Intertidal zone organisms can experience transient freezing temperatures during winter low tides, but their extreme cold tolerance mechanisms are not known. Petrolisthes cinctipes is a temperate mid–high intertidal zone crab species that can experience wintertime habitat temperatures below the freezing point of seawater. We examined how cold tolerance changed during the initial phase of thermal acclimation to cold and warm temperatures, as well as the persistence of cold tolerance during long-term thermal acclimation. Thermal acclimation for as little as 6 h at 8°C enhanced cold tolerance during a 1 h exposure to –2°C relative to crabs acclimated to 18°C. Potential mechanisms for this enhanced tolerance were elucidated using cDNA microarrays to probe for differences in gene expression in cardiac tissue of warm- and cold-acclimated crabs during the first day of thermal acclimation. No changes in gene expression were detected until 12 h of thermal acclimation. Genes strongly upregulated in warm-acclimated crabs represented immune response and extracellular/intercellular processes, suggesting that warm-acclimated crabs had a generalized stress response and may have been remodelling tissues or altering intercellular processes. Genes strongly upregulated in cold-acclimated crabs included many that are involved in glucose production, suggesting that cold acclimation involves increasing intracellular glucose as a cryoprotectant. Structural cytoskeletal proteins were also strongly represented among the genes upregulated in only cold-acclimated crabs. There were no consistent changes in composition or the level of unsaturation of membrane phospholipid fatty acids with cold acclimation, which suggests that neither short- nor long-term changes in cold tolerance are mediated by changes in membrane fatty acid composition. Overall, our study demonstrates that initial changes in cold tolerance are likely not regulated by transcriptomic responses, but that gene-expression-related changes in homeostasis begin within 12 h, the length of a tidal cycle.
Marine intertidal ectotherms exposed to temperatures below their freezing point must either avoid or tolerate extracellular ice formation, and the mechanisms of these tolerances have received some attention (Hawes et al., 2010; Loomis, 1995; Pineda et al., 2005; Storey and Storey, 1996). However, the majority of ectotherms are killed at low temperatures above their freezing points, and the mechanisms underlying this non-freezing cold injury (NFCI) are poorly understood. In insects, NFCI is hypothesised to result from membrane phase transition and/or ion equilibration (Kostal et al., 2004), whereas in marine invertebrates, failure of oxygen delivery to mitochondria has also been implicated (Pörtner, 2010; Pörtner et al., 2006). In comparison to the large number of studies demonstrating short-term cold-hardening in terrestrial arthropods [primarily insects (e.g. MacMillan and Sinclair, 2011; Overgaard et al., 2008)], very few studies have demonstrated that marine or intertidal zone arthropods are capable of acclimation to low temperatures (McAllen and Block, 1997). In an era of rapidly changing climates, there is considerable interest in understanding the speed, effectiveness and constraints of phenotypic plasticity related to temperature sensitivity (Angilletta, 2009; Deutsch et al., 2008; Dillon et al., 2010; Stillman, 2003), and here we extend these investigations in intertidal arthropods to include low temperatures.
To survive exposure to low temperatures, intertidal invertebrates must maintain metabolic homeostasis, and this is likely aided by pre-existing adaptations to hypoxia associated with emersion (Mandic et al., 2009; Sloman et al., 2008). Possible mechanisms of physiological acclimation by arthropods to avoid NFCI at low temperatures include homeoviscous adaptation of membrane fluidity, production of metabolite cryoprotectants and changes in pump, ion or membrane permeability characteristics to maintain ion homeostasis (Hochachka and Somero, 1984; Hochachka and Somero, 2002). Homeoviscous adaptation, whereby membrane fluidity increases with cold acclimation, has been observed in many ectothermic taxa including fish (Logue et al., 2000; Williams and Hazel, 1994), mollusks (Pernet et al., 2007; Williams and Somero, 1996), nematodes (Hayward et al., 2007; Murray et al., 2007), insects (Kostal and Simek, 1998; Purac et al., 2011) and crabs (Cuculescu et al., 1999). Homeoviscous adaptation is often identified through a change in membrane fatty acid chain composition, with a substitution of unsaturated and/or shorter-chain fatty acids resulting in increased fluidity at low temperatures (Williams, 1998). The time frame of homeoviscous adaptation is unclear, although the rapid cold-hardening response in insects has been ascribed to changes in membrane fatty acids (Overgaard et al., 2006; but see MacMillan et al., 2009), which suggests that homeoviscosity could be re-established at low temperatures in a period of minutes to hours. Other biochemical and physiological changes that may enhance cold tolerance, such as the accumulation of metabolite cryoprotectants or the production of chaperones, are assumed to be relatively slow, but could still occur in the time frame of a single tidal cycle (Williams and Somero, 1996).
The proposed mechanisms of cold acclimation are all likely dependent to some extent on the actions of genes or gene products (Gracey, 2007; Gracey et al., 2004; Stillman and Tagmount, 2009). Although a target-gene approach is useful for testing a specific hypothesis, global gene expression studies have great potential to uncover hitherto unexpected mechanisms, and are particularly useful in cases where a clear phenotype is present (in our case, cold acclimation), but the mechanisms are uncertain (Lockwood and Somero, 2011; Logan and Somero, 2010; Stillman and Tagmount, 2009). In previous transcriptomic studies on low temperature in arthropods, acclimation (or selection) treatments have revealed wide suites of genes associated with (among other things) metabolism, stress and cell structural components, and the immune response, indicating that acclimation processes are complex (Clark and Worland, 2008; Zhang et al., 2011). Thus, transcriptomics cannot be expected to provide an answer, but does allow mechanistic hypotheses to be generated. In the case of NFCI, where little is known about the mechanisms of protection or damage, especially in crustaceans, transcriptomic techniques offer a new route to uncovering these mechanisms.
