29, 39-68 (1992)

CRYOBIOLOGY

Characteristics PETER The University Ridge National

MAZUR,

and Kinetics of Subzero Chilling Drosophila Embryos’,*

ULRICH

SCHNEIDER,3

AND ANTHONY

Injury in

P. MAHOWALD*

of Tennessee-Oak Ridge Graduate School of Biomedical Sciences, Biology Laboratory, Oak Ridge, Tennessee 37831; and the *Department of Molecular Cell Biology, Universio of Chicago, Chicago, Illinois 60637

Division, Genetics

Oak and

Drosophila embryos manifest unusually high sensitivity to chilling in that they are killed with increased rapidity by exposure to temperatures between 0 and - 2s”C in the absence of ice formation. Thus, 50% of 15-h eggs succumb in 35, 4, and 1 h at 0, -9, and - WC, respectively. The sensitivity becomes substantially greater in embryos at stages of development earlier than 12 h, especially at 3 and 6 h. The killing kinetics at given subzero temperatures between 0 and -25°C are characterized by a shoulder followed by a more-or-less linear decrease in survival with time. The lower the temperature, the shorter the shoulder and the faster the postshoulder decline. The rate of both components follows Arrhenius kinetics, i.e., plots of log rate vs l/absolute temperature are linear, the slopes being proportional to the activation energy. In both cases the activation energy is high and negative; namely, - 46.5 kcal/mol for the shoulder length and - 24.7 kcal/mol for the postshoulder inactivation. Negative activation energies are unusual, and according to absolute reaction rate theory, they exist only when the entropy of activation is negative, which suggests that the activated state is more ordered, By combining the duration of the shoulder as a function of time and temperature with the rate of postshoulder inactivation, one can compute survival as a function of temperature for embryos cooled at various rates. For those cooled at ~l”C/min, the computed curve of survival vs temperature agrees closely with observed survivals. But for embryos cooled at -lO”C/min, the drop in survival occurs some 7 to lo” above that computed. Embryos exposed to 0°C for >5 min undergo conditioning that renders them more resistant to subsequent exposure to lower temperatures, and those cooled at lW/min presumably lack sufficient time at 0°C to undergo such conditioning; hence the discrepancy between observed and computed survivals. As a test of the possibility that chilling injury is a consequence of the loss of synchrony of coupled reactions involved in embryological development, embryos were rendered anoxic prior to chilling, a treatment that has been shown by Foe and Alberts to reversibly halt development of early stages. Although anoxia somewhat reduced chilling injury in 6-h eggs, it had no effect on 15-h eggs. The sensitivity of embryos to chilling injury below -2s”C cannot be experimentally assessed because differential thermal analysis shows that intact eggs freeze between - 27 and - 32°C with lethal consequences. But death from chilling occurs so rapidly between - 15 and -25°C that the use of orthodox slow freezing procedures to remove intracellular water prior to entering the intraembryonic nucleation zone will not succeed; the embryos will succumb to chilling even before they have reached that zone. Unless ways can be found to reduce the accelerating rate of chilling injury as cooling progresses, the only solution is to “outrace” it by cooling at high rates and hope that intraembryonic ice formation can be avoided by the introduction of solutes that enhance 0 1992 Academic Press, Inc. vitrification.

Received December 28, 1990; accepted May 17, 1991. i The U.S. Government’s right to retain a nonexclusive royalty-free license in and to the copyright covering this paper, for governmental purposes, is acknowledged. ‘This research was supported in the main by the National Science Foundation Grant DCB8520453 and in part by the Office of Health and Environmental Re-

search, U.S. Department of Energy, under contract DE-AC05-840R21400 with the Martin Marietta Energy Systems, Inc. Preliminary reports of this work were presented at the 1988 and 1990 annual meetings of the Society for Cryobiology (Cryobiology 25, 544 and 27, 650). 3 Present address: Deutsches Institut fur Reproduktionsmedizin GmbH, Magdeburger Strasse 13, 3257 Springe , Germany. 39 001 l-2240/92 $3.00 Copyright 8 1992 by Academic Press, Inc. All rizhts of remoduction in anv form reserved.

40

MAZUR,

SCHNEIDER,

Mammalian embryos develop in closely controlled environments whereas insect eggs in nature are subject to large fluctuations in the temperature and humidity of their surroundings. Yet, paradoxically, embryos of most mammalian species can now be frozen to temperatures far below 0°C and returned to normal temperatures with high survivals whereas most insect embryos, such as those of Drosophila, cannot. Aside from the fundamental question of what is responsible for the difference, there is the practical point that the ability to preserve Drosophila embryos at liquid nitrogen temperatures, as can be done with embryos of mice and other mammals, would substantially benefit Drosophila geneticists who currently are forced to maintain more than 10,000 genetic lines by standard breeding and monthly transfers. Not only is this procedure exceedingly time-consuming and costly, but many lines have to be discarded for lack of space and time. Furthermore, the continued breeding of successive generations increases the likelihood of changes from genetic drift. The orthodox approach to the cryopreservation of mouse embryos and other cells requires that the water content of the cells be reduced to a low level before the cells have cooled to the temperatures at which intracellular ice can form. If their water content is sufficiently reduced, ice either will not form in the cells as they traverse the ice nucleation zone or it will form in insufficient quantities to be damaging. There are several strategies to effect the required cell dehydration (16). One is to cool cells slowly enough to allow sufficient time for them to dehydrate osmotically in response to the difference between the chemical potential of the supercooled water in the cells and that of the water and ice in the extracellular solution. Another strategy is to dehydrate the cells prior to cooling by placing them in hyperosmotic concentrations of nonpermeating or poorly permeating solutes. A third strategy is to decrease

AND

MAHOWALD

the required extent of dehydration by introducing high intracellular concentrations of solutes that impede ice crystal formation, or even prevent it by inducing vitrification. A fourth strategy is to induce intracellular vitrification by high cooling rates. The second, third, and fourth strategies are often combined (27a). Although avoiding injurious quantities of intracellular ice is necessary for cell survival, it is not usually sufficient. Most cells also require the presence of cryoprotective solutes, usually at concentrations exceeding 1 molar both outside and within the cells. Glycerol and dimethyl sulfoxide are the most commonly used. The causes of injury when intracellular ice is avoided and the mechanisms by which cryoprotective solutes prevent that injury are a matter of current debate, but injury appears to be a combination of the chemical consequences of high solute concentrations engendered by freezing, the osmotic consequences of those high concentrations, and the mechanical and rheological consequences of the growth of external ice and the concomitant reduction in the size of the liquid channels in which the cells lie (15, 27, 18). All these cryobiological strategies require that the membranes of the cells be permeable to water and to cryoprotective solutes. They also for the most part require that there be little detrimental effect of lowered temperature per se. Unfortunately, Drosophila embryos meet neither stipulation. They are isolated from the outside environment by waxy material in the vitelline membrane that allows gas and vapor exchange but effectively prevents the movement of water or aqueous solutes. And they are highly sensitive to exposure to subzero temperatures in the absence of ice; i.e., chilling injury. The problem of impermeability has been attacked in various ways, most commonly by exposing eggs to alkanes such as octane (13) or hexane (14) to remove the waxy material. The matter of chilling injury has been the

