Various Types of Ghosts Derived from Human Red Cells: Heat Fragmentation and Phase Optics Studies

While the purpose of this paper is to illustrate, rather than to describe exhaustively, the multiplicity of the kinds of ghosts which can be obtained (a) by haemolysis in hypotonic media, (b) by haemolysis in hypotonic media followed by ‘reversal of haemolysis’, (c) by haemolysis by freezing and thawing, and (d) by haemolysis with various lysins, it is convenient to extend the illustrations to include descriptions of the methods by means of which the various kinds of ghost can be obtained reproducibly.

All the observations to be described were made with a phase-optics system which employs an oil immersion (97 x) objective, a 5 × eyepiece, and a green filter. Direct observation of individual fragmenting cells can also be made with phase optics and the heating chamber already described (Ponder, 1950 a). A chess-board micrometer mounted in the eyepiece, the squares corresponding to about 4µ, enables estimates of the sizes of red cells and of ghosts to be made.

The technique for producing and studying fragmentation has already been described (Ponder, 1950 a, b).

These ghosts are prepared by adding washed and incubated ^* human red cells to large volumes of water, allowing the haemolysate to stand at 40 C., and separating the ghosts by centrifuging (Ponder, 1951). The shape of the ghosts prepared by this method depends on how long they have been allowed to stand, either in contact with the haemolysate at 4° C., or separated from it. If less than 24 hr. have elapsed since the commencement of the preparation, the ghosts, which can be seen very clearly with the phase optics, are biconcave or cup-shaped disks. If they have stood for longer times, they tend to be spherical, and may even show spontaneous fragmentation.

The ghosts do not fragment in the medium in which they are suspended, even if they are heated for several minutes to 80° C. If NaCl is added, heating to 52° C. for 3 min. breaks them into many fragments of varying sizes (cf. Ponder, 1950 a). By varying the concentration of NaCl, it can be shown that a concentration of between 0·05 g./100 ml. and 0·1 g./100 ml. (0·0085-0·017 M) is critical; at concentrations above this, fragmentation occurs almost independently of the concentration, while at lower concentrations there is virtually no fragmentation at all. The salt must be present during the period of heating; adding it, washing, and then heating does not result in fragmentation, and heating the ghost suspension and then adding salt is equally ineffective.

The effect of KC1, LiCl, CaCl₂, Na₂SO₄, NaF, NaBr and Nal, all in 0·017M concentration, is indistinguishable from that of 0·017M-NaCl. Fragmentation also occurs in systems containing 0#x00B7;034M glycine or sucrose, although its extent is usually less than in systems containing the same concentration of NaCl, KC1, LiCl, etc. Changes in pH, brought about by phosphate buffers, have a slight effect on the fragmentation observed when the ghosts are heated to 52° C. in 0·017M-NaCl, the fragmentation being somewhat greater at pH 8·6 than at pH 5·6.

Ghosts prepared in a large volume of water and exposed to high concentrations of saponin dissolved in 0·1 % NaCl are not fragmented when heated for 3 min. to 55° C. If the concentration of saponin is less than about 200µg./ml., on the other hand, fragmentation occurs on heating, just as it does in 0·1 % NaCl.

Ghosts prepared in large volumes of water do not undergo a disk-sphere transformation between slide and cover-glass (Ponder, 1942), and the addition of serum or of 1 % serum albumin produces little immediate change of shape. The ghosts in the presence of either serum or serum albumin, however, are less smooth than in water alone ; this may be the result of an osmotically produced volume decrease, or may be related to the spontaneous fragmentation which is seen in the preparations after the lapse of some hours.

When the ghosts suspension containing equal volumes of serum is heated, there is extensive fragmentation with the production of long myelin forms. The extent of the fragmentation depends on the temperature and duration of the heating, and is the same even if 1 % NaCl is added to the preparation in addition to the serum, which itself supplies the electrolyte necessary for fragmentation. When 1 % salt-free serum albumin is added to the ghosts suspension, either with or without the addition of 1 % NaCl, the fragmentation which occurs during heating in the presence of salt is suppressed almost entirely; the ghosts of the heated suspension are shrunken, distorted and agglutinated, but they are not fragmented as they are in systems containing more than 0·1 % NaCl but no albumin. Myelin-form formation, however, is much more conspicuous than in the latter systems, although it differs from the myelin-form formation in systems of ghosts containing serum, the forms being short and shaggy instead of long and thin.