Porcelain crabs, genus Petrolisthes, present an ideal set of species that have differing thermal adaptation across latitudinal (temperate vs tropical) and vertical (intertidal vs subtidal) gradients (Stillman, 2002; Stillman, 2004; Stillman and Somero, 2000). For example, intertidal zone species are more eurytolerant and survive heat and cold thermal extremes better than sympatric subtidal species (Stillman, 2003). Intertidal zone porcelain crabs experience significant thermal habitat heterogeneity associated with tidal and seasonal cycles (Stillman and Tagmount, 2009). In studies of the effects of thermal acclimation on cardiac thermal performance, Petrolisthes cinctipes, a mid–upper intertidal zone inhabitant of the northeastern Pacific that can experience freezing temperatures during wintertime low tides (Stillman and Tagmount, 2009), was observed to survive brief exposure to temperatures between –5 and –6°C following cold acclimation, whereas warm-acclimated individuals did not survive (Stillman, 2003) (J.H.S., unpublished observations). No other porcelain crab species tested survived a temperature exposure below the freezing point of seawater (–1.9°C).
Long-term thermal acclimation is a well-established phenomenon in ectotherms, and in some cases, the mechanisms are understood (Angilletta, 2009; Hochachka and Somero, 2002). However, organisms must also respond to much more rapid changes in temperature that are highly ecologically relevant (Denny et al., 2006). In the case of low temperatures, cold exposure may be experienced over the course of a single tidal or diel cycle (Stillman and Tagmount, 2009). Thus, in this study we have focused on the first 24 h of thermal acclimation, and compare that with ‘endpoints’ of longer acclimation. Our study addressed three aspects of thermal acclimation with particular emphasis placed on cold tolerance. We determined the cold tolerance strategy and capacity for cold and warm acclimation of cold tolerance in the porcelain crab P. cinctipes, we quantified membrane fatty acids, which are the most common signal for changes in membrane composition that could lead to acclimation, and we identified transcriptomic shifts associated with thermal acclimation to allow us to generate novel hypotheses for the mechanisms of acclimation in this species. We have focused on processes occurring in cardiac tissue as we already have data on thermal performance of hearts and changes in gene expression following acclimatization to different habitats (Stillman and Tagmount, 2009).
MATERIALS AND METHODS
Animal collection and maintenance
Adult male and female porcelain crabs, Petrolisthes cinctipes (J. W. Randall 1840), were collected from beneath rocks below the high tide mark at Fort Ross, California (38°30′20″N, 123°13′58″W), on three separate dates. For long-term acclimation and assessment of cold tolerance, crabs were collected on 23 November 2007 and acclimated to 8 and 18°C until February 2008 (10–11 weeks), when tolerance to freezing and supercooling was assessed and samples were taken for lipid analysis. Crabs for the first 24 h cold-tolerance experiment, from which samples were generated for gene expression and lipid analyses, were collected on 6 February 2008. Crabs used in an additional 24 h cold-tolerance experiment were collected on 2 February 2009. Crabs were returned to the Romberg Tiburon Center, where they were maintained in recirculating seawater at 13°C for 7–10 days before the start of thermal acclimation and fed frozen Cyclop-eeze copepods (Argent Laboratories, Redmond, WA, USA) daily. A temperature of 13°C reflects the seasonal average temperature for P. cinctipes (Stillman and Tagmount, 2009), and is between the warm- and cold-acclimated state for this species with respect to cardiac critical thermal maximum (CTmax) (Stillman, 2004).
Cold tolerance was determined by exposure of emersed individual crabs (acclimated to either 8 or 18°C seawater for 10–11 weeks) to sub-zero temperatures (–3.4 to –7.0°C, cooled from 5°C at 0.15°C min–1 in watertight containers in the bath of a Lauda RC6-CP refrigerated circulator; Lauda, Wurzburg, Germany). Crabs were in contact with a 36 gauge type-T thermocouple interfaced to PicoLog recording software via a Picotech TC-08 thermocouple interface (Pico Technology, Cambridge, UK). Thus, it was possible to determine whether each individual crab froze from the presence or absence of an exotherm. Crabs were removed to 13°C and survival after 48 h was noted.
Tolerance of a 1 h exposure to –2°C was measured in crabs collected in February 2008 and 2009. Crabs (N=8 per temperature × treatment combination) acclimated at 8 and 18°C for 1, 3, 6, 8, 12, 18 and 24 h were exposed to –2°C while emersed for 1 h, and then allowed to recover at their acclimation temperature (in 2008) or at 13°C (in 2009). Cold tolerance of crabs that were left in the 13°C acclimation temperature was assessed at the same 0 and 24 h acclimation time points for individuals that had been switched to the other acclimation temperatures. Crabs were checked for survival after 1 and 5 days of recovery. Survival was assessed as responsiveness to probing and observation of active movement of antennules.
Sampling for transcriptome and lipid analyses during the first 24 h of thermal acclimation
On 11 February 2008, 264 specimens were divided into warm (18°C, N=96), cold (8°C, N=96) and control (13°C, N=72) acclimation groups. Crabs were sampled from the 13°C group at 0 h (the start of the experiment) and 24 h (the termination of the experiment). Crabs were sampled from the warm and cold acclimation groups at 6, 12, 18 and 24 h following the start of thermal acclimation. At each time point, heart tissue from 16 crabs from each group was dissected, flash frozen and stored at –80°C.
