Reptile freeze tolerance: Metabolism and gene expression q

Transcription

1 Cryobiology 52 (2006) 1 16 Review Reptile freeze tolerance: Metabolism and gene expression q Kenneth B. Storey * Institute of Biochemistry, College of Natural Sciences, Carleton University, 1125 Colonel By Drive, Ottawa, Ont., Canada K1S 5B6 Received 1 August 2005; accepted 21 September 2005 Available online 29 November Abstract Terrestrially hibernating reptiles that live in seasonally cold climates need effective strategies of cold hardiness to survive the winter. Use of thermally buffered hibernacula is very important but when exposure to temperatures below 0 C cannot be avoided, either freeze avoidance (supercooling) or freeze tolerance strategies can be employed, sometimes by the same species depending on environmental conditions. Several reptile species display ecologically relevant freeze tolerance, surviving for extended times with 50% or more of their total body water frozen. The use of colligative cryoprotectants by reptiles is poorly developed but metabolic and enzymatic adaptations providing anoxia tolerance and antioxidant defense are important aids to freezing survival. New studies using DNA array screening are examining the role of freeze-responsive gene expression. Three categories of freeze responsive genes have been identified from recent screenings of liver and heart from freeze-exposed (5 h post-nucleation at 2.5 C) hatchling painted turtles, Chrysemys picta marginata. These genes encode (a) proteins involved in iron binding, (b) enzymes of antioxidant defense, and (c) serine protease inhibitors. The same genes were up-regulated by anoxia exposure (4 h of N 2 gas exposure at 5 C) of the hatchlings which suggests that these defenses for freeze tolerance are aimed at counteracting the injurious effects of the ischemia imposed by plasma freezing. Ó 2005 Elsevier Inc. All rights reserved. Keywords: Turtles; Lizards; Snakes; Cold-hardiness; Winter hibernation; cdna arrays; cdna library; Antioxidant defenses; Iron metabolism; Serpins q Statement of funding: Supported by a grant from the National Sciences and Engineering Research Council of Canada (OPG 6793). * Fax: address: kenneth_storey@carleton.ca. The ability to endure the freezing of extracellular body fluids is an integral part of winter cold hardiness for numerous vertebrate and invertebrate animals living in seasonally cold environments. Among vertebrates, natural freeze tolerance is well-developed in several species of frogs that hibernate on land and the biochemistry and physiology of vertebrate freeze tolerance has been most extensively studied using the wood frog, Rana sylvatica, as the model animal [25,75,81]. A number of reptile species also survive freezing [22,80]. Whereas it is clear that freeze tolerance is an integral part of winter survival for various terrestrially hibernating frog species, there has been much more debate about whether freeze tolerance is an ecologically relevant part of reptile cold hardiness. A number of reptile species can endure brief freezing exposures at mild subzero temperatures with low amounts of body /$ - see front matter Ó 2005 Elsevier Inc. All rights reserved. doi: /j.cryobiol

2 2 K.B. Storey / Cryobiology 52 (2006) 1 16 ice but die when time stretches beyond a few hours, temperature drops much past 2 C, or equilibrium ice content is achieved. This probably means that some peripheral freezing of skin and skeletal musculature is tolerable but ice penetration through the body core, halting blood flow and the vital functions of internal organs, is not. Among the species that fall into this category are wall lizards (Podarcis muralis, P. sicula), garter snakes (Thamnophis sirtalis), boreal adders (Vipera berus) and hatchlings of some turtles including red-eared sliders (Trachemys scripta elegans), map turtles (Graptemys geographica) and snapping turtles (Chelydra serpentine) (Table 1) [1,2,5,11,13,14,16,22,30]. Several other reptile species show much better freeze tolerance including survival for one or more days at subzero temperatures normally encountered in their hibernacula with ice penetration throughout the body cavity and ice contents that reach equilibrium values of over 50% of total body water. For these, freeze tolerance appears to be ecologically relevant to survival in their natural environment. Species in this category include box turtles (Terrapene carolina, T. ornata) [15,19,83], the European common lizard (Lacerta vivipara) [17,87,89] and hatchlings of several turtle species that spend their first winter within the natal nest including painted turtles (Midland and Western subspecies, Chrysemys picta marginata and C. p. belli, respectively), BlandingÕs turtle (Emydoidea blandingii), and the diamondback terrapin (Malaclemys terrapin) [12,22,29,30,85] (Table 1). Adult box turtles, at about 0.5 kg body mass, are the largest known freeze tolerant animals. Statistics on reptile freeze tolerance to date include hatchling C. picta that variously survived freezing for 3 11 days at 2 to 2.5 C [12,66], E. blandingii hatchlings endured 3 days at 3.5 C [29], L. vivipara endured 1 3 days frozen at 3 C [17,88], and T. carolina endured 2 3 days frozen [15,83]. Survival at equilibrium ice contents of 50% of total body water has been reported in many cases [81]. However, despite significant freezing survival by these species, reptiles appear to use a mixture of strategies to get through the winter and the mixture can vary between geographic populations, between years, within different parts of the winter, and depend on the hibernation site, soil moisture characteristics, insulation/snowcover and weather. This Table 1 Assessment of reptile freezing survival Common name Freezing tolerance Best measured freezing survival References Turtle hatchlings Chrysemys picta marginata Midland painted turtle Good 3 11 days at 2 to 2.5 C [12,84] 1 day at 4 C Chrysemys picta bellii Western painted turtle Good 3 7 days at 2.5 C [12,30,64] 3 4 days at 2 C [64,65] Emydoidea blandingii BlandingÕs turtle Good 3 days at 3.5 C [29,30] 7 days at 2.5 C Malaclemys terrapin Diamondback terrapin Good 7 days at 2.5 C [22,30] Terrapene ornata Ornate box turtle Good 3 7 days at 2.5 C [22,30] Graptemys geographica Map turtle Poor 50% survival 3 days at 2.5 C [2,22,30] Chelydra serpentine Snapping turtle Poor 60% survival 3 days at 2.5 C [22,30] Trachemys scripta elegans Red-eared slider Poor 4 h at 4 C [13,30] 60% survival 3 days at 2.5 C Turtle adults Terrapene carolina Box turtle Good 2 3 days at T b as low as 3.6 C [15,19,82] 2 days at 2 C Lizards Lacerta vivipara European common lizard Good 1 3 days at 3 C [17,87,88] 4 C for 2 h Podarcis muralis Wall lizard Poor Up to 2 h at T b 0.6 to 1.0 [14] Podarcis sicula Italian wall lizard Poor 40 min at 3 C, 60 min at 2.2 C [5] Snakes Thamnophis sirtalis Garter snake Poor 3 5 h at 2.5 C, 6 h at 3.3 C [11,16] 48 h at T b 0.8 to 1.2 Vipera berus Boreal adder Poor 2 3 h at 3.1 C [1] Note. Temperatures are ambient in most cases or body temperature (T b ) where indicated.

