Review Fundamentals of cryobiology in reproductive medicine

Transcription

1 RBMOnline - Vol 9. No Reproductive BioMedicine Online; on web 28 October 2004 Review Fundamentals of cryobiology in reproductive medicine Professor Barry Fuller Professor Fuller was awarded his PhD whilst working in the MRC Division of Low Temperature Biology in London, UK in 1975 for work on organ preservation for transplantation. He has subsequently undertaken research in several areas of cryobiology applied to medicine, including clinical organ perfusion, cryopreservation of hepatocytes, oocytes and embryos, with periods at the Katholieke Universiteit, Leuven, and Michigan State University. He is a UNESCO Professor in the Chair of Cryobiology in the Ukraine Academy of Sciences, Visiting Professor in Cryobiology at the University of Luton, and holds Honorary Appointments in the Department of Obstetrics, University of Wales College of Medicine, and in the Department of Surgical Sciences, Northwick Park Institute, London. He has published widely on topics of low temperature biology both refereed articles and reviews as well as books. Dr Sharon Paynter Dr Paynter gained her PhD in 1990 from Bristol University, UK. Her thesis was entitled Corneal Tolerance of Vitrifiable Concentrations of Cryoprotectant. Her post-doctoral work at Leicester University was on low temperature preservation of pancreatic islets. She was involved with the first clinical transplantation of pancreatic islets to be performed in the UK. Since 1993 she has been working in Cardiff on the cryopreservation of oocytes and ovarian tissue. Her data has enabled the design of cryopreservation protocols in which osmotic stress is minimized and thus post-thaw survival improved. She has also developed protocols for the vitrification of mature murine oocytes producing high levels of development post-thaw. Dr Paynter has acted as expert advisor to the HFEA, the body governing assisted reproduction in the UK, resulting in the granting of the first UK licence for clinical use of cryopreserved human oocytes. Barry Fuller 1,2,3, Sharon Paynter 2 1 Royal Free and University College Medical School, Royal Free Campus, London NW3 2QG; 2 Department of Obstetrics and Gynaecology, Cardiff University Medical School, Heath Park, Cardiff CF14 4XN, UK 3 Correspondence: b.fuller@rfc.ucl.ac.uk Abstract The aim of this review will be to provide a basic understanding of the biophysical processes that accompany the application of cryopreservation in reproductive medicine. The ability to store cells in suspended animation outside the body has become a keystone practice in the development of many modern clinical therapies, and, in fact, the sciences of cryobiology and IVF have developed in parallel over the past 50 years. During this time, some of the underlying principles of the quantitative biophysical aspects of cryobiology have been clarified. Water is the universal biocompatible solvent, but also possesses unique properties for stability of living cells. Whilst low temperatures themselves have defined effects on cell structure and function, it is the phase transition of water to ice that is the most profound challenge for survival. The thermodynamics of dilute aqueous solutions dictate how cells and tissues respond to the freezing process. Current concepts of nucleation, ice crystal growth and solute exclusion from the ice lattice will be discussed to illustrate what cells must negotiate to avoid lethal damage, and the role of cryoprotectants in enhancing recovery. Quantitative formalisms now exist to model and predict how water and solutes move across cell membranes before and during freezing, or how nucleation events will proceed, and these will be outlined. Cryoprotectants have both positive and negative effects on cell function depending on the kinetics of exposure. The concept of tolerable osmotic excursion of cell volume will be discussed, along with the evidence for a pseudo-glassy state for cells during traditional cryopreservation. This will be compared with the recent interest in promoting glassy states in the whole sample using vitrification protocols, outlining the advantages and drawbacks of each approach. Additional methods for controlling ice nucleation have a role to play here, and a brief outline of current technologies will be given. Finally, issues of safety and stability of cryopreserved samples will be set out. 680 Keywords: cryobiology, gamete and embryo cryopreservation, preservation, reproductive technology, vitrification

2 Introduction The ability to store cells in suspended animation outside the body has become a keystone practice in the development of many modern clinical therapies, and none more so than in reproductive medicine. Application of cryopreservation for storage of gametes, embryos, and most recently ovarian tissues, has enabled the logistically acceptable transfer of many of the techniques of in-vitro cell manipulation from laboratory development to routine patient care. In fact, the sciences of cryobiology and IVF have developed in parallel over the past 50 years. Artificial insemination on a large scale only became possible with the pioneering work of Polge et al. (1949), who demonstrated that cells (in this case fowl semen) could be recovered in a viable state from very low temperatures. At about the same time, equally important work suggested that ovarian tissue could be successfully cryopreserved (Deanesly, 1954; Parrott, 1960). These observations remained of only academic note until the recent revival of interest in storage and re-implantation of ovarian tissues in young female patients after cancer therapy. Early work on culture and manipulation of pre-implantation embryos (McLaren and Biggers, 1958) stimulated the search for reliable cryopreservation protocols, which were developed just over a decade later (Whittingham et al., 1972; Wilmut, 1972; Willadsen, 1977). Attempts were even made to study the freezing process in unfertilized oocytes (Smith, 1952), which is now known to be a more challenging developmental stage. These various developments arrived at the beginning of modern cryobiology, when basic understanding of the fundamentals was limited. Over the intervening period, significant progress has been made in comprehension of the biophysical principles involved in successful cryopreservation, and