Calculated Optimal Cooling Rates for Ram and Human Sperm Cryopreservation Fail to Conform with Empirical Observations'

Transcription

1 BIOLOGY OF REPRODUCTION 51, (1994) Calculated Optimal Cooling Rates for Ram and Human Sperm Cryopreservation Fail to Conform with Empirical Observations' M.R. CURRY, J.D. MILLAR, and P.F. WATSON 2 Department of Veterinary Basic Sciences, The Royal Veterinary College, London NW1 OTU, United Kingdom ABSTRACT The permeability coefficient to water (p) and its associated activation energy (E.) were measured for ram (8.47 Lm/min/ atm at 25 C, 1.06 kcal/mol) and human (2.89 m/min/atm at 30 0 C, 1.93 kcal/mol) spermatozoa. By use of these figures, predictive water loss curves were calculated, from published equations, for different cooling rates from 100 C/min to C/ min. The calculated curves show that ram spermatozoa cooled at even the fastest rate would be in osmotic equilibrium by C, and human spermatozoa cooled at rates up to C/min would be in equilibrium by C. If the nucleation temperature for spermatozoa is taken to be between -20 C and -30 C, then ram and human spermatozoa cooled at these rates would apparently not exhibit any intracellular freezing. There is a significant discrepancy between these calculated optimal cooling rates and the published empirically derived optimal rates of 50 C/min for ram and 10 C/min for human. The failure of ram and human spermatozoa to conform with the established and previously successful model for prediction of optimal cooling rates suggests that damage sustained at high cooling rates may be unrelated to intracellular ice formation. INTRODUCTION Spermatozoa were first successfully frozen in the 1940's and 50's. Techniques have been optimized on an empirical basis over the ensuing 45 years, but often without proper account of advances in cryotheory [1]. As cooling proceeds and water is removed from the extracellular solution as ice, the cell must lose water at a sufficient rate to remain in equilibrium and avoid an accumulation of supercooled water. Any supercooling will result in intracellular freezing once the nucleation temperature is reached. The relationship between cooling rate and the extent of supercooling within the cell, and hence the likelihood of lethal intracellular ice formation, has been modelled by use of the cell surface area-to-volume ratio, the membrane water permeability coefficient (Lp), defined as the volume of water that will cross unit area of membrane in unit time at atmospheric pressure, and its activation energy (Ea) as a measure of the change in permeability with temperature [2]. With this model the occurrence of intracellular ice at above optimal cooling rates has been successfully predicted in a number of different cell types including yeast, HeLa and Chinese hamster tissue culture cells, human lymphocytes and red blood cells, rye protoplasts, and mouse ova and embryos [3]. Ravie and Lake [4] used Mazur's model with a number of assumed values to examine fowl spermatozoa and calculated that cooling rates up to C/min produced little likelihood of intracellular ice formation. This result was at variance with the empirically derived optimal freezing rate of only 1 C/min. Duncan and Watson [5] calculated an optimal Accepted July 5, Received March 9, 'Supported by the Medical Research Council (G SB) and NATO (CRG920170). 