Some insects rely on the strategy of freeze tolerance for winter survival. During freezing, extracellular body water transitions from the liquid to the solid phase and cells undergo freeze-induced dehydration. Here, we present results of a thermal analysis (from differential scanning calorimetry) of ice fraction dynamics during gradual cooling after inoculative freezing in variously acclimated larvae of two drosophilid flies, Drosophila melanogaster and Chymomyza costata. Although the species and variants ranged broadly between 0 and close to 100% survival of freezing, there were relatively small differences in ice fraction dynamics. For instance, the maximum ice fraction (IFmax) ranged between 67.9% and 77.7% total body water (TBW). Chymomyza costata larvae showed statistically significant phenotypic shifts in parameters of ice fraction dynamics (melting point and IFmax) upon entry into diapause, cold acclimation and feeding on a proline-augmented diet. These differences were mostly driven by colligative effects of accumulated proline (ranging between 6 and 487 mmol kg−1 TBW) and other metabolites. Our data suggest that these colligative effects per se do not represent a sufficient mechanistic explanation for high freeze tolerance observed in diapausing, cold-acclimated C. costata larvae. Instead, we hypothesize that accumulated proline exerts its protective role via a combination of mechanisms. Specifically, we found a tight association between proline-induced stimulation of glass transition in partially frozen body liquids (vitrification) and survival of cryopreservation in liquid nitrogen.
Insects overwintering in temperate and polar habitats have evolved different cold-tolerance strategies to cope with situations when their body temperature decreases below the equilibrium melting point (Salt, 1961; Lee, 2010). Some insect species show only limited capacity for cold tolerance and are often classified as chill susceptible (Bale, 1993, 1996; Koštál et al., 2011a). The overwintering strategies of truly cold-tolerant insects are categorized as freeze avoidance, i.e. reliance on supercooling of body liquids (Zachariassen, 1985; Renault et al., 2002), or freeze tolerance, i.e. survival after formation of ice crystals inside the body (Sinclair et al., 2003; Sinclair and Renault, 2010). Some insects use the strategy of cryoprotective dehydration, i.e. they lose most of their body water by evaporation and deposition to surrounding ice crystals in their overwintering microhabitat (Holmstrup and Westh, 1994; Holmstrup et al., 2002). Under specific conditions, insect body solutions may also undergo phase transition into a biological glass via the process of vitrification (Sformo et al., 2010; Koštál et al., 2011b).
In this paper, we focused on the strategy of freeze tolerance. The classical view (Lovelock, 1954; Asahina, 1969) is that freeze-tolerant organisms rely on the formation of ice crystal nuclei in the extracellular space. As the ice nuclei grow with decreasing temperatures, the extracellular solutions become more concentrated, which osmotically drives water out of cells. It remains under debate whether and how often intracellular ice formation occurs in vivo and whether it is always lethal (Wharton and Ferns, 1995; Sinclair and Renault, 2010). In this study on dipteran larvae, we adhered to the classical model of extracellular ice formation that is also supported by our direct observations of extracellular ice masses in frozen larvae of Chymomyza costata by scanning electron microscopy of cryo-fractured specimens (Koštál et al., 2011b). Considering the classical model, a freeze-tolerant insect must cope with not only deep sub-zero body temperature but also the multiple deleterious effects linked to freeze-induced cellular dehydration (Muldrew et al., 2004; Sinclair and Renault, 2010). All of these effects threaten the chemical and conformational stability of proteins and other macromolecules (Wang, 1999; Brovchenko and Oleinikova, 2008). In order to alleviate the stresses linked with extracellular freezing, freeze-tolerant insects often synthesize and accumulate a variety of small cryoprotective molecules (Sømme, 1982; Storey and Storey, 1991; Koštál et al., 2016a) and/or macromolecular compounds that regulate the process of ice formation (Zachariassen and Kristiansen, 2000; Duman, 2001, 2015). Under eco-physiological conditions, only a certain fraction of the insect's total body water is freezable (i.e. osmotically active water, OAW; corresponding to maximum ice fraction, IFmax), while the rest is unfreezable (i.e. osmotically inactive water, OIW) because it is non-covalently bound in the hydration shells of charged molecules and ions and, therefore, is not sufficiently mobile to join the growing ice crystals (Franks, 1986; Wolfe et al., 2002; Block, 2003). Although the phenomenon of insect freeze tolerance is widely recognized, the current knowledge about whether and how the relative size of the ice fraction (IF) limits survival remains poor. Available data (summarized in Table S1) are mostly descriptive, although the relationship between IF and freeze tolerance was specifically addressed in some studies (Zachariassen et al., 1979; Ramlov and Westh, 1993; Gehrken and Southon, 1997; Patricio Silva et al., 2013). Collectively, the literature suggests that seasonal change in the IFmax may be one of the adaptive facets of the insect freeze-tolerance strategy (Block, 2003). However, a comprehensive test of this hypothesis is missing.