this paper will set current knowledge and highlight remaining areas of uncertainty. Water, ice and their importance to living cells The ability of low temperatures to reduce the rates of chemical reactions has been studied since the development of modern chemistry, and many of the associated mathematical formulations were set out over the past 300 years. For example, the major principles linking temperature, rates of reaction and activation energy of a particular process were formulated at the start of the previous century (Arrhenius, 1915). Given that life essentially exists through the simultaneous interconnection of a plethora of chemical reactions in non-equilibrium states, which are optimized through homeostatic controls (principally in mammalian cells by temperature, energy and oxygen supply), the only hope to achieve long-term preservation is essentially to reduce these reactions to a negligible level. Attempting to inhibit specific reactions is impossible, so the only practical recourse (apart from some recent experimental concepts which will be discussed later) is to reduce the temperature to very low levels to achieve cryogenic suspension of life. The obvious major hurdle to circumvent is the phase transition of water to ice during cooling. It is the nature and location of that event which has a major impact on cell survival. The thermodynamics of dilute aqueous solutions dictate how cells and tissues respond to the freezing process. These have been reviewed in detail in several articles (Franks, 1982; Mazur, 2004; Muldrew et al., 2004), but there are several points that are worth highlighting here. The rather unique structure of the water molecule (for example, the possession of lone pair electrons from the oxygen atom) permits a degree of structural cooperation between adjacent molecules through hydrogen bonding. It also imparts a distinct polar characteristic, resulting (in situations where the water molecules are free to rotate) in the large dielectric constant. These and the open, connected lattice of aqueous water enable a high degree of solvation of the many ions and small charged molecules upon which life depends. The same properties also dictate how water interacts, via the hydration shells, of many important macromolecular structures within living cells (such as proteins or phospholipid bilayers in membranes) to confer biological stability. When ice nucleates, the nascent ice crystals contain structures where central water molecules are hydrogen bonded to four surrounding water molecules, which are repeated throughout the structure. It is the high degree of hydrogen bonding that results in the high heat of fusion, detected as the released latent heat of crystallization. The equilibrium melting temperature of cell solutions (effectively sodium chloride at about 0.15 mol/l) is 0.55 C, but nucleation is a non-equilibrium stochastic event in practical terms, so ice crystals will not form and grow until lower temperatures are reached. Ice formation is invariably through heterogenous nucleation, in which an available surface (such as a small particle in the solution or irregularity in the holding tube) acts to catalyse the event. Nucleation can also be achieved by adding a small ice crystal to the suspension once it has been cooled below the nucleation temperature, and this is the basis of the so-called seeding process often employed in cryopreservation of embryos or oocytes; in this, clamping the sides of the vial with forceps cooled in liquid nitrogen produces a small, local deep cooling which stimulates ice outgrowth into the solution. Once nucleation has initiated at a sufficiently low temperature, water molecules add on to the growing large ice crystals, which is thermodynamically more favourable than producing new multiple nucleation centres. Homologous ice nucleation is an event in which water molecules, in their co-operative small clusters, are present in sufficient size, low temperature and time to act as self-driven ice nucleating centres. The homogenous nucleation temperature of pure water is about 39 C. In practical terms this is an unusual event, which can only be studied under carefully selected conditions. In samples of the size and nature used in conventional cryopreservation, heterogenous ice nucleation occurs well before the homogenous nucleation temperature is reached. Note here that it is nucleation that has been considered; ice crystal growth can be detected at temperatures tens of degrees lower than these. The growth of ice in the system has several consequences for cells. The first, and perhaps dominant effect during slow cooling, is the exclusion of solutes from the growing ice matrix, as the cooperative properties of the water molecules are altered by the imposed restriction of the ice structure. As cooling proceeds, a decreasing volume of solution exists with a growing solute content, such that by the time temperatures approach 25 C (in a sample without added cryoprotectants), the salinity has increased to around 5 molal (Lovelock, 1953). Biological membranes, and the very small fluid spaces within 681

3 682 cells prevent ice crystal growth in slowly cooled samples, and thus the cells experience a severe, cold, osmotic dehydration as they are also excluded from the ice front and internal water is driven by chemical potential gradient to move to the external hyperosmotic environment. In situations where the freezing process has been visualized directly in a cryomicroscope, this dehydration and cell shrinkage is readily apparent (Leibo, 1979). Early explanations of cell death resulting from the dehydration focused on the irreversible effect this would have on cell volume (Meryman, 1970), but subsequent studies revealed that the damage is multi-factorial. The freezing event has been described in biophysical terms by the set of simultaneous equations of Mazur (Mazur, 1963, 1990). The major factors which dictate how a particular cell type will respond to subzero cooling include (i) those describing the osmotic characteristics of the cell (the water permeability of the cell membrane and the surface area/volume ratio of the cell, and the volume of osmotically active cell water), (ii) the change of water permeability with cooling, (iii) the time and temperature of the cooling event, and (iv) the change in vapour