2 Correspondence: Dr. P.F. Watson, Department of Basic Veterinary Sciences, The Royal Veterinary College, Royal College Street, London NW1 OTU, UK FAX: freezing rate for ram spermatozoa having first estimated the ram sperm membrane permeability coefficient (Lp, 0.22 uim/ min/atm) and its activation energy (Ea, Kcal/mol). Duncan and Watson's [5] value for Lp was low compared with Ravie and Lake's [4] value for fowl spermatozoa (1.94 ILm/min/atm) and an earlier estimate [6] for bull spermatozoa (10.4 ptm/min/atm), but although more in line with Lp values calculated for other cell types, it still yielded a theoretically optimal freezing rate (500 C/min) somewhat higher than that determined in practice (50 C/min) [5]. Spermatozoa are difficult subjects for Lp estimation, and the methods used by Duncan and Watson [5] probably led to inaccuracies in some parameters. Watson et al. [7] developed an alternative methodology for measuring Lp and E a and produced estimates of 2.1 and 10.8 m/min/atm for fowl and bull spermatozoa Lp, which corresponded closely to the existing published values for these species; Ea values were low (fowl 4.4 kcal/mol, bull 3.0 kcal/mol). Noiles et al. [8] used a similar methodology for human spermatozoa (Lp 2.4 m/min/atm, Ea 7.48 kcal/mol) and again found that the calculated optimal cooling rate was considerably faster than had been determined in practice. In this paper we have used the improved methodology for ram spermatozoa in an attempt to establish whether there is a real discrepancy between calculated and empirically established optimal cooling rates. We also compared ram and human spermatozoa to see if there is a difference in permeability that might parallel differences in cryosurvival. MATERIALS AND METHODS Semen Collection and Processing Ram semen was obtained by artificial vagina from Friesland rams held at the Royal Veterinary College (London, UK). Only ejaculates with high wave motion and a high per- 1014

2 OPTIMAL COOLING RATES FOR SPERMATOZOA m 60 U 50, Osmolarity (mosm) FIG. 1. Determination of critical osmolarity of ram (solid line) and human (dotted line) spermatozoa. Values represent means (±+ SEM) of replicates from different individuals (n = 5). Data are presented as percentage of those cells intact under isosmotic conditions. centage of motility were used. Human semen was obtained by masturbation from healthy donors; all samples were judged normo-spermic by World Health Organization criteria [9]. Ram semen was diluted 1:2 with PBS (NaCl 160 mm, Na 2 HPO 4 8 mm, NaH 2 PO 4 2H 2 O 2 mm, ph 7.4, 300 mosm). Human semen was washed with a discontinuous Percoll gradient [10] and resuspended in PBS to a final concentration of approximately 100 x 10 6 /ml. Fluorescent Marker Preparation A stock solution of carboxyfluorescein diacetate (5-CFDA; Sigma Chemical Co., St. Louis, MO) was prepared in dimethyl sulfoxide (0.5 mg/ml). Propidium iodide (PI; Sigma Chemical Co.) was dissolved in distilled water (0.5 mg/ml). Stock solutions were stored frozen in 100-RlA aliquots. Spermatozoa were pre-loaded with CFDA by adding stock solution to a known volume of sperm suspension (10 pzg/ml) and allowing the mixture to stand at room temperature for 15 min. Stock PI was added after treatment (see below) to a final concentration of 5 pzg/ml. Critical Hypotonicity The critical hypotonicity is the tonicity at which 50% of cells will swell beyond their maximum volume-to-surface area ratio and hence rupture. Hypotonic solutions were prepared by diluting PBS with distilled water; actual tonicities were measured with a freezing point depression osmometer (3MO; Advanced Instruments Inc., Needham Heights, MA). For ram semen, 5 Ril of sperm suspension, preloaded with CFDA, was added to PBS (1 ml) at each tonicity on a vortex mixer. For human semen, 10 1 of CFDA-loaded sperm suspension was added to PBS (100 Rl) at each tonicity (final tonicities adjusted accordingly). Stock solution of PI was added to each sample as above. Lysis Time CFDA-loaded ram (10 Al) or human (15 Rl) semen was added to 30 mosm PBS (ram 900 Il, human 450 Rl) on a vortex mixer followed at a variable time interval by a 10- strength concentration of PBS (ram 100 Rl, human 