In addition to the transition of liquid water into a crystalline phase, i.e. ice, we address in this study the transition into an amorphous non-crystalline solid, i.e. glass. The formation of amorphous glass ‘traps’ the structures and/or macromolecules and ‘locks’ them in place and conformation as they are at the moment of solidification. To date, vitrification has been observed in only two cases in insects. The first evidence (Sformo et al., 2010) was obtained in larvae of the beetle Cucujus clavipes puniceus. These larvae avoid freezing by partial dehydration and concomitant accumulation of glycerol, which decreases their supercooling point to as low as −58°C. Some individuals, however, do not freeze at all (to at least −150°C), but vitrify. The second observation comes from partially frozen larvae of Chymomyza costata (Koštál et al., 2011b). It is known that the crystalline phase can coexist with the vitreous phase in partially frozen systems. At very low temperatures, and high concentrations of solutes in the unfrozen phase, the residual solution may vitrify in presence of ice (Taylor et al., 2004). It should be noted that in medicinal cryobiology, the term vitrification is more often used to refer to a cryopreservation technique that attempts to vitrify the fully hydrated system without any ice formation (Fahy and Wowk, 2015), while dehydration or freeze concentration of body fluids played an important role in the two insect examples.
differential scanning calorimetry
fluctuating thermal regime
relative ice fraction
maximum ice fraction
long day length (16 h/8 h light/dark cycle)
osmotically active body water
osmotically inactive body water
proline-augmented diet (50 mg proline per g standard diet)
short day length (12 h/12 h light/dark cycle)
short day length with cold acclimation
total body water
temperature of de-glassing transition
equilibrium melting temperature
temperature of vitrification
temperature of inoculation of body fluids by external ice crystals
temperature of spontaneous freezing (supercooling point)
change in specific heat capacity
Here, we present results of thermal analysis of IF dynamics during freezing of the larvae of two fly species using differential scanning calorimetry. In addition, we verified the occurrence and characterized the parameters of vitrification. The two fly models, Chymomyza costata and Drosophila melanogaster, belong to the same family (Drosophilidae), their larvae are morphologically similar, and we sampled them at a relatively well-defined and comparable ontogenetic stage. The larvae are ecologically similar during the warm season: growing and developing rapidly on decaying plant material. However, the two species inhabit different geographical regions; D. melanogaster originated in the tropics and has spread to mild temperate regions during the last century (Throckmorton, 1975), while C. costata lives in cool temperate and subarctic regions (Hackmann et al., 1970). Larvae of D. melanogaster do not overwinter, are highly chill susceptible, and exhibit only a mild capacity for cold acclimation (Strachan et al., 2011; Koštál et al., 2011a, 2012). In contrast, C. costata larvae are highly seasonal; these larvae overwinter in diapause and exhibit extreme plasticity in freeze tolerance (Koštál et al., 2011b). By acclimating larvae of the two species under different photoperiods, temperatures and dietary conditions, we were able to produce experimental variants that covered the whole conceivable range of insect freeze tolerances. Having these variants in hand, we asked the following specific questions: (i) do IF dynamics differ between the two species?; (ii) does acclimation affect IF dynamics?; (iii) is there a maximum IF that an insect can withstand?; (iv) what is the effect of accumulated cryoprotectants, specifically proline, on IF dynamics?; (v) do we see glass transition in both species and, if yes, under what conditions?; and (vi) does vitrification correlate with survival of freezing?
MATERIALS AND METHODS
Insects and acclimation treatments
We compared two dipterans with contrasting freeze-tolerance capacities: the vinegar fly, Drosophila (Sophophora) melanogaster Meigen 1830, and the malt fly, Chymomyza costata (Zetterstedt 1838) (Diptera: Drosophilidae). Cultures of vinegar flies (Oregon-R strain) and malt flies (Sapporo strain) were maintained in MIR 154 incubators (Sanyo Electric, Osaka, Japan). Vinegar flies were reared as described previously (Koštál et al., 2011a) on a standard cornmeal–yeast–agar diet, and malt flies were reared on a similar cornmeal–yeast–agar diet supplemented with ground malted barley (Lakovaara, 1969; Koštál et al., 2016b). All experiments were conducted with fully grown third instar larvae that were sampled from the diet prior to the onset of wandering behaviour. This ontogenetic stage was chosen because pre-wandering larvae can survive freezing of extracellular body fluids (Koštál et al., 2011b, 2012). To modulate the level of larval freeze tolerance, we applied different photoperiodic and thermal acclimation regimes, and augmented the diet with l-proline (Sigma-Aldrich, St Louis, MO, USA; abbreviated hereafter as proline) as listed in Table 1.
In brief, all D. melanogaster larvae were grown under the same photoperiodic conditions: 12 h/12 h light/dark cycle (as larvae are photoperiodically insensitive) and one of three different thermal regimes (Table 1): constant 25°C, constant 15°C, and constant 15°C followed by 3 days at a fluctuating thermal regime (FTR) of 6°C for 20 h/11°C for 4 h. Under FTR conditions, D. melanogaster larvae enter quiescence (developmental arrest induced directly by low temperature) and increase their cold tolerance (Koštál et al., 2011a, 2016c). The larvae of C. costata reared under constant 18°C are photoperiodically sensitive; they continue direct development (to pupariation) under long day length (16 h/8 h light/dark cycle, LD) but enter diapause (hormonally regulated developmental arrest) under short day length (12 h/12 h light/dark cycle, SD; Koštál et al., 2000, 2016b). Gradual cold acclimation of diapausing (SD) larvae was performed by transferring them to 11°C and constant darkness at 6 weeks of age and, 1 week later, transferring them to 4°C for another 4 weeks. This cold acclimation (SDA) increases proline concentrations in the larvae and enhances freeze tolerance such that larvae survive cryopreservation in liquid nitrogen (Koštál et al., 2011b). We reared some larvae of both species on a proline-augmented diet (50 mg proline per g standard diet, Pro50), which further increases freeze tolerance according to our previous studies (Koštál et al., 2011b, 2012, 2016a).