3 K.B. Storey / Cryobiology 52 (2006) confounds our anthrocentric need to put labels freeze tolerant versus freeze intolerant on species and a rigid categorization may not work for reptiles. Terrestrially hibernating frogs have the highly water-permeable skin of amphibians which dictates two consequences: (a) they must hibernate in moist sites to avoid death by desiccation, and (b) they are highly susceptible to inoculative freezing by contact with environmental ice [80,81]. Furthermore, they have little ability to dig and do not seek refuge deep underground. If ambient temperature at the hibernation site drops below the freezing point of frog blood (about 0.5 C), freezing is virtually unavoidable and, hence, adaptations supporting freeze tolerance are well-developed and prominently expressed in anuran species living in seasonally cold environments [81]. The situation for reptiles is less clear-cut. Reptiles have a less water-impermeable skin that makes them less susceptible to inoculative freezing and may also allow them to winter in drier hibernation sites than could be used by frogs. Hence, a freeze avoidance (supercooling) strategy for dealing with subzero temperatures is realistic in some cases. Indeed, numerous laboratory studies have documented substantial supercooling capacity by reptiles with supercooling points of 10 to 15 C measured in some cases [2,23,59]. Furthermore, apart from hatchling turtles that are restricted to their natal nest over the winter, most other reptiles can modify their cold exposure by situating themselves in thermally buffered hibernation sites (e.g., underground caves, burrows, grass or moss hummocks) where they may not be subjected to microenvironment temperatures below 0 C, despite low subzero ambient air temperatures [3,39,58,60]. However, a mechanism to deal with subzero exposure must always be present in case the choice of hibernaculum turns out to be poor. For example, Bernstein and Black [3] recorded body temperatures of box turtles (T. o. ornata) overwintering in sand dunes and found that 23 animals had body temperatures that never dipped below C over the whole winter but two individuals showed continuous subzero body temperatures ( 2 to 8 C) from late December to mid-february. Variation in weather conditions from year to year can also change the overwintering experience. Nagle et al. [61] reported 45% mortality of hatchling C. picta in a winter when soil temperatures fell to 7 to 9 C without snow cover whereas mortality was <3% in milder winters. Animals may also be unable to prevent inoculative freezing, especially in certain soil types. For example, in laboratory tests, hatchling E. blandingii could remain supercooled at 4 C for 7 days or cool continuously to 14 C in supercooling tests but when placed in moist soil, they cooled to only 1.3 C before freezing [29]. European common lizards illustrate several types of variation that make it hard to assign a single cold hardiness strategy to the species: (a) highland populations show better freezing survival than do their lowland counterparts, (b) adults are more freeze tolerant than juveniles but juveniles are found at a greater vertical depth than adults in the bogs where they hibernate, and (c) viviparous strains show better long term freezing survival than do oviparous strains [17,40,88,89]. Hence, reptile species may use both freeze tolerance and freeze avoidance strategies in whatever combination is needed to optimize their winter survival. Indeed, Grenot et al. [40] retrieved viable individuals of L. vivipara from the same field site in the same winter in both supercooled or frozen states. Supercooling might often be the preferable (default) strategy of reptile cold hardiness as it is the less demanding strategy both physically and biochemically but freeze tolerance provides the animals with the means to survive when environmental conditions cause inoculative freezing. Hence, freeze tolerance has developed as an important facet of cold hardiness for a number of species. The metabolic and gene expression responses that support reptile freeze tolerance are the subject of most of the rest of this review. Winter life of painted turtle hatchlings The ecology, physiology, and biochemistry of winter survival among terrestrially hibernating reptiles has been most extensively studied using hatchling painted turtles, C. picta, as the model animal. Furthermore, the debate over the winter survival strategy used by reptiles freeze tolerance or freeze avoidance has also been most hotly contested by researchers studying this species. Painted turtles lay their eggs in early summer in soil not too far from the edges of rivers or lakes, often on south facing banks. At higher latitudes only one clutch is laid per year and the eggs incubate throughout the summer and then hatch in September. However, instead of digging out of the nest, the hatchlings remain hidden underground over their first winter, living off the remains of a large internalized yolk sac, and emerge in the late spring.