50 1l) to return the spermatozoa rapidly to isotonic conditions. Time 0 controls consisted of the addition of the concentrated PBS before the addition of the sperm suspension. Stock PI was added as above. Temperature Dependence of Lysis Time Lysis time experiments were performed with samples and reagents pre-equilibrated to three different temperatures. For the ram, temperatures of 40, 30, and 18 C were used, and for the human, temperatures of 35, 25, and 15 C. For temperatures below 25 0 C, spermatozoa were cooled at a rate of 0.25 C/min with use of a programmable cooling bath to minimize cold-shock damage. Assessment of Cell Survival Cell survival was assessed on the basis of plasma membrane integrity: cells retaining their green CFDA staining over the entire head and tail were classified as plasma membrane-intact; cells having lost their CFDA staining and having red PI-stained nuclei were designated membranelysed [7, 11]. Cells were assessed either by fluorescence microscopy (Olympus BH2, FITC/Rhodamine filter set, minimum of 200 cells) or by flow cytometer (EPICS Profile Analyzer, minimum of cells). Critical Osmolarity RESULTS Ram spermatozoa showed an initial decrease in the percentage of membrane-intact cells between 300 and 150 mosm (Fig. 1). Membrane integrity was maintained at around 45% between 150 and 100 mosm before undergoing a further steep decline; this second decrease in cell survival is considered to be the result of cells exceeding their critical volume (see Discussion). For the purpose of calculating critical osmolarity, 50% lysis was taken to occur at the midpoint between the plateau region and 0% survival, which gave a value for critical osmolarity of 46 mosm (Fig. 1). Human spermatozoa showed no initial decrease in percentage of membrane-intact cells and maintained a high level of survival down to an osmolarity of approximately 100 mosm (Fig. 1). Below 100 mosm there was a precipitous

3 CURRY ET AL U 7; Jc 1I Z!, cc r Time (sec) FIG. 2. Time to lysis for ram spermatozoa exposed to 30 mosm solution at 40 C (dotted line), 30 C (solid line), and 18 C (dashed line). Values represent means (+ SEM) of replicates from different individuals (n = 5). Data are presented as percentage of those cells intact under isosmotic conditions Time (sec.) FIG. 3. Time to lysis for human spermatozoa exposed to 30 mosm solution at 35 C (dotted line), 25 C (solid line), and 15 C (dashed line). Values represent means (+ SEM) of replicates from different individuals (n = 5). Data are presented as percentage of those cells intact under isosmotic conditions. decline in membrane integrity, with 50% lysis occurring at 67 mosm (Fig. 1). Lysis Time Ram spermatozoa. On exposure to 30 mosm PBS at 40 C, survival fell within 1 sec to approximately 55% of the total number of intact cells under isosmotic conditions (approximately 75% of total cell number were intact at Time 0, i.e., under isosmotic conditions, in all cases), and at 30 C, to approximately 45% of cells intact under isosmotic conditions. Thereafter membrane integrity was maintained for around 3 sec, when cell lysis began to occur. As with the critical osmolarity measurements, the initial cell loss was considered to be unrelated to cells exceeding their critical volume, and 50% lysis was calculated by taking cell survival after the immediate loss as 100%. After the initial difference in survival, which was not significant, the two curves followed an identical pattern. The times for 50% lysis were 4.7 sec at 40 C and 4.9 sec at 30 C (Fig. 2). At 18 0 C, the immediate loss of cell integrity was much greater, with only 15% of the cells intact under isosmotic conditions surviving. Thereafter there was a gradual decline in cell survival over the ensuing 7 sec. The time for 50% lysis in