pressure and solute contents of intracellular and extracellular solutions as the fraction of ice increases. By measuring and computing these factors into the equations, Mazur has shown that the predictive outcomes during cell cooling can, to a reasonable degree, match the observed changes in cells during subzero cooling (Mazur, 1990). If cooling rates are low, and the water permeability of the cell membrane remains in balance with the changing vapour pressure of solution, the cells can maintain effective osmotic balance with the concentrated ice matrix. If the cooling is faster, such that super-cooled water does not have time to leave the cell and remains in the intracellular compartment, then there is a strong likelihood that fraction of water will nucleate as intracellular ice, commonly associated with lethal cell damage. These are the basic tenets which have been put forward to explain the somewhat counter intuitive empirical observations that for most cells during cryopreservation, an optimum cooling rate for survival (with a fall off in viable recovery both at slower and faster rates of cooling) can be described, which varies across different cell types. This has come to be known as the 2-factor hypothesis for cell injury during cryopreservation (Leibo et al., 1970). In simple terms, if cells are cooled slowly, they are exposed for prolonged periods to a hypertonic medium with a salinity many fold higher than isotonic exposure, (such an osmotic stress can cause irreversible breakdown in membrane structures and destabilize proteins); if cells are cooled rapidly, intracellular ice can be formed. Survival depends on a delicate balance to avoid these events in the period before the cells enter a stable state at very low temperatures (see below). Another way to view this is to assume that survival will be greatest if the cell volume response closely tracks those predicted from the equilibrium freezing curve, so that hypertonic exposure is limited to an essential minimum to avoid retention of super-cooled water within the cell. A schematic of such events is shown in Figure 1. It should be pointed out that the optimal level of dehydration for survival can be achieved by means other than by use of a continuous, slow-cooling regime. If cells can be cooled (at relatively fast rates) but then held at an intermediate subzero temperature (in the region of 30 C) for a period of time when water diffusion is still kinetically possible, cells can achieve optimal dehydration before being plunged to very low temperatures [the so-called two-step cooling method (McGann et al., 1976)]. The difficulty with this approach is being able to determine the variables (intermediate subzero temperature and time) for different cell types in a prospective fashion. As cooling proceeds to lower temperatures, at some point, dictated by the contents of the particular solvent solute mixture, the residual liquids solidify. This is a complex event, as yet not fully described, which incorporates the facts that individual solutes (such as sodium chloride) have defined eutectic temperatures (in this case close to 21 C in simple solution) at which they will precipitate. In biological systems, containing complex mixtures of macromolecular proteins, carbohydrates, lipids and nucleic acids, it is unlikely that such clear eutectic points exist. It is more likely that the mixture will assume a highly viscous amorphous state, acting to kinetically restrict the movement of residual water molecules to ice crystals. The current concept is that this complex mixture probably enters a quasi-glassy state in which molecular motions are inhibited below about 100 C. The term quasiglass has been used because there is evidence that such systems may well contain minute ice nucleation centres, from which ice crystals of significant size can grow if conditions are altered to allow this. Measurements of biophysical events by differential scanning calorimetry in the kinds of samples produced in cryopreservation tend to support this concept (Boutron et al., 1986; MacFarlane et al., 1992), in effect a coexistence within the mixture of low temperature glass and ice matrix. Observations from cryomicroscopy suggest the appearance of such a glassy state within the cells (Rall et al., 1984). Freeze substitution studies in electron microscopy also demonstrate the absence of visible ice crystals under these equilibrium conditions of cooling (Walter et al., 1975), but are unable to provide evidence of the biophysical status of the cell constituents. When considering cryopreservation of tissues (such as ovarian tissue), the growth of inter-cellular ice crystals may contribute to considerable structural disorganization, as demonstrated in freeze substitution experiments on muscle tissue (Hunt et al., 1977). Cryoprotectants promoting cell survival during cryopreservation It has been known for more than a century that solutes such as sugars have a protective effect against freezing injury (for example, from the work of Hans Molisch in the 1890s (reviewed by Stout, 1982) on freeze tolerance in hardy plant tissues). The concept of specifically adding cryoprotectants (CPA) in cell biology followed from the work of Polge et al. on the effects of glycerol on sperm freezing (Polge et al., 1949; Smith and Polge, 1950). CPA have been defined as any additive which can be provided to cells before freezing and yields a higher post-thaw survival than can be obtained in its absence (Karow, 1974). In the following years, several other solutes with CPA activity, including dimethyl sulphoxide (Me 2 SO/DMSO), were investigated (Luyet and Keane, 1952; Lovelock 1954; Lovelock and Bishop, 1959). CPA have been particularly successful at protecting cells against slow cooling injury during cryopreservation. One class of CPA, typified by glycerol, can be seen as small, neutral solutes, often poly-hydroxylated and capable of hydrogen bonding with water, capable of permeating across the cell