KINETICS

OF

CHILLING

subject of preliminary study by us (17,20), Myers et al. (23), and Myers and Steponkus (25). Here we present our findings on the extent and kinetics of chilling injury in intact (nonpermeabilized) Drosophila embryos as a function of temperature and developmental stage. In essential agreement with Myers et al., we find that earlier stage embryos (3-6 h) are considerably more sensitive than later stages (12-15 h), but that irrespective of developmental stage, chilling sensitivity increases markedly and progressively as the temperature decreases from 0°C to -25°C. It presumably continues to increase at still lower temperatures, but lower temperatures are inaccessible to experimental study because of the formation of intraembryonic ice at about - 30°C. The sensitivity to chilling injury becomes so high by -25°C as to preclude the possibility of preserving the embryos by orthodox slow freezing techniques. The time spent during cooling at rates low enough to prevent intraembryonic ice will result in death from chilling. Consequently, unless ways can be found to substantially reduce the rate and extent of chilling injury, the only approach to cryobiological preservation appears to be to cool embryos so rapidly that they will reach very low subzero temperatures before chilling injury has had time to be inflicted. To prevent the formation of lethal intracellular ice that ordinarily accompanies such rapid cooling, it will be necessary to introduce sufficient concentrations of glass-promoting solutes to induce vitrification of intraembryonic solution rather than freezing. This is the approach taken by Steponkus et al. (28) in their recent report of the first instance of the survival of Drosophila embryos after cooling to liquid nitrogen temperatures. METHODS

Rearing

AND

MATERIALS

of Flies and Obtaining

Eggs

Flies of the Oregon R-P2 strain of Drosophila melanogaster were maintained by

INJURY

IN

DROSOPHILA

41

the method of Travaglini and Tartof (30) at 25°C with minor modifications. To obtain eggs for experimental use, trays of 2% agar smeared with a paste of rehydrated active dry yeast were placed in a fly cage for three hours. We selected that period for ovoposition so that when we examined the response of nominal 3-, 6-, 9-, 12-, or 15-h embryos to chilling, the collection would include embryos of intervening developmental stages as well. If instead we had used, say, a l-h pulse collection, there was always the possibility that the sensitivity of stages falling outside the pulse period might differ substantially from the chilling sensitivity of eggs collected within the pulse period. At the end of the 3-h collection period, the eggs were washed off the agar trays with room-temperature water and passed through appropriately sized USA Standard Testing Sieves to reduce contamination by adult body parts and yeast. The collected eggs were then spread on moistened filter paper in a petri dish. Staging the Embryos To obtain 3-, 5- to 6-, and 9-h embryos, the collected eggs were maintained at 26 t 1°C for 3 ,5-6, and 9 h from the midpoint of the collection period. To obtain 12- and 15-h embryos, the eggs were held for combinations of times at 26°C (24°C in later experiments), room temperature (22”C), and 18°C calculated to produce 12- and 15-h embryos (as measured from the midpoint of collection) at 8 to 10 AM the following morning. In making these calculations we used growth rate factors of 1, 0.9, 0.75, and 0.5 for maintenance at 26, 24, 22, and 18°C respectively, factors that we obtained from Ashburner and Thompson (1). We checked on the accuracy of the calculations in two ways. The first was the morphological appearance of nominal 12- and 15-h dechorionated eggs compared to photomicrographs published by Wieschaus and NiissleinVolhard (31). The second was to take sam-

42

MAZUR,

SCHNEIDER,

ples of eggs calculated to be at the 15-h developmental stage and determine the time required for 50% of them to hatch when returned to 26°C. Extensive checks have shown that the 12-19 h spent at 18°C have no adverse effect on untreated controls. Indeed their survival was often somewhat better than that of controls incubated at 2C 26°C.

In earlier experiments, samples of about 100-150 eggs at the desired developmental stage were transferred to gridded Nuclepore filters (lo-pm pore size, 13-mm diameter, Cat. No. 110415) and subjected to experimental treatments. In later experiments involving 12- and 15-h embryos, the eggs were transferred to Nuclepore filters immediately after collection and were allowed to develop on the filters. After the embryos had attained the 12- and 15-h stage, the Nuclepore filters were kept at 4°C in a refrigerator or on ice to arrest development during that day’s experiments. This procedure was initiated after we had determined that 12- and 15-h embryos could be maintained at 0°C for >24 h without loss of viability. The eggs were not dechorionated. Prechill Manipulations

In most experiments, surface water was removed from the eggs by flowing air at room temperature (22°C) and humidity (4& 50%) through the Nuclepore filter for 30 s. In other instances, the Nuclepore filters were placed on dry filter paper and exposed to room air for 2.5 min. These two procedures removed less than 1% of the internal egg water (19). In still other instances, the eggs were covered with a few drops of absolute ethanol or 8 M ethylene glycol. Neither is toxic to intact eggs (14, 20). Several experiments involved cooling 15-h embryos to between 0 and - 25°C after 25 or 35% of their water had been removed by exposing the eggs on Nuclepore filters to room-temperature air for 2Yz and 5 h, respectively. The fraction of water lost from the eggs was determined from the weights

AND MAHOWALD

of the filter before adding eggs, after adding 600-1500 pg of eggs (-65-160 eggs) and removing their surface water, and after the 2% or 5 h drying, and from the fact that the normal water content of the egg is 0.75 g/g egg (19). Samples were weighed to 2 1 pg on a Cahn G2 electrobalance. After the brief drying to remove surface water or the more extensive drying just described, a Nuclepore filter with its deposited eggs was placed in the bottom of each of two small plastic cups separated by a strip of aluminum foil. The cups in turn were inserted into a predried polycarbonate centrifuge tube (Fig. 1). The tubes were then sealed with tight fitting plastic caps. In

THERMOCOUPLE

FIG. 1. Sample container used in chilling experiments involving cooling at low or moderate rates. The tube was an IEC No. 2997 polycarbonate centrifuge tube, 28.6 mm o.d., 26.4 mm i.d. 103.5 mm long, sealed with a tight fitting Naglene No. 29 cap. The bottom of the tube was inserted into a close fitting copper cap to provide ballast in the ethanol baths used for cooling. Drosophila eggs on 13-mm Nuclepore tilters were placed in two polyvinyl chloride sample cups (cut from Linbro Multi-well tray 96CV) with base and top diameters of 15 and 18 mm and a height of 10 mm. A small strip of aluminum foil prevented the top cup from nesting into the lower one. The thermocouple was present only in the determinations of cooling and warming rates.

KINETICS

OF

CHILLING

a few experiments involving temperatures of - 8°C and below, a small quantity of the strong drying agent, P,O,, was placed in the bottom of the centrifuge tubes about 1 cm below the eggs. The purpose of these precautions was to minimize the formation of frost on the surface of the eggs by keeping the humidity low. Since intact eggs in air at - 6°C lose water at a rate of only 0.5%/h (19), those at - 8°C or below in the present experiments would lose less than 5% of their water during the maximum exposures to these temperatures. However, in an initial experiment in which eggs were held at 0°C for 27-51 h, the eggs underwent undesired substantial dehydration. To prevent such inadvertent dehydration in subsequent experiments involving long exposures at 0°C we placed a moistened 13-mm Millipore prefilter in the bottom of the centrifuge tubes. With this precaution, the water loss in eggs held for 72 h at 0°C was held to 7-8%, which is not in itself harmful (19). Experiments with anoxic eggs. Several experiments involved the chilling of eggs in which the surrounding air was replaced with high-purity dry nitrogen. Nuclepore filters with their sample of eggs were placed in sample cups in the centrifuge tubes as usual (Fig. l), but the tops of the tubes were closed with rubber stoppers containing an inflow port connected to the nitrogen source and an outflow port. Nitrogen was allowed to flow into the centrifuge tubes for 5 or 15 min, and the ports were then closed off. Chilling was initiated 15 or l-3 min later in the two cases, respectively. Chilling

Some of the chilling experiments involved cooling the eggs in the sealed centrifuge tubes to 0 to - 15°C at either low rates (0.2-l”C/min) or moderate rates (5lO”C/min) and holding them for various times at those temperatures before warming them at about 5-lO”C/min. Other experiments involved cooling samples of eggs at very high rates (-4000”C/min) to - 15,