Freezing and cryopreservation protocols
The inoculation of larval body fluids with external ice crystals at relatively high sub-zero temperatures (close to 0°C) and slow cooling/freezing rates are two essential factors underlying survival in freezing assays (Shimada and Riihimaa, 1988). In our experiments, inoculative freezing was ensured by wrapping the larvae between two layers of moist cellulose, overlain with a small ice crystal (for a more detailed description, see Fig. S1 and Koštál et al., 2016a). The standardization of all steps, including the cooling and warming rates, was achieved by performing all experiments in programmable Ministat 240 cooling circulators (Huber, Offenburg, Germany). Temperature inside the cellulose wrapping was monitored using K-type thermocouples connected to a PicoLog TC-08 datalogger (Pico Technology, St Neots, UK).
The larvae of D. melanogaster show only limited ability to survive freezing and their freezing protocol was optimized previously (Koštál et al., 2012, 2016a). The optimal protocol (Fig. S2) consists of six steps: (i) 20 min of manipulation with larvae at 0°C (washing larvae out of the diet, counting and placing them into tubes); (ii) 10 min of pre-incubation at −0.5°C (with ice crystal added); (iii) slow cooling to −2°C for 180 min (cooling rate, 0.008°C min−1); (iv) rapid cooling to −5°C for 30 min; (v) heating to +5°C for 40 min (heating rate, 0.25°C min−1); (vi) melting at +5°C for 10 min.
The larvae of C. costata survive freezing relatively well and, when appropriately acclimated, can survive cryopreservation in liquid N2 (Koštál et al., 2011b). In this study, we optimized the conditions of freezing and cryopreservation in SDA larvae by exposing them to protocols with modified rates and/or durations of cooling, heating and incubation. A detailed description of all protocols is given in Table S2. The final optimum freezing/cryopreservation protocol for C. costata (Figs S2 and S3A) consisted of six steps: (i) 20 min of larval manipulation at 0°C (washing larvae out of the diet, counting and placing them into tubes); (ii) slow cooling to −30°C (T1; with ice crystal added) for 300 min (cooling rate, r1=0.1°C min−1); (iii) plunging them into liquid N2 for 60 min (or, alternatively, maintaining at T1 for 60 min); (iv) transfer from liquid N2 to −30°C (T2); (v) heating to +5°C for 60 min (heating rate, r4=0.6°C min−1); (vi) melting at +5°C for 10 min.
The larvae of both species and all experimental variants (see Table 1) were exposed to optimal freezing protocols where the rates of cooling and heating were kept constant but the target temperatures (the minimum temperatures) varied. The data on freeze tolerance of D. melanogaster larvae were taken from our previous paper where the target temperatures varied between −2.5°C and −10°C (see table S3A of Koštál et al., 2016a). Results for the freeze tolerance of C. costata larvae were obtained from the present study. The target temperatures for C. costata varied between −5 and −75°C. Because our Ministats were not able to cool the larvae below −40°C, the freeze-tolerance assay at −75°C was conducted by pre-freezing the larvae to −40°C in the Ministat and then transferring them to a freezer (Platinum 370H, Angelantoni, Massa Martana, Italy) where they gradually cooled to a temperature of −75°C (see the temperature record in Fig. S3B). At the end of the freeze-tolerance assay, the unpacked cellulose balls were transferred to fresh standard diet in a tube maintained at constant 18°C. Alive/dead larvae were scored after 12 h recovery. All living larvae were maintained at 18°C for the subsequent 14 days (D. melanogaster) or 2 months (C. costata) and successful pupariation and emergence of adult flies were scored as criterions of survival. Exact numbers of larvae used for each specific experiment are shown in Results. Survival of cryopreservation in liquid N2 was assessed in the SDA variant larvae of C. costata during the optimization of the freezing and cryopreservation protocol (see above).
Differential scanning calorimetry
We conducted the thermal analyses of whole larvae in 50 µl aluminium pans using a differential scanning calorimeter (DSC4000, Perkin Elmer, Waltham, MA, USA), with some modifications (see below) to previously described methods (Block, 1994; Koštál et al., 2011b). The temperature scale of the DSC4000 was calibrated using indium and mercury according to the manufacturer's instructions. The heat flow was calibrated by measuring the areas under the melt endotherms of known masses of ice (Fig. S4). We developed two different DSC protocols, abbreviated as Equi-melt (equilibrium melting) and Ino-freeze (inoculative freezing), to measure the latent heat absorbed/released during the first-order phase transitions (melting and freezing). Using these protocols, we analysed the relative IF size occurring in a partially frozen larva at a given sub-zero temperature.