this case was 5.4 sec (Fig. 2). Human spermatozoa. At 35 0 C and 25 C there was no immediate loss of membrane integrity, and 100% survival was maintained for 4 sec; thereafter there was a steady decline in survival, with 50% lysis at 4.8 sec at 35 0 C and 5.6 sec at 25 C (Fig. 3). At 15 0 C there was a decline in cell survival to approximately 60% of cells intact under isosmotic conditions, a much less severe effect than that seen with the ram spermatozoa. The 50% lysis time was 6.0 sec (Fig. 3). Calculation of Lp Lp was calculated by use of Equation 4 of Leibo [12] LP= e(vi - V 2 ) + V21n [(V - V,)] I (V v2)j (trnt) The equilibrium water volume is calculated as Isotonic osmolarity ehypotonic osmolarit Hypotonic osmolarity x Isosmotic water volume (V 1) However, equilibrium volume is never reached as the fixed cell surface area cannot accommodate the necessary volume increase, and the cell lyses. Hence, for V 2, the volume at the critical osmolarity is calculated as V 2 = Isotonic osmolarity critical osmolarity N is the number of osmoles of solute within the cell calculated as N = isotonic osmolarity (Osm) x (liters/cell) x cell volume (izm 3 ) t is the time taken to reach critical volume or the lysis time assuming the cells behave as perfect osmometers. Parameters used to calculate L. are listed in Table 1. From xv

4 OPTIMAL COOLING RATES FOR SPERMATOZOA 1017 TABLE 1. Parameters used to calculate L,. Parameter Symbol (unit) Ram Human Equil. water volume Ve (m 3) Iso. water volume V (m 3 ) 17.28* 17.0** Crit. water volume V 2 (m 3 ) Lysis time t (min) Surface area A (m 2) 139.0* 120** Gas constant R (Rm 3 /atm/mol/deg) x x 1012 Temperature (K) Osmoles of solute within cell N (Osmol/cell) 5.2 x x 10-5 *Values taken from Duncan and Watson [5]. **Values taken from Noiles et al. [81. these values, for ram spermatozoa Lp = 8.47 Izm/min/atm (at 25 C); for human spermatozoa Lp = 2.89 um/min/atm (at 30 0 C). Calculation of Ea The change in hydraulic conductivity with temperature was calculated by expressing relative hydraulic conductivity as relative permeability (1/lysis time, taking lysis time at 30 0 C for ram and 25 0 C for human as unity) and plotting the natural log of the relative permeability against temperature (Fig. 4). The temperature coefficient of the hydraulic conductivity is given by the slope of the graph. To obtain a value for Ea in Kcal/mol, the temperature coefficient is multiplied by an appropriate constant 1.65 x 10 5 [13]. For ram spermatozoa, the temperature coefficient of hydraulic conductivity was / 0 C, giving an Ea value of 1.06 Kcal/mol. For human spermatozoa, the temperature coefficient was / C and Ea was 1.93 Kcal/mol Calculation of Predicted Water Loss Curves Equation 10 of Fahy [14] was used to calculate the kinetics of intracellular water loss dv g-kg eb(t-t)ar(t )dT BVw I S1 [ (V/n 5 V, + 1) (V/n 5 V. + 1)2 XLn T x 10-5 T 2 Equation 12 of Fahy [14] was used to calculate the equilibrium curve V= Vwns(Xs - - 1) The parameters used are listed in Table 2. Figures 5 and 6 show the plotted equilibrium cooling curves for ram and human spermatozoa together with the effects on relative cell volume of cooling at 100, 1000, , and C/min. The nucleation temperature for spermatozoa is difficult to obtain, but a figure between -20 C and -40 C has been proposed [15]. Figure 5 shows that ram spermatozoa cooled at C/min are in equilibrium by -7C so would avoid intracellular freezing. Even at a cooling rate of C/min, the cells are back in equilibrium by -20C. As shown in Figure 6, human spermatozoa cooled at C/min have reached equilibrium by -15 C so should avoid any intracellular ice formation I Temp (deg. C) FIG. 4. Effect of temperature on the water permeability of ram (open squares) and human (closed