RNA extraction, purification and linear amplification
Preparation and microarray analyses of heart tissue samples were performed as previously described for porcelain crabs (Stillman and Tagmount, 2009). Cardiac tissues from eight specimens per group were shaken in guanidinium thiocyanate-phenol-chloroform extraction buffer [38% phenol in saturated buffer pH 6.6, 0.8 mol l–1 guanidinium thiocyanate, 0.4 mol l–1 ammonium thiocyanate, 0.1 mol l–1 sodium acetate pH 5.0 and 5% glycerol (Chomczynski and Sacchi, 1987)] in a Retsch MixerMill MM300 (Retsch Inc. Newtown, PA, USA) at 30 Hz for 10 min. Total RNA was extracted by standard chloroform and isopropanol precipitation methods, and then purified using Qiagen RNeasy columns (Valencia, CA, USA) and quantitated on a NanoDrop spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE, USA). In some cases, individual cardiac tissues did not yield adequate total RNA to continue, and in one group only five specimens yielded adequate total RNA. A pooled total RNA sample was prepared for each group by mixing equal quantities of total RNA from five individuals in each group in order to have the same amount of biological diversity within each pooled RNA sample. Linear amplification of RNA by in vitro transcription (Van Gelder et al., 1990) was performed as follows. First-strand cDNA was synthesized by heating 3 μg of pooled total RNA with 500 ng of OligodT-T7 primer (5′- GCATTAGCGGCCGCGAAATTAATACGACTCACTATAGGGAGATTTTTTTTTTTTTTTTTTTTTV-3′) (Baugh et al., 2001) in a volume of 15 μl for 5 min at 70°C and then 5 μl 5× first strand buffer [M531A, Promega, Madison, WI, USA; final concentration: 50 mmol l–1 Tris-HCl pH 8.3 at 25°C, 75 mmol l–1 KCl, 3 mmol l–1 MgCl2, 10 mmol l–1 dithiothreitol (DTT)], 1.25 μl 10 mmol l–1 dNTPs, 200 U MMLV reverse transcriptase (M1705, Promega), 15 U RNase inhibitor (bp3225-1, Fisher Scientific, Pittsburgh, PA, USA) and 2.25 μl double distilled water (ddH2O) were added and tubes were incubated at 42°C for 60–90 min followed by a heat inactivation at 70°C for 10 min.
Double-stranded cDNA was produced by adding 75 μl of the following: 1× Escherichia coli DNA polymerase buffer (NEB buffer #2, New England Biolabs, Ipswich, MA, USA), 0.25 mmol l–1 dNTPs, 3 U E. coli RNaseH (TAK 2150A, Takara Mirus Bio, Mountain View, CA, USA) and 40 U E. coli DNA polymerase (M0209L, New England Biolabs). Following 2 h incubation at 16°C, 1 U E. coli DNA ligase (M0205L, New England Biolabs) and 11 μl 10× DNA ligase buffer were added and incubated at room temperature for 15 min. Finally, 6 U T4 DNA polymerase (M0203L, New England Biolabs) were added followed by incubation for 15 min at room temperature. Double-stranded cDNAs were purified using the Qiagen QIAquick PCR purification kit following the manufacturer’s protocol, except that cDNA was eluted from the column in 2× 50 μl elution steps using RNase-free water. Purified double-stranded cDNA was dried by vacuum centrifugation in a speed vacuum and stored at –20°C. For RNA amplification, 40 μl of in vitro transcription cocktail [120 mmol l–1 HEPES, pH 7.5, 28 mmol l–1 MgCl2, 6 mmol l–1 rNTPs (N0466L, New England Biolabs), 1 mmol l–1 spermidine (19485201, MP Biomedicals, Inc., Solon, OH, USA), 0.1 mmol l–1 acetylated bovine serum albumin (BSA) (R3961, Promega), 40 mmol l–1 DTT (9779, Sigma-Aldrich, St Louis, MO, USA), 0.2 U pyrophosphatase (EF0221, Fermentas, Hanover, MD, USA), 200 U T7 RNA polymerase (EP0113, Fermentas) and 15 U RNase inhibitor (bp3225-1, Fisher Scientific)] were added and incubated at 37°C for 14 h. Amplified RNA was purified (Qiagen RNeasy), eluted 2× with 50 μl RNase-free water, concentrated by speed vacuum and quantified from absorbance at 260 nm. For the seasonal comparison, a second round of amplification was performed, except that: (1) random hexamer primers were used in the first strand synthesis and (2) 500 ng of oligodT-T7 primer were added to the second strand reaction.
Reverse transcription of antisense RNA (aRNA) was performed by incubating 3.25 μg aRNA, 1 μg random hexamer primers and water to a total volume of 15 μl for 5 min at 65°C, then adding 10 μl of aa-dNTP cocktail [1× first strand buffer, 1 mmol l–1 DTT, 1.25 mmol l–1 amino allyl dNTP mix (10 mmol l–1 dATP, dGTP, dCTP, 3.33 mmol l–1 dTTP, 6.66 mmol l–1 amino allyl dUTP), 200 U MMLV reverse transcriptase, 5% v/v RNase inhibitor] and incubating at 42°C for 70 min. aRNA was hydrolyzed by adding 10 μl of 1 mol l–1 NaOH (0.22 mol l–1 final concentration) and 10 μl 0.5 mol l–1 EDTA, pH 8.0 (0.11 mol l–1 final concentration) at 65°C for 15 min, and then neutralized by addition of 25 μl HEPES, pH 7.4. Amino-allyl cDNAs were purified (Qiagen QIAquick PCR purification kit) and cDNA was eluted from the column in 2× 30 μl elution steps using pH 8.0 adjusted ddH2O, and dried.
