Cryobiology and anhydrobiology of cells

Introduction

Freezing is worse than cold

Further, cryogenic temperatures (those near the boiling point of nitrogen) are not dangerous per se. At liquid nitrogen temperatures, almost all biochemistry and physiology is so slow that nothing (including damage!) happens at those temperatures. Cooling to cryogenic temperatures may kill, and warming up may kill---we'll discuss these later---but little damage occurs in the cryopreserved state.

Freezing, however, is often deadly. On the scale of organs, the formation of ice can cause mechanical damage by expansion, or rupture as pointed ice crystals grow through the tissue. Further, if the ice forms inside the cell, the cell almost always dies. There have been a few recent reports of cells surviving intracellular ice formation (IIF) in vitro, but IIF is usually taken by working cryobiologists as indicating cell death.

Ice is a poor solvent. When ice forms in an aqueous solution, most solutes are excluded from the ice, and remain in a concentrated unfrozen solution. So concentrations become very large and may be toxic. High concentrations of salts affect electrical or ionic interactions, including those that help stabilise the native state of enzymes. The unfolding and denaturation of enzymes is often irreversible. Further, ice and water interact differently with the surfaces of membranes and macromolecules, which is important because these interactions (via the surface tension of water or the hydrophobic effect) are also involved in maintaining the healthy state of the cell's internal anatomy or ultrastructure.

Before we go on to look at freezing damage, consider the question: How to avoid intracellular ice formation? It may be asked and answered at the physiological, ecological and medical levels. Keeping warm is the simplest answer, although this strategy is unavailable to many living things in most environments. Four other strategies are supercooling, freezing point depression, dehydration and vitrification,. These occur to varying extents in both the artificial process of cryopreservation and in environmental freezing. In all cases, the integrity of the cell's plasma membrane is critical. An intact membrane is necessary to prevent extracellular ice from nucleating intracellular ice. But an intact membrane is vital for other reasons too (e.g. maintaining the different compositions of intra- and extracellular solutions) and rupture of the membrane is also widely used as an indicator of cell death.

Supercooling

Figure: A very simplified sketch shows a cell before and after the freezing of its tissue or supporting medium. The elevated osmotic pressure causes a large reduction in the aqueous volume of the cell. Consequently, all of the non-aqueous components are brought into close proximity. In this state, stacks of membranes begin to resemble lamellar phases.

Many species of plants and animals, and especially their seeds and spores, survive freezing temperatures by cellular dehydration. Dehydration increases the osmotic pressure of the intracellular solution (the cytoplasm) which depresses its freezing temperature and promotes vitrification---both inhibit intracellular ice formation. Equilibration with ice at about -20 °C or with an atmosphere of about 80% relative humidity requires that a cellular interior have a composition of about 10 osmolal. Osmolal means roughly 'osmotically effective moles of solute per kg of water'. The amount of dehydration required to achieve this depends on the initial composition. If all solutes were ideal, and if the initial composition were 1 osmolal (a typical value for a plant cell, and a few times higher than that of most animal cells), then only 10% of the initial intracellular water would remain. If the initial composition were 2 osmolal, then 20% would remain. In practice, these water contents are underestimates because the osmotic pressure of many solutions increases more than linearly with concentration, and because of colligative properties of the cellular ultrastructure. Dehydration itself can cause damage, which we discuss later in some detail. The next figure shows the relationship among several variables that depend upon the chemical potential of water, which is a function of the freezing temperature or the atmospheric humidity.

Molecules in a liquid undergo random, Brownian motion. For freezing to occur in a supercooled liquid, the diffusing molecules must spontaneously form a small cluster (called a nucleus or embryo) of molecules that temporarily has a structure similar to that of ice. In a supercooled liquid such clusters form and dissipate rapidly. If however the cluster is larger than some critical size, it becomes energetically favourable for other diffusing molecules to join the structure, and it grows through the sample (crystallising, or freezing). This process is called nucleation and crystal growth. Nucleation can be either homogeneous (as described above, and in the figure below) or heterogeneous, where an impurity (or the container wall) forms a substrate upon which nuclei can grow.

Figure: A simplified sketch of the effects of solutes on the nucleation of freezing. Water molecules and solutes are represented by the symbols w and s respectively. The arrows represent diffusion, and the length of the arrows indicates the speed of diffusion. The large circles represent the critical nucleation radius. In (a), only water is present. For a critical nucleus to form, the water molecules in the volume represented by the circle must spontaneously arrange themselves into a regular ice-like structure. If this regular lattice is larger than critical size, then the crystal will grow. (b) shows the situation in the presence of solutes (the solute:water molar ratio is 1:4). First, solutes increase the viscosity, so diffusion is reduced (hence smaller arrows in (b) than (a)). Second, in order to form a critical nucleus, a volume equal to or greater that the critical radius must be completely free of solute molecules. This is not the case here. As the concentration of solutes increases, this effect becomes stronger, further reducing the chance of nucleation.

The probability of nucleation in a supercooled liquid depends on several factors: it increases with the volume of the sample and the degree of supercooling; it decreases with increasing solution concentration; and it also increases in the presence of impurities that can act as heterogeneous nuclei. A pure liquid in a small volume with no impurities can be supercooled a long way below its equilibrium freezing point. Small volumes (microlitres) of pure water, for example, can be cooled to about -40 °C.

At very low temperatures, the viscosity of a solution rises sharply, and molecular diffusion is reduced. If cooling is fast, then the viscosity rises rapidly, hindering nucleation. If cooling is sufficiently fast, the viscosity can become so large that molecular diffusion is effectively halted, and the probability of nuclei formation becomes negligible. The sample is then said to be a glass or vitreous solid, and the process is called vitrification. A glass is amorphous (unlike a crystal, it has no long range order) but has the mechanical properties of a solid. A material is said to be a glass if its viscosity reaches 10¹⁴ Pa.s (Franks, 1982). A glass is by definition in a state of very long-lived non- equilibrium.

Figure: A simplified and incomplete flow chart for cryopreservation showing some of the steps and some of the dangers. The black rectangles represent cell death. Not shown on this chart is the possibility of vitrification of small volumes in the absence of cryoprotectants, using extremely rapid cooling.

The cytoplasm has a higher solute concentration, because of a combination of dehydration and the addition of cryoprotectants that can permeate through the membrane. This, together with the rapid cooling rate, may together allow vitrification as discussed above. Higher cooling rates allow lower doses of the toxic cryoprotectants, but cooling rates are often limited in practice by heat conduction in the samples being cooled, particularly for macroscopic organs.

Once the cell is vitrified, "suspended animation" is nearly achieved: all processes of metabolism and injury are slowed to almost zero. Provided that the sample avoids mechanical shocks (glasses are brittle!), the next threat it faces is crystallisation during warming. The vitreous state is unstable with respect to ice plus concentrated solution, but it is prevented from achieving the stable state by its very high viscosity. As the temperature rises, the viscosity falls, molecular motion becomes less slow, and water molecules may diffuse and rotate into the configurations required to nucleate ice, or to add to existing nuclei. The chance of large scale ice nucleation and/or growth occurring depends on the time the sample is exposed to lower viscosities while it is below the equilibrium freezing temperature. Thus successful warming should cross this temperature range quickly (Rall et al., 1984). As is the case with cooling, warming rates are usually limited by conduction. Microwave heating has been proposed, but the rapidly changing absorption spectrum makes this more difficult (Baudot, 1997). There is the further problem that rapid but inhomogeneous heating can produce dangerous mechanical stresses in macroscopic tissues.

Figure: A simplified flow chart for freeze-thaw damage in some species that are capable of acclimation including solute accumulation. The black rectangles represent cell death.

A range of symptoms of damage have been reported in the severely dehydrated state for both model membrane systems and the membranes of living cells. (i) In the lipids that are the major molecular component of cell membranes, the gel-liquid crystal phase transition occurs at higher temperatures (e.g. Crowe et al., 1988, Tsvetkov et al., 1989, Koster et al., 1994 and references in these papers). This is important because coexistence of the phases has been associated with reduced semipermeability. (ii) Membranes may undergo topological changes. Electron micrographs show membranes associated with arrays of long cylinders which resemble the inverted hexagonal phase formed by some lipid-water dispersions at very low hydration. In this phase, the water is found in long narrow cylinders on a hexagonal array, each cylinder surrounded by the hydrophilic moiety of the lipids. This geometry renders them unsuitable for forming a semipermeable barrier. Other topological changes have also been reported (Fujikawa 1995, Gordon-Kamm and Steponkus 1984, Uemura et al., 1995, Webb et al., 1993). (iii) Lateral phase separations may occur in the fluid state. At low hydrations, large areas of protein free membrane are observed in electron micrographs. Further, phase separation may occur to produce one phase rich in highly hydrating lipid species, and another rich in weakly hydrating species. The significance here is that the inverted hexagonal phases can most easily be formed by weakly hydrating lipids in the absence of protein.

Osmotic effects

Reduction in mechanical stress

The proviso about partitioning is very important: many solutes, especially polymers, are excluded from the inter-membrane regions. As a result, their osmotic effect may have no direct effect on reducing the membrane stress, and it may even increase it (Koster et al, 2000). Solutes which remain external to the cell or to the vesicles in model systems will also give no direct reduction to membrane stress. At temperatures above freezing, the addition of the solute to the external solution can, in sufficiently high concentration, dehydrate the cell or vesicle sufficiently to increase the stress. Permeating cryoprotectants (such as DMSO) partition readily into the inter- membrane space. In model systems, it is not simple to produce high concentrations of non-permeating solutes in the inter-membrane space, so comparisons among different experiments should be made carefully, unless the inter-membrane concentration, rather than the total sample concentration, is measured.

Yoon et al. (1998) studied lipid-solute-water systems at freezing temperatures and used the nuclear magnetic resonance signal of the water to determine the distribution of solute and solvent between lamellar phases and a concentrated bulk solution phase in equilibrium with ice (they used either D₂O or deuterated solutes). For the small solute molecules (DMSO and sorbitol) they found that the phase behaviour was close to that expected using the effects discussed above and assuming no specific effects. They found that the disaccharides sucrose and trehalose (which have about twice the volume of the others) increased the hydration less than would be expected from their osmotic effects alone. This effect was consistent with a model in which these molecules were excluded from a very thin layer of water closest to the lipids. Yoon et al. found that all of the solutes studied decreased the intra-membrane stress, but that the disaccharides decreased it more than the smaller solutes. This is only one reason why sucrose and trehalose may occur so widely as natural cryoprotectants, however. Other reasons concern vitrification and crystallisation.

In principle, solutes could affect the membrane stress in specific ways. If the solutes bound to the membrane surface, for example, one would expect a modification in the hydration force. One study of inter-membrane forces using the Surface Forces Apparatus found no specific effects on the inter- membrane force due to DMSO, sorbitol or trehalose, however this study was limited for technical reasons to concentrations of about 1 kmol.m^{- 3} (Pincet et al., 1994).