squares) spermatozoa. Regression analysis was used to determine the line of best fit; R 2 was 0.97 for ram and 0.94 for human. DISCUSSION The value of 8.47 jum/atm/min for ram spermatozoa L, is considerably larger than the previous estimate of 0.22 Ilm/atm/min [5]. The difference in values for the ram is a result of large changes in two of the parameters used. Duncan and Watson [5] in calculating critical osmolarity took the 50% lysis point as being in the middle of the plateau region of their curve, thereby using the whole curve. We now believe that the initial loss of cell integrity at relatively high osmolarities is unrelated to cells exceeding their critical volume. Curry and Watson [16] have shown that ram spermatozoa taken to the same final osmolarities in a series of 25 mosm steps rather than in a single step do not show

5 1018 CURRY ET AL. TABLE 2. Parameters used to calculate the equilibrium curve. Parameter Symbol (unit) Ram Human Cell water volume V 0 (m 3 ) Temperature T (C) Moles solute in cell ns (mol) 2.7 x x Partial molar vol H 2 0 V, (m 3 /mol) x x 1012 Constants* I 1 1 S 0 0 Gas constant R (Am 3 /atm/{mol deg}) x x 1012 Cell surface area A (m 2 ) Reference temperature T (C) Temp. coeffic. of K b (/ C) Hydraulic conduct.** Kg(lm/min/atm) Cooling rate C/min variable variable *1 and S are constants determined by the identity of the cryoprotectant and its concentration before freezing. For the purposes of this calculation all solutions are assumed to behave ideally [28]. **Fahy [14] uses the symbol K for hydraulic conductivity, which may be considered here as equivalent to LP. the same cell loss at osmolarities between 300 and 150 mosm but exhibit a survival curve very similar to that of the human spermatozoa (Fig. 1). This loss of membrane integrity appears to be related to the rate at which water enters the cell rather than the final volume attained. These cells exhibiting loss of membrane integrity at relatively high osmolarities would appear to constitute a sensitive subpopulation within the ejaculate since repeated cycles of exposure to a 120 mosm stress does not result in further cell rupture after the initial exposure [16]. The size of this subpopulation reflects the numbers of sperm that fail to survive cryopreservation even under optimal conditions [17]. The true 50% lysis point is therefore at the midpoint of the curve after the plateau region. Using the data presented here, this interpretation of the graph gives a figure for critical os- molarity of approximately half that used by Duncan and Watson [5], 46 mosm rather than 109 mosm. The curve of Duncan and Watson [5] yields a comparable figure if the present logic is applied. Critical tonicity was measured at only a single temperature in these experiments. Noiles et al. [8] observed a small increase in critical tonicity for human sperm between 22 and 8 C; however, the effect on Lp calculation is relatively small, and other membrane temperature effects, evidenced by the rapid loss of membrane integrity of ram sperm at 18 C, are probably more significant. In addition, Duncan and Watson [5] measured lysis time from the time taken for sperm tails to coil and straighten in distilled water. Only a relatively small proportion of the cells were recorded as behaving in this way, and the time of 15.3 sec obtained may represent a small and particularly 1 4 U.) Temp (deg C) FIG. 5. Calculated water loss curves for ram spermatozoa cooled infinitely slowly so as to remain in osmotic equilibrium (E) and at rates of 100 C/min (1), 1000 C/min (2), C/min (3), and C/min (4). I Temp (deg C) FIG. 6. Calculated water loss curves for human spermatozoa cooled infinitely slowly so as to remain in osmotic equilibrium (El and at rates of 100 C/min (1), 1000 C/min (2), C/min (3), and C/min (4).