INJURY

IN DROSOPHILA

43

- 20, - 25, and - 30°C and holding them for various times before warming them to room temperature at even higher rates. Slow cooling. To cool samples slowly, the capped polycarbonate centrifuge tubes were placed in ethanol in small thermoelectric coolers or in controlled cooling rate refrigerators (FTS Systems). When cooling was to be at l”C/min, the initial temperature of the ethanol was + 10°C. In experiments involving cooling at 0.2”Umin the samples were generally cooled from 10 to 0 or - 10°C at l”C/min and then at O.TC/min to lower temperatures. As the samples reached the desired temperatures, the centrifuge tubes were removed and were either held at those temperatures for desired times in other baths or were immediately warmed by exposure to room-temperature air. Cooling at moderate rates. To cool samples at moderate rates to - 8 to - 15°C the capped centrifuge tubes were immersed to within 2 cm of the cap in stirred ethanol in an FTS bath precooled to about 0.2 to 0.5”C below the desired minimum exposure temperature to compensate for the measured temperature gradient between the bath and the bottom of the sample cup (Table 1). (In preliminary experiments involving short exposure times, the low-temperature baths were simply 4-liter dewars containing prechilled ethanol.) Cooling to 0°C at moderate rates was achieved by burying the centrifuge tubes to their tops in ice. After the desired exposure time, samples were warmed by placing the centrifuge tubes in room-temperature air. Table 1 summarizes the cooling and warming rates produced by these procedures, as measured by a 36-gauge thermocouple inserted through a small hole in the cap into the bottom of a sample cup, and it gives the temperature ranges over which these rates apply. The cooling rates and warming rates depend, of course, on the temperature range over which they are calculated and on the temperature of the bath into which the sample tubes are immersed

44

MAZUR,

SCHNEIDER,

AND

MAHOWALD

TABLE 1 Moderate Cooling Rates and Warming Rates Using the Sample Tubes Described in Fig. 1

Bath temperature (“C) 0 -8 - 15 -20 -25

Cooling ratea (“C/min) 5.2 6.0 8.4 11.9 12.5

Temperature range” (“Cl 20 to 0 to 0 to 0 to 0 to

2 -5.5 -12.8 - 17.8 -22.5

Steady-state temperature difference (sample bath)b (“C)

Warming rate’ (“C/mitt)

Range’ (“Cl

0.1 0.2 0.4 0.4 0.3

3.1 7.8 8.2 8.8 11.0

oto 15 -8to0 -15too -20 to 0 -25 to 0

LISample tubes were transferred from room temperature to an ice bath, or to stirred ethanol baths at - 8 to - 25°C. The listed cooling rates apply to the indicated ranges. b Differences between the temperature indicated by the sample thermocouple and that of the ice or ethanol bath after a steady state was attained. c Sample tubes transferred from the low-temperature baths to a beaker at room temperature.

or from which they are removed. But since the rates to and from - 15, -20, and -25°C (the temperatures at which chilling injury becomes critical) average close to lO”C/min, we shall sometimes refer to the rates produced by these procedures as being lO”C/min. Ultrarapid cooling and warming. A copper cylinder (38 mm diameter, 62 mm high) was immersed to within 3-4 mm of its top in ethanol in an FTS bath and precooled to the desired temperature as measured by thermocouples cemented into small holes drilled in the side of the cylinder (below the surface of the ethanol) and in its top (above the surface of the ethanol), and by a thermocouple in the ethanol bath. All three agreed to 90% of untreated control survivals. b The cooling rate to 0 and - 9°C was mostly 5&C/ min. The cooling rate to - 15, -20, and -25°C was -4OOOYYmin.

computed by least-squares lit, one can calculate an apparent activation energy, Ed* from the equation ln(D,/D,)

= EJ/R[lIT,

- l/T,],

[I]

where R is the gas constant (1.987 caYmo1 de&. The value of Ed4 is very large ( -46.5 kcaYmo1) reflecting the very strong effect of rol

3.5

FIG. 10. Thermogram of a sample of surface-dry 15-h embryos that were cooled to -24.3”C at 0.4”C/ min and then warmed at - l”C/min.

Duration of high-survival” plateau

Temneratureb

i6

3.7

3.8 l/T(x

3.9 lo3 K)

4.0

41

42

FIG. 11. Arrhenius plot of the logarithm of the duration of the high-survival shoulder in 12- to 15-h embryos as a function of the reciprocal of the absolute temperature at which they were held. The apparent activation energy ES is - 46.5 kcahmol.

KINETICS

OF CHILLING

temperature, and it is negative reflecting the fact that the rate of shoulder disappearance increases with decreasing temperature . As embryos are cooled, survival will remain near 100% as long as some of the shoulder remains. But the shoulder will be “used up” at an increasing rate as the temperature drops, and the lower the cooling rate the more shoulder will be used up upon cooling to a given temperature. To calculate the temperature at which the shoulder disappears in embryos cooled at given rates, we proceeded as follows: Method I

51

INJURY IN DROSOPHILA

“Using-up” Time (min) 0

0.1 0.2 0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.34

TABLE 4 of High-Survival Shoulder with Falling Temperature Temperature (“C) 0

-1 -2 -3 -5 -7 -9 -11 -13 -15 -17 - 19 -21 -23 -23.4

Resid. Shoulder (mm) 840 613 446 324 170 87.8 44.9 22.7 11.3 5.47 2.55 1.09 0.383 0.0384 0

1. For initial conditions we take the shoulder duration D, to be 840 min at 273.16 K(T,) (Table 3). Note. Cooling rate lO”C/min. Temperature decre2. The temperature is dropped instantament (AT) = O.l”C. neously to T2 by a small increment (AT) of 0.1 or 0.01 K. 100,000”C/min, survival will remain near 3. We compute II,, the shoulder duration 100% to -17.0, -35.3, and -46.2”C, reat T, from Eq. [l], the Arrhenius relation spectively . derived from Fig. Il. 4. The time the embryos spend at T2 is Method ZZ taken to be the time (t) that would have An alternative and perhaps more easily been spent cooling from T, to T2 at the selected cooling rate B; i.e., t = AT/B. visualized approach to computing the tem5. That time t is subtracted from D, to perature dependence of the kinetics of the obtain the time of shoulder remaining shoulder disappearance is to consider it as (TMSR) at the end of t min at T2. an amount = rate X time problem. Assume 6. We redefine T, as equal to T2 and D, as equal to TMSR, and then loop back to step 2. 7. The loop is repeated until the shoulder disappears; i.e., until TMSR = 0 min. An example of the computation is shown in Table 4 for embryos cooled at lO”C/min using a temperature decrement of 0.1 K. The shoulder is computed to disappear at - 23.4”C, as is the case with a AT of 0.01 K. Such computations were run at cooling rates ranging from 0.3 to lo5 Wmin, and in FIG. 12. Effect of cooling rate on the computed temFig. 12 we plot the temperature of shoulder perature (Tn) at which the high-survival shoulder of disappearance (T,,) as a function of the 12-to 15-h embryos becomes “used up.” The equation cooling rate B. The plots shows, for exam- of the line is Tn = - 17.142 - 5.924 log B, where B is ple, that for embryos cooled at 1, 1000, or the cooling rate in Wmin.

52

MAZUR,

SCHNEIDER,

that the embryo contains a quantity Q of a native molecule essential to normal development, and that during low-temperature exposure those molecules are converted to an inactive or denatured form at a rate that increases with decreasing temperature according to a variant of the Arrhenius relation in Eq. [l]; namely, k, = k. exp[ -23424(1/273.16 - l/T)], where k,, the rate at O”C, is l/840 or O.O0119/min. Survival remains high as long as Q > 0. The problem then is to compute the temperature at which Q falls to 0 as a function of cooling rate. We assume further that the rate of disappearance of Q at a given temperature is constant; i.e., it is independent of the amount remaining. We abruptly lower the temperature from 0°C by a small increment AT, for example by l”C, and hold 1 min (equivalent to a cooling rate of l”C/min). The conversion rate at - 1°C from the above equation is O.O0163/min, so that during the 1-min hold, 0.00163Q is converted and (1-0.00163)Q remains. We lower the temperature another degree. The conversion rate at - 2°C increases to O.O0224/min. That amount is converted to the denatured form in 1 min, so that the amount remaining after the hold is (0.99837-0.00224)Q. And so forth until Q reaches 0. The temperatures for disappearance computed by this method are identical to those computed by the preceding method. Indeed, as shown by Mitchell in the Appendix the two methods are mathematically equivalent. The Appendix also derives an exact differential formulation for computing the temperature of shoulder disappearance. The values computed by it agree with those plotted in Fig. 12 to within O.Ol”C. It should be noted in our incremental method, that if AT and At are small it makes essentially no difference whether one first lowers the temperature by AT and then holds At, or does the reverse. Once the shoulder is used up, survival will then drop with increased exposure time during cooling. Figures 3 and 6 show that over most of the range, the postshoulder

AND

MAHOWALD

0.01

)

I 3.8

3.6

I 4.0 l/T

1 I 4.2 4.4 x lo3 (K-l)

/ 46

I 4.8

I

I

I

I

I

/

I

0

-10

-20

-30

-40

-50

-60

"C

FIG. 13. Arrhenius plot of the rate of postshoulder killing of 12- to 15-h Drosophila embryos as a function of the temperature at which they are held. The equation of the least-squares best fit relation between killing rate, k, (%/min), and temperature is k, = 9.53 X lo-*‘e - (-‘244o)“, and the apparent activation energy E& is - 24.7 kcal/mol.

survival drops approximately linearly with time at given temperatures (although less so at - 15°C). We define the postshoulder killing rate (kT) as the slope of the lines at 50% survival. And in Fig. 13 we plot the log of the killing rate k, as a function of l/T. From the best fit curve to the live points and the Arrhenius equation (Eq. [l]), we compute an apparent activation energy EkS of - 24.7 kcal/mol. Although this is only about half the activation energy for the disappearance of the shoulder, it is still a high value that corresponds approximately to a tripling of the killing rate for every 5°C drop in temperature. Experimental determinations of the kinetics of chilling injury are not possible below - 25°C because of the occurrence of intraembryonic freezing; consequently, the dashed portion of the line in Fig. 13 is an extrapolation.

KINETICS

OF CHILLING

INJURY IN DROSOPHILA

53

Computation of the Survival of Unfrozen Embryos as a Function of Cooling Rate and Temperature We are now in a position to combine the effects of temperature and cooling rate on the shoulder duration with their effects on postshoulder killing, and thereby iteratively compute the effect of chilling alone on the survival of embryos cooled at given rates to given final temperatures. We proceeded as follows: 1. For a given cooling rate B, use the equation of the line in Fig. 12 to compute the temperature T, at which the highsurvival shoulder is used up. At zero time at that initial temperature, survival is taken to be 100%. 2. Instantaneously drop the temperature by a small interval AT (e.g., 0.1 or 0.01 K). 3. Compute the killing rate k, at that lower temperature from the experimentally derived Arrhenius relation (Fig. 13). 4. Compute the time t for the temperature to drop AT at cooling rate B. 5. Assume that the embryos remain at temperature T for time t, and compute the percentage killed as (kT) (t). Compute survival S, at the end of that time as S2 = S1 k,. t. 6. Redefine S, as equal to S, and loop back to step 2. 7. Repeat until survival reaches zero. Figure 14 gives the results of such computations for embryos cooled at rates ranging from 0.3 to 100,000”C/min. We reemphasize that this plot considers only the contribution of chilling injury and ignores the potential effect of ice formation or vitrification on survival. The plot also assumes that the increase in the rate of the disappearance of the shoulder and the increase in the rate of postshoulder killing with falling temperature continue unabated as the temperature falls below the experimentally accessible region. In other words, the plot assumes that the activation energies calculated from data between 0 and

FIG. 14. Computed survival of 12- to 15-h Drosophila embryos as a function of the cooling rate to indicated temperatures, assuming that chilling is the only factor affecting survival and ignoring the effect of time spent during warming.

-25°C apply to lower temperatures. And finally, the plot neglects the effect of time spent during warming. With these caveats in mind, the plots show the following: 1. The avoidance of intracellular ice by orthodox slow freezing requires cooling at rates low enough to permit cells to lose sufficient water to be in near osmotic equilibrium with the outside medium prior to their reaching their nucleation temperature (16). Studies by Myers et al. (24) show that the required rate in permeabilized Drosophila embryos is less than O.S”C/min. Figure 14 predicts that such low cooling rates will not be successful, for the embryos will be killed by around -26°C i.e., they will be killed by chilling injury before they even reach the upper temperature boundary for intraembryonic freezing. This prediction is in accord with the observed 0% survival in DTA runs terminated at -23 or -24°C (Fig. 10). 2. Mortality from chilling can be sup-

54

MAZUR,

SCHNEIDER,

AND

pressed to lower and lower temperatures by increasing the cooling rate. Thus, as the cooling rate is increased from 100 to 1000 to 10,000 and to 100,0OO”C/min, the temperature at which 50% of the embryos remain alive decreases to -49, - 58, -66, and - 74°C respectively. 3. Although the chilling injury can be temporarily “out run” by increasing the cooling rate, it will eventually catch up at sufficiently low temperatures and destroy the embryos. That is, it will do so if the increase in the rate of chilling injury with decreasing temperature continues unabated. We will say more on this point in the Discussion. 4. Figure 14 shows the computed survival at the instant the embryos arrive at the indicated temperatures. It does not take into account the effects of time spent at a given temperature or the time spent during warming. If we assume that the effects of time during warming are simply a mirror image of the effects of time spent cooling, we can assess the contribution of warming by assigning an effective cooling rate (ECR) which when followed by infinitely rapid warming exposes the embryos to subzero temperatures for the same length of time as the actual cooling and warming rates selected. If CR and WR are the actual cooling and warming rates, then ECR = (l/CR

1 + 1IWR) ’

PI

For example, the computed survival curve vs temperature for embryos cooled and warmed at 2000YYmin is indistinguishable at the scale drawn from the 1000”C/min (ECR) curve shown in Fig. 14. Experimental vs Computed vs Temperature

Survival

The curve in Fig. 15 labeled “hydrated” shows the observed survivals of 15-h eggs slowly cooled to various temperatures and then warmed at some 4-ll”C/min. Two

MAHOWALD

-5

-10 MINIMUM

-!5 TEMP. PC)

-20

-25

FIG. 15. Normalized survival (solid curves) of slowly cooled 15-h Drosophila embryos as a function of the minimum temperature attained. The embryos were either fully hydrated (0) or had 23% (0) or 35% (A) of their water removed by air drying. The embryos were either cooled at l”C/min to indicated temperatures and then held 3 min before initiating warming, or they were cooled from + 10°C to indicated temperatures at 0.2”Clmin and immediately warmed. The survivals for the two procedures are pooled. The survival of unchilled controls was 85%. The dashed curve shows the computed survivals vs temperature for embryos cooled at 0.2Wmin.

slight variants in procedure were used. In one experiment, the eggs were cooled to 0, - 10, - 15, -20, or -25°C at l.O”C/min with a 3-min hold at each temperature. In the other experiment, the eggs were cooled at l”C/min to - 10°C and then at 0.2”C/min to each final temperature at which point they were immediately warmed. In both cases survival remained near 100% to - IX, and then dropped abruptly to near 0% by - 25°C. Note once again that the killing is total at temperatures above those at which intraembryonic freezing first occurs, as evidenced by the DTA measurements. The dashed curve in Fig. 15 gives the survival as a function of temperature of em-

KINETICS

OF CHILLING

INJURY

IN DROSOPHILA

(00---------l--,,

55

I bryos cooled at 0.2”C/min, as computed by the procedures used to construct Fig. 14. The agreement with observed survivals is close. At the scale drawn there is no distinguishable difference in computed survival between including or not including the contribution from the time spent during warming at -lO”C/min. J The experiments of Fig. 15 were repeated 15 hr -o-c, using a moderate cooling rate instead of +ooc, slow cooling. The capped tubes containing the embryos were immersed in baths at 0, -10, -13, -15, -20, and -2YCfor 15 or k \ 18 min, and then warmed in air. This proI \\ I\‘\1 duced cooling rates of about 1OYYmin to - 15 to - 25°C a holding period at the given temperatures of some 11 or 14 min, and a warming rate of about lO”C/min (TaL! , ble 1). The three dashed curves in Fig. 16 -30 -5 -10 -15 -20 -25 0 TEMP. (“Cl show the computed survivals for this proFIG. 16. Normalized survival (solid curves) of 12cedure. Curves 1 and 2 are calculations for and 15-h embryos cooled at a moderate rate (--lo”C/ a holding period of 11 min; curve 3 is for a mitt to - 15°C and below) to indicated temperatures, holding period of 14 min. Curve 2 incorpoand held 1l-14 min before warming at some lW’C/min. rates the effect of time spent during warm- The notation “ + 0°C” means the embryos were held at ing after an 11-min hold; the other two ig- 0” to + 4°C between the time they reached the desired developmental stage and some 5 min before the chillnore that contribution, which is small. ing to lower temperatures was initiated. The notation Thus, the main contributor to the computed “-0°C” means such refrigeration was omitted; i.e., drop in survival at these temperatures is the the embryos were maintained at room temperature. ll- or 16min holding period. To incorpoThe three dashed curves give computed survivals as a rate the effect of holding period into the function of temperature for a cooling rate of lO”C/min computations, we first compute the dura- for, 1, embryos held 11 min and either warmed infinitely rapidly or, 2, warmed at lO”C/min, and, 3, emtion of high-survival shoulder remaining bryos held 14 min and warmed infinitely rapidly. (TMSR) upon first reaching the final temperature. For example, after cooling at 10”Clmin to - 15°C TMSR is 5.5 min. We calculations proceed a little differently. subtract that time from the holding time to First, we use the procedures used to conobtain the postshoulder exposure time struct Fig. 14 to compute the percentage (PST) at the given temperature; i.e., 1l-5.5 killed during cooling to the hold temperaor 5.5 min, in our example. Next we use the ture. For example, for embryos cooled at Arrhenius relation (Legend, Fig. 13) to de- lO”C/min to -27°C that figure is 3.3%. We termine the killing rate k, at that temperathen again calculate the postshoulder killing ture, and finally we calculate the kill during rate at the hold temperature (here -27”(J), PST as (kT) (PST), and subtract that from and use that rate to compute the additional 100%. loss in viability that occurs during the holdIf the high-survival shoulder has already ing period. At - 27°C) the additional death disappeared upon initially reaching the during an 11-min hold is 94.1% to yield holding temperature (which at a cooling 3.6% survival at its conclusion. rate of lO”C/min occurs by -23.4”C), the We checked on the validity of Eq. [2] by

601fl~cJ$+j

56

MAZUR,

SCHNEIDER,

AND

using another approach to incorporate the computed contributions to chilling injury of the time spent warming; namely, after every degree of cooling, we interjected a loop between the two parts of step 6 of the algorithm used to construct Fig. 14. In that loop, steps 2-6 of the Fig. 14 algorithm were performed with the temperature increasing in small increments rather than decreasing, and with the times at each increasingly higher temperature corresponding to the warming rate rather than the cooling rate. These computed survivals agree with those obtained using Eq. [2]. The experimental results are shown by solid curves in Fig. 16. The left-most pair of experimental curves show the survival of 15-h embryos cooled at some lO”C/min to indicated temperatures and held 11 or 14 min. In one case prior to the experiment the embryos had been held at 4°C after reaching the proper developmental stage (“ +O”C”). In the other case (“ -O’C”), the maintenance at 4°C was omitted. In both cases survival dropped abruptly between about - 13 and - 20°C some 7-9°C above that computed. The right-hand pair of experimental curves represent analogous experiments, but with 12-h rather than 15-h eggs. Again an abrupt drop in survival occurred, but in this case, it occurred between -20 and - 25°C rather than between - 13 and -20°C. In the case of the 12-h eggs, the temperature range over which survival

MAHOWALD

drops coincides reasonably closely with that computed in the dashed curves. The experimental points are pooled data for holding periods of 11 and 14 min. The three additional minutes of exposure had no effect on 12-h embryos but did have some effect on 15-h embryos; namely, for eggs maintained at 0 to 4°C prior to exposure to lower temperatures, survivals after holding for 11 and 14 min were 85 +- 6 and 50 +- 9% at - 15°C and were 21 -+ 8 and 7 + 3% at - 20°C. As this paper was being written, Myers and Steponkus (25) reported that 13- to 14-h embryos that were held some 5 to 30 min or more at 0°C suffer less killing after subsequent holding at - 15 or - 20°C than do embryos in which the prior exposure to 0°C is omitted. Furthermore, they report that the “conditioning” from the 0°C exposure is lost if the embryos are returned to room temperature for ~-5 min prior to subsequent exposure to subzero temperatures. In light of these findings, we repeated portions of the experiments just described in which we controlled with some precision the temperature regime prior to subzero chilling to - 15 to - 25°C. The results are given in Table 5. Four temperature regimes were used. In the first, the embryos were exposed to - 15, -20, or -25°C without any prior refrigeration. In the second, the embryos were held at 0 to 4°C for l-5 h and then chilled to - 15, - 20, or - 25°C without any warming in between, To ensure lack of

TABLE 5 Effect of Prior Exposure to 0°C on the Survival of 12- and 15-h Embryos Subsequently Cooled to - 15, -20, or -25°C at 1OWmin and Held 12-14 min Treatment prior to subzero treatment Held at 18-22°C Cooled to and held at 0°C l-5 h Cooled to 0°C l-5 h; warmed to room temperature for -5 min Cooled to O”C, l-5 h; warmed to room temperature for 20 min

% Survival 12-h eggs - 15°C 98 k 3 -

% Survival 15-h eggs

- 20°C

- 25°C

-15°C

- 20°C

- 25°C

94 * 3 85 k 6

922 72 k 3

37 * 5 88 -t- 6

9k4 91 f 1

421 19 * 9

98 k 1

28 2 3

6427

12 2 4

-

78 + 10

17 f 3

71 ? 8

37 f 16

1 * 0.5

KINETICS

OF

CHILLING

warming, all manipulations of the Nuclepore filters carrying the eggs were carried out in a 0°C cold room, and the polycarbonate tubes were prechilled to 0°C. In the third treatment, which is the “ + 0°C” data in Fig. 16, the refrigerated Nuclepore filters and embryos were in most cases dried for 2.5 min at room temperature and in all cases were transferred to sample cups and polycarbonate tubes at room temperature prior to initiating chilling to - 15°C and below. Thus, in this case there was an interval of some 5 min at room temperature between refrigeration and subzero chilling. In the fourth treatment, the intervening room temperature exposure consisted of 15 min on moist paper, 2.5 min on dry filter paper, and l-2 min to load the Nuclepore filters into the polycarbonate tubes-some 20 min in all. With the 12-h embryos there were no substantial differential effects of the four prechill treatments on survival after 14-min exposure at -2O”, but there were substantial differences after exposure to -25°C; namely, in the absence of prior refrigeration 9% survived whereas with uninterrupted refrigeration 72% survived. When some 5 or 20 min at room temperature were interspersed between refrigeration and chilling to -25”C, survival dropped to 28 and 17%, respectively. With 15-h embryos the relative effects of the four prechill treatments were comparable, but they were manifested at - 20°C rather than at - 25°C. Thus, two factors are operating. One confirms Myers and Steponkus’ (25) report that a period at 0°C makes the embryos more resistant to subsequent subzero temperatures and that this conditioning is rapidly lost upon return to room temperature. We do not, however, confirm their finding that when embryos are returned from 0°C to room temperature they become more sensitive to subsequent subzero chilling than those that are never refrigerated. The second factor is the stage of development: To produce comparable degrees of chilling in-

INJURY

IN

57

DROSOPHILA

jury, 12-h embryos have to be exposed to about a 5” lower temperature than 15-h embryos. However, even the fully conditioned 12-h embryos succumb at - 25°C. Thus, increasing their time at -25°C from 14 to 17 min lowered survival from 72 to 54%. Similarly, fully conditioned 15-h eggs are decimated by 1Cmin exposure to - 25°C (19% survival). In other words, conditioning delays the onset of subzero chilling injury but does not prevent it. Effect of Partial Dehydration on Chilling Znjury During Subzero Cooling

Successful freezing procedures require that cells be at least partially dehydrated before they are cooled to temperatures at which intracellular ice formation can occur. We wished to see, therefore, whether partial dehydration would affect the chilling sensitivity of Drosophila embryos. Airdrying for several hours at room temperature was used to remove 23-35% of water (as determined on a Cahn electrobalance) from 15-h embryos. The embryos were then cooled directly from room temperature to temperatures between 0 and -25°C in the same manner as the unconditioned fully hydrated embryos in Figs. 15 and 16. The resulting survivals, which are shown in Fig. 15 and Fig. 17, are quite similar to those for the fully hydrated embryos except that the partially dehydrated embryos undergo a given degree of chilling injury at somewhat higher temperatures. We do not know whether development continued unabated during the 21/2-5 h of drying. DISCUSSION

Although a number of cell types exhibit injury as a result of cooling to below 0°C in the absence of ice formation (22), intact Drosophila embryos are unusually sensitive . Chill Sensitivity of Drosophila

vs Developmental Embryos

Although 3- to 15-h developmental

Stage

stages

58

MAZUR,

-to

TEMP. (‘CL)

9--t-20

SCHNEIDER,

-30

FIG. 17. Effect of the removal of 25 or 37% of egg water on the survival of 15-h embryos vs the minimum temperature to which they were cooled at some lO”C/ min. The partially dehydrated embryos were held at the several temperatures for 11 min before warming was initiated. The corresponding curve from Fig. 16 for fully hydrated embryos is shown for comparison. Embryos were not subjected to refrigeration prior to chilling.

all succumb to exposure to 0°C and below, the earlier stages are more sensitive (Fig. 2). Indeed, 3- and 5-h embryos are so sensitive that it is not feasible to use storage at 0°C to keep the developmental stage constant during the course of a day’s experiments. The high sensitivity of these early stages might also play a role in limiting the geographic distribution of the insect. Thermal inertia in nature is so great that when the temperature drops to 0°C in the vicinity of early-stage eggs, that temperature is likely to be maintained long enough to kill them. One characteristic of the chilling injury is that it is preceded by a high-survival shoulder or plateau. The greater sensitivity of the early stages is reflected more in a shortened duration of this shoulder than it is in the subsequent rate of killing. One possibility is

AND MAHOWALD

that during the time represented by the shoulder, faulty enzymatic products accumulate as concatenated biochemical reactions lose their synchrony at low temperature because different reactions possess different activation energies. If so, the sensitivity might be reduced if development could be reversibly halted prior to chilling. Consequently, we were intrigued by Foe and Albert’s (6) report that anoxia produces such a halt. When we subjected 6-h embryos to exposure at 0°C the duration of the high-survival shoulder was in fact extended appreciably in those rendered anoxic (Fig. 5). However, anoxia had no effect on 6-h embryos at - 8 or - 15°C (Table 2), and it had no beneficial effect on the chill sensitivity of 15-h embryos at 0 or -8°C (Fig. 4). Since aerobic 15-h embryos were still more chill-resistant than anoxic 6-h embryos, we did not investigate anoxia further. Still, further study of the role of anoxia might be warranted. The 12- to 15-h embryos are not only more chill-resistant than early stage embryos, they are also more resistant to partial dehydration (19). They are thus clearly the preferable stage on which to concentrate efforts to develop cryopreservation techniques, which of necessity involve both a reduction in temperature and dehydration. The cited papers of Steponkus and colleagues indicate that they have reached the same conclusion. Chilling Znjury in 12- to 15-h Embryos as a Function of Temperature The sensitivity of embryos increases rapidly as the temperature is lowered below 0°C so that at - 25°C all are killed in a matter of minutes. Whether chilling sensitivity continues to increase at still lower temperatures cannot be assessed by direct experiments since intraembryonic freezing occurs between - 27 and - 33°C as shown by the DTA experiments reported here and by Mazur et al. (19) and by the cryomicros-

KINETICS

OF

CHILLING

copy and differential scanning calorimetry experiments of Myers et al. (24). Several lines of evidence support the view that the death of intact unpermeabilized eggs between 0 and - 25°C is a consequence of low temperature per se and is not a consequence of the presence of ice either inside or on the surface of the eggs. 1. In most of the experiments, the eggs were subjected to a stream of room air (RH 40-50%) prior to initiating chilling. Weighings on sensitive Cahn electrobalances have shown that such drying removes all evidence of surface water, and does so without removing more than a few tenths percent of the intraembryonic water (19). 2. The filters serving as a support for the eggs were then placed in tubes which had been carefully dried beforehand and which were closed with tight-fitting caps prior to the initiation of cooling. Some tubes in addition contained a powerful drying agent, phosphorous pentoxide. As a result of these precautions the quantity of frost inside the tubes should have been minuscule or zero. 3. DTA runs in which cooling was halted at - 22 to - 24°C showed no evidence of an exotherm, with the exception of one small exothermic spike in one run that probably reflected the freezing of a single egg. Our apparatus can detect the heat released by the freezing of 67 kg of water. 4. In most of the experiments involving very rapid cooling to - 15 to - 25°C the top of the copper block on which the Nuclepore filter was placed and the eggs themselves were covered with a thin film of ethanol or 8 M ethylene glycol to preclude the possibility of ice formation. 5. The eggs in our experiments had normal chorions and vitelline membranes and, because of the latter, they were impermeable to liquid water. Therefore, it is unlikely that minute amounts of ice on the egg surface, if present, could adversely influence the embryo within. On the other hand, Myers et al. (24) have shown that the in-

INJURY

IN

DROSOPHILA

59

traembryonic freezing of dechorionated permeabilized eggs is shifted to considerably higher temperatures by the presence of external ice, as is the intracellular freezing of other cells that have normal high permeabilities to liquid water (14a). Components Drosophila

and Kinetics of the Embryo Chilling Injury

Investigators in the field of low temperature nonfreezing injury generally distinguish two types of chilling injury (22). One is cold shock in which the injury is manifested at high cooling rates but not at low rates. The other type of chilling injury is independent of the cooling rate, and that is the type applicable to 12- to 15-h Drosophila embryos, at least over the range of cooling rates that we examined (0.2 to 4000”Clmin). Injury in Drosophila is thus a function of the exposure time at low temperatures and is not a consequence of the act of cooling to a given temperature. If the exposure time at given subzero temperatures is sufficiently short, survival is 100%. This is also the case of 3- to 6-h embryos cooled to 0°C. The kinetics of the Drosophila embryo chilling injury are characterized by an initial shoulder or plateau during which survival remains 100% and a postshoulder decline in survival that is approximately a linear function of time. The duration of the shoulder in 12- to 15-h embryos decreases and the rate of postshoulder killing increases dramatically as the temperature is decreased from 0 to - 25°C. Both components follow Arrhenius kinetics in that the logarithm of rate is a linear function of the reciprocal of the absolute temperature. The slope of that line multiplied by the gas constant (1.987 cal/deg mol) yields the apparent activation energy, the values of which are - 46.5 and - 24.7 kcaVmo1 for shoulder disappearance and postshoulder killing, respectively. Both values are unusual in two respects: they are large and they are negative. An ES of - 46.5 kcal/mol corresponds

60

MAZUR,

SCHNEIDER,

to about a 25fold increase in rate with a 10°C decrease in temperature. Most biological processes show a 2- or 3-fold change in rate with a 10” change in temperature, and nearly all biological processes and chemical reactions exhibit increases in rate with increases in temperature and thus have positive activation energies.

AND

MAHOWALD

injury where AES and AH+ have large negative values, ASS must have a large negative value to make AF$ positive. That is to say, the entropy of the activated state has to be lower than that of the initial state. Johnson et al. (9) and Eagland (5) cite a number of other examples in which AS$ is calculated to be negative. What might such negative values mean? The interpretation that Johnson et al. give The Nature of the Chilling Injury is that only a small proportion of activated In their treatise on the kinetics of biologmolecules have the proper conformation to ical and chemical processes, Johnson et al. proceed to the final state. Another interpre(9) point out that the Arrhenius formulation tation is that the activated molecules are has difficulty accounting for a negative ac- more ordered than the initial molecules, or tivation energy, because it would mean es- that the act of activation induces more orsentially that the activated state has lower der (lowered entropy) in the whole system. potential energy than the energy of the re- An example of the former might be that acactants and would thus lie in an energy well tivation on the road to damage involves in which it would be trapped. They argue crystallization of essential lipid molecules. instead that the kinetics of reactions are de- An example of the latter might be that activation involves the unfolding of essential termined not by ES the energy of activation (which is very close to AHS, the enthalpy of proteins so that normally buried nonpolar activation), but rather by AF$, the free en- residues become exposed to water and as a consequence induce increased order (lowergy of activation. As with AF in equilibered entropy) in the water. There is a large rium thermodynamics, AFS can be resolved body of data starting with the work of ininto an enthalpic and an entropic component such that vestigators like Frank and Evans (7) and Klotz (11) that the introduction of nonpolar AF$ = AH+ - TASS. 131 groups into water does in fact increase the order of the water, which is the reason the The rate of a reaction is given by transfer is unfavorable from a free energy k = A exp( - AFflRT) = A point of view. This phenomenon led Kauzexp( - AHflRT) exp(AS$/R), t41 mann (lo), Nemethy and Scheraga (26), and Tanford (29) among others to the concept in which A contains Boltzman and Planck’s constants and represents the fixed rate of that an important factor in the stability of decomposition of the activated state. native proteins is their assuming a conforThis formulation is able to account for mation in which maximal numbers of nonnegative AH$. or negative ES. They argue polar or hydrophobic groups are buried out and present considerable evidence that for of contact with water. The term given to reactions to occur at measurable rates be- this factor is hydrophobic bonding or hytween 0 and 50°C AF$ has to lie between drophobic interactions. +20 and 30 kcal/mol. They further show Brandts (2, 3) was the first to point out that in cases such as high-temperature pro- that proteins in which hydrophobic bonds tein denaturation in which AE$ or AH+ are play an important role in stabilizing the naoften above 100 kcaYmo1, AFS is still only tive conformation can undergo denatur20-30 kcal/mol because the entropy of ac- ation at sufficiently low temperatures as tivation ASS has a large positive value. In readily as at high temperatures. Low temperatures weaken the hydrophobic bond by the contrasting case of Drosophila chilling

KINETICS

OF

CHILLING

making it less energetically unfavorable for nonpolar residues to be exposed to water as a consequence of unfolding. In most proteins the temperatures at which the hydrophobic interactions become weak enough to permit denaturation are too far below 0°C to be experimentally accessible because of ice formation, but Brandts et al. (4) using high pressures to suppress the freezing point of solutions and Franks and Hatley (8) using special techniques to supercool emulsions of protein solutions have demonstrated reversible unfolding of chymotrypsinogen at about - 12°C. It is tempting to suggest that Drosophila eggs represent another special case, provided by nature, whereby the protein solutions in the embryo can be exposed to as low as - 25°C without ice formation and thus may undergo low-temperature denaturation. One obvious caveat is that Drosophila embryos are not a single purified protein, but a system involving hundreds of enzymes many of which must act in a coordinated or concatenated fashion for proper function. This coordination could well be lost at reduced temperatures if different enzymes have different activation energies. The increased rate of inactivation with decreasing temperature, then, may somehow be a reflection of increased loss of coordination rather than a reflection of a negative activation energy for a single reaction. McGrath (21) has proposed a different mechanism of chilling injury in the absence of freezing based on the fact that reduced temperature will result in membranes attempting to thermally contract laterally around an essentially incompressible aqueous interior. The contraction will cause the efflux of cell water, but the efflux of water will concentrate the cytoplasmic solution and reduce its chemical potential, thus setting up an osmotic force to drive water into the cell. Equilibrium will be attained when the contracting membrane exerts sufficient hydrostatic pressure on the cell contents to raise the chemical potential of the cell water to the value of the chemical potential of

INJURY

IN

DROSOPHILA

61

the external water. The result will be a finite stress or tension on the membrane. That stress will induce an area strain in the plane of the membrane in that the membrane area will be greater than its area would be in a stress-free state. If that discrepancy exceeds 2-3%, it will exceed the strain that most membranes can tolerate. McGrath calculates that temperature decreases of lO-20°C will produce tensions that are known to damage red blood cells and plant protoplasts. He concludes further that the tensions developed in membranes surrounding small vesicles with diameters ~0.2 pm will also depend on the water permeability (L,) of the vesicle and the cooling rate, because L, will set a limit on the rate at which the contraction-induced water efflux can occur relative to that required for equilibrium as the cells are cooled at various rates. His calculations indicate that these rate effects should not be manifested in objects the size of cells. We did not observe cooling rate effects in our study. We do not wish to leave the impression that the above discussion constitutes a complete survey of possible mechanisms of Drosophila chilling injury. A number of other hypotheses of chilling injury in general are reviewed in books by Morris and Clarke (22a) and Grout and Morris @a). Cryobiological

Implications

Knowledge of the kinetics and activation energies of the two components of chilling injury in Drosophila embryos (the duration of the high-survival shoulder and the rate of postshoulder killing) permits one to compute survival as a function of temperature for embryos cooled and warmed at various rates. In the experimentally accessible range (0 to -2X), the predicted functional relation agrees well with that experimentally observed for embryos cooled at O.TC/min. Our analysis indicates that cooling at this rate will result in the death of all embryos by -25°C which is several degrees above the highest temperatures at

62

MAZUR,

SCHNEIDER,

which intraembryonic freezing occurs in intact embryos. That latter prediction is confirmed by experiment (Figs. 10 and 15). The usual approach to cryopreservation is to cool cells slowly enough to allow them to osmotically dehydrate close to the water content required for osmotic equilibrium prior to their traversing the intracellular nucleation zone. Cells cooled at such low rates will not undergo intracellular ice formation when they enter the nucleation zone, or if intracellular ice forms, there will be too little to be damaging (16). Myers et al. (24) have shown that “a low enough rate” for permeabilized 12- to 13-h Drosophila embryos is about O.YC/min in D-20 physiological salt solution with or without 1 M DMSO, although even at that low rate some 40% freeze intracellularly and presumably are killed. But our data and computations indicate that even the unfrozen embryos will die at that cooling rate from chilling injury per se, unless chilling injury is substantially mitigated by the presence of high concentrations of cryoprotectants such as those used by Steponkus et al. (28). Consequently, unless other ways can be found to reduce the chilling sensitivity markedly or increase the subzero temperature at which it ceases, the only solution is to “out-race” it by cooling at high rates and hope that a way can be found to prevent intraembryonic ice formation or its lethal consequences, a conclusion drawn by Mazur et al. (20) and later by Steponkus et al. (28). Our computations in Fig. 14 predict that as the cooling rate is raised from l”C/ min to 1000, 10,000, and 100,000”C/min, the median lethal temperature (LT,,) from chilling injury alone will decrease from -28°C to -58, -66, and -74°C respectively (ignoring the contribution of the time spent warming). But the computations also indicate that even if the cooling rate were 100,000”C/min (40 ms to cool from 0 to -7O”C), the chilling injury will eventually catch up and destroy the cells by -76°C. The curves in Fig. 14 and the conclusions

AND

MAHOWALD

above assume that the chilling sensitivity continues to increase as the temperature drops below -25°C and that the activation energies computed from data between 0 and -25°C are applicable to indefinitely lower temperatures. But this assumption must eventually break down; i.e., there must be subzero temperatures below which the rate of inactivation slows and approaches zero. Below - 120°C for example, liquid water becomes glassy and its viscosity rises to such high values (1013 poises) that molecular motion for practical purposes ceases. The critical question is whether the temperature at which chilling injury ceases is high enough or can be made high enough to permit its being attained at achievable cooling rates prior to the embryos being killed during cooling to that temperature. If, for example, the chilling injury ceases abruptly below -6O”C, a cooling and warming rate of some 2OOOW min would suffice for the survival of half the embryos. But if the temperature has to drop another 10°C before chilling injury ceases, survival of half the embryos would require cooling and warming rates some IOto 30-fold higher. Some Complications The curves in Fig. 14 and the underlying numerical analysis for predicting death from chilling as a function of cooling rate and minimum temperature were based for the most part on data from experiments in which embryos were cooled rapidly to given subzero temperatures and held various times. The results in Fig. 15 were based on other experiments in which the embryos were cooled slowly to indicated subzero temperatures and then warmed immediately or within 3 min of attaining those temperatures. Computed survivals based on the approach in Fig. 14 agreed with the experimentally observed survivals. Figure 16 shows the results of somewhat different experiments in which the embryos were cooled at a moderate rate of some 1OWmin

KINETICS

OF

CHILLING

to indicated subzero temperatures, held 1l14 min, and then warmed at about lO”C/ min. When the numerical analyses underlying Fig. 14 were applied to these experimental conditions, the agreement between computed survivals and those experimentally observed for 15-h embryos was not good; i.e., the temperature at which half the embryos were killed was about 10°C above the computed temperature (- 15°C vs - 25°C). The explanation for the discrepancy appears to lie in the findings of Myers and Steponkus (29, which we have largely confirmed (Table 5), that exposing embryos to near 0°C for some 5 or 30 min enables them to survive for longer times at - 15 or - 20°C than those not so “conditioned” at 0°C. In the experiments involving very slow cooling (Fig. 15), the time spent traversing the region near 0°C was presumably sufficient to induce conditioning. In the experiments involving cooling at 10”Clmin from room temperature (Fig. 16), the time spent traversing the region near 0°C was presumably not sufficient. This latter conclusion is supported by the fact that when we purposely induced conditioning by 21-h exposure at 0°C (row 2 of Table 5), survival following cooling at lO”C/min to - 20 and - 25°C was much higher than that of embryos not so conditioned and was comparable to that observed for the very slowly cooled embryos in Fig. 15. Figure 16 also shows a curve for 15-h embryos that were held at 0°C for 2 1 h prior to cooling; yet, despite the presumed conditioning, their susceptibility to subsequent cooling to - 15 to - 25°C was nearly as high as in those not held at 0°C. The explanation probably lies in the fact that the former were warmed back from 0°C to room temperature for several minutes before cooling to - 15 to - 25°C was initiated. Both Myers and Steponkus (25) and we (Table 5, rows 3 and 4) find that a brief intervening room-temperature exposure causes conditioning to be lost. Conditioning chiefly affects the survival

INJURY

IN

DROSOPHILA

63

of embryos subsequently cooled to - 15°C or below, and the data in Fig. 6 for - 15 to - 25°C which formed the principal experimental underpinning for the calculation of activation energies and the computations of survival, were performed on inadvertently conditioned embryos, inadvertent in that the refrigeration was used to ensure constancy of the developmental stage during the day of the experiment. In summary then, predicted survivals for conditioned embryos cooled at given rates to given subzero temperatures, held for given times, and then warmed agree reasonably well with the observed survivals of conditioned embryos subjected to those treatments. Although conditioning lowers by some 10°C the temperature at which chilling injury commences during cooling at a moderate rate and the temperature at which a given fraction of the embryos are killed, it does not appear to substantially affect the slope of the ensuing curve of death vs temperature. Our guess then is that if the computed curves in Fig. 14 were to be based on experiments with unconditioned embryos, one would find the curve for each cooling rate shifted to the left by perhaps 10 degrees. Put differently, conditioning may delay the onset of chilling injury but it does not change the conclusion that very high cooling rates will be required to outrace the rapid increase in the rate of chilling injury as cooling progresses. Of course, the conditioning treatments studied by Myers and Steponkus (25) and by us may not be maximally effective. Perhaps, for example, + 5 or - 5°C might be more effective than 0°C. The comments just made with respect to conditioned vs nonconditioned embryos probably also apply to the differences noted in Fig. 16 and Table 5 with respect to 12-h vs 15-h embryos; namely, the shift in the survival curve for 12-h embryos to somewhat lower temperatures is not likely to be significant in ameliorating the need to achieve extremely high cooling rates to outrun the accelerating chilling injury at low

64

MAZUR,

SCHNEIDER,

temperatures. We note, however, that these differences between 12- and 15-h embryos were not noted in the experiments in Fig. 6 that formed the principal basis of the curves in Fig. 14. Extrapolation to Cryobiological Preservation Our experimental data on chilling injury are for chorionated unpermeabilized eggs subjected to temperatures between 0” and - 25°C. Cryobiological preservation will require that dechorionated eggs be permeabilized and that they be cooled to far lower temperatures. It also will almost certainly require that the embryos contain and be surrounded by high concentrations of glasspromoting solutes to permit vitrification of the cytoplasm and the surrounding medium (27a). There is some direct and inferential evidence that these added requirements for cryopreservation may not appreciably alter the nature and kinetics of chilling injury described here for intact eggs. An experimental study of chilling injury in permeabilized eggs without added cryoprotectants is pre-

AND MAHOWALD

cluded by the fact that permeabilization substantially raises and greatly broadens the temperatures for the onset of intraembryonic ice formation (24); consequently, in the critical range of - 15 to -25°C death from chilling injury cannot be distinguished from death from intraembryonic freezing. But Myers et al. (24) have also shown that with a cooling rate of 4”C/min, the introduction of ethylene glycol suppresses the temperature at which 50% of permeabilized eggs undergo intraembryonic freezing by about 12°C for each unit increase in molarity. Thus, in 2 M ethylene glycol, half the embryos remain unfrozen at - 33°C. Table 6 summarizes data reported by Leibo et al. (12) and Steponkus et al. (28) on the survival of permeabilized embryos cooled at rates between 0.2 and 6”C/min to -20 to -43°C in the presence of 1.5 or 2 M ethylene glycol, or after exposure to 2 M and then 4 M ethylene glycol. The data in Myers et al. (24) suggest that the incidence of intraembryonic freezing is low under these sets of conditions so that the major determinant of survival in Table 6 is probably

TABLE 6 Comparison of Observed Survival Data of Leibo et al. (12) on Permeabilized Embryos in Ethylene Glycol with that Predicted in Figs. 14 and 15 for Intact Eggs Cooling rate (“C/min)

Minimum temperature (“Cl

1.5-2

0.2 to 0.5

-20 -35

0 0