Cy3 or Cy5 fluorescent dyes (PA23001, PA25001, Amersham Biosciences, GE Healthcare, Piscataway, NJ, USA) were coupled to cDNAs resuspended in 9 μl 0.1 mol l–1 NaHCO3 buffer, pH 9.0. Five microliters of dye dissolved in DMSO (final concentration 1.14 mmol l–1) were added and incubated at room temperature in the dark for 1 h. Dye coupling was quenched by addition of 4.5 μl of 4 mol l–1 hydroxylamine (final concentration 0.97 mol l–1). Samples to be hybridized together on a microarray (supplementary material Fig. S1, see below) were combined and diluted to a total volume of 100 μl with water. cDNA was purified (Qiagen QIAquick PCR purification) and eluted in 2× 50 μl ddH2O. Thirty micrograms of polyA RNA was added to each tube, and tubes were speed vacuum dried.
For microarray hybridizations, we used 25 slides in an incomplete loop design where each sample was hybridized five times and labelled with each Cy dye two to three times (supplementary material Fig. S1). Because we used pooled RNA samples to make aRNA, all of our replication assessed technical variation in estimating the means of the five individuals in each group.
Custom microarrays were printed on poly-l-lysine coated glass slides (Erie Scientific, Portsmouth, NH, USA) at the UCSF Center for Advanced Technology using 24,748 PCR-amplified cDNAs, each representing a unique consensus sequence from a library of 61,440 cloned cDNAs (Stillman et al., 2006; Tagmount et al., 2010). Before use, spots were rehydrated by incubating slides in humidified air, snap dried at 100°C and cross-linked by 200 mJ UV exposure in a cross linker (Spectronics Co., Westbury, NY, USA). Slides were shaken in 0.2% sodium dodecyl sulfate (SDS), washed in copious amounts of ddH2O and incubated at 42°C for 30–45 min in pre-warmed blocking solution [0.1% SDS, 5× saline-sodium citrate (SSC) and 1% BSA] to block free poly-l-lysine molecules. Slides were washed in five to six changes of ultrapure water, denatured by washing in 95°C ddH2O for 2 min and centrifuged to dry at 125 g for 3 min. Dried cDNA + polyA RNAs were resuspended in 40 μl hybridization buffer (50% deoinized formamide, 5× SSC and 0.1% SDS), heated 2 min at 94°C, chilled on ice, and pipetted underneath an Erie LifterSlip placed on the microarray slide and hybridized for 12–16 h at 42°C. Following hybridization, coverslips were removed and slides were washed for 5 min with gentle shaking at 42°C two times in 2× SSC and 0.1% SDS, two times in 1× SSC and two times in 0.1× SSC. Slides were dried during gentle centrifugation (500 g) for 5 min.
Hybridized slides were scanned at 10 μm resolution (GenePix 4000B and GenePix Pro v.6.0, Axon Instruments, Union City, CA, USA) while adjusting photomultiplier voltage to balance signal intensity in the red and green channels. Features that were not usable were flagged, and flagged features were excluded from the results.
Statistical analysis of microarray data
Differentially expressed features were identified using MAANOVA [version 1.8.1 (Wu et al., 2003)] in Version 2.6.2 of the R statistical environment (http://www.r-project.org/). Median fluorescence values for each feature were log transformed (Draghici et al., 2003), and a weighted linear least-squares regression LOWESS normalization was conducted on log2-transformed data using the joint LOWESS function (Wu et al., 2003). GenePix results files and log2-transformed LOWESS normalized data have been submitted to the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE35307). Microarray images and sequence data are also available at the Porcelain Crab Array Database (PCAD; http://array.sfsu.edu).
A fixed ANOVA model with array, dye and sample (=treatment group) terms was fit using restricted maximum likelihood (REML), and a permutated F-test (N=500 permutations) was used to identify features that significantly differed (tabulated P≤0.001) among items in the sample term using the FS statistic at α=0.001. The FS statistic uses overall gene expression variation to calculate a gene-specific variance (Cui et al., 2005; Wu et al., 2003) and has a much higher stringency than other F or t statistics. To control for false positives, we employed the q-value false discovery rate correction (Storey, 2002; Storey, 2003; Storey et al., 2004; Storey and Tibshirani, 2003). For those features selected as significant by the FS-test, we exported log2 expression fold change data for filtering and k-means clustering using Cluster v.3.0, and for visualization as heat maps using Treeview software version 1.60 (http://rana.lbl.gov/EisenSoftware.htm). Data were filtered to remove flagged features. Mean expression values and standard deviations for each cluster were calculated in Microsoft Excel and data were plotted using DeltaGraph v5.7.5 (Red Rock Software, Salt Lake City, UT, USA).
Gene homology for each microarray feature was determined from consensus sequences submitted to the following searches: BLASTx (GenBank, UniProt, SwissProt, Daphnia pulex v1.1 Filtered Models protein set), InterPro Scan (13 different protein identification algorithms) and tBLASTx (Daphnia pulex v1.1 transcript set) (Stillman et al., 2006; Tagmount et al., 2010). Using all of the available data, we ascribed gene identity for each feature where there was a match with a score of ≤1e–5. Putative functions were determined from the UniProt database.