6 OPTIMAL COOLING RATES FOR SPERMATOZOA 1019 resistant subpopulation within the ejaculate. The figure of 4.9 sec used here still does not represent the entire ejaculate since around 50% of the cells appear to lyse immediately on mixing with the 30 mosm solution, probably the same cells that showed loss of membrane integrity under relatively mild hyposmotic conditions. The value for human spermatozoa L obtained here, 2.89 plm/atm/min, is in good agreement with the previously published value of Noiles et al. [8] of 2.4 ILm/atm/min. The similarity of the human values gives a degree of confidence in our value for ram spermatozoa. Ea was calculated by using the temperature coefficient rather than directly from an Arrhenius plot because temperature coefficient (b) is the value substituted into Fahy's equation [14]. However, Mazur et al. [13] have shown that above -20 C the two methods of calculation yield very similar results. The calculated Ea values are low, reflecting the very small changes in lysis time recorded at different temperatures. The increasing percentage of cells undergoing very rapid membrane rupture (within 1 sec) seen with ram spermatozoa as temperature was decreased means that the lysis time recorded at 18 C was based on only a small percentage of the cells in the ejaculate (18%). This increased sensitivity prohibits the use of lower temperatures to extend the temperature range used to calculate Ea. The comparatively large standard errors associated with the human lysis time values indicate that there is no significant difference in the 50% lysis times at the different temperatures. While these considerations require the absolute quoted values for Ea to be treated with a degree of caution, it is clear that the true value is small and that, over the given range, temperature has little effect on the permeability coefficient. Duncan and Watson [5] found a relatively small discrepancy between their calculated optimal freezing rate and their experimental results, and proposed that a more accurate calculation of Lp and Ea would reconcile the figures. In practice, the optimal cooling rate for ram spermatozoa is C/min [5] and for human spermatozoa 1-10 C/min [18]. The results presented here for ram spermatozoa, far from reconciling the differences between calculated and empirical optimal cooling rates, greatly exacerbated the problem, with a cooling rate of C/min reaching equilibrium by C. For human spermatozoa, the results predict that a cooling rate of C/min would regain equilibrium by -15 C, an even greater discrepancy than the fold difference reported by Noiles et al. [8]. A number of possible explanations have been proposed to account for the observed discrepancy. The parameters necessary to calculate L and Ea are difficult to obtain for spermatozoa and involve exposing the cells to extremely large osmotic stresses. It is possible that the derived values may still be erroneous. However, as figures from different laboratories and for different species accumulate in the literature [4, 5, 7, 8], we are forced to conclude that spermatozoa are extremely permeable cells with very low activation energies, even if there are minor errors in the absolute values calculated. We have calculated that in order to reconcile the predicted and observed optimal cooling rates, Lp would need to be 400 times lower and Ea 10 times higher, changes that greatly exceed any experimental error associated with calculation of the values. Spermatozoa are frozen/thawed in the presence of permeating cryoprotectants that alter cell water volumes and the number of moles of solute within the cell, and may also significantly decrease Lp and increase Ea. There is some evidence for this with human red blood cells, where glycerol lowers the cooling rate required to cause hemolysis by a factor of 10 [19]. Moreover, in mammalian embryos, intracellular cryoprotectant reduces LP by a factor of 2 or more [20], and in human spermatozoa 1 M glycerol reduces L by 33% [21]. None of these recorded effects of cryoprotectant on Lp would be of sufficient magnitude to account for the observed discrepancies. It should also be noted that glycerol concentrations used to freeze spermatozoa are comparatively low compared to those used for other cell types. Lp and Ea may change significantly at temperatures below zero. Noiles et al. have shown a small decrease in L of human spermatozoa at -7 C with a corresponding increase in Ea [8]. There may be greater changes at lower temperatures, but this is difficult to test below the freezing