4 membrane (in the case of glycerol, at a rather slow rate), and non-toxic during exposure to cells in concentrations of between about 1 5 mol/l, depending on the cell type and conditions of exposure. Lovelock (1954) recognized these various properties and also the stress of high salinity encountered in freezing mixtures, and formulated his theory of colligative action. In this, he proposed that because of the molar depression of freezing point produced by mixtures of solutes in solution, at any given temperature during the cooling process, the cells would experience less salt-induced stress than under the same conditions if no CPA were present. This salt buffering effect would prevent the establishment of critically high solute concentrations in the residual frozen fraction until the system was cooled to the very low temperatures where all molecular activity was inhibited. Whilst these basic properties of colligative CPA have withstood the test of time, other important aspects of CPA activity have been subsequently defined. The first of these was the discovery that certain high molecular weight colloids also possess CPA properties. For example, hydroxyethyl starch (HES) was shown to be useful in cryopeservation of blood cells and a limited range of nucleated cells (Connor and Ashwood-Smith, 1973; Sputtek and Körber, 1991). It has been suggested that such polymers promote non-ideality of freezing behaviour in solutions by hydrogen bonding with water, increasing viscosity and limiting on a kinetic basis the amount of ice, and thus salt concentration, during slow cooling, and possibly by masking nucleation sites in the solution (Franks et al., 1977). However, this is a non-equilibrium situation, and success is best achieved using relatively rapid cooling rates (Sputtek and Rau, 1992), thus not allowing sufficient time for the ice crystals to grow. Given the complexity of biological systems requiring protection for integrated cell survival, it is likely that CPA activities will be mutifactorial. As long ago as 1966, Nash attempted to explain CPA function on a combination of factors including ability to modulate hydrogen bonding, interact with water, achieve a high aqueous solubility, and molecular volume. He also recognized a toxicity factor, which he calculated on the oil/water partition of the CPA. Later studies have examined other properties of CPA, including the ability of the agents to act as stabilizing agents for biomembranes, or proteins (Anchordoguy et al., 1987; Crowe et al., 1990). These studies have collectively demonstrated that under the extreme dehydration during ice formation, the water stabilizing the amphipathic nature of membrane bilayers is progressively removed, resulting in destruction of the bilayer structure, loss of membrane material or fusing of normally separate envelopes of membranes. These collectively lead to rapid loss of membrane permeability and selectivity on thawing, which have been frequently documented as damaging effects of cryopreservation. Other secondary effects of CPA have been attributed to scavenging oxygen free radicals (Fleck et al., 2000) which appear to be produced by some as yet not fully identified biophysical processes associated with freezing. One important feature of high concentrations of CPA, identified by Nash (1966), was their ability to inhibit ice nucleation completely (on visual inspection) during the time course of freezing studies, and this will be discussed further. More recently, a new approach to quantifying CPA toxicity, based on the average water hydrogen bonding of the polar groups within the molecular structures (or qv * ) has been proposed (Fahy et al., 2004). This has the advantage that it can be used to quantify the water binding variable within multi-component mixtures of CPA, such as used in vitrification solutions. Initial Figure 1. Curves depicting water loss during the freezing process (as represented by volume change) from mouse oocytes exposed to 1M dimethyl sulphoxide (as cryoprotectant) and cooled at various rates, computed from the equations of Mazur (1990). The equilibrium curve (EQ) is that predicted for oocytes which loose sufficient water to keep the water potential of their internal compartments in exact balance with that of the surrounding medium. The vertical line at 33 C represents the median temperature at which intracellular ice may be visualized in cells by cryomicroscopy, from the work of Rall et al. (1983). It will be seen that oocytes cooled at rates of 4 C/min and faster are predicted to form significant amounts of intracellular ice (lethal to the cells). This fits with the visual observations and the high mortality of oocytes cooled in this way. Cells cooled at 1 C/min are not exactly on the equilibrium curve, and form small amounts of intracellular ice, but these cells can be rescued from lethal damage if the warming rates are rapid (from Mazur, 1990, by permission of Humana Press). 683

5 684 studies have supported the use of qv * to design less toxic CPA mixtures with enhanced post-thaw recoveries in a limited number of model systems, including mouse oocytes (Fahy et al., 2004), but more detailed work will be necessary to confirm this. Other important CPA functions have been documented by studying some natural systems, where organisms have evolved strategies to survive freezing. These are generally restricted to insects, lower vertebrates, microbes and plants, and current knowledge about these can be found in several recent reviews (Lee et al., 1996; Duman, 2001; Zachariassen and Zachariassen, 2000). These include: ice nucleating agents which stimulate nucleation (Lee et al., 1996), and can be helpful in producing consistent extra-cellular ice crystal formation, facilitating equilibrium cooling and dehydration and avoiding intracellular ice formation; anti-freeze proteins which can mask growing ice nucleation sites (DeVries, 1992) and inhibit ice crystal growth (these may be particularly helpful during re-warming, where ice may grow to damaging proportions from small ice nuclei present at the end of slow cooling; and compatible solutes which include a range of sugars and amino acids (Hochachka and Somero, 2002; Block, 2003), produced during winter hardening in the cold hardy species, and which have a high propensity to stabilize biological structures under threat from denaturation by the high salinity in the residual unfrozen fraction. Some of these compatible solutes, such as the sugar trehalose, are attractive for use in laboratory cryopreservation because of their low toxicity, but suffer from problems of poor permeability in mammalian cells. In reproductive medicine, specific techniques, such as microinjection into oocytes, are