Lipid extraction and separation
Crab hearts (N=8 per acclimation temperature and time point) were placed in a –80°C freezer until assays were performed. All samples were weighed using a microbalance (Mettler Toledo MX5, Columbus, OH, USA), and then homogenized in 2 ml of Folch reagent (1:1 chloroform:methanol). Homogenate was then rinsed into a 20 ml glass culture tube with an additional 2 ml of Folch reagent. Glass homogenizers were cleaned between each sample by rinsing with 2 ml of Folch reagent and 1 ml of distilled water. Homogenates were then centrifuged at 2000 g for 15 min. Following centrifugation, samples were filtered through #1 Whatman filters (VWR, Mississauga, ON, Canada) into clean, dry culture tubes, where 5 ml of Folch (2:1) reagent was added. Samples where then vortexed and filtered again into clean tubes, into which 3 ml of 0.25% KCl was added. The tubes were incubated at 70°C for 10 min then cooled. The aqueous layer was then removed and discarded, while the remaining organic layer was dried under a nitrogen stream at 70°C.
Dried samples were re-suspended in 100 μl of pure chloroform. Prior to loading samples, 1 ml Supelclean LC-NH2 SPE tubes fixed to the Supelco Visiprep DL vacuum block were conditioned by adding 2 ml of hexane. Each sample tube was then rinsed with 2×100 μl of chloroform and added to the appropriate column. Column and sample tube were then centrifuged together for 1 min at 1300 g. To elute neutral lipids from the loaded columns, two aliquots of 0.9 ml of chloroform:isopropanol (2:1) were added, and the column and tube were centrifuged as above. Eluting non-esterified fatty acids was done in a similar fashion using two washes of 0.8 ml of isopropyl ether:acetic acid (49:1), followed by centrifugation. Finally, phospholipids were eluted from the column with 3×0.9 ml of methanol. To this separated phospholipid sample, 20 μl of a 19:0 fatty acid in hexane (0.31 mg ml–1) was added as an internal standard. All samples were then evaporated under a nitrogen stream.
Dried samples were re-suspended in 2 ml of 1 mol l–1 acetyl chloride in cold methanol, capped tightly and incubated at 90°C for 2 h. Samples were then evaporated under nitrogen and redissolved in 1 ml methanol to remove residual HCl. Samples were evaporated one last time, and redissolved in 100 μl dichloromethane, and transferred to gas chromatography (GC) vial inserts.
Phosopholipid samples were run on an Agilent Technologies 6890N network GC system (Agilent Technologies, Mississauga, ON, Canada) equipped with a J&W 122-2332 30.0 m×250 μm×0.25 μm capillary column (Agilent Technologies). Helium was used as the carrier gas at an average velocity of 42 cm s–1. The inlet temperature was set at 250°C with an injection volume of 1 μl under splitless mode and the flame-ionization detector temperature was set at 280°C. Initial column temperature was set to 80°C for 2 min, and was then ramped up to 180°C at 5°C min–1, held for 3 min, and heated from 180 to 200°C at 1.5°C min–1. Fatty acids were identified by comparing retention times with standards (Sigma-Aldrich). Peak areas were corrected using a molecular weight correction factor (Christie, 1989) and were then used in calculating the molar percent of each fatty acid.
Analysis of fatty acids was performed in SAS (v. 9.1, SAS Institute, Cary, NC, USA). Samples with extreme outliers (two or more fatty acids that were more than 50% larger or smaller than the mean concentration of that fatty acid) were discarded, and the mean number of double bonds per molecule and ratio of saturated:unsaturated fatty acids were calculated for each sample. These parameters (which approximated normal distributions) were compared among sexes, acclimation temperatures and times using a repeated-measures ANOVA using PROC MIXED in SAS, and a stepwise model building approach where the Akaike information criteria of the models were compared and the best explanatory model was chosen.
In this study we examined changes in organismal tolerance to freezing and supercooling, organismal tolerances to extreme cold, transcriptome profiles of cardiac tissue, and phospholipid composition that occurred during the first 24 h of thermal acclimation. Our results from cold tolerance assays suggest that the process of thermal acclimation begins within 6 h of transfer to a new temperature, but no gene expression differences were observed until 12 h following transfer to a new temperature. There were no differences in phospholipid composition with respect to acclimation temperature.
Regardless of thermal acclimation, no crabs survived internal ice formation (N=21). However, acclimation for 10–11 weeks in cold (8°C) or warm (18°C) conditions had a significant impact on survival of chilling at sub-zero temperatures. All 18°C-acclimated crabs that remained unfrozen (N=12) died following exposure to temperatures below –2°C, but 37.5% (N=8) of 8°C-acclimated crabs survived exposure to temperatures below –3.5°C, provided they did not freeze.
Following 10–11 weeks of thermal acclimation to warm (18°C) or cold (8°C) temperatures there were marked differences in survival following a 1 h exposure at –2°C (Fig. 1A,B). Warm-acclimated crabs had approximately 10% survival whereas all cold-acclimated crabs survived exposure to –2°C (Fig. 1A,B). Thermal acclimation for only a few hours at warm or cold temperatures was sufficient to alter survival following exposure to –2°C for 1 h (Fig. 1). Survival of 13°C-acclimated crabs was generally 40–60% 1–5 days following the exposure to –2°C (Fig. 1). Acclimation at 8°C enhanced survivorship to between 60–100%, whereas acclimation to 18°C reduced survivorship to below 50% in most cases (Fig. 1). There was not a large difference in survival of crabs that recovered at 13°C following the –2°C exposure (Fig. 1C) compared with crabs that recovered at their acclimation temperatures (Fig. 1A). Survival patterns 5 days following the –2°C exposure (Fig. 1B) were not systematically different to survival observed 1 day following the –2°C exposure (Fig. 1A).