In the Equi-melt protocol (Fig. S5), the larva was hermetically sealed in an aluminium pan and inserted into the DSC4000 together with an empty reference aluminium pan. The larva was then rapidly cooled, and freezing occurred spontaneously at the larva's innate TSCP. Maximum ice fraction (IFmax) was allowed to form as the larva reached a sufficiently deep sub-zero temperature (−30°C). Next, a portion of body ice was melted by heating the larva to a specific Equi-melt temperature (TEM). The larva was equilibrated at TEM for 30 min, after which the remaining ice was melted. The equilibrium IF for each specific TEM was derived from the melt endotherm and related to total body water (TBW). TBW was measured as the difference between the total mass of the freshly sealed larva inside the aluminium pan and the total mass of the same pan, punctured after DSC analysis and dried for 3 days at 60°C.
In the Ino-freeze protocol (Fig. S6), the pre-weighed larva was put into the instrument (placed in the aluminium pan that contained external ice crystals) during the run of a temperature program, exactly at −1°C. The larva was inoculated with external ice crystals at a specific temperature (TINO) and the IF gradually increased during slow cooling/freezing at a rate of 0.1°C min−1 (simulating the conditions in freeze-tolerance assays) to a specific target temperature (used to analyse the presence/absence of vitrification; see below). The amount of ice formed during gradual freezing was estimated from the area under the freeze exotherm and related to TBW.
The thermal curves obtained by running the Equi-melt and Ino-freeze DSC protocols were analysed using Pyris Software (Perkin Elmer, Waltham, MA, USA) (see Figs S5 and S6 for examples of analysis and more details). The fraction of freezable (‘free’), osmotically active water (g OAW g−1 TBW) was calculated from the Boltzmann sigmoid curves fitted to summarized Equi-melt data. The fraction of unfreezable (‘bound’), osmotically inactive water [g OIW g−1 dry mass (DM)] was calculated analogously (Figs S7 and S8).
The results of Ino-freeze protocols that were run with different target temperatures were used not only to derive the IF from inoculative freeze exotherms but also to observe (upon rapid heating) the occurrence and parameters of de-vitrification (de-glassing) phase transition (a second-order phase transition without absorbing/releasing any latent heat). Vitrification typically happens over a ca. 10°C temperature interval centred on a glass transition temperature (Tg) when the viscosity rises by a factor of 1000, and heat capacity, thermal expansivity and compressibility suddenly fall from liquid values to near those of a crystal (Wowk, 2010). Vitrified matter maintains the structure, energy and volume of a liquid, but the changes in energy and volume with temperature are similar in magnitude to those of a crystalline solid (Kauzmann, 1948). In this study, an inflection point of the de-glassing transition was read as the temperature of de-vitrification (Tdg). The change in specific heat capacity (ΔCp) was derived from a difference in heat flow between the onset and the end of the de-glassing transition. The temperature of vitrification (Tg) during slow freezing (at a rate of 0.1°C min−1) was estimated based on the presence/absence of the Tdg transition after reaching each specific target temperature of slow freezing (see Fig. S6 for more details).
Metabolomic analysis was performed for C. costata larvae only, while the comparative data for D. melanogaster were taken from our previous studies (Koštál et al., 2012, 2016a,c). Whole larvae of C. costata (pools of 5 larvae in each of 4 replicates) were sampled from the diet, weighed to obtain fresh mass, plunged into liquid nitrogen and stored at −80°C until analysis. The dry mass and water mass were measured in a parallel sample of 20 larvae weighed individually and dried at 65°C for 3 days. The pools of larvae were homogenized in 400 μl of methanol:acetonitrile:water mixture (volumetric ratio, 2:2:1) containing internal standards (p-fluoro-dl-phenylalanine, methyl α-d-glucopyranoside; both at a final concentration of 200 nmol ml−1; both from Sigma-Aldrich). The TissueLyser LT (Qiagen, Hilden, Germany) was set to 50 Hz for 5 min (with a rotor pre-chilled to −20°C). Homogenization was repeated twice and the two supernatants from centrifugation at 20,000 g for 5 min at 4°C were combined. The extracts were subjected to targeted analysis of major metabolites using a combination of mass spectrometry-based analytical methods described previously (Koštál et al., 2016a,c).
Low molecular weight sugars and polyols were determined after O-methyloxime trimethylsilyl derivatization using a gas chromatograph with flame ionization detector (GC-FID-2014) equipped with an AOC-20i autosampler (both from Shimadzu Corporation, Kyoto, Japan). Profiling of acidic metabolites was done after treatment with ethyl chloroformate under pyridine catalysis and simultaneous extraction in chloroform (Hušek and Šimek, 2001). The analyses were conducted using a Trace 1300 gas chromatograph combined with single quadrupole mass spectrometry and a Dionex Ultimate 3000 liquid chromatograph coupled with a high-resolution mass spectrometer (Q Exactive Plus; all from Thermo Fisher Scientific, San Jose, CA, USA). All metabolites were identified against relevant standards and subjected to quantitative analysis using a standard calibration curve method. All standards were purchased from Sigma-Aldrich. The analytical methods were validated by simultaneously running blanks (no larvae in the sample), standard biological quality-control samples (the periodic analysis of a standardized larval sample – the pool of all samples), and quality-control mixtures of amino acids (AAS18, Sigma-Aldrich).
Experimental variants cover the whole range of insect freeze tolerance
By exposing larvae of two drosophilid species to various acclimation conditions, we produced experimental variants that cover the broadest range of insect freeze tolerance: from intolerance (100% mortality) of mild freezing at −2.5°C for a few minutes in D. melanogaster larvae (25°C variant) to almost 100% survival of larvae and high production of adults in C. costata (SDA variant) exposed to −75°C for a few hours (Fig. 1) or cryopreserved in liquid N2 for 18 months (Table S2). The two species differed in the overall ability to tolerate freezing but, more importantly, we found that freeze tolerance is a highly plastic trait in both species. The actual level of freeze tolerance was strongly influenced by acclimation conditions (Fig. 1).
IF dynamics are similar among species and acclimation variants
We measured the IF inside the larval body at different temperatures using two DSC-based methods. First, we measured the IF that remains in the larval body after heating the frozen larva to TEM (Equi-melt method). Boltzmann sigmoid curves fitted well to the empirical data (Figs S7 and S8), which allowed us to interpolate the IF at any given temperature. The final curves describing the dynamics of IF formation at decreasing temperature for all experimental variants are presented in Fig. 2A,E. In the Equi-melt method, however, the ice crystals grow inside the larval body very rapidly at relatively low temperatures (at the larva's TSCP, ranging between approximately −16 and −20°C). Freezing at the TSCP is lethal in many insects including C. costata and D. melanogaster larvae. Therefore, we additionally used the Ino-freeze method to measure the IF that gradually grows in the larval body after inoculative freezing at mild sub-zero temperatures (TINO) (Figs S9–S11). The two methods gave similar estimations of IF (see overlap of Equi-melt lines and Ino-freeze ellipses in Fig. 2A,E). The two fly species and different experimental variants differed relatively slightly (when compared with large differences in freeze tolerance) in various parameters of the IF dynamics. Thus, the mean MP varied from −0.4°C to −1.0°C in C. costata variants, and from −0.4°C to −0.6°C in D. melanogaster variants. The IFmax varied between 67.9% and 76.1% of TBW in C. costata variants, and between 75.0% and 77.7% of TBW in D. melanogaster variants. The temperature at which 99% of IFmax was reached varied from −5.9°C to −10.5°C in C. costata variants, and from −3.9°C to −6.3°C in D. melanogaster variants (for details, see Figs S7 and S8).
The two fly species differed in TBW, which ranged between 2.26 and 2.98 mg mg−1 DM in C. costata variants (Fig. 2B), and between 3.63 and 4.10 mg mg−1 DM in D. melanogaster variants (Fig. 2F). The OIW was relatively low in C. costata LD, SD and SDA variants (ranging between 0.67 and 0.85 mg mg−1 DM), while it was slightly higher in the C. costata LD Pro50 variant and all variants of D. melanogaster (ranging between 0.90 and 0.96 mg mg−1 DM; Fig. 2B,F). The inter-species difference in TBW was also reflected in the maximum ice content, which ranged between 1.5 and 2.3 mg mg−1 DM in C. costata variants (Fig. 2C) and between 2.7 and 3.2 mg mg−1 DM in D. melanogaster variants (Fig. 2G).
Association between IF and freeze tolerance
Fig. 3 shows the relationships between IF (derived from Fig. 2A,E; see also Figs S7 and S8) and survival at a given sub-zero temperature (taken from Fig. 1). Three basic patterns were observed: (i) the freezing event was lethal (D. melanogaster, 25°C and 15°C acclimation variants); (ii) lethal effects of freezing occurred when the IF reached a specific threshold (D. melanogaster, 15°C FTR and 15°C FTR Pro50 acclimation variants; C. costata, LD and LD Pro50 acclimation variants); the specific threshold for tolerable IF lay close to the IFmax, which corresponds to the OAW fraction (or, vice versa, which is limited by the OIW fraction); (iii) the freezing event was always survived irrespective of the size of the IF (C. costata, SD and SDA acclimation variants).
Chymomyza costata larvae accumulate high concentrations of proline
We analysed concentrations of 23 amino compounds, 6 polyols, 3 sugars, 4 intermediates of the TCA cycle, and lactate (37 metabolites in total) in the larvae of C. costata (for details, see Table S3). The concentrations of the three most abundant metabolites (proline, glutamine and trehalose), plus the sum concentration of all other quantified metabolites, are presented in Fig. 2D. In order to make a direct comparison between the two species, similar data for D. melanogaster are shown in Fig. 2H (Koštál et al., 2012, 2016a,c). The two species dramatically differed in their ability to accumulate proline. The larvae of C. costata naturally accumulated 339 mmol kg−1 TBW of proline during cold acclimation (SDA variant). Feeding the non-diapause larvae of C. costata a proline-augmented diet (LD Pro50 variant) increased the concentration of proline up to 487 mmol kg−1 TBW. In contrast, D. melanogaster larvae naturally accumulated only 9 mmol kg−1 TBW of proline (15°C FTR variant) and feeding them a proline-augmented diet increased the concentration of proline to only 57 mmol kg−1 TBW (15°C FTR Pro50 variant).
Glass transition occurs only in C. costata
Vitrification (revealed as the presence of a de-glassing transition) was never registered in any DSC thermal analysis of D. melanogaster larvae (all experimental variants), irrespective of whether larvae were frozen rapidly (at the innate TSCP) or gradually (after inoculation by external ice at TINO), and irrespective of the target temperature (ranging between −5 and −70°C). When heating frozen larvae back to 20°C at a fast rate of 10°C min−1, typical melt endotherms were always observed, but the characteristic de-glassing transition was absent in all cases.
In contrast, the de-glassing transition was detected in DSC Ino-freeze thermal analyses of C. costata larvae for all experimental variants (see an example in Fig. S6). However, the presence/absence and parameters of de-glassing were strongly affected by acclimation and freezing conditions (Fig. 4). De-glassing never occurred in larvae that were frozen to target temperatures higher than −30°C. At the target temperature of −30°C, de-glassing transitions were observed only in SDA larvae (50% of cases) and LD Pro50 larvae (20% of cases) but not in LD or SD larvae. With decreasing target temperature, the frequency of occurrence of the de-glassing transition rapidly increased in all experimental variants. Nevertheless, the rates of this increase differed among experimental variants in the order: SDA>LD Pro50>SD>LD (Fig. 4A). Similar ordering of experimental variants was also observed in the other two parameters of de-glassing: Tdg and ΔCp. The mean Tdg varied from −15.8°C to −20.1°C and was variant specific (SDA<LD Pro50<SD<LD), but independent of target freezing temperature within each variant (slopes of linear regressions did not deviate from zero; Fig. 4B). The ΔCp was variant specific (SDA>LD Pro50>SD>LD) and it increased as the target temperature decreased (slopes of linear regressions deviated from zero; Fig. 4C).
Differences in IF dynamics among species and acclimation variants
We found that the IF dynamics are very similar in two drosophilid species [a response to our Introduction question (i)]. This close similarity contrasts with the profound differences observed in freeze tolerance between the two species. Temperature-dependent IF dynamics in a larva is mathematically described by two partially interlocked parameters, the MP and the OAW (Wang and Weller, 2011). As the MP ranged only moderately between −0.4 and −1.0°C in our study, its effect on IF dynamics was relatively small (Figs S7 and S8). The OAW affects the overall shape of IF dynamics more profoundly, as it directly sets the IFmax. It is known that the total volume of OAW may undergo relatively massive and rapid changes in response to changing ambient conditions (Wharton and Worland, 2001; Block, 2002; Hadley, 1994), especially in small soil invertebrates that rely on the overwintering strategy of cryoprotective dehydration (Holmstrup and Westh, 1994). However, we did not see any dramatic changes in OAW in our model species. The OAW (IFmax) values measured in D. melanogaster and C. costata larvae, ranging between 67.9% and 77.7% TBW, fall within the scope of published records for other insects, which range between 64% and 84.5% TBW (Table S1).
The two model species differed mainly in their ability to phenotypically modulate IFmax in response to acclimation. [Introduction question (ii)]. The C. costata larvae showed decreases of IFmax upon entry into diapause, cold acclimation and feeding on a proline-augmented diet. As discussed above, these shifts in IFmax were driven by concomitant decreases in MP and OAW fraction, both metrics intimately related to increasing the osmolality of body fluids, which was driven by accumulation of proline. We will discuss later whether this phenotypic modulation of IFmax driven by proline may have some adaptive meaning. In contrast, the IFmax slightly increased in D. melanogaster larvae upon cold acclimation and entry into quiescence, while feeding on a proline-augmented diet had no effect on IFmax.
Increasing the OIW fraction has been proposed by Storey et al. (1981) as a potentially important adaptive mechanism for insect freeze tolerance. They observed that warm-acclimated larvae of Eurosta solidaginis had 0.193 mg OIW mg−1 DM, which considerably increased to 0.633 mg OIW mg−1 DM after a 6 week stepwise acclimation to −30°C. The hypothesis about the adaptive meaning of increasing OIW rests fundamentally on a concomitant reduction in OAW that causes a reduction of IF at any given temperature and, consequently, mitigates the deleterious effects linked with freeze-induced cell dehydration. In this study, the OIW fraction was relatively constant in all acclimation variants in both species ranging between 0.67 and 0.96 mg mg−1 DM. This amount of bound water was safely above the anhydrobiotic threshold of 0.2–0.4 mg mg−1 DM that limits normal functionality of cellular structures and enzymes (Brovchenko and Oleinikova, 2008; Ball, 2008). Moreover, we have seen that the more freeze-tolerant larvae of C. costata exhibit a lower amount of OIW (expressed per mg DM) than the less freeze-tolerant larvae of D. melanogaster. These results lead us to question whether relatively small differences in OIW fraction explain such large variation in freeze tolerance in C. costata and D. melanogaster larvae.
Association between IF, unfreezable water and freeze tolerance
The seemingly trivial question (iii) posed in the Introduction – ‘is there a maximum IF that an insect can withstand?’ – appears as difficult to address empirically as it is to answer generally. Most likely, it is not the IF itself but rather the detrimental effects linked to freeze-induced dehydration that limit freeze tolerance. These effects include mechanical stress caused by growing extracellular ice crystals and shrinking of the cell, decreased activity of water molecules, increased ionic strength, acidity and concentrations of potentially toxic intermediates of metabolism, increased viscosity and increased packing of macromolecules (Muldrew et al., 2004). The association between IFmax, OAW/OIW and larval survival is depicted in Fig. 3. Our data suggest that strongly freeze-tolerant insects (such as cold-acclimated diapausing larvae of C. costata) survive after the formation of IFmax inside their body. The maximum IF that an insect can withstand thus obviously exists only for the insects falling into the categories of partial or moderate freeze tolerance (sensu Sinclair, 1999; such as warm-acclimated active larvae of C. costata or quiescent and proline-fed larvae of D. melanogaster). The existence of maximum tolerable IF was previously estimated in other invertebrates (Zachariassen et al., 1979; Ramlov and Westh, 1993; Gehrken and Southon, 1997; Patricio Silva et al., 2013). However, the follow-up question on what mechanism sets this maximum tolerable IF remains open. We can only speculate why the maximum tolerable IF occurs relatively close to the IFmax not only in our study but also in the other studies: in adult tenebrionid beetles, Eleodes blanchardi, the IFmax represented approximately 75% TBW in both cold- and warm-acclimated specimens, which died when reaching a threshold IF of 62% and 65% TBW, respectively (Zachariassen et al., 1979). In New Zealand wetas, Hemideina maori, the IFmax was 82% TBW but they died when exposed to temperatures below −7°C, which corresponds to approximately 81% TBW (Ramlov and Westh, 1993). In adult chrysomelid beetles, Melasoma collaris, the IFmax varied between 77% and 84.5% TBW (cold- and warm-acclimated specimens, respectively), and the lower limit of freeze tolerance was associated with an IF of 73–75% TBW (Gehrken and Southon, 1997). In a freeze-tolerant potworm, Enchytraeus albidus, the IF was manipulated by exposing them to various sub-zero temperatures and environmental salinities. For two different populations of potworms, it was found that lethal effects of freezing occurred when IF reached a sharp threshold between 56% and 57% TBW, while IFmax was similar in the two populations at 58.8% and 61.4% TBW, respectively (Patricio Silva et al., 2013). One potential hypothetical explanation for such commonality (proximity of maximum tolerable IF to IFmax) could be that some (moderately) freeze-tolerant insects can tolerate the loss of most of their OAW relatively well, while greater losses impair the integrity of the OIW pool and, consequently, lead to irreversible changes in macromolecular conformation (Wang, 1999; Ball, 2008; Brovchenko and Oleinikova, 2008) and ultimately mortality. Other (partially) freeze-tolerant insects can be sensitive to even relatively small losses of their OAW associated with relatively mild cell dehydration.
Cryoprotective role of accumulated proline
In response to our Introduction question (iv) regarding the effects of proline on IF, we can say that in both fly species, the accumulated proline affected the empirically measured parameters of IF dynamics according to its theoretically expected colligative effects on biological systems. The MP, the relative OAW fraction and the IFmax all decreased with increasing proline concentration. The classically proposed mechanistic model of cryoprotection is based just on these colligative effects. This model posits that at any given cryogenic temperature T, the amount of extracellular ice and, consequently, the magnitude of deleterious cellular freeze-dehydration, is lower in the system with accumulated cryoprotectant than in the system without cryoprotectant (Lovelock, 1954; Salt, 1961; Meryman, 1971; Zachariassen, 1985; Storey and Storey, 1988; Lee, 2010). Our results suggest that the colligative effects linked to proline accumulation are detectable in fly larvae and may thus theoretically contribute to larval freeze tolerance. However, these colligative effects should not be used as sole and straightforward predictors of freeze tolerance. The insect cold-tolerance literature agrees on a view that a single molecule, such as proline, may play more than one (colligative) mechanistic role in building the insect's freeze tolerance. Moreover, a whole complex of other mechanisms, in addition to accumulation of cryoprotectants, needs to be taken into account in order to fully explain the resulting freeze tolerance (for review, see Storey and Storey, 1988, 1991; Sinclair, 1999; Lee, 2010).
We are of the view that proline exerts its protective role in C. costata by a combination of mechanisms, and that the importance of individual mechanisms may gradually change with the acclimation state of the insect during the course of its entry into dormancy, cold acclimation, cooling and freezing. Thus, proline might be actively involved in entry into diapause via its diverse regulatory functions in sensing the energy status and production of reactive oxygen species (Phang et al., 2010; Liang et al., 2013). Proline can also scavenge free radicals (Kaul et al., 2008). Proline accumulation continues as the ambient temperature decreases. Finally, proline levels elevate to 339 mmol kg−1 TBW in cold-acclimated, diapausing larvae (C. costata, SDA variant), which represents as much as 499 mmol kg−1 OAW. Our previous analysis (Koštál et al., 2011b) most probably underestimated the concentration of proline in cold-acclimated C. costata larvae (showing only 147 mmol kg−1 TBW). The higher value (339 mmol kg−1 TBW) is correct, as we have verified in several generations of larvae since the original publication. It is plausible to propose that high concentrations of proline help to reduce partial unfolding of proteins in chilled and supercooled larvae (prior to freezing) via the mechanism of preferential exclusion (Arakawa and Timasheff, 1985; Timasheff, 1992, 2002; Bolen and Baskakov, 2001). Upon freezing, proline can serve as a molecular shield and prevent aggregation of partially unfolded proteins or fusion of membranes in tightly packed organelles (Bryant et al., 2001; Hoekstra et al., 2001; Hoekstra and Golovina, 2002; Ball, 2008). At high levels of freeze-dehydration, proline concentrations exceed>1 mol kg−1 TBW. At such extremely high concentrations, proline is known to form specific supramolecular aggregates (Rudolph and Crowe, 1986; Samuel et al., 2000) which may interact with the hydrophobic surfaces presented by partially unfolded proteins, thereby stabilizing the folding intermediates and preventing their aggregation (Samuel et al., 1997; Ignatova and Gierasch, 2006; Das et al., 2007). In addition, the supramolecular structures of proline may promote the formation of amorphous biological glass – vitrification (Rudolph and Crowe, 1986).
Association between glass transition and freeze tolerance
We observed glass transition in all acclimation variants of C. costata but found no indication of glass transition in D. melanogaster larvae [Introduction question (v)]. Our study provides no direct mechanistic explanation for why this difference between the two species exists. Although proline is known to stimulate glass transition (Rudolph and Crowe, 1986), we assume that some additional mechanisms should be involved in the stimulation of glass transition in C. costata [compare: no glass transition in D. melanogaster, 15°C FTR Pro50 variant (proline, 57.2 mmol kg−1 TBW) versus Tdg≈−17°C in C. costata, LD variant (proline, 31.4 mmol kg−1 TBW)].
We found that de-vitrification occurred at around −20°C, while vitrification was completed in all larvae of the SDA variant of C. costata at approximately −30°C (Fig. 5). Interestingly, we found that these larvae can survive plunging into liquid N2 only when slowly pre-frozen to below −20°C (corresponding to Tdg), and that they can survive and metamorphose into adults only when slowly pre-frozen to below −30°C (corresponding to a completion of glass transition). Assuming that Tg must lie somewhere in the interval between Tdg and the completion of glass transition, our results reveal a strikingly tight association between glass transition and the dramatic increase of survival of cryopreserved SDA C. costata larvae (Fig. 5). Based on these results, we formulate the answer to our last Introduction question (vi): correlative evidence was obtained showing that vitrification of the residual solution after formation of IFmax is associated with the survival of SDA C. costata larvae cryopreserved in liquid N2. We suggest that vitrification may further stabilize the structures of macromolecular complexes and protect them against thermomechanical stress (Rubinsky et al., 1980) linked to rapid changes of temperature during plunging into liquid N2 and re-warming.
Using DSC thermal analysis, we found that temperature-dependent IF dynamics are very similar in the larvae of D. melanogaster and C. costata (each species analysed in four different acclimation variants). The two species differed mainly in their ability to phenotypically modulate the parameters of IF dynamics in response to acclimation. The C. costata larvae decreased MP, OAW and IFmax upon entry into diapause, cold acclimation and feeding on a proline-augmented diet. The maximum tolerable IF changed according to the acclimation state in both species: from freeze intolerance to a specific tolerable threshold in D. melanogaster, or from specific tolerable threshold to independence of IF in C. costata. The specific threshold (maximum tolerable IF) was situated very close to IFmax. The phenotypic shifts in IF dynamics were associated with colligative effects caused by accumulated proline. In addition to colligative effects, accumulated proline probably affected freeze tolerance of C. costata larvae by a combination of other mechanisms. We have not detected any glass transitions in D. melanogaster larvae exposed to temperatures as low as −70°C. In contrast, the glass transitions of the residual solution after formation of IFmax occurred at temperatures below −30°C in all acclimation variants of C. costata. We found a tight correlation between the occurrence of glass transition and the dramatic increase of survival in liquid N2-cryopreserved C. costata larvae of the SDA acclimation variant.
We thank Irena Vacková, Anna Heydová, Iva Opekarová and Helena Zahradníčková (all from Biology Centre CAS) for assistance with insect rearing, sample preparation and analyses. We thank Lauren Des Marteaux (Biology Centre CAS) for commenting on an earlier version of the paper.
Conceptualization: J.R., V.K.; Methodology: J.R., M.M., P.S., V.K.; Validation: V.K.; Formal analysis: J.R., M.M., P.S., V.K.; Investigation: J.R., M.M., P.S., V.K.; Writing - original draft: V.K.; Writing - review & editing: J.R., M.M., P.S., V.K.; Supervision: V.K.; Project administration: V.K.; Funding acquisition: V.K.
This work was supported by Grantová Agentura České Republiky (grant 16-06374S to V.K.).
The authors declare no competing or financial interests.