4 4 K.B. Storey / Cryobiology 52 (2006) 1 16 Wintering in the nest appears to be a strategy for minimizing time under high risk conditions (cold weather, low food availability, and susceptible to predation) and delays emergence until conditions are conducive to rapid juvenile growth [35]. The strategy is common to a number of turtle species over both warm and cold regions of their range but, in northern regions, is complicated by nest temperatures that can fall below 0 C and the consequent need for a strategy of cold hardiness. Painted turtles in their second and subsequent years of life overwinter under water where strategies of subzero survival are not an issue but where anoxia tolerance is critical for these lung-breathers. Indeed, freshwater turtles of the Chrysemys and Trachemys genera are the premier facultative anaerobes among vertebrates and they are widely used model species for studying the molecular mechanisms that support natural anoxia tolerance and in medical studies that seek to improve the ischemia/anoxia resistance of mammalian organs [52,53,73,77]. The mechanisms of cold hardiness used by painted turtle hatchlings have been extensively studied by three groups: ourselves, Costanzo and Lee (University of Miami, Ohio) and Packard and Packard (Colorado State University). Work by the Packard laboratory champions the idea that freeze avoidance is the primary means of overwintering and that freeze tolerance makes little or no contribution to winter survival in the natural environment [65]. They point to nest temperatures that can fall to about 10 C for up to 2 weeks at a time at their study sites, whole animal supercooling points that decrease in the field over the autumn months (and in the laboratory with cold acclimation), poor long term freezing survival in laboratory studies at temperatures lower than 2 C, and survival in artificial nest situations of only those animals that did not show a freezing exotherm [63 65]. They also linked improved resistance to inoculative freezing over the autumn with the accumulation of a dense lipid layer in the a-keratin layer of the epidermis of the limbs which are in contact with soil in the nest [91]. However, studies by ourselves and the Costanzo/ Lee laboratory have shown good freeze tolerance of C. picta hatchlings and conclude that adaptive strategies for freeze tolerance are an important part of winter cold hardiness for this species, at least over parts of the species range [12,21,22,85]. Indeed, Ontario hatchlings (from Algonquin park near the northern limit of the species range) that were tested immediately after excavation from nests in the spring showed the least supercooling capacity ( 1 C) and greatest freeze tolerance (11 days at 2.5 C) that has ever been reported for C. picta [12]. Costanzo et al. [21] reported improvement in freeze tolerance over time with few turtles withstanding freezing when sampled from nests just after hatching but most animals sampled later (December to April) readily surviving freezing trials (3 days at 3 C). Our analysis of Ontario turtles also reported the presence of specific proteinaceous nucleators in plasma of cold acclimated hatchlings, a common observance for freeze tolerant species [84], whereas studies by Packard and Packard [65] reported a clearance of nucleators from body fluids over the autumn months, a common part of a freeze avoidance strategy. Studies by Costanzo and Lee acknowledge that both strategies can contribute to hatchling cold hardiness [21,22]; indeed, they found that supercooling capacity, resistance to inoculative freezing, and freeze tolerance all improved during cold acclimation or during the autumn months under field conditions. The thermal minimum for freezing survival appears to be 3 to 4 C but turtles can survive at temperatures as low as 12 C if they remain supercooled although supercooling is not always possible due to contact with ice crystals or ice-nucleating agents (INAs) in nesting soil [21]. Given that the risk of inoculative freezing is strongly influenced by moisture content, texture, and other characteristics of nesting soil as well as yearly variation in ambient temperature and snow cover [21], C. picta hatchlings may need to be prepared to use either freeze tolerance or supercooling strategies to deal with the particular conditions of any given winter. Freeze tolerance is probably of greatest importance during wet winters when the high likelihood of inoculation by environmental ice or INAs precludes supercooling. Hence, painted turtle hatchlings show significant freeze tolerance and they are a valid model with which to investigate the physiological, biochemical, and genomic mechanisms that support vertebrate freezing survival. Apart from wood frogs, they are the second most extensively studied freeze tolerant vertebrate species. We have analyzed the process of whole animal freezing in hatchling C. p. marginata using proton magnetic resonance imaging. Freezing was initiated at a peripheral site (probably due to inoculation via ice forming on the skin) and the freezing front propagated in a directional manner though the turtleõs body with ice forming first in extraorgan spaces such as the brain ventricles and the abdominal space

5 K.B. Storey / Cryobiology 52 (2006) [70]. This is the same pattern as was seen in freeze tolerant wood frogs [71]. Interestingly, in both species, thawing was not directional but melting occurred uniformly throughout the body core and organs melted more rapidly than did the extraorgan ice that surrounded them. Metabolic responses to freezing in freeze tolerant reptiles Natural cold hardiness, including both freeze avoidance and freeze tolerance strategies, typically involves various metabolic adaptations. Two of the common ones are the proliferation of proteins that help the organism to manage ice and the accumulation of high levels of low molecular weight osmolytes that provide colligative antifreeze action to the whole animal in the case of freeze avoidance or to the intracellular milieu in the case of freeze tolerance. What is the status of these in cold hardy reptiles? Ice management is provided by either antifreeze proteins in freeze avoiding species or ice nucleating proteins (or sometimes nonspecific nucleators) in freeze tolerant species. As mentioned previously, supercooling capacity and resistance to inoculative freezing by C. picta hatchlings increases seasonally and with cold acclimation (as does freeze tolerance) [21,24,91]. The seasonal enhancement of supercooling capacity involves elimination of endogenous ice nucleating agents in the feces (nucleators may have been derived originally from ingestion of nesting soil) [20]. However, tests for thermal hysteresis activity in plasma, which would be indicative of the production of specific antifreeze proteins, were negative for both C. picta and L. vivipara [84,89] suggesting that reptiles do not invest in these proteins that are commonly found in coldwater marine fish and terrestrial invertebrates [97]. The plasma of C. picta exhibited considerable ice nucleating capacity that maintained its potency up to a dilution of 1:500 [84]. Both low ph and heat treatments destroyed the activity suggesting that it is proteinaceous. Hence, specific ice nucleating proteins may be present in this species. The dichotomy between an enhancement of resistance to inoculative freezing and the elimination of gut nucleators (both strategies for supercooling) versus the presence of proteinaceous plasma ice nucleators (a strategy for freeze tolerance) is more apparent than real. A freeze tolerant species (or a species that uses either strategy depending on circumstances) benefits from controlling the freezing process, particularly the propagation of ice through its microvasculature. Hence, a reduction or elimination of modes of nonspecific nucleation is not at odds with the use of specific nucleators in selected extracellular compartments to initiate and manage crystal growth in a noninjurious manner. The other common adaptation for cold hardiness that occurs across phylogeny is the seasonal or freeze-responsive accumulation of high concentrations of low molecular weight metabolites that provide colligative action to resist whole body (freeze avoidance) or intracellular (freeze tolerance) freezing. For example, plasma levels of glucose can rise to mm in freeze tolerant wood frogs (control values are 1 5 mm) [81] whereas glycerol levels can reach 2 M or more in freeze avoiding insects. The search for colligative protectants in cold-hardy reptiles has been unimpressive. Polyhydric alcohols such as glycerol are typically insignificant in reptiles (<1 mm), urea accumulates only when individuals are under water stress [21], and the total amino acid pool of plasma is low (2 10 mm) [12,24]. Glucose and lactate levels do rise in plasma and tissues of both supercooled and frozen reptiles but the levels are low overall and not unlike the values seen when the same species are challenged by hypoxia or anoxia [18,30,55,67]. Indeed, anoxia exposure at 4 C resulted in plasma lactate levels in hatchling turtles that were actually 3- to 6-fold higher than in the same species frozen at 2.5 C [30] which argues against a purposeful synthesis of lactate as a cryoprotectant. Glucose levels do not reach values that would provide substantial colligative action. For example, our studies of painted turtle hatchlings found plasma and organ levels of glucose of only about 15 mm in frozen animals [10,12,85] whereas Costanzo et al. [21] reported midwinter net increases in plasma glucose ranging from 2 to 40 mm depending on year and geographic location of nests sampled. Blood glucose levels rose to 25 mm in freeze exposed L. vivipara [89] and to 20 mm in box turtles [83] but species with poor freeze tolerance, garter snakes and T. s. elegans hatchlings, showed a much small increase to only mm glucose in liver [11,13]. Blood and organ lactate reached 5 38 mm in C. p. marginata hatchlings frozen in the laboratory [10,85] and plasma lactate was 2 6 mm in hatchlings sampled from nests in midwinter [21]. Plasma lactate was mm in freeze tolerant hatchlings of C. picta, E. blandingii, and M. terrapin after 3 7 days frozen

6 6 K.B. Storey / Cryobiology 52 (2006) 1 16 at 2.5 C [30] and rose only as high as 7 mm in organs of 24 h frozen box turtles and 10 mm in 4 h frozen T. s. elegans organs [13,83]. Lactate also accumulated in supercooled painted turtle and map turtle hatchlings [2,18,42], apparently as a result of oxygen limitation at subzero temperatures; supercooled painted turtle hatchlings showed a decline in cardiac activity below 0 C with heart beat ceasing at about 9 C [4]. Overall, these amounts of glucose and lactate are not impressive for providing colligative resistance to cell volume reduction during freezing and there is no evidence that the production of these metabolites is been specifically enhanced to provide cryoprotection in response to freezing as is the case for glucose production in freeze tolerant frogs [81]. Furthermore, plasma osmolality changes very little with cold acclimation, freezing or over the winter months in reptiles (except under dry conditions where a loss of body water occurs) [21,83,88] so it is apparent that there are no unidentified osmolytes present that could act as cryoprotectants. However, the relatively low level of cryoprotection offered by colligative osmolytes in hatchling turtles might be augmented by a noncolligative method of cellular water retention. Microscopy of liver slices from C. picta hatchlings that were frozen at 4 C on a directional stage showed a lower relative amount of extracellular ice (36%) than did comparable slices from skeletal muscle or heart (both 61%) [70]. Instead, micrographs of frozen liver slices showed large amounts of water remaining inside hepatocytes, most of it appearing as spherical droplets around dark granules. These granules appeared to be glycogen which is known to be a highly hydrated macromolecule. Hence, it may be that the high glycogen content of hepatocytes is exploited to achieve significant noncolligative resistance to cell water loss during freezing. Although glycogen granules were not prominent in the muscle tissues, significant intracellular water appeared to remain in muscles along the myofibrils which is where glycogen is known to be localized. Hence, this putative noncolligative mechanism of water retention may represent a significant but previously unappreciated mechanism of natural cryoprotection. Indeed, image analysis of the micrographs showed considerably less ice in all three tissues than was predicted from mathematical calculations based on the osmolality of turtle body fluids [70]. Intermediary energy metabolism during freezing has been examined in garter snakes (T. sirtalis) and hatchling C. picta and T. scripta [9,10,13]. Changes in the levels of glycolytic intermediates in liver were consistent with the activation of liver glycogen breakdown as the carbon source for the glucose and lactate accumulated. The amount of glycogen phosphorylase in the active a form rose sharply in liver in response to freezing in painted turtles and box turtles to catalyze glycogenolysis and supply glucose for both endogenous needs and for export [12,83]. Phosphorylase is undoubtedly activated via a b-adrenergic pathway (as also occurs in freeze tolerant frogs) [81] since C. picta liver plasma membranes showed substantial levels of b 2 -adrenergic receptor binding but virtually no a 1 -adrenergic receptors [44]. Changes in the substrate (fructose-6-phosphate) and product (fructose-1,6-bisphosphate) levels of phosphofructokinase (PFK) in liver from frozen animals were consistent with inhibitory control placed on glycolysis at this locus [10]. Similar responses by glycolytic intermediates (elevation of hexose phosphates, little or no change in other intermediates) were seen in liver of T. s. elegans hatchlings and garter snakes [9,13]. Inhibition of PFK is commonly seen in vertebrate liver under conditions where glucose is being produced for export (such as in response to hypoxia/anoxia) and is prominent in wood frog liver during freeze-stimulated glucose synthesis. Inhibition of PFK is also a component of anoxiainduced metabolic rate depression which serves to adjust the rate of glycolytic ATP output in anoxia to match the greatly reduced rates of ATP-consuming processes to achieve a strong net reduction in metabolic rate [73]. For example, adult C. picta and T. s. elegans submerged in cold water (10 C) show a very strong drop in metabolic rate to 10% of the corresponding aerobic rate and this is a key factor in their long term survival during winter hibernation under water [46]. Although it has never been directly measured, a comparable freeze-induced suppression of metabolic rate (responding to the anoxia of the frozen state) is strongly suspected to be an important factor in long term organ viability in the frozen state for all freeze tolerant animals [75]. Antioxidant defense Good antioxidant defenses have been identified as an important component of freezing survival among freeze tolerant frogs [54,81] and the addition of antioxidants is known to improve

7 K.B. Storey / Cryobiology 52 (2006) hypothermic and freezing preservation of cells and tissues in cryomedical applications [8]. Antioxidant defense is critical in situations where oxygen availability varies widely and rapidly. For example, situations of ischemia (interrupted blood flow) in mammals result in metabolic damage due to ATP limitation when oxygen-based metabolism is interrupted but the reintroduction of oxygen after ischemia is equally damaging [8,51,76]. Reperfusion injuries occurring when oxygen is reintroduced result from a burst of reactive oxygen species (ROS) generation that can overwhelm existing antioxidant defenses. Organisms that endure frequent anoxic or ischemic episodes in nature are typically prepared with high constitutive activities of antioxidant enzymes (e.g., superoxide dismutase, catalase, glutathione peroxidase, glutathione S-transferase, and peroxiredoxin), proteins (e.g., thioredoxin), and metabolites (e.g. ascorbate, glutathione) [50,51]. Adult freshwater turtles (T. s. elegans) that are excellent facultative anaerobes and able to survive for as long as 3 months in deoxygenated water at low temperature [53] have the highest antioxidant enzyme activities among cold-blooded vertebrates that have been examined [92,93]. Freeze tolerant frogs (R. sylvatica) that undergo cycles of ischemia/reperfusion with each freeze/thaw event also exhibit high antioxidant enzyme activities, much higher than activities in the same organs of freeze intolerant leopard frogs (Rana pipiens) [54]. The antioxidant enzyme, c- glutamyltranspeptidase, also increased significantly (by 2.5-fold) during freezing in liver of both R. sylvatica and C. picta [43,44] and catalase activity increased strongly in liver of hatchlings of several turtle species in response to either anoxia or freezing exposure [30]. Interestingly, the magnitude of the catalase response by four out of five species was similar under both stresses but the effect was most pronounced in species with low freeze tolerance [30]. This could suggest that constitutive activities in liver of freeze tolerant species were already largely sufficient to deal with oxidative stress during freezing/anoxic excursions. Indeed, inducible defenses are more commonly seen in species that deal with infrequent situations of anoxia/ ischemia exposure [51]. For example, freezing exposure of garter snakes (T. sirtalis), resulted in significant increases in the activities of catalase and glutathione peroxidase in skeletal muscle whereas anoxia exposure strongly increased superoxide dismutase in liver [48]. Signal transduction for gene regulation The implementation of adaptive metabolic responses to deal with the consequences of freezing requires efficient signaling mechanisms that translate environmental signals into metabolic and gene expression responses. For example, signals arising from b-adrenergic cell surface receptors trigger the camp-dependent protein kinase (PKA) in wood frog liver which in turn coordinates the activation of glycogenolysis and the production and export of glucose as the cryoprotectant [81]. The same system appears to act in C. picta liver [44]. The signaling mechanisms involved in freeze-responsive gene expression are still largely unconfirmed in freeze tolerant animals but multiple protein kinases have been implicated. For example, protein kinases C and G appear to mediate the up-regulation of the novel freeze-responsive proteins, FR47 and Li16, respectively, in freeze tolerant wood frogs [75]. Some other freeze responsive genes are known to be under the control of specific transcription factors such as the hypoxia-inducible factor (HIF-1) (see below) so that they may be regulated via well-established signal transduction pathways for mediating low oxygen signals in vertebrate systems. The mitogen-activated protein kinases (MAPKs) are serine/ threonine kinases that play a major role in the regulation of gene expression in cells responding to numerous growth factors, death signals, and environmental stresses [26]. Three MAPK modules are known: the extracellular signal-regulated kinases (ERKs), the c-jun N-terminal kinases or stress-activated kinases (JNK or SAPK), and the p38 kinases [26]. We analyzed responses to freezing and anoxia by the three MAPK modules in both wood frogs (R. sylvatica) and adult and hatchling turtles T. s. elegans [36 38]. ERKs did not respond to freezing in wood frogs, to freezing or anoxia exposures in hatchling T. s. elegans, or to anoxia in adult turtles. JNK activities were not affected in wood frog organs over a 12 h freezing exposure but activities were reduced by 40 50% in turtle liver and heart over a 4 h freeze [36,38]. JNK activity rose during survivable anoxia exposure in tissues of both adult and hatchling turtles; activity peaked after 5 h of anoxic submergence in both cases but fell with longer exposure [36,37]. This suggests that JNKs may have a natural role to play in coordinating gene expression during the hypoxia transition period as animals prepare for full anaerobic function. However, a role for JNKs in freeze-responsive gene

8 8 K.B. Storey / Cryobiology 52 (2006) 1 16 expression does not appear to occur although JNK activity had increased in wood frog organs after thawing which implies that it may act to aid recovery after freezing [38]. The active form of the p38 MAPK increased strongly in wood frog liver and kidney as an early response to freezing [38]. This kinase is well-known to mediate cell responses to a number of environmental stresses including osmotic stress [26] and, as such, is of great continuing interest to us as a potentially central signaling enzyme in freeze-responsive gene activation in freeze tolerant animals. It will be interesting to compare and contrast p38 responses to freeze/thaw and identify the genes up-regulated by p38 in freeze tolerant frogs and turtles to expand our understanding of the genes/proteins involved in natural freeze tolerance. Stress-induced gene expression Protein synthesis is one of the greatest energy consuming activities in all cells (e.g., using about 36% of total ATP turnover in normoxic turtle hepatocytes) [52] for it requires over 4 ATP equivalents per peptide bond formed [62]. Cell systems under stress typically suppress protein synthesis as an early response to energy limitation and our studies of freezing and/ or anoxia tolerant organisms have shown that global suppression of transcription and translation is a critical part of the metabolic rate depression that supports anoxia/freezing survival [75,77,82]. However, against this background of general suppression, our studies in recent years have shown that both freezing and anoxia exposure trigger the up-regulation of a selected small number genes whose protein products appear to address specific functional needs for stress endurance. In particular, we have analyzed freeze-responsive gene expression in wood frogs (R. sylvatica) and anoxia-responsive gene expression in both vertebrates (adult red-eared slider turtles T. s. elegans) and invertebrates (marine snails Littorina littorea) using techniques of cdna library screening and DNA array screening (reviewed in [56,75,77]). Interestingly, our work on these systems, as well as on gene expression in mammalian hibernation [74,76], is highlighting several consistent themes in gene expression that recur across phylogenetic lines and across multiple forms of stress-induced metabolic arrest. As discussed below, our analysis of freeze-responsive gene expression in hatchling C. p. marginata concurs with and extends our understanding of both freeze and anoxia responsive gene expression in other systems. Survival of anoxia or freezing undoubtedly requires a broad suite of protein adjustments addressing multiple cellular needs. For example, one of the known protein responses to anoxia in adult turtles is enhanced expression of selected heat shock proteins in an organ-specific manner [7,68]; these chaperones may aid the long term stability of other proteins over extended periods of hypometabolism. Our initial expectation when we began to search for freeze-responsive genes was that we would find a considerable number of genes whose protein products would be associated with some of the known physiological and biochemical mechanisms that had previously been identified as important to freeze tolerance (e.g., genes for proteins involved in cryoprotectant metabolism and distribution, or in cell volume regulation, etc.). In some cases we were correct (e.g. some genes associated with glucose metabolism are up-regulated by freezing exposure in wood frogs [75]) but in other cases our screening results surprised us by revealing genes/proteins that had never before been associated with freeze tolerance [75]. These suggest that many different and previously unsuspected areas of metabolism require adjustment to create coordinated cell, organ, and organismal survival of freezing. For example, in wood frogs the genes identified to date as being freeze responsive suggest that freezing survival requires alterations to blood clotting capacity, membrane transporters, ribosomal function, adenosine receptor signaling, hypoxia tolerance, natriuretic peptide regulation of fluid dynamics, defenses against the accumulation of advanced glycation end products, and improved antioxidant defenses [75]. Our initial analysis of stress-induced gene expression in turtles used differential screening of cdna libraries to identify anoxia-responsive genes in organs of adult T. s. elegans [6,94]. Interestingly, the results included multiple examples of the up-regulation of genes encoded on the mitochondrial genome during anoxia exposure. Screening of a heart library made from anoxic turtles (20 h submergence in N 2 -bubbled water at 7 C) showed up-regulation of Cox1 that encodes cytochrome c oxidase subunit 1 (COX1) and Nad5 that encodes subunit 5 of NADH-ubiquinone oxidoreductase (ND5) in anoxia, as compared with aerobic controls [6]. Cox1 and Nad5 transcript levels in heart rose within 1 h under anoxic conditions to 4.5- and 3-fold higher than control values, respectively, and remained high over 20 h of anoxia before declining during aerobic

9 K.B. Storey / Cryobiology 52 (2006) recovery. Both genes also responded to anoxia in red muscle, brain, and kidney [6] whereas other mitochondrially encoded genes (Nad4, Cytb that encodes cytochrome b) were anoxia responsive in liver [94]. Furthermore, when Cox1 and Nad5 expression were analyzed in liver of control versus frozen hatchling C. p. maginata, we found that these two mitochondrially encoded genes were also freezeresponsive [6]. This suggests that some gene expression responses to freezing by turtle hatchlings are responses to the anoxia/ischemia that develops due to plasma freezing (a concept that is further supported by the results of DNA array screening reported below). Indeed, this idea was first put forward from our studies of freeze tolerance in wood frogs [75]. Two of the major consequences of freezing are cellular dehydration due to water exit into extracellular ice masses and ischemia due to the freezing of extracellular body fluids. We mimicked each of these stresses by separately exposing wood frogs to either dehydration or anoxia at 5 C and found that some freeze responsive genes were also up-regulated under dehydration stress (fibrinogen subunits, inorganic phosphate carrier, FR10) whereas others responded to anoxia stress (ADP- ATP translocase, FR47, Li16, P0) [28,75,95]. Note that FR10, FR47 and Li16 are freeze-responsive novel proteins not found in other vertebrates and whose functions have yet to be identified whereas P0 is a protein in the 60S ribosomal subunit that is essential for translation. P0 up-regulation also occurred in response to anoxia in freeze intolerant leopard frogs [95] and its levels peaked during periods of low oxygen consumption in diapausing insects [27] which confirms its link with ribosomal function under oxygen-limited conditions. DNA array screening is an extremely powerful technique that is allowing major advances to be made in our understanding of the gene expression responses that underlie animal adaptation to environmental stresses. Array screening has key advantages including (a) thousands of genes, representing hundreds of different cell functions, can be screened simultaneously, (b) coordinated responses by families of genes or by functional groups of genes can be identified, and (c) the breadth of gene response to a particular stress can be viewed, often identifying genes and gene families that have never before been considered as contributing to stress tolerance [79]. We have made extensive use of human 19,000 gene cdna arrays produced by the Microarray Center at the University of Toronto to provide overviews of gene expression changes in multiple animal systems including hibernating mammals, freeze tolerant frogs and turtles, and anoxia tolerant turtles and snails [74 79]. The use of heterologous probing means that cross-reaction is never 100% but after optimizing binding/washing conditions, a high percentage hybridization can typically be achieved. Of the total 19,200 spots on the array, 2% showed an increase in signal intensity of 1.5-fold or more