starting to be used to investigate their possible CPA activity (Eroglu et al., 2002; Wright et al., 2004). The positive effects of CPA on cryopreservation have so far been stressed, but it is easy to see how such solutes, which permeate intracellular spaces and affect water interactions in general, can also have toxic effects within the highly organized molecular machinery supporting cell life processes. A marked toxicity (both osmotic and chemical) can be detected if the exposure conditions are not optimized (Ashwood-Smith, 1987; Fahy et al., 1990). The osmotic toxicity can be understood in that cells, permeated by high concentrations of CPA (with a membrane permeability several orders of magnitude slower than that for water) during cryopreservation, experience reverse osmotic damage or dilution shock when returned to isotonic conditions after thawing (Fuller and Bernard, 1984). Cells may only tolerate a small degree of over-swelling produced by these conditions and can be lethally damaged if this is exceeded (Armitage and Mazur, 1984). Problems of permeation and dilution of CPA into complex tissues introduce another degree of complexity, where cells are densely packed, and several different cell types are present, as when applying cryopreservation to ovarian or testicular tissues. However, it does seem that by using thin slices of tissues, exposed for times in the region of min, successful cryopreservation can be achieved (Gosden, 2002; Hovatta, 2004) with recovery of some tissue elements (in the case of ovarian tissues, the small primordial and primary follicles; Hovatta, 2003, 2004). Further work will be required to optimize such techniques. The importance of the composition of the carrier solution that contains the added CPA has recently been highlighted by the work of Stachecki, Willadsen and colleagues. Here, the ionic composition of the solution has been manipulated by replacing the greater part of the sodium content with choline (Stachecki et al., 1998a), which may have significant consequences for the development of salt-induced cryopreservation injury. However, there may also be other contributions to the protective effect (such as increase in viscosity at low temperatures), as indicated by survival of choline-exposed oocytes when plunged to liquid nitrogen from relatively high sub-zero temperatures (e.g. from 20 C), than seen in traditional sodium-containing media (Stachecki et al., 1998b; Stachecki and Cohen, 2004). There are also indications of an underlying chemical toxicity of certain CPA, which has proved difficult to fully define. There have been suggestions that agents such as Me 2 SO may bind to protein side chains or enhance disulphide bridge formation (Fahy et al., 1990). In the same study, the authors attempted to identify chemicals that could act as counteracting solutes or toxicity neutralizers, and found a weak positive response to some agents, notably formamide. Other CPA may have a membrane chaotropic effect, resulting in membrane blebbing and disorganization; for example, butane diol was shown to produce membrane blebbing during exposure in mouse oocytes (Todorov et al., 1993). Other studies (on mammalian oocytes) determined that CPA can have direct effects on organelle structures such as microtubules and microfilaments (Johnson and Pickering, 1987), causing disassembly which could be reversed if the CPA exposures were of relatively brief duration, or at lower temperatures. The area of CPA toxicity remains one of considerable uncertainty, and is ripe for re-evaluation using modern molecular techniques. As cryopreservation becomes increasingly important in reproductive biology and banking of genetic resources, there is a growing interest in investigating phenotypic and genotypic responses to the techniques, a topic area which is more highly developed in plant cryopreservation (Harding, 2004), but which must be transferred to reproductive medicine, and the role for CPA toxicity (if any) in these has not yet been clearly identified. For example, Thompson-Cree et al. (2003) have investigated sperm nuclear DNA fragmentation after cryopreservation, and reported an increase in DNA fragmentation, particularly after sow cooling protocols. There was some correlation between lower levels of DNA fragmentation and better pregnancy rates, but once pregnancy was established, there were no detectable differences between use of fresh or frozen spermatozoa when considering the subsequent cumulative embryo scores or numbers of blastomeres. Studies on cell biology allied to cryobiology the role of predictive modelling From the description above, it will be obvious that to be able to approach cryopreservation in a predictive fashion, there are several important physiological properties of cells that need to be studied. These include the permeabilities of the cell membrane to water (the hydraulic conductivity or L p ) and solutes, particularly the cryoprotectants (CPA) used to prevent

6 freezing damage (see later), designated P s, the volume of osmotically active water (the inverse of the non-osmotically active volume of the cell, or V b ), the surface area and volume relationships of the cell type, and the temperature relationship of these various parameters. In practice, these can be calculated from the Kedem Katchalsky formalism (Kedem and Katchalsky, 1958; Mazur, 1990) by measuring the volume changes of cells on exposure to different solutes (both permeating and non-permeating solutes) equations, and from the Arrhenius relationship by performing the studies over a range of temperatures. Full descriptions of this approach can be found in the reviews of Kleinhans (1998) and Mazur (2004), but to demonstrate the inter-relationships of these, the equations are shown here as: dv w+s /dt = L p ART{M n e M n i ) + σ(m s e M s i } where the change in cell water and solute volumes with time resulting from solute flux is computed, with V w+s representing the water and solute volume, A the cell surface area, and M is osmolality of the solution of intracellular (i) and extracellular (e) nature, combining permeating (s) and non-permeating (ns) solutes. The factor σ is introduced to allow expression of the interaction between solute and water during flux. The change in intracellular permeating solute (predominantly, in cryopreservation, the chosen CPA) is expressed