some macromolecules are constantly being formed during the life of the organism of which

30

RALPH W. G . WYCKOFF

they form a part. The development of fibrin from fibrinogen is an example of this. The collagen fibrils of connective tissue are also being formed in adult animals, a t least in response t o damage which the animal suffers. But the growth of many macromolecular structures, such as the cell walls of plants and the muscle and tendon of animals occurs primarily during the period of initial formation of the organism; their investigation is thus a chapter in a macromolecular embryology t h a t will gradually take shape as these studies proceed. A few fragmentary observations have already been made. Thus a start has been made towards seeing how fibrin (9, 14, 27) forms from fibrinogen under the action of thrombin. These experiments have shown that relatively concentrated thrombin and a consequent rapid clotting is required t o give the striated fibrin already described; with slower clotting the fibers are devoid of internal detail. They also have suggested t h a t fibrin results from some sort of side-by-side association of thread-like molecules of fibrinogen, but it is not now clear how this happens or how the striae arise through this process. An approach has also been made t o the fascinating problem of how the elementary filaments of cellulose are produced by plant cells. Imperfectly formed cellulose fibrils have been found in the slimy coats of certain seeds (23), all stages in the development of primary and secondary walls of plant cells can be seen in the coleoptiles (22) of sprouting seeds, and with cultures (21) of Acetobacter xylinum i t has been possible t o follow the actual formation of cellulose fibrils. When these bacteria start t o grow in a fresh medium they first become heavily encapsulated; and then this capsular stuff disappears as cellulose begins t o appear in the culture. The first fibrils of cellulose appear in the extracellular capsular material and rapidly increase in number once they start t o form. It seems evident that in this case cellulose is polymerized extracellularly from the apparently amorphous capsular material. The slimes enveloping many swollen seeds contain cellulose fibrils like those from other sources, but those of some seeds, such as linseed, have only clusters of short particles which may be incompletely-formed cellulose. It is interesting that if these are freed of the polyuronic acids in which they are embedded, they associate together into longer fibrils that approach ordinary cellulose in appearance. The formation of cellulose and its organization into primary and secondary wall structures can be followed by examining the coleoptiles (22) of sprouting seeds. This has been done using fragments from the mechanical disintegration of the coleoptiles ; still more information will beyond doubt be obtained now that it is becoming possible t o prepare sections of suitable thinness for electron microscopy. The mechanically

T H E ELECTRON MICROSCOPY OF MACROMOLECULES

31

disintegrated material has not shown partly-formed microfibrils. I n the very young shoots few filaments are seen, but they are already of the usual thickness and lengths. Older sprouts furnish pieces of primary and secondary wall that are identical under the electron microscope with those observed in fragments of commercial fibers and of wood. 2. Viruses

Though it is not too difficult to visualize the elementary particles of a number of viruses within the cells they invade, a reconstruction of the mechanism of their growth will not be simple. Viruses thus far examined illustrate several of the reasons for this. Thus with bacteriophage a complete understanding of what is seen under the electron microscope is made difficult by the great rapidity with which bacteriophage is formed, by the complexities and differences that exist in bacterial protoplasm and by the modified patterns of bacteriophage production that result from these differences. Bacteriophage, and the influenza virus as well, also offer complications due to their own diverse morphologies. Both have shown filamentous forms (4,15, 19, 45) in addition to the more familiar sperm-like units of bacteriophage and spherical elementary bodies of influenza. Besides the filaments of influenza virus there are also t o be seen in infected tissues connective tissue elements of similar shape which can only be distinguished from virus by their smaller diameters, by their striations which can be seen only in good preparations, and by their different sites of origin. This is an illustration of the fact that healthy tissue often contains macromolecules of the same general size and shape as virus. Such macromolecules have already been noted during attempts to purify various plant viruses as well as the virus of poliomyelitis and they are proving to be a serious problem in the identification of spherical viruses within sectioned plant tissues. A large measure of experience with tissue in both healthy and diseased states is a necessary, if time-consuming, preliminary to sound progress in this field of investigation. Another important problem is brought to the fore when we examine tobacco mosaic virus in the leaf cells of diseased plants (2). In these plant cells, and the same thing is true of many infected animal cells, virus is associated with much obscuring, non-infectious macromolecular material of smaller molecular size. It interferes with the recognition of small amounts of virus and thus complicates visualization of early stages of infection, especially where small viruses are concerned. The basic problem in the study of virus proliferation, however, involves learning how to preserve with minimum of damage the fine structure of cells and how to gain through experience the ability to inter-

32

RALPH W. G . WYCKOFF

pret what appears in such improved preparations. With the growth of this knowledge we will inevitably find ourselves in a position to pursue with increasing vigor an inquiry into how “healthy” macromolecules, such as those described earlier in this chapter, are produced in nature.

REFERENCES 1. von Ardenne, M. (1940). Elektronen-Uber-Mikroskopie, Springer, Berlin. 2. Black, L. M., Morgan, C., and Wyckoff, R. W. G. (1950). Proc. Sac. Exptl. Biol. &fed.73, 119. 3. von Borries, B. (1949). Die ubermikroskopie, Saenger, Berlin. 4. Dawson, I. M., and Elford, W. J. (1949). J . Gen. Microbial. 3, 298. 5. Eriksson-Quensel, I.-B., and Svedberg, T. (1936). Biol. Bull. 71, 498. 6. Gross, J. (1949). J . Exptl. Med. 89, 699. 7. Gross, J., and Schmitt, F. 0. (1948). J. Exptl. Med. 88, 555. 8. Hall, C. E. (1949). J. Am. Chem. Soc. 71, 1138. 9. Hall, C. E. (1949). J. Bzol. Chem. 179, 857. 10. Hall, C. E. (1949). J. Am. Chem. Sac. 71, 2951; (1950) J. Biol. Chem. 186,45. 11. Hall, C. E., Jakus, M. A., and Schmitt, F. 0. (1945). J. Applied Phys. 16, 459. 12. Hall, C. E., Jakus, M. A., and Schmitt, F. 0. (1946). Biol. Bull. 90, 32. 13. Hambraeus, G., and Ranby, B. (1945). Nature 166, 200. 14. Hawn, C. Z., and Porter, K. R. (1947). J. Exptl. Med. 86, 285. 15. HerEik, F. (1950). Experientia 6, 64. 16. Jakus, M. A., and Hall, C. E. (1947). J. Biol. Chem. 167, 705. 17. Kinsinger, W. G., and Hock, C. W. (1948). Ind. Eng. Chem. 40, 1711. 18. Morgan, C., Rossa, G., Szent-Gyorgyi, A., and Wyckoff, R. W. G. (1950). Science 111, 201. 19. Pvlosley, V. M., and Wyckoff, R. W. G. (1946). Nature 167,263. 20. Muhlethaler, K. (1949). Biochim. et Biophys. Acta 3, 15. 21. Muhlethaler, K. (1949). Biochim. et Biophys. A d a 3, 527. 22. Muhlethaler, K. (1950). Biochim. et Biophys. Acta 6, 1. 23. Muhlethaler, K. (1950). Exptl. Cell Research 1,341. 24. Oster, G., and Stanley, W. M. (1946). Brit. J. Exptl. Path. 27, 261. 25. Pease, D. C., and Baker, R. F. (1949). Am. J. Anatomy 84, 175. 26. Polson, A., and Wyckoff, R. W. G. (1947). Nature 160, 153. 27. Porter, K. R., and Hawn, C. Z. (1949). J. Exptl. Med. SO, 225. 28. Porter, K. R., and Vanamee, P. (1949). Proc. SOC.Exptl. Biol. Med. 71, 513. 29. Pratt, A. W., and Wyckoff, R. W. G. (1950). Biochim. et Biophys. Acta 6 , 166. 30. Preston, R. D., Nicolai, E., Reed, R., and Millard, A. (1948). Nature 162, 665. 31. Fbzsa, G., Ssent-Gyorgyi, A., and Wyckoff, R. W. G. (1949). Biochim. et Biophys. Acta 3, 561. 32. Rozsa, G., Szent-Gyorgyi, A., and Wyckoff, R. W. G. (1950). Exptl. Cell Research 1, 194. 33. Smith, K. M., and Wyckoff, R. W. G. (1950). Nature 166, 861. 34. Snellman, O., and Erdos, T. (1948). Biochim. et Biophys. Acta 2, 650. 35. Snellman, O., Tenow, M., and Erdos, T. in Pederson, K. 0 . (1948). Ann. Reu. Biochem. 17, 192. 36. Stanley, W. M., and Anderson, T. F. (1942). J. Biol. Chem. 146, 25. 37. Straub, F. B. (1948). Hung. ActaPhysiol. 1, 150. 38. Svedberg, T., and Stamm, A. J. (1929). J . Am. Chem. Soe. 61,2170.

THE ELECTRON MICROSCOPY O F MACROMOLECULES

33

Watson, J. H. L., Heller, W., and Wojtowicz, W. (1949). Science 109, 274. Williams, R. C., and Wyckoff, R. W. G. (1945). Nature 166, 68. Wolpers, C. H. (1947). K l i n . Wochschr. 24-26, 424. Wyckoff, R. W. G. (1947). Biochim. et Biophys. Acta 1, 139. Wyckoff, R. W. G. (1948). Acta Crystal. 1, 292. Wyckoff, R. W. G. (1948). Biochim. el Biophys. Acta 2, 27, 246. Wyckoff, R. W. G. (1950). Ezperientia 6, 66. 46. Zworykin, V. K., Morton, G. A., Ramberg, E. G., Hillier, J., and Vance, A. W. (1945). Electron Optics and the Electron Microscope. Wiley, New York. 39. 40. 41. 42. 43. 44. 45.