These effects may be summarized by saying that the addition of serum to suspensions of ghosts prepared in large volumes of water results in their being fragmented by heat and in the production of one type of myelin form, whereas the addition of serum albumin protects the ghosts from heat fragmentation and leads to the production of a morphologically different type of myelin form.

Ghosts can be prepared by haemolysing human red cells in volumes of water which vary from about 5 to over 100 vol., the watery haemolysate being brought, after various lengths of time at various temperatures, to isotonicity by the addition of salts (‘reversal of haemolysis’). By varying the kind of salt added, the reversal can be brought about at various values of pH, etc. The ghosts in such systems can be centrifuged down, examined, heated, frozen, and so on, and the properties of the ghost vary with the method of preparation.

A series of solutions which, when diluted to a concentration of 0·1M, give pH values from 5·5 to 7·3, can be prepared from mixtures of 1·5M-NaH₂PO₄ and 1·5M-Na₂HPO₄. Suspensions of red cells haemolysed in 15 (or more) vol. of water can be brought to isotonicity at any pH within the range by adding 1 (or more) vol. of these solutions; if the cells are haemolysed with less than 15 vol. of water, the phosphate mixtures are diluted proportionally before being added to the haemolysate. The pH of the systems is limited by the low solubility of Na₂HPO₄, but one can reach pH 8·0 by using 0·6M-Na₂HPO₄ and adding proportionately larger volumes of the buffer mixtures to bring about reversal of haemolysis.

The red cells of heparinized human blood can be quickly washed three times with saline to make a suspension of the same volume concentration as that of blood, and then haemolysed in 15 vol. of water at 25° C. If the haemolysate is allowed to stand for 15-20 min., 1 vol. of 1·5 M phosphate buffer being added at the end of the period of standing, ghosts can be thrown down (2 × 10³g. for 15 min.). These ghosts are flat, coarsely crenated disks, which do not show spontaneous fragmentation in times up to 2 hr. If the pH of the added buffer is 5·5, heating to 53-55° C. converts the cells into spheres but does not fragment them. If the pH is higher than about 6·5, the cells become spherical and some of them break into two or three fragments; there is, however, no extensive fragmentation and very little appearance of myelin forms.

If the same suspension of human red cells is kept at 4° C. for 12 hr. or more, repetition of the same process as that described above results, at all values of pH, in a yield of thin, coarsely and finely crenated ghosts. The fine crenations are most conspicuous at the edges. These ghosts have no marked tendency to spontaneous fragmentation, although some of the small crenations may break off to form tiny beads at, or separated from, the cell edges. If the pH of the added buffer is 5·5, heating to 53—55° C. results in spherical forms which do not fragment, but if the pH is 6·2 or more, heating results in the ghosts breaking into innumerable tiny spherical fragments accompanied by many long myelin forms. Comparison of this result with that obtained with freshly prepared ghosts shows that heat fragmentation is favoured by an increase in pH and also by the time during which the preparation stands, i.e. by the ‘age’ of the preparation.

The buffer system described by Michaelis (1931) has the advantage that it gives pH’s from 2·6 to 9·6, the concentration of the salts (sodium acetate, sodium veronal, NaCl and HC1) being such as to have a constant ionic strength at all values of pH, and to be isotonic with plasma. The buffer system is somewhat limited by its low solubility at low pH. Mixtures above pH 8 can be prepared with a tonicity of 5 times that of 1 % NaCl; mixtures at pH less than 8 are prepared with a tonicity of twice that of 1 % NaCl. Reversal of haemolysis is brought about by adding 1 vol. (above pH 8), or 4 vol. (below pH 8) of the buffers to freshly prepared washed red cell suspensions (volume concentration 0-4) haemolysed with 4 vol. of water.

Between pH 5 and 8, the ghosts produced by the reversal are disks with a moderate amount of coarse crenation. At pH 4 they tend to be smooth and cupshaped, with an accentuated rim; at pH 9 they tend to be crenated spheres. The effect of heating to 53° C. for 3 min. depends on the pH of the added buffer. At pH 4, there is no fragmentation, shrunken irregularly-shaped ghosts of from 4 to 6 µ in diameter being aggregated into masses. At pH 5, the ghosts are dull objects with almost no surface reflectivity ; fragmentation apparently takes place, but the fragments either do not separate completely or stick to each other after separation in small irregular clusters about the size of a red cell. At pH 6, there is fragmentation into spheres which have a low surface reflectivity; many minute myelin forms can be seen at their edges, and their dull appearance may be due to the surface being covered with small myelin forms. As the pH is increased, the fragmentation and the development of myelin forms increases, so that many of the fragments are flat dull objects to which extruded myelin forms are attached and which break down, as time goes on, into networks of myelin forms of many shapes and sizes.

As compared with the results obtained with freshly prepared ghosts after reversal of haemolysis with phosphate, the formation of myelin forms in veronal systems is quite remarkable; it resembles the myelin-form formation which is seen in phosphate systems only if the cells have stood at low temperatures for 24-48 hr. The appearance of the non-reflective masses of fragments after heating at pH 5, and the extensive fragmentation and myelin network formation after heating at pH 8 and 9 are peculiar to the veronal systems.

Ghosts made by Parpart’s (1942) method, which consists in haemolysing washed red cells with water, adding them to a large volume of cold CO2-saturated water, and washing repeatedly with this same medium, are almost Hb-free and tend to agglutinate into masses. Seen singly with phase optics, they are very thin and tenuous bodies, roughly circular, and flat or cup-shaped. They do not have the distinct rim which ghosts prepared in large volumes of water have, nor do they have as great a volume.

On heating them to between 52 and 56° C., either in the medium in which they are suspended or with added 1 % NaCl, no fragmentation is observed. There is a great deal of agglutination into masses of distorted objects, which, since they have a diameter of about 5µ, are probably roughly spherical. The surfaces of these ghosts are covered with shaggy irregularities which are probably myelin forms and which no doubt contribute to the great adhesiveness between the ghosts. When the preparation is pressed on, many of the myelin forms become dislodged, and float in the medium as tiny droplets or fine threads. Except for the appearance of the myelin forms, the ghosts formed in CO2-saturated water resemble those found after phosphate reversal of haemolysis by phosphate at pH 5·5.

There is independent evidence of a difference between the ghost formed by haemolysis in large volumes of water and those which are subsequently flocculated out with CO2. The electrophoretic mobility of the latter is only about 80 % of that of the former (Furchgott & Ponder, 1941), and the latter do not disintegrate into myelin forms when lyotropic agents are added (Furchgott, 1940).

While red cells which are frozen to—20° C. for 12 hr. and then thawed undergo haemolysis, they do not fragment. The form of the ghost is that of a sphere. Heating the ghosts to between 50 and 6o° C. does not fragment them.

Ghosts prepared in large volumes of water are not fragmented by freezing and thawing, provided that they are relatively fresh, i.e. that not more than 72 hr. has elapsed since the beginning of the process by which they are prepared. This is true whether the form of the ghost is that of a disk or that of a sphere. After freezing and thawing, however, many of the watery ghosts show a vacuolation phenomenon which looks very similar to that described by Dervichian & Magnant (1947) in haemoglobin-nucleinate-myristylcholine coacervates. Sometimes one or two, but more often several round or oval vacuoles appear in the substance, or in the interior, of the ghost, which itself is often so large (12µ in diameter and apparently spherical because it is never seen on edge) as to suggest that either imbibition or coalescence phenomena are involved. Heating the ghosts of these frozen and thawed preparations does not produce fragmentation, either in the absence or in the presence of salt.

If the ghosts are not relatively fresh, i.e. if they have been kept for more than 96 hr. either in the large volume of water in which they are prepared or even separated from it, the discoidal form tends to be replaced by the spherical form, and both tend to undergo spontaneous fragmentation. Freezing and thawing of such ghosts increases the amount of fragmentation considerably. Heating the ghosts, either with or without salt, however, does not increase the fragmentation appreciably. The conclusion is that freezing and thawing, whether of red cells or of watery ghosts, produces a ghost which is not fragmented by heat. The properties of the watery ghost must therefore be changed, by freezing and thawing, into those of the ghost produced from the red cell by freezing and thawing.

The ghosts resulting from the action of different haemolysins on human red cells are different in appearance and in their ability to be fragmented by heat. Their appearance and behaviour are probably dependent on the concentration of lysin used, but the following descriptions will illustrate the principal differences observed. In each case the haemolytic system was composed of i ml. of washed red cells of fresh human blood, suspended in 1 % NaCl in a volume concentration of 0·4, with the addition of 4 ml. of the lysin dissolved in saline in the concentration given. The concentrations were such as would produce complete lysis within 5 min. at 25° C. ; the results are accordingly those for systems containing the lysins in concentrations which are neither very great nor very small. The ghosts are separated by centrifuging, a little saline being used to suspend them. The descriptions of the effects of heating refer to a 3 min. heating to 53° C.

The ghosts are pale, round or spheroidal bodies measuring 5—6µ in diameter. Their edges are sharp; there is no substantial ‘rim’ which would suggest that the surface of the cell encloses a considerable volume. No spontaneous fragmentation or development of myelin forms occurs. After heating, there is no noticeable change in shape, and there is no fragmentation. If the concentration of saponin in the system is reduced to about 200µg./ml., however, fragmentation takes place on heating (see above).

The ghosts are discoidal, cup-shaped, and sometimes biconcave bodies, very thin, but with a clearly observable ‘rim’. The ghosts tend to stick together in masses. Heating results in extensive fragmentation into discrete, round fragments of all sizes, down to those of a ‘dust’. No long myelin forms, however, are to be seen. When the concentration of digitonin in the system is some 5 times greater, however, heating does not produce fragmentation. As in the case of saponin, the occurrence or non-occurrence of heat fragmentation depends on the concentration of the lysin.

The ghosts are disks, usually cupshaped, but occasionally biconcave, not so thin as the ghosts in digitonin systems, and with a clearly visible ‘rim’. They measure about 8µ in diameter, and from 0·5 to 1·5 µ. in thickness. After heating, the disks are replaced by spheres of various sizes (2-4µ in diameter). These do not appear to be the result of a fragmentation process, but rather of the disks of varying volumes having assumed the spherical form. It is tempting to think of these spheres as being composed of the ‘fixed framework’ of the ghost in the form of a solid mass; since the diameters of the spheres vary from 2 to 4µ, their volumes vary from about 4µ³ to about 30 µ³, and something between these two values would be a likely enough figure for the volume of the ‘fixed framework’ of the ghost. Many of the small spheres have small, shaggy myelin forms attached to their surfaces.

The ghosts are small spheres with a diameter of between 2 and 4µ. After heating they are replaced by many tiny spheres with diameters between 1 and 2 µ. The change in size probably involves a fragmentation or a disintegration phenomenon, and the surfaces of many of the little spheres is irregular and covered with minute myelin forms.

As an extension of these observations, suspensions of ghosts were prepared by the method described in §1, and lysins in various concentrations were added. After observing the changes produced by the lysin, each system was heated for 3 min. to 53° C. and re-examined. The phenomena observed were much the same as those observed in systems containing red cells and lysins. Ghosts prepared in large volumes of water, although discoidal and readily fragmented by heat when salt is present, are not fragmented by heat when treated with saponin in saline in concentrations greater than about 200µg./ml. Fragmentation is also prevented by treating the watery ghosts with concentrated solutions of digitonin, although heat fragmentation and myelin-form formation occurs if the concentration of digitonin is less than about 50µg./ml. Treatment with sodium dodecyl sulphate converts the discoidal watery ghosts to small spheres which are exceedingly difficult to see even with phase optics; heating of these results in still smaller spheres, probably as a result of shrinkage or the loss of myelin material into the surrounding fluid. Treatment of watery ghosts with oleate results in the formation of spheres, some with myelin forms attached; heating of these result in still more myelin-form formation, but not in any real fragmentation.

The generalization which seems to hold is that the ghosts produced by the action of these haemolysins, or by the action of these haemolysins on the ghosts produced by water, may be either discoidal or spherical, but that they are not fragmented by heat if the concentration of the lysin is greater than a certain critical concentration. In all these systems, spontaneous fragmentation and myelin-form formation are not at all conspicuous, at least when the cells and ghosts are derived from freshly drawn heparinized blood.

It will be helpful to condense the foregoing detailed descriptions into a table (Table 1) and a list of conclusions. The table will show how the shape of the ghost and the effect of heat in producing fragmentation, myelin forms, and shape changes depends on the method of preparation; the list which follows will summarize the principal generalizations which can be arrived at from the observations :

• (1) The ghost which results from haemolysis in large volumes of water does not fragment when heated unless one of the number of substances, which includes NaCl, glycine and sucrose, is added in sufficient concentration.

• (2) This fragmentation is prevented by serum albumin.

• (3) Fragmentation of ghosts by heat is favoured by a high pH and by increasing ‘age’ of the preparation; this seems to be true of ghosts however formed, provided that they fragment at all.

• (4) Different methods of preparation result in the formation of ghosts with different properties, e.g. the ghosts formed in CO₂-saturated water have a low electrophoretic velocity, and do not fragment when heated; the ghosts formed in veronal systems of pH greater than 6 develop extensive myelin forms when heated.

• (5) Ghosts prepared by freezing and thawing do not fragment when heated, and freezing and thawing of ghosts produced by haemolysis in water renders them non-fragmentable by heat even in the presence of salt.

• (6) When ghosts are formed by the action of chemical lysins such as saponin and the soaps, their shape may be either discoidal or spherical, but, if the concentration of lysin is great enough, they do not fragment when heated.

To begin with the first observation on the list, an obvious possibility is that the fragmentation or non-fragmentation of ghosts, on being heated in the systems described, is determined by their shape, discoidal ghosts, like discoidal red cells, being easily fragmented, whereas spherical ghosts, like spherical red cells, are fragmented by heat with greater difficulty or not at all (Ponder, 1950 a). This simple idea is untenable in view of the fact that ghosts prepared in large volumes of water are discoidal if less than about 24 hr. elapse between the beginning of the process of preparation and the separation of the ghosts from the water, and that these ghosts do not fragment on heating to 53° C. unless salt is added. If prepared by reversal of haemolysis by phosphate, indeed, discoidal ghosts do not fragment even if salt is present provided that they are freshly prepared from fresh red cells.

Another simple idea is that fragmentation occurs only when the ghost contains a considerable concentration of surplus Hb, i.e. that the ghost prepared in a large volume of water does not fragment when heated because the concentration of Hb, although considerably in excess of that in the surrounding medium, is not great enough; a still greater concentration of surplus Hb would be, on this hypothesis, the necessary condition for fragmentation, and would be produced only when the volume of the ghost is reduced by the addition of NaCl or other osmotically active substances. It is certainly true that the processes which result in ghosts which are not fragmentable by heat (e.g. lysis by saponin or digitonin, or lysis by freezing and thawing) are processes which leave very little surplus Hb.^* If the concentration of surplus Hb is a factor which enters into the situation, the way in which it does so is obscure. On the view that fragmentation by heat occurs at an unstable stage in the transition from the initial state of the ghost ultrastructure to its final state of a viscous fluid (Ponder, 1950 a), the surplus Hb, which is now regarded as a component of a lipoprotein-Hb complex (Ponder, 1951), could possibly supply the viscous element, but the idea that the concentration of surplus Hb determines the ease and extent of fragmentation is not entirely adequate as a simplifying hypothesis even when it is combined with the idea of the discoidal shape as a determining factor. As has already been remarked, ghosts prepared from fresh red cell suspensions by reversal of haemolysis, do not fragment when heated even though they are disks and contain relatively large amounts of surplus Hb. Again, ghosts prepared by reversal of haemolysis by the addition of phosphate or veronal buffers do not fragment when heated if the pH of the system is sufficiently low, although they are discoidal and although they contain various and considerable amounts of surplus Hb. Finally, the hypothesis does not account for the action of serum albumin in preventing heat fragmentation which would occur in its absence, ^† nor can it be extended to cover subsidiary occurrences such as the appearance or non appearance of myelin forms and their variety, or the effect of ‘ageing’. Factors in addition to those of shape and surplus Hb concentration seem to be involved, and the situation is apparently too complicated to be explained in a simple way.

Although not entirely satisfactory as an explanation, the idea that fragmentation of ghosts depends largely on the concentration of surplus Hb is interesting because it leads to a new set of relations. The hypothesis that fragmentation occurs at a stage in a transition towards a viscous fluid can be restated by saying that fragmentation depends on the plasticity or cohesiveness of the cell as a whole. Now plasticity may be expected to have a special value corresponding to some special value for the cell volume, for if the cell volume is too small, the Hb molecules will be too closely packed to provide plasticity, whereas if the cell volume is too large, the Hb molecules will be so far apart that plasticity will be lost. We may accordingly accept plasticity as a function of cell volume and of the concentration of surplus Hb fragmentation, which depends on plasticity, being a function of tonicity, which regulates volume. The fragmentation-tonicity relation should, on this argument, show a maximum value, i.e. there should be one tonicity which is optimal for fragmentation. It is too difficult to investigate the fragmentation-tonicity relation for ghosts because fragments of ghosts cannot be readily enumerated, but the form of the relation, as well as that of other similar relations which present maxima or minima at tonicities which are optimal for a phenomenon, can be easily determined in systems which contain intact red cell.

The two phenomena in which the dependence on tonicity shows the clearest maxima are the heat fragility of red cells and their resistance to mechanical haemolysis.

The washed red cells of fresh human blood are suspended in 1 % NaCl in a volume concentration of 0·4. One volume of this suspension is added to 4 vol. of 3, 2, 1·5, 1·0, 0·8 and 0·6 % NaCl; this gives systems whose tonicity T is approximately 2·75, 1·87, 1·43, 1·00, 0·83 and 0·65. After standing for 30 min., samples of each system are heated to a known temperature (between 52 and 56o C. in these experiments) for a known time (3 min. in these experiments). The number of cells No in the systems before heating is found by counting; after the heating, the number N of cells plus fragments is counted. The fractional amount of haemolysis, p, in each system is found photometrically ; the fragmentation is then measured by

haemolysis in these systems being substantially all-or-none (see Ponder, 1950a).

The results of an experiment of this type are shown in Table 2 and in Fig. 1.

The value for the fragmentation f is at a maximum in a tonicity of about 1·5, becoming small at high and low tonicities. The value of p increases with decreasing tonicity. The maximum at T= 1·5 is also observed in systems containing plasma (unwashed red cells being used in place of washed red cells). In a tonicity of 1·5, the volume of the human red cell is about 20 % smaller than it is in a tonicity of 1·0, and the Hb is accordingly about 1·25 times as concentrated as in the normal red cell. This appears to be the concentration (about 37 g. %) at which the viscous and plastic properties of the constituents of the red cell are optimal for the production of stable fragments. It is understandable that there should be an optimal value of T which provides a state intermediate between that in hypotonic media, with low Hb concentration and low cohesion, and that in hypertonic media, with Hb so concentrated as to approach paracrystalline or even crystalline arrangements.

The washed red cells of fresh human blood are suspended in 1 % NaCl in a volume concentration of 0·4, and are added to solutions of NaCl of various tonicity as described in the foregoing section but in the proportion of 1 ml. of suspension to 8 ml. of each NaCl solution. After standing for 30 min., the tubes containing the systems are centrifuged gently, and 8 g. of the supernatant fluid are removed into a weighed vessel. The 1 ml. (approximately) which remains contains the red cells in volume concentrations which are a little smaller, or a little greater, than 0·4, according to whether the red cell volume has decreased or increased as a result of the tonicity change.^* The red cells in each tube are gently resuspended in the medium surrounding them, ^† and 0·5 ml. of each suspension is transferred to the flasks of the mechanical fragility apparatus.^* Each flask contains three glass beads. The flasks are rotated for 45 min. at a rate of 30 rev./min., at the end of which time the suspension in each is transferred to small tubes and centrifuged. The Hb in each supernatant fluid is determined photometrically, and expressed as a fraction of the total Hb of the system.

The results of an experiment of this kind are shown in Fig. 2. The mechanical fragility of the cells, as measured by the fractional haemolysis p, is at a minimum in a tonicity of 1·0, increasing as the tonicity is made greater or smaller. This suggests that when the cell has the volume corresponding to T =1·0 (86 µ³), the cohesion of its constituents is such as to present a maximum resistance to the somewhat obscure forces to which it is exposed in the mechanical fragility apparatus. These forces must be of quite a different nature from those involved in heat fragmentation, a phenomenon which is at a maximum in about the same tonicity as that at which mechanical fragility is a minimum. It is known, in this connexion, that the lysis produced mechanically is all-or-none, and that it does not involve fragmentation. It has been suggested that a swollen red cell is less able to be distorted mechanically without its surface being stretched than a normally shaped red cell is, and that this may be the explanation for the increased mechanical fragility in hypotonic media; the observation that the fragility-tonicity relation has a minimum, however, suggests instead that the resistance to mechanical stresses depends on properties such as plasticity and the special cohesions which exist between the molecules of an ordered structure.

It is understandable that some of the properties inherent in a structure which is built in some special manner will be at a maximum or a minimum when the structure is neither compressed nor expanded, and it is interesting to examine a number of phenomena to see which of them are functions of tonicity, and which of them present maxima or minima. Phenomena which show maxima and minima may be suspected of being related to an optimal arrangement or separation of the molecules of the red cell ultrastructure, and if we distinguish somewhat arbitrarily between the surface ultrastructure, with dimensions of l², and the ultrastructure of the interior, with dimensions of l³, the phenomena in question will be more likely related to the latter than to the former. The phenomena of heat fragmentation and mechanical fragility, which may obviously involve properties of plasticity and cohesiveness (l³ properties) are good examples of this idea.

Looked at in this way, it is not surprising that disk-sphere transformations, whether produced by distearyl lecithin or sodium tetradecyl sulphate, are found to have no tonicity dependence when examined quantitatively by methods which have already been described (Ponder, 1947). The disk-sickle shape transformation, on the other hand, is greatly affected by tonicity. The typical sickle form occurs only at a tonicity in the neighbourhood of 1·0; in hypotonic media, it is replaced by a large atypical sickle with rounded outlines and little filament formation, while in hypertonic media (T=3) the typical sickles are replaced by ‘holly wreath’ forms with multiple foci of distortion. These results are understandable in terms of the sickling shape change having a dependence on l³ and accordingly showing a tonicity maximum, whereas the disk-sphere transformations have a dependence on l² and have no maximum or minimum in their tonicity dependence ; indeed, they do not seem to have any tonicity dependence at all.

Two other phenomena which have a well-marked maximum in their tonicity dependence have been found in the course of a search for phenomena of this type. The first occurs in systems containing human red cells and the lysins sodium dodecyl sulphate or sodium tetradecyl sulphate ; the velocity of haemolysis by these substances, in concentrations which approach their asymptotic concentrations, is at a maximum in tonicities between 1·0 and 1·5, and becomes much less in more hypertonic or hypotonic systems. There are many examples of lysins which have lytic effects with a tonicity dependence (e.g. the lytic effect of saponin decreases with decreasing tonicity and the lytic effect of lysolecithin increases with decreasing tonicity (Ponder, 1937; Wilbur & Collier, 1943); the maximum found in systems containing the detergents is very unusual. The most unexpected minimum in a tonicity dependence, however, is that in the K-Na exchange which occurs in human red cells kept for about 24 hr. at 40 C. This exchange is at a minimum at a tonicity of about 0·7; in a tonicity of 3, it is almost double its minimum value, and at a tonicity of 0·5 it is slightly greater than the minimum value.^* The presence of the minimum suggests that the exchange has a dependence of the l³ type rather than one of the l³ type, and could be used as an additional indirect argument against its being a simple diffusion phenomenon taking place along concentration gradients.

1. A large number of kinds of ghost, each with its own characteristic properties, can be prepared from human red cells. These ghosts, which differ from each other in their appearance as seen with phase optics, in their heat fragmentation, and probably in other respects also, are the result of the procedures used to prepare them (lysis by water as opposed to lysis in dilute saline, lysis followed by ‘reversal of haemolysis ‘by phosphate or veronal buffers at various pH or by saturating the haemolysate with CO₂, lysis by freezing and thawing, lysis by chemical lysins, etc.). In view of this, it is meaningless to speak of the ‘red cell ghost’ and of its structure without further specification.

2. No entirely satisfactory explanation for the differences in heat fragmentation of the various kinds of ghost has been found. The best one is that fragmentation into stable fragments occurs when the ghost is discoidal and when it has a certain plasticity, the plasticity depending on the concentration of residual Hb. It is shown that heat fragmentation of red cells occurs maximally in a tonicity of about 1·5, i.e. when the Hb molecules are a certain optimal distance apart.

3. A number of other phenomena which occur optimally at certain values of tonicity, i.e. when the Hb molecules are neither too closely packed nor too widely separated, are described. Among these phenomena are mechanical fragility, the formation of typical sickles, resistance to certain lysins, and K-Na exchange. The various forms of disk-sphere transformation seem to have no tonicity dependence.