point. Putative changes in L below the freezing point of the extracellular medium may in fact be of little relevance since the very high initial permeability means that the bulk of the cell water will have left the cell before any decrease in permeability occurs. Neither of these hypotheses, the effect of cryoprotectant or the possibility of nonlinear abrupt changes in Lp, can be conclusively discounted at this time, but the magnitude of the changes required would need to be considerable to account for a difference of four orders of magnitude in critical cooling rate of ram spermatozoa. Frozen/thawed spermatozoa have been shown to exhibit the inverted "u"-shaped survival curve with cooling rate [5, 18] previously observed for a number of other cell types [22]. As cooling rate is increased, cell survival also increases up to a maximum, after which any further increase in cooling rate results in a decline [22]. This curve has been interpreted as resulting from two opposing factors affecting cell survival [22]. Increasing cell death at low cooling rates has been attributed to so-called solution effects, while at high cooling rates there is an increasing tendency for lethal intracellular ice formation. However, it now seems questionable that the decline in sperm survival with increased cooling rate illustrated by the right-hand portion of the graph is a result of intracellular ice formation, although there is at present no direct confirmation of the absence of intracellular ice at high freezing rates. In the absence of intracellular freezing, it is not clear what the lethal mechanism might be. A possible mechanism

7 1020 CURRY ET AL. of membrane damage as a consequence of exceeding a critical gradient in osmotic pressure across the membrane has been suggested by Muldrew and McGann [23, 24] whereby the osmotically driven water flux exerts a frictional force on the plasma membrane that, if sufficiently large, will result in membrane rupture. This hypothesis is consistent with our own observation [16] that a subpopulation of ram spermatozoa are sensitive to the rate of water passage into the cell under relatively mild hyposmotic conditions. Although water movements across the plasma membrane are of foremost importance in determining cell survival during a freeze/thaw cycle, it is by no means clear how water actually traverses the membrane. There are two possible routes for water passage, either through the bilayer or via polar transport sites (pores). The data available from other cell types, particularly erythrocytes, kidney proximal tubule, and vasopressin-sensitive tight epithelia, have established criteria to distinguish between these two routes [25]. Compared with pure phospholipid membranes, porous membranes generally have a high osmotic water permeability coefficient but a low activation energy. The ratio of the osmotic water permeability coefficient to the diffusional water permeability coefficient in pure phospholipid membranes is equal to one, but is greater than one where membrane pores are present [25]. Membrane pores are also subject to inhibition by mercurial compounds such as mercuric chloride and p-chloromercuriphenylsulphonic acid [25]. These are empirically derived characteristics for proposed water channels, and there is still little information as to the molecular basis of water transport, although it is generally agreed in the literature that water passage is essentially a passive process [26]. Although we have been unable to demonstrate any inhibitory effect of mercurial compounds on the water permeability of ram spermatozoa (unpublished observations), and similarly Liu et al. [27] have shown no effect on human spermatozoa, the L v and Ea values for spermatozoa are consistent with a porous membrane. It is difficult to account for the high water permeability exhibited by spermatozoa, in terms of their physiology or their biological function. Other cells with comparably high water permeabilities tend either to have functions concerned with water regulation in the urinary system, e.g., renal epithelial cells, or to be exposed to large fluctuations in their osmotic environment, e.g., erythrocytes during their passage through the kidney medulla. Spermatozoa do experience a change in osmolarity on leaving the relatively hyperosmotic epididymis, at the time of ejaculation, but for most species this change would hardly seem sufficient to justify such a large L value. Whatever the reason for this high water permeability, it has profound implications for sperm cryopreservation