Transcriptome changes during thermal acclimation
A total of 6209 microarray features were statistically significantly differentially expressed across acclimation treatments and times (Table 1). Expression patterns fell into seven k-means clusters: two clusters contained features solely upregulated in warm-acclimated crabs, two clusters contained features solely upregulated in cold-acclimated crabs, and three clusters had mixed expression patterns across the sample groups and did not vary systematically with acclimation temperature (Fig. 2, Table 1). In general, there was considerable variation across acclimation time points in expression levels (Fig. 2). In all clusters there were no differences in expression observed at 6 h of acclimation, nor were the 6 h acclimation expression levels different from the expression levels of 13°C-acclimated crabs sampled at the start of the acclimation experiment (0 h) (Fig. 2). Large differences in gene expression between warm- and cold-acclimated crabs were observed at 12 and 24 h of acclimation (Fig. 2B–F), although at 18 h of acclimation the expression differences often either disappeared (Fig. 2B,C,E,F) or were the opposite of what was observed at the 12 and 24 h time points (Fig. 2D). In all but one cluster (Fig. 2A), there was no change in expression level in 13°C-acclimated crabs between the start (t=0 h, when crabs had been acclimated to 13°C for 7–10days) and the end (t=24 h)of the acclimation experiment (Fig. 2). Expression patterns were divided into four groups (Table 1): (1) ‘mixed’, where at some time points expression was higher in 8°C-acclimated crabs and at other time points expression was higher in 18°C-acclimated crabs (Fig. 2A,D); (2) ‘no difference’, where there was no systematic difference between 8°C- and 18°C-acclimated crabs (Fig. 2G); (3) ‘warm acclimation specific’, where expression was elevated only in 18°C-acclimated crabs (Fig. 2B,C); and (4) ‘cold acclimation specific’, where expression was elevated only in 8°C-acclimated crabs (Fig. 2E,F).
For all seven clusters, approximately 52% of the features had some annotation where the sequence matched a known gene, and approximately 73% of those annotations were to unique genes (Table 1), thus approximately 38% of the features within each cluster had a unique functional annotation. As the purpose of this study was to identify genes with acclimation-temperature-specific expression profiles, our paper focuses on the four clusters where there was specific induction in warm- (Fig. 2C,D) or cold-acclimated (Fig. 2E,F) specimens. For each cluster, gene functions were ascribed to 15 categories (Table 1), including one category ‘other’ for which approximately 24% of all annotated unique genes were placed (Table 1). Functions were assessed using the UniProtKB database (http://pir.uniprot.org/uniprot/) as well as primary literature sources. Expression data, CloneIDs and functional annotations for all features within each cluster are available in supplementary material Table S1. For the subset of the annotated features in each cluster from which we could extract KEGG IDs (Tagmount et al., 2010), KEGG pathway representations have been provided and are hyperlinked to the KEGG database (supplementary material Table S2).
Genes induced by acclimation to warm temperatures (clusters B and C)
The largest expression changes exhibited an average 34-fold upregulation (Table 1) and ranged from 7- to 131-fold upregulation (supplementary material Table S1), and were observed in warm-acclimated crabs (Fig. 2B). This cluster (cluster B) was relatively small and contained only 99 features, 55 of which were annotated to 34 unique genes (Table 1). Immune response protein-encoding genes comprised 35% of the 34 unique genes (Table 1, cluster B), and included boophilin, carcinin, complement and von Willebrand Factor genes (supplementary material Table S1). Genes encoding proteins involved in extracellular processes comprised 21% of the 34 unique genes (Table 1, cluster B), and included L-selectin, cerebellin and other cell wall and epithelial proteins (supplementary material Table S1). KEGG pathways identified from more than one of the features from annotated genes in cluster B (supplementary material Table S2B) included complement and coagulation cascades (KO04610), and neuroactive ligand-receptor interaction (KO04080), supporting both immune response and extracellular processes as being upregulated.
Features that were statistically significantly upregulated in warm-acclimated specimens, but that had only a small increase in expression (twofold), are illustrated in cluster C (Table 1, Fig. 2C). Cluster C contained 643 features, 58% of which were annotated to known genes, and 80% of those represented unique genes (Table 1), for a total of 46% of features in cluster C representing unique known genes. Like cluster B, immune response genes were represented, but only 6% of the unique annotations matched immune and stress responsive genes (Table 1). These genes encoded immune response proteins such as macroglobulins, Anti-lipopolysaccharide factor, carcinin, crustin, clotting protein, clumping factor, Kunitz-type proteinase inhibitor 5 II, lysozyme, macrophage mannose receptor, Plasma kallikrein and complement, and encode stress response proteins including thioredoxins, the 70 kDa heat shock proteins, heat shock protein 90, hypoxia up-regulated protein 1 and PCNA associated factor (supplementary material Table S1). One of the most highly represented KEGG pathways from cluster C (supplementary material Table S2C) involved immune response (complement and coagulation cascades, KO04610). Extracellular/intercellular processes were highly represented in cluster C, with 9% of the unique annotated features. Genes for cuticular and extracellular proteins such as cell wall protein DAN4, chitin binding peritrophin-A and –48, endocuticle structural glycoprotein SgAbd-8, FRAS1 and a number of additional transmembrane and translocon associated proteins were observed, indicating modification of cell surface and intercellular interaction (supplementary material Table S1). A number of N-acetylglucosaminidases and N-acetylglucosamine binding proteins were also observed (supplementary material Table S1). KEGG pathways represented by multiple features in cluster C that involve extracellular and intercellular actions include gap junction (KO04540), focal adhesion (KO04510), antigen processing and presentation (KO04612), neuroactive ligand-receptor interaction (KO04080), and pathways involving synthesis and degradation of extracellular matrix materials including glycosphingolipids, glycosaminoglycans and other glycans (supplementary material Table S2C).
The largest fraction of unique annotated genes represented in cluster C (17% of unique features; Table 1) encoded proteins involved in modification and regulation of nucleic acids and proteins, representing processes of DNA replication, transcriptional regulation and translation initiation (supplementary material Table S1). DNA replication and chromatin modifying proteins encoded by these genes included DNA replication licensing factor MCM5 and MCM6, histones, histone methytransferase, histone acetyltransferase and DNA topoisomerase (supplementary material Table S1). Proteins involved in transcription and processing of mRNA included kruppel-like factor 6, zinc finger proteins, RNA helicases, DNA-directed RNA polymerases, RNA binding proteins, small nuclear RNA-assocated proteins and mRNA splicing proteins such as crooked neck-like protein (supplementary material Table S1). Proteins involved in translation included eukaryotic translation initiation factors (EIFs) EIF2S2, EIF2S3 and EIF4G2 (supplementary material Table S1). Most of these genes were not annotated with KEGG IDs, thus the set of KEGG pathway maps for cluster C (supplementary material Table S2C) does not contain a large set of pathways representing DNA replication, transcription, translation, protein synthesis and/or protein degradation.
The second largest percentage of unique annotated features in cluster C included genes that encoded for proteins involved in the processes of protein synthesis, modification and degradation (13%; Table 1). Genes for proteins involved in synthesis, folding and modification of new polypeptides included the chaperone Hsp40, dolichyl-diphosphooligosaccharide protein glycosyl transferases, protein O-mannosyl-transferase-2, elongation factor 1B, splicing factors, protein disulfide isomerases and protein kinases (supplementary material Table S1). The set of unique genes for protein degradation processes was larger and included a number of proteases, ubiquitin-related proteins, lysosomal proteins and proteasome proteins (supplementary material Table S1). Similar to what was observed for cluster C features encoding genes involved with transcription and translation, protein modification and degradation KEGG pathways were not highly represented (supplementary material Table S2C).
Glucose transport proteins were encoded by 5% of the uniquely annotated features in cluster C (Table 1), and included genes for alpha-endosulfine, which stimulates insulin secretion, insulysin, which degrades insulin, endoglucanase A, Na+/glucose cotransporter 4 and sugar phosphate exchanger 2 (supplementary material Table S1). Additional biological processes represented in genes upregulated by warm acclimation in cluster C include lipid modification (2%; Table 1), structural and cytoskeleton proteins (4%; Table 1), cell cycle proteins (5%; Table 1), amino acid metabolism (4%; Table 1) – which may be involved principally in aminoacyl-tRNA biosynthesis (supplementary material Table S2C) – and carbohydrate metabolism (1%; Table 1).
Genes induced by acclimation to cold temperatures (clusters E and F)
There were three to four times more features upregulated only in cold-acclimated crabs. These were clustered by strong expression differences (average fourfold difference) between warm- and cold-acclimated crabs (N=404; Fig 2E, Table 1) or weak expression differences (average twofold difference) across acclimation temperature groups (N=1927; Fig. 2F, Table 1).
Cold-acclimated crabs had elevated expression of genes encoding proteins involved in the net production of glucose. Of a total of 36 uniquely annotated features for glucose production or transport genes, 26 (72%) were observed in the clusters that were upregulated by cold acclimation (Fig. 2E,F, Table 1), and 15 of those were strongly induced (10- to 12-fold induction; Table 1, Fig. 2E). Cold acclimation strongly induced glucose production and transport genes including glycogen phosphorylase, phosphoenolpyruvate carboxykinase, glucose-6-phosphatase, glucose transporter, glucose repression mediator and sugar transporters such as osmotin and solute carrier family 2 facilitated glucose transporter (supplementary material Table S1). The insulin (KO04910), adipocytokine (KO04920) and peroxisome proliferator-activated receptor (PPAR; KO03320) signaling pathways were the most highly represented of features strongly induced in cold-acclimated crabs (supplementary material Table S2E), indicating upregulation of pathways that induce cellular uptake of glucose. Related pathways regulating carbohydrate metabolism including starch and sucrose metabolism (KO00500), the citrate cycle (KO00020), and pyruvate metabolism (KO00620) (supplementary material Table S2E) were also highly represented in the features strongly induced by cold acclimation (Fig. 2E). Features weakly induced by cold acclimation (Fig. 2F) also included genes encoding for proteins that increase intracellular glucose, including alpha amylase, β-hexoseaminidase, glucose-6-phosphatase, mannosyl-oligosaccharide glucosidase and glucose repression mediator protein (supplementary material Table S1). Many of the same KEGG pathways involving glucose regulation and carbohydrate metabolism were identified in the set of features that were weakly induced (Fig. 2F), but were not the principal pathways identified in cluster F (supplementary material Table S2F).