by: dn s /dt = (1-σ)(1/2)(M s e + M s i ) dv w+s /dt + P s A(M s e M s i ) where N s is the number of osmoles of solute in the cell and P s is the membrane solute permeability coefficient. It will be seen that by measuring cell volume changes under conditions where known concentrations of non-permeating solutes or permeating solutes are added, and by measuring or assuming intracellular solute concentrations and available volume of water which responds osmotically, these equations can be solved, and from them L p and P s. Cell volume changes can be measured in real time, for example by use of Coulter Counter techniques if large numbers of cells are available for study, as in the case of spermatozoa (Gao et al., 1992), or by direct observation of single units, as in the case of embryos or oocytes (Leibo, 1980; Hunter et al., 1992). An example of such measurements, using immobilized oocytes, is shown in Figure 2, where the volume histories are shown over time during exposure to ethylene glycol at different temperatures (Paynter et al., 1999). Here, the initial rapid decline in volume is predominantly (although not entirely) an expression of L p as water leaves the oocyte in response to exposure to the more slowly permeating ethylene glycol, whilst the later reexpansion is affected mainly by the value of P s, which is lower in value (slower volume recovery) with diminishing temperature. By deriving the values at different temperatures, and plotting an Arrhenius relationship, it is possible to derive activation energies for the different factors. There is a degree of uncertainty when extrapolating the relationship down to very low temperatures, because it is difficult to make measurements in the presence of ice. However, where this has been attempted, there is reasonably good correlation with values derived from above-freezing temperatures (Toner et al., 1990). Refinements in these predictive models continue to be made, and there are debates about particular approaches for deriving the values (Kleinhans, 1998). The question is also often raised Figure 2. Curves depicting mean volume changes of mature mouse oocytes exposed to 1.5 mol/l ethylene glycol at 30 C (white squares), 19 C (black diamonds) and 10 C (black squares) as measured by direct microscopy. The solid lines in each case represent the predicted osmotic response, computed from the membrane permeabilities to solute and water. The initial rapid volume loss represents water outflow in response to exposure to the high concentration of ethylene glycol. The slow recovery in each case represents inward diffusion of the ethylene glycol. In oocytes exposed at lower temperatures, the expected slower inward diffusion of the solute was clearly observed (from Paynter et al., 1999, by permission of Elsevier). 685

7 686 about the suitability of animal models for predicting the responses of human cells. Of course, a blanket answer cannot be given across the range of species and cell types involved, but by judicious choice of the model, it is possible to obtain information which, when cross-checked, can be very close to that obtained with the human. For example, in Figure 3 are shown osmotic response curves for groups of fresh murine or human oocytes, exposed to CPA (in this case, Me 2 SO) under identical conditions of concentration and temperature. It can be seen (unpublished data) that oocytes from both species behave in a manner kinetically indistinguishable when tested statistically. Another area of modelling in cryobiology relevant to reproductive medicine is that for predicting intracellular ice formation. Using information on L p and cooling velocity, coupled with visual observation of intracellular ice formation in mouse oocytes, Toner et al. (1993) produced a predictive model with good agreement to experimental results. This model was used to demonstrate that a theoretically optimized protocol for mouse oocyte cryopreservation could be applied to yield improved post-thaw recovery by avoiding potentiation of intracellular ice formation at high subzero temperatures (Karlsson et al., 1996). Sophisticated formalisms have been developed to predict the rate of ice nucleation within cells (Karlsson et al., 1994), dictated by the temperature and chemical composition of the intracellular solution (determined by the initial conditions and the value for V, the volume of osmotically active intracellular solution). The general ability to fully model all the complex events during cryopreservation is still at an early stage, but as these models achieve a greater degree of surety with the biological realities, they will make a significant contribution to the development of enhanced cryopreservation protocols. One further area of solution physical chemistry that deserves mention is that of the study of the phase diagrams for mixtures of solutes, particularly CPA. It is possible, principally from calorimetric studies, to discern the behaviour of aqueous mixtures of solutes during cooling to very low temperatures as ice forms and the mixture approaches the glass transition (at which the mixture achieves amorphous solidification at high concentration and viscosity at very low temperatures. Using this information, it is possible to predict the concentrations of CPA required to achieve a glassy state during the process of vitrification (see below), during which the solid amorphous state is attained. In Figure 4 are shown the various states of water, ice and amorphous glass for a solution of CPA (in this case Me 2 SO or propane 1,2-diol), plotted in a format known as a supplemented phase diagram displaying both equilibrium and non-equilibrium information (Taylor et al., 2004). A full description of the relationships between solution, ice and the amorphous state can be found in other reviews (Fahy et al., 1984; McFarlane et al., 1992; Fahy, 1998) but a brief summary of the important points will be helpful to the present discussion. The equilibrium freezing curve (T m or, in fact, the equilibrium melting curve) is defined by the concentration of solute (in this case CPA) which is reached at progressively lower temperatures without the presence of any supercooled water. As discussed above, equilibrium freezing is invariably initiated by heterogenous nucleation under conditions of slow cooling compatible with calorimetric