as evidence increasingly points to water movement across sperm membranes as an important source of cryoinjury. ACKNOWLEDGMENTS The authors thank Dr. P. Mazur (Oak Ridge National Laboratory, Oak Ridge, TN) and Dr. J.K. Critser (Cryobiology Research Institute, Methodist Hospital of Indiana Inc.) for helpful discussion of these results. REFERENCES 1. Watson PF. The preservation of semen in mammals. In: Finn CA (ed.), Oxford Reviews of Reproductive Biology. Oxford: Oxford University Press; 1979: Mazur P. Kinetics of water loss from cells at subzero temperatures. J Gen Physiol 1963; 47: Mazur P. Freezing of living cells: mechanisms and implications. Am J Physiol 1984; 247:C125-C Ravie O, Lake PE. Prediction of ice formation in fowl spermatozoa at particular cooling rates. Cryo Lett 1982; 3: Duncan AE, Watson PF. Predictive water loss curves for ram spermatozoa during cryopreservation: comparison with experimental results. Cryobiology 1992; 29: Drevius L-O. Permeability coefficients of bull spermatozoa for water and polyhydric alcohols. Exp Cell Res 1971; 69: Watson PF, Kunze E, Cramer P, Hammerstedt RH. A comparison of critical osmolarity and hydraulic conductivity and its activation energy in fowl and bull spermatozoa. J Androl 1992; 13: Noiles E, Mazur P, Watson PF, Kleinhans FW, Critser JK. Determination of water permeability coefficient for human spermatozoa and its activation energy. Biol Reprod 1993; 48: WHO. Laboratory Manual for the Examination of Human Semen and Semen- Cervical Mucus Interactions. Cambridge: Cambridge University Press; Mortimer D. Semen analysis and sperm washing techniques. In: Gagnon C (ed.), Controls of Sperm Motility: Biological and Clinical Aspects. Boston: CRC Press; 1990: Garner DL, Pinkel D, Johnson LA, Pace MM. Assessment of spermatozoal function using dual fluorescent staining and flow cytometric analysis. Biol Reprod 1986; 34: Leibo SP. Water permeability and its activation energy of fertilized and unfertilized mouse ova. J Membr Biol 1980; 53: Mazur P, Rall WF, Leibo SP. Kinetics of water loss and the likelihood of intracellular freezing in mouse ova. Cell Biophys 1984; 6: Fahy GM. Simplified calculation of cell water content during freezing and thawing in nonideal solutions of cryoprotective agents and its possible application to the study of "solution effects" injury. Cryobiology 1981; 18: Watson PF, Gao DY, Mazur P, Critser JK. Ice nucleation temperature of spermatozoa and causes of sperm cell loss below 0 C. Cryobiology 1991; 28: Curry MR, Watson PF. Osmotic effects on ram and human sperm membranes in relation to thawing injury. Cryobiology 1994; 31: Watson PF, Noiles EE, Curry MR, Mazur P, CritserJK, Hammerstedt RH. Response of spermatozoa to hyposmotic stress reflects cryopreservation success. 12th Int Congr Animal Reprod. The Hague; 1992: Henry MA, Noiles EE, Gao D, Mazur P, Critser JK. Cryopreservation of human spermatozoa. IV. The effects of cooling rate and warming rate on the maintenance of motility, plasma membrane integrity, and mitochondrial function. Fertil Steril 1993; 60: Mazur P, Leibo SP, Miller RH. Permeability of the bovine red cell to glycerol in hyperosotic solutions at various temperatures. J Membr Biol 1974; 15: Mazur P. Equilibrium, quasi-equilibrium, and nonequilibrium freezing of mammalian embryos. Cell Biophys 1990; 17: Noiles EE, Mazur P, Kleinhans FW, Critser JK Water permeability of human sperm in the presence of intracellular glycerol. In: Proceedings of the Annual Meeting of the Society for Cryobiology. Ithaca, NY: Mazur P, Leibo SP, Chu EHY. A two-factor hypothesis of freezing injury. Exp Cell Res 1972; 71: Muldrew K, McGann LE. Mechanisms of intracellular ice formation. Biophys J 1990; 57: Muldrew K, McGann LE. The osmotic rupture hypothesis of intracellular freezing. Cryobiology 1993; 30: Finkelstein A Water movements through lipid bilayers, pores and plasma mem-

8 OPTIMAL COOLING RATES FOR SPERMATOZOA 1021 branes: theory and reality. In: Distinguished Lecture Series of the Society of Gen- water channel protein in highly water-permeable human spermatozoa. Cryobioleral Physiologists, vol 4. New York: Wiley; ogy 1993; 30: Verkman AS. Mechanisms and regulation of water permeability in renal epithelia. 28. Fahy GM. Analysis of "solution effects" injury: equations for calculating phase Am J Physiol 1989; 257:C837-C850. diagram information for the ternary systems NaCI-dimethyl sulfoxide-water and 27. Liu C, Gao D, Preston GM, Mark L, Critser B, Agre P, Critser JK. Absence of CH1P28 NaCI-glycerol-water. Biophys J 1980; 32: