Freezing or supercooling: how does an aquatic subterranean crustacean survive exposures at subzero temperatures?

Groundwater ecosystems are generally described as very poor in nutrients and oxygen (Spicer, 1998; Malard and Hervant, 1999), and are also characterized by cool but highly buffered temperatures throughout the year [11±1°C in French plains(Ginet and Mathieu, 1968)]. Thus, most of the aquatic crustacean species living in these environments should exhibit stenothermal characteristics(Huey and Kingsolver, 1989). However, we recently found that the hypogean amphipod Niphargus rhenorhodanensis paradoxically survived -2°C for 60 days, whereas they never normally endure such thermal conditions(Issartel et al., 2005a). This ability to survive low-temperature exposure may correspond to a relict adaptation dating back to the European quaternary glaciations(Issartel et al., 2005a; Issartel et al., 2005b; Lefébure, 2005). During that cold palaeoclimate period, sub-zero temperatures presumably occurred,perhaps even in sub-surface groundwater habitats(Tweed et al., 2005). Thus if organisms such as N. rhenorhodanensis did face sub-zero temperatures,they had to avoid freezing by extensive supercooling or tolerate ice formation in their tissues (Salt,1961).

In freeze-avoiding species, the freezing temperature of body fluids, i.e. the lowest temperature they can endure, is depressed (supercooled state)particularly by accumulating large amounts of cryoprotectants [e.g. polyols,sugars, free amino acids or antifreeze proteins (for reviews, see Salt, 1961; Storey, 1997; Ramløv, 2000)]. By their hydrophilic nature, cryoprotectants bind water molecules and reduce the probability of them forming an ice embryo(Ramløv, 2000). In freezing-tolerant ectotherms, ice-nucleating agents are synthesized and trigger nucleation at high sub-zero temperatures(Lee and Costanzo, 1998). This mechanism prevents the extensive supercooling of cells and thus reduces the probability of lethal intracellular freezing(Holmstrup and Zachariassen,1996). Moreover, ice progression results in a strong increase of the extracellular fluid concentration and causes a water loss from the cell to the extracellular compartment(Zachariassen and Kristiansen,2000). In order to prevent the deleterious effects of `freezing'dehydration (causing membrane and protein denaturation) in cells,cryoprotective substances, such as glycerol, trehalose and free amino acids may also be accumulated (Ramløv,2000). In N. rhenorhodanensis, large accumulations of trehalose and amino acids were found during low-temperature acclimation(Issartel et al., 2005b); but until now no studies have accurately investigated whether this subterranean species is freeze tolerant or freeze avoiding.

In addition, some authors have studied the changes in the free water/bound water ratio during low temperature acclimation in cold-hardy ectotherms. The bound water is the water that is so closely associated with cellular or other components in an organism that it is not available to participate in the freezing processes (Hazelwood,1977). The bound water content of the freeze-tolerant larvae of Eurosta solidaginis increased with cold acclimation(Storey et al., 1981), and this was due to changes in water binding by cryoprotectants and macromolecules(mainly glycogen and proteins). In freeze-tolerant species, bound water will not participate in ice formation and this results in non-freezable shells of water surrounding cellular components, protecting them from the denaturation due to freezing dehydration (Storey et al., 1981).

The data dealing with freezing survival in invertebrates are overwhelming,but very few studies have investigated the problem of freezing in aquatic invertebrates (Moore and Lee,1991; Frisbie and Lee,1997; Lencioni,2004). When water from aquatic environments freezes, the physical constraints differ significantly from terrestrial ones: aquatic invertebrates may be subjected to anoxia or mechanical stress due to external ice(Frisbie and Lee, 1997; Lencioni, 2004). Moreover,contact with external ice may trigger ice growth inside the body, which strongly increases the probability of freezing occurring. As a result,supercooling in aquatic invertebrates may not be a likely strategy(Frisbie and Lee, 1997).

In this study, we compared the responses of two freshwater amphipod crustaceans, the hypogean (i.e. subterranean) N. rhenorhodanensis and the morphologically close epigean (i.e. surface-dwelling) Gammarus fossarum when exposed to subzero temperatures. Thus, in both species we investigated the influence of cold acclimation on (1) the supercooling point,(2) the freezing resistance by inoculation, (3) the ice contents and (4)biophysical parameters such as the bound water content determined by a non-invasive method.

Specimens of Niphargus rhenorhodanensis (Shellenberg)(subterranean aquatic amphipod, 12-13 mg fresh mass) were collected from an interstitial aquatic environment (Chalamont, Dombes Forest, France,46°0.4′N, 5°10′E), with traps sunk into the sediment. Gammarus fossarum (Koch) (epigean aquatic amphipod, 32-34 mg fresh mass) was collected from a swiftly flowing river (La Verna,Hyères-sur-Amby, France, 45°48′N, 5°17′E) with a net. All species were placed in the dark, in separate tanks kept in thermostated chambers as described elsewhere(Hervant et al., 1997a; Hervant et al., 1997b). The tanks with N. rhenorhodanensis contained clay and stones. Tanks with G. fossarum contained leaves. Both sets of organisms were fed once a week. They were first maintained at 12°C for 15 days. Then they were separated into three groups: the first group was kept at 12°C for 6 months, the second group was acclimated to 3°C for 6 months. The third group was first acclimated to 3°C for 6 months and next acclimated to-2°C for 2 weeks as described by Issartel et al.(Issartel et al., 2005b). For acclimation at -2°C, crustaceans were individually put in 6 ml plastic tubes containing 3 ml of filtered rearing tank water. For all acclimation groups, water was changed twice a week and no physico-chemical parameters changed during acclimation. Food was removed from experimental tanks 1 week before the measurements to ensure that the presence of food in the gut would not affect the results.

The glycogen content was determined by standard enzymatic methods as described by Hervant et al. (Hervant et al., 1995; Hervant et al.,1996). Total proteins were extracted according to published methods (Elendt, 1989; Barclay et al., 1983), and then determined using specific test-combinations. All assays were performed in a recording spectrophotometer (Beckman DU-6) at 25°C. Enzymes, coenzymes and test-combination substrates used for enzymatic assays were purchased from Boehringer (Mannheim, Germany) and Sigma Co. (St Louis, USA).

To determine the crystallisation temperature (T_c) of body fluids, we used a differential scanning calorimeter DSC₇system (Perkin-Elmer). The experiments were conducted using standard hermetically sealed aluminium pans (Perkin-Elmer, 0219-0062) designed for volatile samples. In order to check the insulation of the sealed pans, their masses were measured at the end of the experiments and compared with the masses obtained before the DSC₇ measurements. We used a microbalance (Sartorius, type 1712 001; accuracy ±0.1 mg). A single individual was removed from the rearing tanks and placed briefly on a filter paper to remove excess water as described elsewhere(McAllen and Block, 1997). It was then placed in the pan before sealing and weighing. The pan was placed in the DSC₇ oven manually. Each sample was run against an empty sealed aluminium pan for reference. Temperature and heat flow DSC₇calibration were evaluated from the melting of the ice of deionized water(T=0°C and latent heat of fusion of ice ΔH=333.88 J g^-1) and from the crystallographic transition of cyclohexane to its solid state (T=-87.1°C). Temperature values were found to be reproducible within ±0.5°C. The samples were cooled from 20°C to -15°C at a rate of 1°C min^-1. Thermograms were recorded on a computer (Pentium II), and T_c was obtained using the Pyris 3.7.A software.

To emphasize the possible role of acclimation, the experiment was run on both crustacean species acclimated to 12°C, 3°C and -2°C. The animals were tested out of the water because of the unnatural mechanical stress that ice-filled containers can produce on the organisms(Frisbie and Lee, 1997). Thus,according to the methods described by Frisbie and Lee, individuals were placed in contact with a thermocouple connected to a Consort data logger. Then the animal and the thermocouple were wrapped in a water-saturated strip of paper towel. The wrapped animal was closely fitted into a 5 ml pipette cone. The cone was then lowered into a 15 ml plastic tube immersed in an alcohol-filled low temperature bath the temperature of which was adjusted to -2°C. Ice crystal formation was initiated in the wet paper towel by contact with a metal rod cooled in liquid nitrogen. The inoculative freezing of the water in the paper was verified by observing an exotherm (heat release during the freezing process). Once the temperature had again reached the set temperature (i.e. 2 h after the onset of the exotherm), the structure containing the animal was heated to 3°C. To check the effect of inoculative freezing, control structures were not inoculated, organisms were cooled and reheated in the low-temperature bath at the same time as the inoculated structures. Survival of the organisms was noted 24 h after the end of the experiment.

To determine the ice content of frozen individuals, we used the whole-body calorimetry technique (Layne and Lee,1987; Layne and Lee,1991). The calorimeter consisted of an insulated flask that was imbedded in a block of Styrofoam insulation and fitted with a Styrofoam plug that fitted down into the flask, leaving a space of only about 200 μl at the bottom of the flask. Thawing was done in a volume of 200 μl water for all the animals. A thermocouple was positioned below the water surface and connected to a digital thermometer. The change in water temperature caused by thawing the crustacean was recorded. Calculations of body ice content used experimentally determined values for our system which were: F factor for the calorimeter=1.17, the percentage of body mass that is water for N. rhenorodanensis and G. fossarum is 73.44±0.36% and 76.10±0.38%, respectively (values do not vary with acclimation),specific heat of the dry mass measured by calorimetry (±s.d.)=0.18±0.04 and the melting point of body fluids as estimated from osmolality determinations= -0.54°C for both species. Ice content was expressed in percentage of total body water.

where A_i represents the relative weight of the exponential decay with the time constant T_i. Choosing 12 components with times equally spaced in log scale between 2 and 500 ms gave excellent fits of the data (χ² lower than 2×10^-4) and relatively robust distribution curves with a bimodal shape. The relative weight p of the peak of the short-time constant thus reflects the relative amount of bound water in the specimen. It was checked on one specimen that the interecho time tcp chosen in the experiment did not influence the bimodal aspect of the decay curve and the resulting value of P, as compared to inter-individual variations.

All results are presented as mean ± s.e.m. The intra- and inter-specific differences in metabolite concentrations, T_c, and bound water contents were investigated by a two-way ANOVA. When significant differences were found, the Tukey's HSD post-hoc test was performed. Data were log or square-root transformed to homogenize variances when homoscedasticity was not observed. Statistical analyses were performed with Statitistica 6 (StatSoft Inc., Tulsa, USA).

The hypogean N. rhenorhodanensis had body protein content of 0.134±0.008, 0.133±0.005 and 0.128±0.005 g g^-1FM (fresh mass) when acclimated at 12, 3, and -2°C, respectively(Fig. 1). The surface-dwelling G. fossarum showed quite similar values with protein amounts of 0.139±0.003, 0.139±0.007 and 0.129±0.004 g g^-1FM after being acclimated at 12, 3, and -2°C. No intra- and inter-specific differences in protein content were found (P>0.3).

Whatever the acclimation temperature, body content of glycogen was statistically higher in the hypogean N. rhenorhodanensis than in the epigean G. fossarum (P<0.001).

In N. rhenorhodanensis, the amount of glycogen in the body rose significantly by 21% (P<0.001), changing from 60.8±2.4μmol glycosyl g^-1 FM at 12°C to 73.5±1.5 and 73.3±3.2 μmol glycosyl g^-1 FM when acclimated at 3°C and -2°C, respectively (Fig. 1). In G. fossarum, acclimation at 3°C did not affect the glycogen level (29.8±1.3 μmol glycosyl g^-1 FM at 12°C, and 32.1±1.4 μmol glycosyl g^-1 FM at 3°C),but a significant decrease of 32% (P<0.001) was measured at-2°C (21.8±1.3 μmol glycosyl g^-1 FM; Fig. 1).

The T_c values are presented in Table 1. The mean crystallisation temperature was statistically lower in N. rhenorhodanensis than in G. fossarum, whatever the acclimation temperature (P<0.05).

Cold acclimation induced a significant increase in the T_c in both species: the T_c rose by 37%(P<0.001) in N. rhenorhodanensis and by 40%(P<0.01) in G. fossarum.

No survival was observed in either species after thawing.

Survival values are presented in Table 2. In G. fossarum, no survival was observed after inoculation, whatever the acclimation temperature. In N. rhenorhodanensis, no survival was observed after inoculation in 12°C-acclimated individuals. After acclimation at 3°C and -2°C survival rose to 90% and 100%, respectively. No mortality was recorded in control (non inoculated) individuals of either species.

The percentages of crustaceans' body water transformed into ice after inoculation are presented in the Fig. 2. No variation of the ice content was apparent in G. fossarum whatever the temperature (54.40±6.41, 58.52±3.80 and 61.25±7.59% at 12°C, 3°C and -2°C, respectively). N. rhenorhodanensis had ice contents of 62.07±4.30,52.72±4.32 when acclimated at 12°C and 3°C, respectively. The percentage of ice is significantly lower in specimens acclimated at -2°C(40.07±4.1%; P<0.05) than in the control group.

The relative bound water contents in N. rhenorhodanensis and G. fossarum when acclimated to 12°C, 3°C and -2°C is shown in Fig. 3. No variation in bound water was observed in the epigean G. fossarum whatever the acclimation temperature. After being acclimated to 3°C and -2°C, the subterranean N. rhenorhodanensis showed a significant increase in its bound water content (P<0.001).

Traditionally, two main strategies of cold-hardiness are used in ectotherms that are exposed to temperatures below the freezing point of their body fluids(Salt, 1961; Vernon and Vannier, 2002). Ectotherms are either freeze-tolerant or freeze-avoiding species, depending on their ability to survive the formation of extracellular ice(Bale, 1987; Lee, 1989). In freeze-avoiding species, temperature of crystallisation (T_c) is relatively low, often below -10°C, whereas many freeze-tolerant animals have limited abilities to supercool, i.e. ice formation occurs at relatively high temperatures [T_c above -10°C(Bale, 1996; Sinclair, 1999; Sømme, 1999)]. Moreover, cold acclimation generally leads to a T_cdecrease in freeze-avoiding species (Danks,1978; Duman et al.,1991). In our study, amphipods exhibited a relatively high T_c, and cold acclimation induced a T_cincrease in both G. fossarum and N. rhenorhodanensis, but no survival was observed after thawing whatever the acclimation temperature. A higher T_c was also reported in animals acclimated at the lowest temperature in the intertidal copepod Tigriopus brevicornis(McAllen and Block, 1997). In addition, although it was increased, the T_c remained very low (about -20°C) when the animals were acclimated at 0 or 10°C. Such a T_c elevation in G. fossarum and N. rhenorhodanensis when cold acclimated is thus probably non-adaptive but may result from endogenous or exogenous ice nucleating agents.

If temperature falls below 0°C, N. rhenorhodanensis and G. fossarum probably encounter external ice crystals and thus become vulnerable to inoculative freezing (IF). In such cases, supercooling is not likely to work as a strategy (Frisbie and Lee, 1997). We observed distinct patterns in the two crustaceans:after they were in contact with ice crystals, only the 3°C and -2°C acclimated N. rhenorhodanensis survived whereas all others, including the 3°C and -2°C acclimated G. fossarum, died. Specimens that survived IF showed a recovery time of a few hours whereas all control organisms (exposed at -2°C but non-inoculated) were immediately active when reheated at 3°C. This survival is probably linked to the lower ice contents endured by cold-acclimated N. rhenorhodanensis that does not exceed 53% ice within body, unlike the other groups. Contact with external ice induced inoculative freezing of body fluids as the external ice lattice can propagate through a body orifice or directly through the cuticle(Salt, 1963; Lee and Hankinson, 2003). The very thin cuticle characteristic of amphipod gills (allowing gas diffusion)may be the preferential sites from which ice will propagate through the body.

A number of terrestrial arthropods that live in wet habitats require IF in order to survive extracellular ice formation, since if contact with external ice is prevented, they will supercool and die when spontaneous freezing occurs(Lee et al., 1996). Cold-acclimated N. rhenorhodanensis survived freezing if nucleation occurred after an inoculation at high subzero temperature. A similar feature was previously reported in the centipede Lithobius forficatus: it survived freezing only when nucleation was initiated at temperatures of almost-1°C by inoculative freezing (Tursman et al., 1994). In nature, if temperature drops to almost -1°C, N. rhenorhodanensis will experience inoculative freezing before reaching its T_c, as it will be surrounded by ice. Consequently, the tolerance to inoculative freezing seems to be an adaptive trait in these organisms (Tursman et al.,1994).

The physiology of cold tolerance of many arthropods is based on water and its activity at low temperatures. Water content influences the supercooling capacity of freezing-susceptible species, and in freezing tolerant ones a proportion of body water remains unfrozen in order to allow a low level basal metabolism (Block, 2003). Thus,one of the key features that has been rarely studied in arthropods is the capacity to bind water molecules (Storey et al., 1981; Storey,1983). In our study, we used an original non-invasive protocol to determine the relative bound water content in crustacean bodies by proton NMR transverse relaxation measurements performed on the whole live organisms. The hypogean N. rhenorhodanensis contained 25% more bound water when cold-acclimated, whereas no changes occurred in the epigean G. fossarum. Adaptations that increase the amount of bound water are used to ensure that the lethal limit is not exceeded(Storey and Storey, 1989). Our results are in agreement with these findings, as ice contents decrease with increasing bound water in N. rhenorhodanensis. Furthermore, the present results confirm previous work showing that both low-molecular weight compounds (LMWs; mainly polyols and sugars) and high-molecular weight compounds (mainly glycogen and proteins) participate in this phenomenon(Storey et al., 1981; Storey, 1983). Indeed,cold-acclimated N. rhenorhodanensis accumulate both glycogen (this study) and amino-acids and trehalose(Issartel et al., 2005b). An increase in glycogen is rather paradoxical as numerous studies have reported a decrease in glycogen during cold acclimation of ectotherms: glycogen being generally used as a fuel for synthesis of polyols and sugars. In G. fossarum, which exhibited no changes in the amount of bound water,glycogen remained stable. Furthermore, glycogen levels are twice as high in N. rhenorhodanensis as in G. fossarum, which may partly explain the larger bound water content found in the former. However, even if the increased glycogen in cold-acclimated N. rhenorhodanensis may be partly responsible for the increased bound water (together with increase of amino acids and trehalose), its function in the freeze tolerance adaptation still remains unclear and needs further investigations.

Storey and Storey (Storey and Storey,1989) reported that the most important mechanism for controlling the freezing process is the accumulation of LMWs. Moreover, total levels of polyols and sugars are usually significantly lower in freezing tolerant species than in freeze-avoiding ones. In N. rhenorhodanensis, we found a significant accumulation of the total free amino acids (from 58.93±3.88 to 98.63±6.89 μmol g^-1 FM at 12°C and -2°C, respectively) and trehalose [from 1.19±1.2 to 19.66±5.2 μmol g^-1 FM at 12°C and -2°C,respectively (Issartel et al.,2005b)]. These findings may also explain the increased bound water found in N. rhenorhodanensis and we may hypothesize that accumulated LMWs are used for controlling the amount of ice in the body rather than for supercooling. On the other hand, LMW concentrations measured in both species are probably too small to involve a decrease of the glycogen content as it is usually observed in cold hardy invertebrates. From an adaptive standpoint, the amount of bound water has been found to vary in direct proportion to cold hardiness (Danks, 1978; Storey et al., 1981, Ring, 1981). By increasing the amount of water-binding micro and macromolecules, a greater fraction of intracellular water can exist as bound water, and therefore the probability of intracellular freezing (which is lethal for organisms) is strongly decreased(Storey et al., 1981; Ramløv, 2000). However,the osmotic water loss from the cell during extracellular freezing exposes the intracellular components to a dramatic dehydration stress(Zachariassen and Kristiansen,2000). It is hypothesised that unfrozen water shells surrounding sub-cellular components could prevent irreversible protein denaturation due to freezing desiccation and cold temperatures(Hazelwood, 1977; Storey et al., 1981).

Thus, the presence of such adaptations in the subterranean N. rhenorhodanensis may explain its survival capacity when exposed to inoculative freezing.

According to the data in the literature, Sinclair(Sinclair, 1999) proposed that freezing tolerance is divisible into four groups according to T_c and lower lethal temperature: partially freeze-tolerant, moderately freeze-tolerant, strongly freeze-tolerant and freeze-tolerant. Partially-freezing tolerant species survive the conversion of a small proportion of their body water into ice, but do not survive if ice formation reaches an equilibrium at or above the T_c, which is visually represented by the total completion of the exotherm at a given temperature (Sinclair, 1999). The epigean crustacean G. fossarum that does not survive nucleation whatever its acclimation, is a freezing intolerant species; it belongs to the chill-susceptible species (species that die after brief chilling to high sub-zero temperatures). The subterranean crustacean N. rhenorhodanensis exhibited responses to subzero temperatures similar to those found in freeze-tolerant species that survive IF. However, it appears from our results that N. rhenorhodanensis can neither be classified as a partially freeze-tolerant nor as a moderately freeze-tolerant species since: (i) survival was observed after the total completion of the exotherm which is lethal in partially freeze-tolerant individuals(Sinclair, 1999), and (ii) no survival was observed after the animals reached the T_c,whereas moderately freeze-tolerant species survive after T_c is reached. Thus, like numerous arthropods showing similar characteristics (see Lee et al.,1996), N. rhenorhodanensis seems to belong to a still indeterminate category.

The presence of such complex adaptations in an organism that currently never endures cold during its life cycle seems at first very paradoxical. However, from recent biogeographical and phylogenetic studies, there are now several proofs that the subterranean amphipod N. rhenorhodanensissurvived the quaternary glaciations at the limit or within the nunataks, i.e. the mountain tops surrounded by ice never covered by the glaciers(Lefébure, 2005). Thus,in such palaeo-environments, freshly-melted water coming from the glacier [at temperatures near, or even just below 0°C(Tweed et al., 2005)] may have infiltrated the sediment and considerably influenced the subterranean temperatures. As a result, the hypogean crustacean N. rhenorhodanensis may have encountered sub-zero temperatures and ice, and may thus have been subjected to inoculative freezing.

To conclude, our results agree with this evolutionary scenario, and the possible `near-glacial' survival of N. rhenorhodanensis during that period may explain (i) the cold-induced accumulation of cryoprotectants(Issartel et al., 2005b), (ii)the bound water increase, and the resulting inoculative freezing tolerance(this study).

This research was supported by funds from University Claude Bernard-Lyon I and the National Centre for French Scientific Research (CNRS). The authors thank Dr C. Romestaing for her valuable assistance in protein assays, M. D. Hürlimann, for providing access to a fast Laplace inversion routine and Prof. E. Pattee for his assistance in checking a first version of the manuscript.