cRY0B10L0GY
12, 293-308
Effect
( 1975)
of Cooling
G. RAPATZ,
Cryobiology
and Rewarming Rates Human Erythrocytes ‘r2 B. LUYET,”
AND
on Glycerolated
A. MACKENZIE
Research Imtitute, American Found&on for Biological Rt. 5, Box 137, Madison, Wisconsin 53704
Research,
more injurious than slow rewarming. Leibo It might be said with confidence that the greater number of freezing studies et al. (9) observed that the survival of mouse marrow stem cells cooIed slowly completed during the last 25 yr indicated the superior value of a rapid thawing (1.S deg C/min) to -196°C in the pres(1-3, 5, 9, 14-18, 20-22, 24, 25, 30-32, ence of 1.25 M glycerol was the same re34-37). This being the case, a further in- gardless of whether the suspensions were spection wouId no doubt reveal the parthawed at 2 deg C/ min or 900 deg C/min. ticular value of the rapid thawing where When the same cells were cooled rapidly the cells in question first experienced and (200 deg C/min), rapid rewarming gave survived a very rapid freezing. Much evi- significantIy better survival than slow redence has been accumulated to support warming. Studies by Whittingham et al. these rather general conclusions. Rela(40) on the survival of mouse embryos tiveIy simple theories relating survival to after freezing to -196 or to -269°C freezing and thawing veIocities were de- showed that sIow cooling (0.3-2 deg C veloped in a number of laboratories. min) and slow rewarming (4-25 deg C/ Only recently have several findings min) were required for the recovery of viable specimens, Rapid rewarming (215 offered seemingly contradictory evidence. Meryman (26), in a study on freezing and and 450 deg C/min) proved Iethal regardrewarming rates in the range -10 to less of the rate at which the embryos had -4O”C, found that human erythrocytes been frozen. It would appear, in essence, suspended in a sodium-free buffered salt that diverse evidence exists to support solution molar with respect to glycerol the contention that slower thawing is to were least damaged when cooled and re- be preferred in certain circumstances. We now report a repetition and extenwarmed at rates of approximately 0.3 deg studies (26) to lower C/min. He also observed that rapid re- sion of Meryman’s warming of rapidly frozen specimens was freezing bath temperatures and to higher and lower cooIing and rewarming rates. We will show that the greater value of Received April 17, 1975. the slower thawing is Iimited according to 1 Research Supported in part by NIH Grants the temperature to which the erythrocytes HE-06470 and HE-11380. ZThe results of this investigation were prewere cooled after freezing. It will be seen sented at the 14th Annual Meeting of the Biothat the findings cannot be explained enphysical Society, Baltimore, Maryland, February tirely on the basis of existing theories of 1970 . freezing injury. s Deceased hgarch 30, 1974. 293 Copyright 0 1975 1)~ Academic Press, Inc. All tights of reproduction in a,ny form reserved.
RAPATZ,
294
LUYET
AND MACKENZIE
1 mm above the surface of the cooling bath, and we touched the wall of the tube Preparation of Material at this point with a thin steel spatula preVenous blood was collected from cooled in liquid nitrogen. As soon as freezhealthy volunteer donors and immediately ing began we lowered the tube so that the mixed with ACD anticoagulant in a ratio upper limit of the column of blood was at of 100 parts of blood to 15 parts of ACD least 1 cm below the surface of the cooling solution. (The ACD solution contained bath, 2.2 g sodium citrate, 0.8 g of citric acid, Moderately rapid cooling. Moderately and 2.45 g of dextrose in 100 ml of solu- rapid cooling was accomplished by direct tion.) The erythrocytes were washed three immersion, for 5 min, of test tubes 10 mm times, by centrifugation, in a sodium-free in diameter and 75 mm in length, conbuffered salt solution containing glycerol taining 2 ml of the blood cell suspension, in 1 M concentration. (This buffered solu- in baths maintained at the desired temtion contained 0.02 moles of dibasic po- peratures. If spontaneous nucleation and tassium phosphate, 0.02 moles of mono- freezing did not occur within 1 min after basic potassium phosphate, and 0.06 moles the tube was immersed in the cooling bath of potassium citrate. ) One part of the we “seeded” the specimen as described in washed packed cells was then suspended the previous paragraph. in four parts of buffered salt solution with Vey rapid cooling. Specimens were glycerol; the celI suspension thus obtained cooled very rapidly by direct immersion of served as the experimental material. 50 mm3 of the cell suspension, contained in capillary tubes, into an isopentane coolMethod of Freezing ing bath, for 5 min, at the desired temSpecimens were frozen to various tem- perature. The same “seeding” procedure peratures at cooling rates we have des- was again employed to initiate freezing ignated as slow, moderately rapid, or in those samples in which spontaneous nucleation was not observed, This method, very rapid. Slow cooling. Two milliliters of the cell while the same as that used for moderately suspension were placed in glass test tubes, rapid cooling, gave considerably higher rates because of the smaller sample size. 10 mm in diameter and 75 mm in length; or 50 mm3 of the suspension were transferred from a calibrated micropipette to a Method of Thawing glass capillary tube with an outside diamThe frozen specimens were thawed eter of 1.5 f 0.05 mm and a wall thickness slowly, moderately rapidly, or very rapidly. of 0.25 f 0.05 mm. The specimens were To thaw the material slowly, the cooling immersed in an isopentane cooling bath, bath containing the frozen specimen was after the ends of the capillary tubes had rewarmed at a rate of 0.3 deg C/min. The been sealed with Vaseline and the open rate of rewarming of the bath was reguends of the test tubes closed with tightlated by the cam-actuated pneumatic temfitting corks. The bath was cooled at a perature controller, Moderately rapid rate of 0.3 deg C/min to the desired temthawing was accomplished by direct imperature; the cooling rate was regulated mersion of the test tubes, each containing by a cam-actuated pneumatic temperature controller. To prevent unpredictable super- 2 ml of frozen material, in a well-stirred cooling, we “seeded” the specimen at water bath at 37°C. Very rapid thawing -3°C. For that purpose we raised the was achieved by abrupt immersion of the capillary or the tube so that the upper capillary tubes containing the frozen speciend af the column of blood was about mens in the water bath at 37°C. In these MATERIALS
AND
METHODS
COOLING
0 A. SLOW
AND
REWARMING
IO
20
OF
HUMAN
30 40 TIME IN MINUTES
ERYTHROCYTES
SO
295
60
COOLING
TIME IN MINUTES
B. MODERATELY
RAPID
COOLING
TIME IN SECONDS
C. VERY RAPID COOLING
FIG. 1. Copies of cooling curves recorded when blood cell suspensions were frozen at various rates. A. Simultaneous recordings of the cooling of a blood cell suspension contained in a capillary tube (Curve C) and a test tube (Curve T), and of the cooling of the bath (Curve B) in which the specimens were frozen at an average cooling rate of 0.3 degree C per minute, B. Cooling curve obtained when a blood cell suspension contained in a test tube was frozen by abrupt immersion into a cooling bath at -30°C. C. Cooling curve recorded when a blood cell suspension containedin a capillary tube was frozen by very rapid immersion into a cooling bath at -30°C. See text for practical details.