under experimental (freezing or anoxia) conditions compared with controls and 0.45% showed 2.0-fold or greater signal when C. p. marginata tissues were analyzed (for a description of the methodology see the legend of Table 2). Enhanced expression of a variety of genes was seen in response to freezing or anoxia with binding intensities of labeled probe from experimental samples that were 1.5- to 3.5-fold higher than in controls. In particular, our analysis sought commonalities between the two organs examined (heart, liver) and between the two stresses (freezing, anoxia) tested. Selected genes meeting these criteria fell into three categories: (1) iron binding proteins, (2) enzymes of antioxidant defense, and (3) serine protease inhibitors (serpins). These are listed in Table 2 and discussed below. Heterologous array screening requires confirmatory methods to validate the data for the species under study. We have used two methods to do this and in numerous studies have shown that results from heterologous array screening hold up when examined by other methods [75,77,78]. One of these methods is the screening of a species-specific cdna library to find up-regulated genes followed by Northern blotting to assess mrna expression levels in different experimental tissues/conditions. The other is a PCR-based approach involving the design of DNA primers based on a consensus sequence of the gene of interest (put together from available sequences in GenBank), use of the primers to retrieve the species-specific cdna via PCR and then design and use a species-specific probe to evaluate organ-, time-, and stress-specific gene expression via Q-PCR. Western blotting can also be used to directly determine whether the protein products of putatively up-regulated genes are also elevated. For the data on freeze responsive genes we turned to a cdna library that we had constructed from tissues of frozen hatchlings using the Strategene Lambda Zap kit, as described previously [6]. A pooled cdna library was constructed from a combination of five tissues (muscle, heart, liver, kidney,

10 10 K.B. Storey / Cryobiology 52 (2006) 1 16 Table 2 Genes identified by DNA array screening as freeze and anoxia up-regulated in liver and heart of hatchling painted turtles, C. p. marginata Expression ratio, stress:control (fold up-regulation) Liver Heart Iron binding proteins Ferritin heavy chain Ferritin light chain Transferrin receptor Hemoglobin a Hemoglobin b Antioxidant proteins Glutathione peroxidase Glutathione S-transferase M Glutathione S-transferase A Peroxiredoxin Serpins Serpin C Serpin D1 2.3 Serpin G1 1.6 Eggs collected in Algonquin Park, Ontario in June were laboratory-incubated until hatching in September. Hatchlings were transferred to plastic boxes containing damp sphagnum moss and temperature was lowered in stages to 5 7 C where turtles were maintained for 4 months (mean animal mass 4.34 g ± 0.44 SD, n = 43). Experiments done in winter compared three groups. Control animals were sampled directly from the 5 7 C incubator. For freezing, turtles were held at 2.5 C in trays lined with damp paper toweling and were sampled 5 h post-nucleation (based on a mean chilling time before nucleation of 30 min as determined from thermistors attached to the plastrons of some individuals). Mean body temperature was 1.2 C at sampling and ice was evident in the extremities and brain cavity. For anoxia, plastic jars with a layer of damp paper towel on the bottom were set in crushed ice and then flushed with nitrogen gas (introduced via syringe port) for 2 min. Turtles (four per jar) were added and N 2 flushing was continued for 20 min followed by sealing the jars and returning to the 5 7 C incubator for 4 h. Nitrogen flushing was reconnected while individuals were removed for dissection. Excised tissues were frozen in liquid nitrogen and transferred to 80 C for storage. RNA extraction and quality assessment were as described previously [32] and then samples were sent to the Microarray Center of the University Health Network (UHN; Toronto, ON) for screening. RNA samples were labeled by the UHN indirect (amino-allyl) labeling protocol (10 lg total RNA per sample) and hybridized to UHN Hum19k6ss arrays. Images were acquired using the ScanArray4000 (Perkin Elmer) scanner and quantified using ArrayVision (v8.0, Imaging Research). The data were imported into Gene- Traffic (Iobion Informatics) for normalization (LOWESS, subgrid), filtering and visualization. The expression ratio, stress:control, indicates the fold up-regulation indicated by the screening. Hemoglobin a was also identified as freeze responsive in C. p. marginata by cdna library screening. and brain) sampled from C. p. marginata hatchlings given 1, 4, 12 or 20 h freezing at 2.5 C. The pooled method is highly effective for gaining a broad picture of prominently up-regulated genes in control versus stressed tissues, especially when tissue sample size is limiting, as it is with tiny hatchlings. Tissue-specific and time-specific expression of key stress-responsive genes are subsequently sorted out by Northern blotting. One result from the screening of this cdna library (performed as in [6]) was the prominent freeze-responsive up-regulation of the gene encoding the hemoglobin a chain. Northern blots confirmed its up-regulation (prominently in liver) and validated the results from array screening. Iron binding proteins Heart and liver of hatchling C. p. marginata showed enhanced expression during freezing and anoxia of the mrna coding for iron binding proteins including ferritin heavy (H) and light (L) chains, transferrin receptor 2 (TfR2), and hemoglobin a and b chains. Iron is a vital component of many functional proteins in cells but free iron in the ferrous state (Fe 2+ ) participates in the Fenton reaction with hydrogen peroxide and lipid peroxides to generate highly reactive hydroxyl radicals and lipid radicals [45,47]. Hence, intracellular free iron levels are kept low by storing the metal in ferritin, a huge protein consisting of 24 H and L subunits that surround a core of up to 4500 iron atoms locked in a low reactivity ferrihydrite state [45]. In plasma, iron is bound tightly to another protein, transferrin (Tf), and iron uptake into cells involves endocytosis of Tf after docking with the cell surface transferrin receptor (TfR) [45]. Acidification of the endosomes releases the iron and Tf and TfR are recycled. The TfR1 isoform is ubiquitous but TfR2 in mammals is apparently restricted to liver and erythroid cells. Ferritin, Tf, TfR and hemoglobin are induced by hypoxia in mammals and the latter three genes are under the control of HIF-1 [72,86,90]. HIF-1 up-regulates genes whose protein products contribute to two main responses to low oxygen: (a) enhancing capacity for anaerobic ATP production (e.g., up-regulation of glycolytic enzymes and glucose transporters), and (b) enhancing oxygen delivery to tissues (e.g., increased synthesis of red blood cells, stimulation of capillary growth). These gene responses to freezing or anoxia in the hatchlings suggest that an activation of HIF-1 occurs under both stresses.