measurements, but should this be circumvented by altered cooling conditions, homogenous nucleation will proceed at lower temperatures. At sufficiently high solute concentrations and low temperatures the kinetics of the process become so slow that T h is difficult to detect, or, at very high solute concentrations, inhibited. On further cooling, molecular translational and rotational motion is halted and the system retains an amorphous structure in a glassy state (MacFarlane et al., 1992). This phase transition (T g ) can be detected using physical techniques such as differential scanning calorimetry (DSC) or differential thermal analysis (DTA). The transition is also dependent upon other characteristics of the system, including the cooling rate. With CPA concentrations of about 40%, it is possible to cool samples through the T h curve without apparent freezing, producing what has been termed a doubly unstable glass (Angell et al., 1981). This means that the amorphous system almost certainly contains small ice nuclei which have the potential to grow during warming, producing the somewhat counter-intuitive concept of freezing during warming (Rall et al., 1980), also termed devitrification, and is signified by the curve T d in Figure 3. Only if warming is so fast as to prevent the growth of these ice crystals can this phenomenon be avoided. These concepts may seem esoteric, but since for all known CPA, severe toxicity is encountered when exposing cells to concentrations beyond about 30% w/w, it will be obvious that knowledge of the precise minimum concentration of CPA which is needed to achieve the quasi-glassy state but still avoid significant ice formation during warming, is essential to achieve cell survival. Whilst many advances have been made in the ability to model the cryopreservation process, there still remain sufficient uncertainties (especially about the nature of the intracellular matrix existing at very low temperatures, and the effects of the extreme viscosity changes on movement of water or solute molecules) to restrict the accuracy of current models for truly prospective prediction of cell responses. However, when models have produced unexpected or difficult concepts, these have been seen to reflect novel biological possibilities. For example, work modelling the membrane permeability to water in sperm produced unexpectedly high values for L p (Curry et al., 1994) which were difficult to reconcile with the slow cooling regimes developed empirically for maximum recovery from cryopreservation. However, cryopreservation is performed in the presence of CPA, and it may be that for spermatozoa, these agents are having a significant effect on L p that dictates use of slow cooling. Also, there are differences in values for L p produced using different measuring techniques (Curry et al., 2000). In fact, spermatozoa have recently been shown to be capable of recovery by vitrification techniques without added CPA, using very high cooling rates (Isachenko et al., 2004) that could depend to some degree on this innate high membrane water permeability. (No other reproductive cells have so far been vitrified without addition of CPA.) There continues to be a need for constant cross-talk and re-evaluation between experimentally observed results and model predictions to refine the capabilities to quantify the many facets of biophysical change during cryopreservation.

8 Figure 3. A comparison of osmotic responses between murine (black triangles) and human (black squares) oocytes during exposure to 1.5 mol/l Me 2 SO at 24 C, using the technique described in Paynter et al., Values are shown as means (n = 5) for measurements in each group. It can be seen that the kinetics of water loss (the first, rapid decrease in cell volume), the minimum volumes attained, and the permeation by CPA (the later cell re-expansion) overlap each other in the two species. However, the volume reduction on initial shrinkage was slightly greater with the human oocytes, which, on computation from a larger group of observations resulted in a statistically higher (P < 0.05) permeability to water (L p of 0.70 μm/min per atmosphere) than in the mouse oocytes (0.41 μm/min per atmosphere). However, this is a relatively small difference and as a generality, mouse oocytes can be used as satisfactory models for the biophysical aspects of oocyte cryopreservation research. a b Figure 4. Supplemented binary phase diagrams for aqueous mixtures of (a) dimethyl sulphoxide (Me 2 SO/DMSO), and (b) propane-1,2-diol, showing the principal events and phase changes associated with cooling and heating, combined from previous data (MacFarlane, 1987; Fahy, 1998; Taylor et al., 2004). T m is the equilibrium melting point curve, T h is the homogenous nucleation curve, T d is the devitrification curve, and T g is the glass transition curve. With progressive ice formation on slow cooling (the T m curves), the concentration of solute increases. At sufficiently high solute concentrations and low temperatures, ice formation is kinetically inhibited and the mixtures enter an unstructured glassy state (T g ). At intermediate, high solute concentrations (in the region of 50% w/v), a quasi-glassy state or doubly unstable state (Angell et al., 1981), achieved on cooling, can undergo devitrification (T d ) and ice formation during warming. The stepped arrows above the T m curve in (a) represent a scheme for incremental equilibration of a sample with sufficient cryoprotectant at progressively lower temperatures to avoid ice formation during cooling (Taylor et al., 2004). Reproduced from Taylor et al. (2004), with permission from CRC Press. 687

9 688 Vitrification as a technique for conservation of reproductive cells As discussed above, the recent interest in the technique of vitrification in reproductive biology stems from the avoidance of ice during the cooling to very low temperatures. The concept has been attractive to cryobiologists for some time (Luyet and Gehenio, 1940), but the problem has been in achieving it in a way that does not first kill the cells by extreme CPA toxicity. The vitreous state in this context can be considered as a solidified amorphous liquid state of extreme viscosity at low temperatures, obtained by specific conditions of cooling and solute concentration that inhibit ice crystal nucleation and growth. Full details of the underlying biophysical principles can be found in Taylor et al. (2004). The technique was successfully re-investigated and applied to embryo cryo-conservation in the pioneering studies of Rall and Fahy (1985). As pointed out, the doubly unstable amorphous state can only support viable cells if ice formation is avoided both on cooling and on warming, so research over the past 20 years has focused on attaining the maximum achievable nontoxic mixtures (about 45% w/v) which can be exposed to the cells for a minimum of time and which can be used in such small aqueous volumes, that any tiny ice nuclei which do form, do not have time to grow during warming. For example, sample volumes of less than 1 μl are achieved for vitrification of oocytes or embryos using the cryo-loop technique (Lane et al., 1999), whilst it has been calculated that ice crystal growth of sufficient size to be biologically significant is effectively inhibited (in kinetic terms) below about 40 C (Pegg, 1988). The issues of using mixtures of CPA in vitrification solutions in reproductive medicine to reduce overall toxicity, and enhancement of glass formation by inclusion of polymers, has been reviewed recently (Kasai and Mukaida, 2004; Stachecki and Cohen, 2004). To achieve this, technologies such as plunge on electron microscope grids (Martino et al., 1996), cryoloops (Lane et al., 1999), and open pulled straws (Vajta et al., 1998; Kuleshova et al., 1999) have been developed for storage of embryos and oocytes. These techniques may not only improve percentage of cells recovered. For example, some reports using these new methods conclude that molecular cell structure (such as the meiotic spindle of oocytes) is better preserved (Chen et al., 2000). More recently, vitrification has been applied to sperm (Isachenko et al., 2004). In this case, the interesting observation was made that spermatozoa may be successfully vitrified without addition of CPA as long as the cooling rate was high enough. Additional manoeuvres, such as inclusion of natural antifreeze proteins, have been applied in an attempt to suppress the growth of any small ice nuclei which might be present in the vitrification mixture and cause problems on rewarming (O Neil et al., 1998). This approach is still being evaluated with new, synthetic ice nucleating inhibitors (Taylor et al., 2004). In general, the concept of vitrification is seductively attractive, because samples are handled rapidly and expensive cooling equipment is not required. However, caution should be applied, because unless precise and continual monitoring is made of all the manipulations involved, it is easy to obtain highly variable results between different batches. At a more futuristic level, the concept of cell conservation in the glassy state at ambient temperatures is just starting to be investigated (Puhlev et al., 2001), but here, the challenges are even more formidable and it is expected to be some time before this is applied in reproductive medicine (but see also Kaneko et al., 2003 below). Safety issues in cryopreservation The routine availability of the cryogen, liquid nitrogen, has greatly facilitated the use of cryopreservation techniques across a broad range of applications. However, whilst it is easy to handle and safe when used with appropriate care, there are considerations of safety and cross-infection, which need to be kept in mind. There are obvious dangers when handling cryopreserved specimens that relate to skin burns resulting from touching extremely cold surfaces, or potential explosions of vials or containers caused by rapid expansion of the liquid nitrogen into the gas phase on warming. These problems can be readily overcome by wearing safety gloves and eye protection. Where large volumes of liquid nitrogen are used in multiple containers, consideration must be given to potential oxygen deprivation in the local atmosphere if the liquid cryogen spills and vaporizes. Good ventilation and use of oxygen detectors (set to 18% oxygen) in the storage room minimizes any risk to staff. Another issue is the need to ensure that samples, stored at low temperatures, cannot be infected with agents, especially viral agents, inadvertently released into the storage tanks from other (infected) samples. There has been a report of viral transmission between samples of bone marrow stored in the same liquid nitrogen storage tank (Tedder et al., 1993), which may have arisen from failure of the filling ports on the storage bags. Viruses in general may survive immersion in liquid nitrogen. This has been studied specifically in reproductive medicine where there have been concerns that some types of plastic containers, such as freezing straws, may be susceptible to small leakages of potentially infectious samples during freezing (Letur-Konirsch et al., 2003), due to some steps in the handling procedures, particularly the sealing procedure of the straws. Experiments have demonstrated that virally contaminated liquid nitrogen can certainly transmit the infectious agents to embryo samples in receptacles such as open-pulled straws (which are sometimes used for vitrification methods), whilst sealed cryo-vial retained their barrier capabilities (Bielanski et al., 2000). Thus, as a general rule, staff handling cryopreserved samples should treat them with the same precautions as fresh samples, and remain vigilant about procedures for filling and rewarming the storage containers to avoid cross-contamination. There are currently moves to develop safety straws and improved cryo-vials, which may overcome the problems, but which remain to be fully evaluated. For example, high security ionomeric resin straws have been developed, and when these were tested in comparison with conventional straws during simulated cryopreservation with virally loaded specimens, they exhibited a greater level of safety (Letur-Kornisch et al., 2003). It has also been suggested that storage in the vapour phase (rather than in liquid nitrogen-filled tanks) may reduce chances of cross infection. However, such use of vapour phase storage requires very good temperature monitoring and increased frequency of observation to ensure that there are not risks of inadvertent warming and loss of viability in samples. This may be a particular risk where there is a high frequency of access to the storage tanks.