latter two cases the size of the sample determined the warming rate. Slow cooling, slow warming, rapid thawing experiments. Two milliliter volumes of the cell suspension were placed in test tubes (10 mm diam, 75 mm length) and cooled at 0.3 deg C/min to -35°C. (The rate of cooling was controlled as described above in the method of slow cooling.) When a temperature of -35°C was obtained the frozen specimens were rewarmed at a rate of 0.3 deg C/min to various given temperatures and then thawed by abrupt immersion of the test tube into a well-stirred water bath at 37°C. Cooling and Rewarming Curves Cooling and rewarming curves were obtained during the freezing and thawing of some of the specimens to give a com-
parison of the rates obtained with the various procedures employed. In the case of slow cooling and rewarming (0.3 deg C/mm) the temperature of the bath was recorded simuhaneously with that of the specimens in the capillary and the test tube. Three-mil copper-constantan thermal junctions, placed as nearly as possible at the geometric centers of the specimens, served to sense the temperature in the latter during freezing and rewarming. Measurement of Hemolysis To measure the extent of hemolysis, we suspended a 50 mm3 aliquot of the thawed blood in 10 ml of the sodium-free buffered salt soIution containing glycerol in 1 M concentration, and removed the intact cells by centrifugation. Then a measurement was made, with a Klett-Summerson photoelectric calorimeter, of the color density
RAPATZ,
296
-30
1 0
A. SLOW
LWYET
I 10
1 20 TIME
AND MACJSENZIE
I 30 IN MINUTES
t 40
WARMING
-30 0
I 0.5
I I .o TIME
B. MODERATELY
RAPID
I I.5
IN MVWTES
OR VERY
RAPID
WARMING
FIG. 2. Copies of warming curves obtained when frozen blood cell suspensions were rewarmed at various rates. A. Simultaneous recordings of the warming of the frozen suspensions contained in a test tube (Curve T) and a capillary tube (Curve C), and of the rewarming bath (Curve B) in which the specimens were warmed at a rate of 0.3 degree C per minute, B. Simultaneous recordings of the warming of frozen suspensions contained in a capillary tube (Curve C) and a test tube (curve T) in which the specimens were thawed in each case by abrupt immersion into a water bath at 37°C.
resulting from the presence of the hemoglobin in the supernatant fluid. The percentage of hemolysis corresponding to a given calorimeter value was read directly from a calibration curve constructed from measurements obtained with 50 mm3 aliquots of mixed suspensions containing known quantities of intact and completely hemolyzed erythrocytes taken from the stocks used in the respective experiments. RESULTS
I. Cooling and reumrming cumes. Representative freezing curves obtained with the three different means of cooling are presented in Fig. 1. Figure 1A gives the temperatures recorded in the freezing bath (curve B), in the 50 mm3 of cell suspension contained in the capillary (curve C ), and in the 2 ml of suspension in the test tube (curve T) at cooling rates of 0.3 deg
C/min. The three curves merge into one throughout their length, except in the freezing range. Freezing of the suspension contained in the test tube continues for a considerably longer time (approx 8 min, compared with less than 1 min) and over a greater temperature range (2.5 deg compared with less than 0.3 deg) than in the capillary. The curves recorded when 2 ml of the cell suspension in test tubes or 50 mm3 of suspension in the capillary were frozen by abrupt immersion in a freezing bath at -30°C are shown, respectively, in Fig. 1B and 1C. The rate of cooling in the capillary tube is more than 15 times that in the test tube (less than 0.1 of a minute to reach -30” in the capillary to more than 1.5 min in the test tube). Likewise, the length of time at the freezing point of the solution is considerably reduced in the
COOLING
AND
REWARMING
OF
capiIlary tube (compare the freezing pIateaus of the curves in Fig. 1B and 1C). Representative curves obtained with the three different methods of rewarming are given in Fig. 2. As in the case of slow cooling we recorded simultaneously, during slow rewarming, the temperature of the warming bath (curve B ), the specimen in the capillary tube (curve C ), and in the test tube (curve T), the results of which are given in Fig. 2A. Like the cooling curves, these curves merge throughout their length except, in this case, in the thawing range, at which time a definite endotherm is recorded in the specimen contained in the test tube (curve T). We did not record an endotherm during melting of the specimen in the capillary tube, probably because of the small size of the sample and the limitations of our temperature sensing system. A comparison of the rewarming rates obtained when frozen specimens contained in capillary tubes and test tubes were thawed by abrupt immersion in a water bath at 37°C is given in Fig. 2B. The rate
HUMAN
ERYTHROCYTES
of rewarming is approximately 15 times higher in the capillary than in the test tube (1 set to warm from -30 to 0°C in the capillary and 15 set for the test tube). II. Slow cooling and warming in combination with moderately rapid cool@ and warming. The effects of the four com-
binations, namely, slow cooling and slow warming, slow cooling and moderately rapid warming, moderately rapid cooling and slow warming, and moderately rapid cooling and moderately rapid warming are summarized in Fig. 3. In the temperature range from -15 to -35”, SIOW cooling combined with either slow warming or moderateIy rapid warming was less damaging than either treatment invoIving rapid cooling. Further, in the same temperature range, the combination: slow cooling/ slow warming was significantly less injurious than the sequence: slow cooling/moderately rapid warming. In the temperature range from -15 to -25”C, moderately rapid cooling/moderately rapid warming was the most damaging of the four combinations.
80
DEGREES
FIG. 3. Percentage of hemolysis in blood rates. The Z-ml vol were frozen and thawed rapid cooling, moderately rapid warming; triangles: slow cooling, moderately rapid warming. See text for meaning of terms Each plotted point represents an average of ten determinations; the vertical lines illustrate
297
CENTIGRADE
cell suspensions frozen and thawed at various in 10 X 75 mm test tubes. Squares: moderately moderately rapid cooling, slow warming; circles: warming; diagonal crosses: slow cooling, slow used to designate cooling and rewarming rates. the percentage hemolyses obtained in not less than the experimental variations.
RAPATZ,
298
LUYET
Each point plotted in Fig. 3 represents an average of at least ten individual experiments with at least three different lines repblood samples. The vertical resent the extent of the experimental variation, In the temperature intervals just considered, there was no ovedapping of the data obtained with the various freezing treatments. While a statistical analysis was not conducted, it would appear that the differences are nonetheless highly significant, When the specimens were frozen at lower temperatures, the combination of moderately rapid cooling and warming became gradually less injurious at gradually 100
r
80
20 IO 0
0
-10
-20
-30
DEGREES
-40
-50
-50
-70
CENTIGRADE
FIG. 4. Percentage of hemolysis in blood cell suspensions frozen and thawed at various rates; 0.05-ml vol were frozen and thawed in glass capillary tubes f 1.5 * 0.05 mm o.d,, wall thickness: 0.25 1 .05). Squares: very rapid cooling, very rapid warming; circles: very rapid cooling, slow warming; triangIes: slow cooling, very rapid warming; diagonal crosses: slow cooling, slow warming. See text for meaning of terms used to designate cooling and rewarming rates. Each plotted point represents an average of the percentage hemolyses obtained in not less than ten determinations; the vertical lines illustrate the experimental variations.
AND
MACKENZIE
lower temperatures. This same cooling and warming combination, which was the most damaging in the range from -10 to -25”, became the least injurious in the range from -50 to -140°C. At temperatures below -3O”, moderately rapid cooIing/slow warming was the most injurious sequence. The beneficial effect of slow warming after slow coohng decreased below -5O”, but the average hemolysis was still below that obtained by slow freezing and moderately rapid warming. There was, however, considerable variation and an overlapping of individual experiments so that the latter differences are probabIy not as significant as those at higher freezing temperatures. III. Slow cooling and wu~ming in combination with very rapid cooling and warming. The results of the four combinations, slow cooling and slow warming, slow cooling and very rapid wamling, very rapid cooling and sIow warming, and very rapid cooling and very rapid warming are given in Fig. 4. An examination of these data shows four main differences from those presented in Fig. 3, in which slow and moderately rapid cooling were used in combination with slow and moderately rapid rewarming: ( 1) Very rapid cooling and rewarming minimizes hemolysis associated with exposure to higher freezing temperatures. For example, while moderately rapid freezing to (and thawing from) -25°C resulted in 85% hemolysis, very rapid freezing to (and thawing from f the same temperature resulted in the hemoIysis of only 37% of the cells. (Compare Figs. 3 and 4). Likewise the maximum hemolysis was reduced in the very rapidly cooIed specimens; 44% compared with 87% at temperatures of -20 and -3O”C, respectively. (2) In the temperature range -10 to -30°C very rapid thawing was only slightly less effective than slow thawing in promoting recovery of slowly frozen specimens. Moderately rapid rewarming was, in contrast, much less effective than
COOLING
AND
REWARMING
slow rewarming where similar slow freezing was employed. (3) When specimens were cooled slowly to temperatures below -40°C very rapid warming was significantly less injurious than slow rewarming. These results are in direct contrast to those obtained when slowly cooled specimens were warmed at moderately rapid rates in this temperature range. (4) Very rapid cooling/slow rewarming was the most damaging of all combinations at all temperatures except where cooling was limited to -IO”C; the highest freezing temperature employed. IV. Slow ux&ng, sloes warming, moderately rapid thawing expe&ments. The temperature range over which sIow rewarming exerted a protective effect on specimens slowly cooled to a final temperature of -35°C is shown in Fig. 5. One notes a decrease in hemolysis (from 72 to 39 s ) as the frozen specimens are rewarmed to gradually higher temperatures in the range -30 to -lO”C, prior to abrupt thawing. Slow rewarming to temperatures above -10°C did not improve recovery. Evidently, the protection afforded by slow warming is exerted over nearly the entire rewarming range. Effects specific to particular temperatures are not indicated.
OF
HUMAN
ERYTHROCYTES
‘4 7 301
I
I
-30 -25 -35 TEMP. T O WHICH
I
-20 SPECIMEN
I
I
I
I
-15 -10 -5 0 WAS SLOWLY REWARMED
FIG. 5. Hemolysis as a function of the temperature to which blood cell suspensions slowly frozen to -35°C were slowIy warmed prior to thawing by abrupt immersion into a warm water bath at 37°C. Slow cooling and slow warming were effected at 0.3 degree C per minute. The 2-4 vol were frozen, rewarmed, and thawed in 10 X 75 mm test tubes.
sent a scheme designed to supplement, and to incorporate the so-called two-factor hypothesis.
DISCUSSION
It would be tempting to consider the various curves in Figs. 3 and 4 one at a time and to conclude from such an analysis that the freezing/thawing hemolysis is explained in each case according to a “classical” two-factor argument (23). That the basis for the observed behavior requires additional explanation will become apparent. It will be seen that, taken two at a time, the plots in Figs. 3 and 4, respectively, argue the importance of the state in which the cell is exposed to Iowered temperatures and concentrated electrolyte and, afterwards, to higher temperature and isotonic medium. The same notions wilI be seen to account for the findings expressed in Fig. 5. We will pre-
Slow and Moderately Rapid Cooling in Combination with SEow and Moderately Rapid Warming of 2 ml Speciwns Figure 3 shows, as we said, that the hemoIysis resulting from each of the four specified treatments increases rapidly with exposure to necessarily more concentrated salts. The steeply rising part of any one of the four plots in Fig. 3 could, therefore, be attributed to so-called salt damage, this despite the presence of the gIycero1. Three of the four plots involve prolonged exposure at intermediate subzero temperatures; their trends to higher hemolysis with Iower freezing temperature are explained in qualitative terms. The very distinct maximum in the fourth curve sug-
RAPATZ,
300
LVYET
b-
., -*kc..
--6o COOLING
BATH TEMPERATURE
FIG. 6. Ratios of hemolyses obtained cells suspended in 24111 vol were thawed at indicated rates to and from peratures. Data from Fig. 3. MRC: rapid cooling; SC: slow cooling; MRW: rapid warming; SW: slow wamling. sion for further information.
..^..I--1 -I20
-I40
(~1
after blood frozen and various temmoderately moderately See Discus-
gests, in line with the foregoing, that it is in the exposure to intermediate freezing temperatures that the cells are effectively destroyed. Rapid passage to (and from) the lowest temperatures employed s&ices to reduce hemolysis to near-minimum values. The results of the moderately rapid freezing/moderately rapid thawing experiments are in complete accord with those of Lovelock (11, E), Rapatz and Luyet (32, 34), Gehenio et al, (B), and Rapatz et al. (35), The cooling velocities attained in the freezing of the 2 ml specimens being no more than several hundred degrees C per minute, the possibility of an intracellular freezing is excluded ( 19, 39). It wil1, in consequence, be seen that the question of the contribution of the second of the “two factors” need not be raised in the discussion of any of the four curves in Fig. 3.
AND
MACKENZIE
A further inspection of Fig. 3 reveals a series of curious relationships. The fmdings confirm and extend Meryman’s results in (the range the range -10 to -25°C within which he completed the fourfold comparison). They demonstrate a totally reverse order best seen at -20 to -25°C. Broken only by the aforementioned descent of the rapid/rapid curve, the reverse order is maintained, albeit with increasing digerences, to - 120°C. The curves illustrate the failure of the exposure per se to determine hemolysis at the higher freezing temperatures. Slow cooling/slow warming hemolysis is insignificant until freezing is extended to -30”; slow cooling/moderately rapid warming, moderately rapid cooling/slow warming, and moderateIy rapid cooling/moderately rapid warming hemoIyses have each become significant at -2O”C, being the more remarkably distinguished at -25°C. Inasmuch as “sah injury” would seem to leave hemolysis above -30°C unexplained, we must search for one or more other contributing factors. To undertake a quantitative comparison of individual curves in Fig. 3, it will be convenient to refer to Fig. 6 in which all but two possible comparisons express the ratios of the hemolysis obtained by one freeze/thaw routine to that resulting from another. The comparisons are limited to cases where one of the two sequential procedures hoIds in common (it would be difficult to compare the results of a moderately rapid cooling and a slow warming with those of a slow cooling and a moderately rapid warming, and just about ashard to compare slow coohnglslow warming with moderately rapid cooling/moderately rapid warming, save to remark again that the last of these cycles is in fact the most damaging from -15 to -3O”C, the slow/ slow combination allowing the highest recovery). Figure f3a compares hemolyses at common cooling rates; Fig. 6b alIows comparisons at common warming rates. We see from Fig. 6a that greater dam-
COOLING
AND
REWARMING
age resulted from moderately rapid warming where (a) the cells were subjected to moderately rapid cooling to temperatures not lower than -30°C (more time in the supposedly damaging zone during warming leading to less damage), (b) the ceIIs were cooled slowly to temperatures down to -50°C. Our moderately rapid warming procedure proved, in contrast, less destructive than our sIow warming where cells were cooled moderately rapidly to -30°C and to lower temperatures (less time in the presumably damaging zone resulting in these latter cases in less damage). We may note that the twofactor hypothesis would place both the plots entirely below the unit center line in Fig. 8a. Comparing the one plot with the other, we note essentially the same behavior at the higher temperatures ( above -30°C). The marked difference in the plots below -40°C is explained in Iarge part by the much higher survival obtained after moderately rapid cooling and moderately rapid warming, the three other treatments securing a near-complete hemoIysis wi.th decreasing temperature. It is the similarity of the plots (of the ratios describing the differences between the respective treatments) where the celIs are frozen to the higher temperatures that is the more remarkable. Slower warming is, in these instances, shown to be less damaging. We see from Fig. 6b that greater damage also resulted from moderately rapid cooling where (a) the cells were subjected to moderately .rapid rewarming from temperatures not lower than -40°C (less time in the supposedly damaging zone during cooling leading to higher hemolysis), (b) the cells were warmed sIowly from temperatures as low as -40°C (the shorter exposure during cooling resulting in a much increased hemolysis), ( c) the cells were warmed slowly from temperatures in the range -40 to -120°C (much reduced differences being observed). Comparing the one plot with the other, we find
OF
HUMAN
ERYTHROCYTES
the same behavior down to -40”~. The greater damage is associated in each case with a faster coohng. The slow warming peak is displaced to a temperature a little lower than that associated with moderately rapid warming. Apparently the the slower warming alIows the damaging reaction a greater opportunity to proceed to completion at lower temperatures (where the measured differences begin to disappear); the moderately rapid warming has already contributed to a higher hemolysis during freezing to, and thawing from the same temperature (regardless of the cooling rate). Slower cooling to the higher and the intermediate freezing temperatures is, therefore, less damaging, this despite the certainty that the faster cooling was much too slow to permit an intracellular freezing. We will attempt to develop these findings very shortly. Slow and Very Rapid Cooling in Combinution with Slow and Very Rapid Warming of 0.05 ml Specimens A glance at Fig. 4 reveals the rising hemolysis resulting from increasing exposure to necessariIy more concentrated salts. Much as in Fig. 3, the steeply rising portion of any one of the four plots couId be attributed to salt damage. The slow cooling/slow warming and the slow cooling/very rapid warming plots could be explained in their entirety, one at a time, without recourse to other possible factors. Intracellular freezing intervenes to complicate the very rapid cooling/very rapid warming and the very rapid cooling/sIow warming curves. The drop in hemolysis by very rapid cooling/very rapid warming to and from temperatures below -20°C corresponds to the rapid passage of the specimen to temperatures at which the ceIIs, dehydrated by extracelIular freezing, are less and less rapidly damaged during the 5-min holding period. The upswing in -50°C correIates with hemolysis below the onset of intracelluIar freezing. The results are in very good agreement with
302
RAPATZ,
LUYET
2 . .I 1
I -10
r -20 COOLING
r -30 BATH
I -50 TEMPERATURE
I -60
-40
I -70
(“Cl
FIG. 7. Ratios of hemolyses obtained after blood cells suspended in 0.05-ml vol were frozen and thawed at indicated rates to and from various temperatures. Data from Fig. 4. SC: slow cooling; VIE: very rapid cooling; SW: slow warming; VRW: very rapid warming. See Discussion for further information.
those of Gehenio et al. ( S), Luyet et al. ( 17)) and Rapatz and Luyet (34) on erythrocytes and those of Mazur and his colleagues (22, 24, 25) on other cells, Very rapid cooling/slow warming becomes, and remains very destructive during freezing to lower and Iower temperatures, slow warming from the lowest freezing temperatures allowing the recrystallization of intracellular ice. The second of the “two factors” is, by general agreement, essential to the discussion of situations typified by the very rapid/very rapid plot and, by further inference, the very rapid cooling/slow warming plot in Fig. 4. Comparing one curve in Fig. 4 with another, we encounter some of the same curious interrelationships we found in Fig. 3. Slow cooling/slow warming hemolysis rises the least during freezing to, and thawing from temperatures as low as
AND
MACKENZIE
-30°C. Very rapid cooling/slow warming causes by far the greatest damage in the same interval, Neither finding is explained on the basis of a supposed salt damage. A corresponding quantitative comparison of individual curves in Fig. 4 is reproduced in Fig, 7. The form follows that adopted in Fig. 6. Figure 7a reveals relatively little that was not obtained by a direct inspection of Fig. 4. Higher recoveries by very rapid cooling/very rapid warming are, nonetheless, clearly presented. Slow warming apparently offered no advantage to the 0.05ml specimens at the higher freezing temperatures. Greater significance is more readily attached to Fig. To. We note first that, where cells are slowly warmed, very rapid freezing to temperatures in the range - 10 to -35°C is far more damaging than slow freezing. We see in the lower curve that, where the 0.05ml specimens are subjected to very rapid warming, much higher hemolysis results from very rapid cooling to temperatures between -15 and -25°C. It wouId appear that here, as in the 2.0-ml specimens, slower cooling to the higher freezing temperatures is very much less damaging. We will examine one further series of determinations before attempting an inviting correlation. Slow Freezing, Slow Warming, and Rapid Thawing of 2-ml Specimens The experiments in which erythrocytes were subjected to slow cooling to -35”C, slow warming to various intermediate temperatures, and rapid subsequent thawing showed that slow warming to gradually higher temperatures in the range -30 to -10°C permitted a gradually higher recovery. Slow warming to temperatures higher than -10°C did not result in further benefits. The findings support the results reproduced in Fig. 3 in which an uninterrupted slow warming was shown to be less damaging than a moderately rapid warming. Much like Fig. 3, Fig. 5 contradicts the notion that the damage
COOLING
AND
REWARMING
resulting from freezing and thawing derives from the freezing where thawing was rapid. Noting the plateau in the range -10 to 0°C in Fig. 5, we distinguish two modes of damage. It would appear that slow freezing to -35°C causes the destruction of 39% of the cells present regardless of the warming rate. Raising the total hemolysis to 7070, moderately rapid warming from -35°C (or, for that matter, from -30°C) would seem to have caused a 31% hemolysis avoided by slow warming. Given the circumstances, the presence of the glycerol, and the rather high subzero temperatures, it is hard to see how hemolysis cannot have resuhed, in part at least, from too rapid a change in some critical variable. These, and the conclusions drawn from Figs. 6 and 7 prompt the presentation of the following expIanation.
OF
HUMAN
ERYTHROCYTES
303
red cell membrane (and, very possibly, other cell membranes) exists in different states at different temperatures, and in the presence of salts and other sohtes at different tonicities. This being the case a change in the milieu will cause the membrane to seek a new most stabIe state. It is assumed that the times required fall within the time scaIe of the freezing/ thawing experiments. Too fast a cooling wilI cause the cell membrane to reach the final freezing temperature in the state in which it was most stable prior to freezing. Too fast a warming will, where a slow cooling permitted the conversion to the most stable state at a Iower temperature, render the membrane metastable after thawing. The possibilities are illustrated in Table 1. Slow cooling/freezing allows State A to convert to State B, the most stable state at the lowest temperature, Slow warming/ thawing permits a reconversion to State A A Tentative Hypothesis with a minimum hemolysis. Rapid warmWe wish to propose a description of ing/thawing causes the membrane to freezing injury based on the probability persist in State B to higher temperatures that the cell membrane tends to assume at which it may tend (i) to decompose, (ii) to return to State A, Rapid freezing different most stabIe states according to retains State A to Iower temperatures at the circumstances. Noting the frequent inseparability of the thermal and osmotic which it may (i) decompose, (ii) convert components in freezing/thawing experito State B. It is assumed that the rates at ments, we wilI base the analysis on the which States A and B decompose and interconvert are strongly dependent on the notion of ‘%hermal/osmotic” shock. The composition and the concentration of the reasoning derives in some respects from the work of Lovelock (14) and Leibo and suspending medium and the temperature. Mazur (10). Meryman raised the question We do not propose relationships between of thermal shock, discarding it, apparently, the rates in question. Nor do we seek to to propose a theory relating freezing in- identify series of intermediate states which jury to dehydration (26). Farrant and co- we would have to consider in a more exworkers reported extensive evidence sug- haustive analysis. gesting a rather specific role for thermal Major features characterizing Figs. 3-7 shock (4, 7, 28). The following description are interpreted according to the following is deliberately general in nature. schemes, Higher high subzero hemolysis Strong evidence was recently presented caused by slow cooling/rapid warming (Figs, 3, 4, 6, and 7) resuIts from the to support the notion that the structure of the red cell ghost depends on the temcomparative instability of the state assumed at the lower temperture, State B, perature (6), Steck (38) reviewed the current knowledge of the human red cell at the higher temperature, after thawing. membrane in very considerable detail. We Still higher hemolysis resulting from rapid assume for our purposes that the human cooling/ slow warming reflects the greater
RAPATZ,
304
LUYET
AND MACKENZIE
TABLE SUQQESTED Initial
state
“z2”
BASIS
Intermediate statew
FOR THERMAL/OSMOTIC WrtIming rate
A
Slow
B (+B’)
Slow
A
Slow
B (+B’)
Rapid
A
Rapid, to higher freezing temperatures
A (-+A’), B (-+B’)
Siowc
A (-+A’), B (+B’)
Rapid
A
A
A
Rapid, to higher freezing temperatures Rapid, to lower freezing temperatures Rapid, to lower freezing temperatures
1 SHOCK Final state (9)’
A, B”b 13 (+B”‘),
B’ (+B”“)
A”, 3”
A’ (+A”‘),
B’ (+B”“)
Ultimate ~O~~~“~ll~~8
Comparatively littIe hemolysis More hemolysis Still more hemolysis
Comparatively exten&e hemolysis
A
Slowc
A” B” ,
Comparatively extensive hemolysis
A
Rapid
A
Complete recovery
LLThe contribution made by the transitions A + B --f A has been omitted from all but the first row, though a percentage of the cells may survive as a result of it, depending on the cooling and warming rates. b States denoted by one or more primes deteriorate irreversibly. e Destructive changes proceeding during slow warming (i) augment changes at higher minimum temperatures, (ii) decrease recoveries maintained during rapid cooling to lower minimum temperatures.
instability of the membrane in its original state. State A must, therefore, have been less stable in the frozen state than State B during and after thawing. The highest hemolysis in the range -10 to -30°C (Figs. 3 and 6) combines the damage associated with the passage of the cell in State A to lower temperatures and the further damage resulting from the passage of cells reaching the lower temperature in the more stable state (State B) to the higher temperature prior to reconversion. The decreased hemolysis by rapid cooling/ rapid warming to, and from lower and intermediate temperatures (Figs. 3 and 4, respectively) is explained by the temperature dependence of the reaction: A + A’, sticiently low temperatures preventing State A from undergoing irreversible decomposition. We hasten to note that Figs. 3 and 4 are not explained in their entirety by thermal/ osmotic shock. The slow cooling/slow
warming curves indicate an unavoidable hemolysis (still slower cooling, etc., does not reduce these values) demanding an explanation best described, perhaps, as salt/dehydration damage for the present. It will be clear from Fig. 5 that certain circumstances allow contributions to total hemolysis to be traced to salt/dehydration injury, and, after that, to thermal/osmotic shock (on warming, in Fig. 5). Protection by rapid cooling to lower temperatures can, likewise, be traced to conditions in which damage resulting from exposure to concentrated salts and to dehydration is arrested. We doubt that the further notion that hemolysis results, in certain circumstances, from intracellular freezing, will be challenged by the introduction of the concept of thermal/osmotic shock. It is our hope that the several factors can be combined to yield, in due course, a more nearly comprehensive theory of freezing injury.
COOLING
AND
REWARMING
OF
HUMAN
305
ERYTHROCYTES
Other Factms It will be seen that the proposed thermal/osmotic shock leaves certain obvious differences between the behavior of the 2- and the 0.05”mI specimens unexplained. The Iower peak in the very rapid cooling/ very rapid warming plot (compare Fig. 4 with Fig. 3) is especiaIIy striking. The better to permit a quantitative comparison of Figs. 3 and 4 we have plotted the ratios of the hemolyses obtained when the 2and the 0.05mI specimens were subjected to corresponding freezing to, and thawing from the sarne temperature. The results are reproduced in Fig. 8. Given no difference in behavior related to the size of the the plots should each have specimen, taken the form of a straight line parallel, at the unit level, to the abscissa. Where a very rapid cooling and/or a very rapid warming resulted in greater damage from thermal/osmatic shock than a moderateIy rapid cooIing and/or moderately rapid warming, a ratio less than one would have been obtained. The moderately rapid cooling/moderately rapid warming/very rapid cooling/ very rapid warming pIot (Fig. 8a) reveals the extent of the departure from the behavior we might have expected, the deviation being the greatest, it appears, We suggest, between -30 and -50°C. for the present, that the very rapid cooling, despite the likelihood of a finer-grained freezing, fails to dehydrate the red ceils as completely as the moderately rapid cooling, The further water transfer from the cells during 5 min at the freezing temperatures in question proceeding, most probably, at a much reduced rate, the cells will have been subjected to less concentrated salts. We would, in this connection, question Meryman’s assertions concerning the absence of unfrozen freezable water (27). The less completely dehydrated cells would, of course, be subjected to a lesser thermal/osmotic shock. The superiority of the moderately rapid cool-
COOLING
BATH
TEMPERATURE
(“Cl
FIG. 8. Ratios of hemolyses obtained after blood celIs suspended in 2- and O-05-ml vol were frozen and thawed according to corresponding procedures to and from various temperatures. Data from Figs. 3 and 4. Squares, circles, triangles, and diagonal crosses retain their former respective significance.
ing/slow warming over the very rapid cooling/ slow warming procedure between -15 and -3O”C, and the absence of a difference below -30°C (Fig. 8a, again) support the concept of a thermal/osmotic shock in the temperature interval in which it was suggested by the body of the evidence presented. The better sIow cooling/slow warming recoveries in the larger specimens at the higher freezing temperature (Fig. 8b) might be explained on the basis of a longer freezing pIateau in the larger specimens (Fig. la) and a corresponding reduction in thermal/osmotic shock during the growth of the larger fraction of the ice formed. Such an argument requires, however, that we consider thermal/osmotic rather shock in the range -3 to -10°C than from-10 to -30°C where it appears more likely. Nor does the slow cooling/
306
FiAPATZ,
LUYET
moderately rapid warming/sIow cooling/ very rapid warming curve (Fig. 8b, again) support the further contention. It is, perhaps, more likely that the differences seen in Fig. 8b derive from geometric factors, ice crystal size, red cell distribution, and physical stress each playing a part. Nei (29) raised useful questions concerning the location of the red cells with reference to the crystalline phase. Rapatz and Luyet (33) described the recrystallization of ice in aqueous glycerol at different temperatures. According to their findings, slow coohng and/or slow warming supported a recrystallization of ice the nature and extent of which must have been determined in the present study by the sizes and shapes of the specimens. Convective mixing of the cell-bearing medium and water resulting from the melting of the larger ice crystaIs could, perhaps, become a factor in certain circumstances, The subject would seem to deserve a careful microscopic analysis. SWMMARY
Human erythrocytes suspended in a so* dium-free buffered saIt solution containing glycerol in 1 M concentration (1 part of packed cells to 4 parts buffered salt solution) were frozen by slow, moderately rapid, or very rapid cooling to various subzero C temperatures. The frozen specimens, after a 5-min storage period at a given temperature, were thawed at low, moderately high, or very high rates. The hemolysis in the frozen and thawed samples was measured by a calorimetric determination of the hemoglobin released from the damaged cells. At -lO”C, the highest freezing temperature empIoyed, nearly IOOY, recovery of intact erythrocytes was obtained irrespective of the cooling and rewarming conditions. The extent of the hemolysis after exposure to lower freezing temperatures depended upon the cooling and rewarming conditions. Moderately rapid and very rapid freezing to, and thawing from temperatures below -40°C
AND
MACKENZIE
permitted significantly higher recoveries of intact cells than the other freezing/ thawing combinations. In the temperature range -15 to -30°C the combination slow cooling and SIOW rewarming afforded maximum protection. Very rapid freezing/ SIOW thawing was the most damaging combination throughout the entire freezing range. The results were interpreted in part by a conventional two-factor analysis, Iower cooIing rates allowing concentrated salts to determine hemolysis, higher cooling rates destroying the cells by intraceIIular freezing. Apparent anomalies were expIained in terms of a generalized “thermal/osmotic” shock according to which the erythrocytes were subject to greater ,hemoIysis the higher the rates of cooling and/or warming. ACKNOWLEDGMENTS
The authors gratefully acknowledge the competent technical assistance of Peter Amundson, Raymond Keener, and Linda Rapatz. REFERENCES 1. Albrecht, R. M., OmdorfF, G. R., and MacKenzie, A. P. Survival of certain microorganismssubjected to rapid and very rapid freezing on membranefilters. C7yobiology
lo, 2xsz39
( 1973).
2. Ashwood-Smith,M. J., Warby, D., Connor, K. W., and Becker, G. Low-temperature preservationof mammaliancells in tissue culture with polyvinylpyrrolidone (PVP) , dextrans,
cryobiology
and
hydroxyethyl
9, 441-449
starch
(PIES
).
( 1972).
3. Barnhart, E. R., and Terry, C. E. Cryobiology of Neurosporu c4a.w~. I. Freeze responseof
NeurospoTacrassaConidia. Cyobiology 4.
5.
6.
7.
8,
323-327 ( 1971). Daw, A., Farrant, J., and Morris, G. J. Membrane leakage of solutes after thermal shock or freezing. Cryobiology 10, 126-133 (1973). Deal, Paul H. Freeze-thaw behavior of a moderately Halophilic Bacterium as a function of salt concentration. Cryobiology 7, 107-112 ( 1970). Elgsaeter, A., and Branton, D. Intramembrane particle aggregation in erythrocyte ghosts. 1. The effects of protein removal. J. Cell Biol. 63, 1018-1030 (1974). Farrant, J., and Morris, G. J. Thermal shock
COOLING
8.
9.
10.
11.
AND
REWARMING
and dilution shock as the causes of freezing Injury. CsyobioIogy 10, 134-140 ( 1973). Gehenio, P. M., Rapatz, G. L., and Luyet, B. J. Effects of freezing velocities in causing or preventing hemolysis. Biodylaumica 9, 77-a ( 1963). Liebo, S. P., Farrant, John, Mazur, Peter, Hamra, Jr., M. G., and Smith, L. H. Effects of freezing on marrow stem cell suspensions: Interactions of cooling and warming rates in the presence of PVP, sucrose, or glycerol, Cryobiology 6, 315-332 ( 1970). Leibo, S. I’., and Mazur, P. Freezing of bacteriophage T4: Temperature and rate effects as a function of salt concentration. Biophys. J. 9, A-258 ( 1969). Lovelock, J. E. The haemolysis of human red blood-cells by freezing and thawing.
Biodaim. (1953).
Biophys.
12. Lovelock, J. E. The tective action of lysis by freezing
Actu
10,
273-280
( 1955).
17. Luyet, B. J., Rapatz, G. L., and Gehenio, P. M. On the mode of action of rapid cooling in the preservation of erythrocytes in frozen blood. Bbdynamica 9, 95-124 ( 1963) 18. MacKenzie, A. P. Death of frozen yeast in the course of slow warming. In “The Frozen Cell” (G. E. W. Wolstenholme and Maeve O’Connor, Eds.), pp. 89-98. J. & A. Churchill, London, 1970. 19. Mandell, M. S., and Diller, K. R. Intracellular freezing in human erythrocytes. Cryobiology 11, 540 (Abstract) (1974). 20.
Mazur, Peter. Physical and chemical basis of injury in single-cehed micro-organisms subjected to freezing and thawing. In “Cryobiology” ( H. T. Meryman, Ed. ),
307
ERYTHROCYTES
Exp. Cell Res. 71, 346355 24.
25.
(1953).
13. Lovelock, J. E. Physical instability and thermal shock in red cells. Nature (ton&m) 4406, 659-661 ( 1954). 14. Luyet, B. Ultra-rapid freezing as a possible method of bIood preservation. In “The Preservation of Formed Elements and Proteins of Blood,” pp. 141-149. American Red Cross, 1949. 15. Luyet, B. Effects of ultra-rapid and of slow freezing and thawing on mammahan erythrocytes. Biodynamics 6, 217-233 (1949). 16. Luyet, B. J., and Gehenio, P. M. Effect of the rewarming velocity on the preservation of rapidly frozen blood. Biodynamics 7,
HUMAN
pp. X4-310, Academic Press, New York and London, 1966, 21. Mazur, Peter. CryobioIogy: The freezing of biological systems. Science 168, 939-949 (1970). 22. Mazur, Peter, Farrant, John, Leibo, S. P., and Chu, E. H. Y. Survival of hamster tissue culture cells after freezing and thawing. Interactions between protective solutes and cooling and warming rates. Cyobiology 6, 1-9 ( 1969). 23. Mazur, P., Leibo, S. P., and Chu, E. H. Y. A two-factor hypothesis of freezing injury.
414426
mechanism of the proglycerol against haemoand thawing. Biodaim.
Biophys. Actu 11, 28-36
OF
26.
27.
28.
29.
30.
31.
32.
Mazur,
Peter,
Leibo,
S.
(1972).
P., Farrant, John, Chu, E. H. Y., Hanna, Jr., M. H., and Smith, L. H. Interactions of cooling rate, warming rate and protective additive on the survival of frozen mammalian cells. In “The Frozen Cell” (G. E. W. Wolstenholme, and Maeve O’Connor, Eds. ), pp. 69-88. J, & A. Churchill, London, 1970. Mazur, Peter, and Schmidt, Janice J. Interactions of cooling velocity, temperature, and warming velocity on the survival of frozen and thawed yeast. CryobioZogy 5, 1-17 ( 1968). Meryman, H. T. The relationship between dehydration and freezing injury in the human erythrocytes. In “Cellular Injury and Resistance in Freezing Organisms” (E. Asahina, Ed.), pp. 231-244. Institute of Low Temperature Science, Sapporo, Japan, 1967. Meryman, H. T. Absence of unfrozen freezeable water in rapidly frozen red cells. Cryobiology 7, 252-255 ( 1971). Morris, G. J., and Farrant, J. Effects of cooling rate on thermal shock hemolysis. Cyobiology 10, 119-125 (1973). Nei, T. Mechanism of haemolysis of erythrocytes by freezing at near-zero temperatures. In “The Frozen Cell” (G. E. W. Wolstenholme, and M. O’Connor, Eds.), pp. 131147. J. & A. Churchill, London, 1970. Nei, Tokio, and Asada, Minnru. Changes appearing in the rewarming process of rapidly frozen erythrocytes. Low Temp. sci. Ser. B. 30, 45-63 (1972). Pert, J, H., Schork, P. J., and Moore, R. Low Temperature preservation of human erythrocytes. Biochemical and clinical aspects. In “Proceedings of the Tenth Congress of the International Society for Blood Transfusion,” pp. 674-682. Stockholm, 1964. Karger, 1965. Rapatz, G., and Luyet, B. The problem of .. _ _ _ the effect of intracellular ice iu hemolysis.
308
33.
34.
35.
38.
RAPATZ,
LLJYET AND MACKENZIE
In “Progress in Refrigeration Science and Teclmology,” pp. 1587-1572. Proceedings of tha XIth International Congress of Refrigeration, Munich, 1963. Pergamon Press, New York, 1985. Rapatz, G., and Luyet, B. On the instability of frozen glycerol solutions stored at various low temperatures. Biodynumica IO, 81-93 (1967). Rapatz, G., and Luyet, B. Combined effects of freezing rates and of various protective agents on the preservation of human erythrocytes. Cryobiology 4, 215-222 (1968). Rapatz, G., Sullivan, 1. J., and Luyet, B. Preservation of erythrocytes in blood containing various cryoprotective agents, frozen at various rates and brought to a given final temperature. CyobioZogy 5, IS25 (1988). Rowe, Arthur W., Eyster, E., and Keimer, A.
37. 38. 39.
40.
Liquid nitrogen preservation of red blood cells for transfusion. A low glycerol-rapid freeze procedure. C~yobioIogy 5, 119-128 (1988). Sakai, A. Some factors contributing in the survival of rapidly cooled plant cells. Cyobiolugy 8, 225-234 ( 1971). Steck, T. L. The organization of proteins in the human red blood cell membrane. J. Cell Biof. 82, 1-19 ( 1974). Toscano, W. M., Cravalho, E. G., Silvares, 0. M., and Huggins, C. E. The thermodynamics of intracellular ice nucleation in the freezing of erythrocytes. ASME Publication 74-WA/Bio-11, pp. l-16. ASME, the Professiona division, New York, 1974. Whittingham, D. G., Leibo, S. P., and Mazur, P. Survival of mouse embryos frozen to -198’ and 269°C. Science 1’78, 41.1-414 (1972).