11 K.B. Storey / Cryobiology 52 (2006) This is also supported by our early studies of the hatchlings that found a freeze-responsive increase in the activity of two glycolytic enzymes known to be HIF-1 responsive (pyruvate kinase, lactate dehydrogenase) in liver and skeletal muscle [85]. Thus, it appears that HIF-1 mediated responses to oxygen limitation are being activated by freezing and anoxia exposures in the hatchlings. This differs from the response by adult turtles which do not show up-regulation of genes under HIF-1 control during anoxic submergence; indeed, a gene coding for an inhibitor of vascular growth is one of the prominently up-regulated genes in adult turtles under anoxia [77]. However, our array screening of freeze-responsive genes in wood frogs showed prominent up-regulation of the HIF-1 a subunit and ferritin light chain [75] indicating consistency in the response to freezing. The differences between the hatchling and adult turtle responses to anoxia probably reflect the constitutive enzyme/protein levels at the two life stages. Metabolic adaptations supporting anoxia tolerance are fully developed in adults that experience frequent hypoxia/anoxic excursions while diving or hibernating underwater whereas anoxia tolerance may be less fully developed in hatchlings that have not yet experienced hypoxic/anoxic excursions. Hence, the hatchlings respond to anoxia or freezing with a rapid up-regulation of genes that will enhance their survival at low oxygen. Ferritin and TfR2 up-regulation under freezing or anoxia (or during recovery) in hatchling turtles would result in increased sequestering of iron. This could help to supply the iron needs for hemoglobin synthesis but may also have another function. In recent studies with marine snails (Littorina littorea) we also found anoxia-responsive up-regulation of ferritin H chain mrna and protein as well as enhanced levels of ferritin mrna in the polysome fraction during anoxia. Hence, ferritin is one of just a few actively translated proteins in anoxia [57]. However, snails do not have hemoglobin and so enhanced iron sequestering in anoxia must have another purpose. Two seem possible: (a) to sequester free Fe 2+ to minimize iron-mediated generation of ROS, particularly when oxygen is rapidly reintroduced into tissues when the anoxic excursion ends, or (b) to store excess iron during hypometabolism when the net rate of biosynthesis of iron-containing proteins is low. For hatchling turtles, the first possibility seems quite likely especially when put together with the simultaneous up-regulation of antioxidant genes in turtle organs, as discussed next. Antioxidant defenses Array screening revealed up-regulation of the genes for several antioxidant enzymes (AOEs) during freezing or anoxia exposure in heart and liver of hatchling C. p. marginata. Under anoxia this included glutathione peroxidase (GPX) isozyme 1 and 4, glutathione S-transferase (GST) isozymes M5 and A2, and peroxiredoxin 1 [79]. With the exception of GPX4, the same genes were up-regulated in response to freezing (Table 2). Array screening also revealed freeze-responsive up-regulation of antioxidant genes in wood frog heart [75] including thioredoxin (a low molecular weight peptide reducing agent), metallothionein (a metal binding protein that may also act directly as an antioxidant), GST and glucose-6-phosphate dehydrogenase, which generates the NADPH needed to maintain reduced pools of glutathione and thioredoxin. The enzymatic activity of GST (and GPX) also increased significantly in both heart and liver of wood frogs during freezing [54] and GST activity rose in heart and liver of adult turtles during recovery after anoxia [93]. This strongly indicates that freeze-responsive changes in mrna expression in hatchling organs, as identified from array screening, will result in functional changes in the activities (protein levels) of these enzymes. Metabolic damage resulting from hypoxia or ischemia in mammals can often be traced to ROS attack on cellular macromolecules. Damage is particularly pronounced when oxygen is rapidly reintroduced to tissues after a hypoxic/ischemic episode. Such reperfusion damage occurs because cellular antioxidant defenses can be overwhelmed by a burst of ROS generation when oxygen levels rise rapidly [41,47,51]. Thus, species that exhibit natural hypoxia/anoxia tolerance require adaptations that deal not only with the problems of the anoxic state (low ATP availability, toxicity of accumulated end products, etc.) [73,77] but also adaptations that deal with the insults caused by the reintroduction of oxygen into the system [41,51]. Anoxia tolerant species have two ways of dealing with oxidative stress during transitions from low to high oxygen availability: (a) maintenance of high constitutive defenses, both AOEs and metabolites, and (b) stress-induced enhancement of antioxidant defenses [50,51]. The latter is typically triggered in the low oxygen state itself and so appears to be an anticipatory response to the need for elevated AOE activities when oxygen levels rise again.