Cellular surface and cellular structure related genes were strongly induced by cold acclimation (Fig. 2E, supplementary material Table S1). Genes encoding cell surface receptors and ligands and genes involved in membrane structural protection (e.g. mucin) showed the strongest response to cold, representing 15% of the genes upregulated in cold-acclimated crabs (Table 1), and exhibiting expression differences of up to 20-fold (Fig. 2E, supplementary material Table S1). Other structural genes that could be related to the cytoskeleton and/or sarcomere structure (e.g. I-connectin, titin and projectin) exhibited a sevenfold to 13-fold increase in induction in cold-acclimated crabs (Fig. 2E, supplementary material Table S1).
Compared with warm acclimation, cold acclimation induced more changes in the expression of genes encoding proteins involved in DNA and RNA binding and modification involved in the regulation of transcription and translation. In warm-acclimated crabs, 6% (N=2 unique annotations) of the strongly induced uniquely annotated features encoded genes involved with DNA and RNA processes (Table 1, Fig. 2B), whereas in cold-acclimated crabs, 24% (N=53 unique annotations) of the uniquely annotated features involved DNA and RNA processes. Percentage-wise, the number of weakly induced features involving DNA and RNA binding was similar in warm- and cold-acclimated crabs (17 and 23%, respectively), but the number of features was much greater in cold-acclimated crabs (141 vs 50; Table 1). DNA and RNA binding proteins strongly induced (up to fivefold) by cold acclimation (Fig. 2E) include transcription factors such as high mobility group proteins HMG-I and HMG-Y, zinc finger proteins and MYB (supplementary material Table S1). Chromatin remodeling proteins such as ATRX, histone deacetylase, histone mRNA 3′-exonuclease and chromodomain-helicase-DNA-bidning protein Mi-2 homolog (supplementary material Table S1) were also induced by cold acclimation, and interestingly may regulate chromatin structure in a fashion opposite of what was observed in warm-acclimated crabs (condensation in warm acclimation vs expansion in cold acclimation). Additional genes encoding proteins involved with transcriptional processes that were strongly induced by cold acclimation included RNA binding protein 28, RNA polymerase, RNA helicase, polymerase associated protein LEO1 and immediate-early protein, among others (supplementary material Table S1). These transcriptional and translation regulatory proteins could be related to activity of a number of signaling pathways that were induced by cold acclimation, including MAPK (KO04010), phosphatidylinositol (KO04070), calcium (KO04020) and GnRH (KO04912) signaling pathways represented by multiple features in cluster E (supplementary material Table S2E).
There was a lower representation of genes involved with protein turnover in cold-acclimated crabs compared with warm-acclimated crabs (Table 1). Strongly induced (5%, N=10) and weakly-induced (10%, N=61) uniquely annotated genes involved with protein synthesis and degradation included relatively fewer genes encoding proteins involved in protein degradation, protein synthesis and inhibition of protein degradation than in warm-acclimated specimens (supplementary material Table S1). Protein degradation proteins such as trypsin, and a peptidase were strongly induced by cold acclimation (Fig. 2E, supplementary material Table S1), and cathepsin I, ubiquitin ligase, carboxypeptidase, proteasome proteins and zinc metalloproteinases were weakly induced by cold acclimation (Fig. 2F, supplementary material Table S1). Proteins that inhibit protein degradation that were strongly induced by cold acclimation include ubiquitin thioesterase OTU1 and ubiquitin-like modifier-activating enzyme (Fig. 2F, supplementary material Table S1), and protease inhibitors such as 4 disulfide core proteins, antistasin, serine protease inhibitor dipetalogastin proteins and ubiquitin carboxyl-terminal hydrolases 16 and 36, which could suppress ubiquin mediated proteolysis, were weakly upregulated by cold acclimation (Fig. 2E, supplementary material Table S1).
Although not strongly represented by numbers of unique annotations, cold acclimation induced nine genes encoding proteins involved with of osmotic and ionic regulation (Table 1), including Na+/K+/2Cl– cotransporter, Na/Ca exchanger, angiotensin converting enzyme, canalicular multispecific organic anion transporter 2, Na+/K+ ATPase alpha subunit, T-type Ca2+ channel subunit alpha G and solute carrier proteins (supplementary material Table S1). Although these genes did not represent a large fraction of those differentially expressed, it is worth noting that no genes encoding for proteins involved in osmotic or ionic regulation were specifically induced by warm acclimation (Table 1, supplementary material Table S1).
Genes involved in lipid binding, lipid modification and lipid transport were observed both in warm-acclimation-specific induction clusters (Fig. 2C) and cold-acclimation-specific induction clusters (Fig. 2E,F). In the warm-acclimated crabs, lipid modification genes (N=6) that were upregulated included those involved in phospholipid biosynthesis, phospholipid elongation and lipid absorption (supplementary material Table S1). In contrast, cold-acclimated crabs strongly induced genes encoding proteins involved with phospholipid degradation and emulsification, including hydroxyacyl-CoA dehydrogenase and lipase (Fig. 2E, supplementary material Table S1). Lipid modification genes weakly induced in cold-acclimated crabs (Fig. 2F) included a wider array of functions, e.g. included fatty acid biosynthesis (e.g. acyl-CoA synthetase family member 3 and trans-2-enoyl-CoA reductase), steroid hormone biosynthesis and modification (e.g. estradiol 17-beta-dehydrogenase 8 and estrogen sulfotransferase) and sphingolipid synthesis and modification (e.g. Phosphatidylcholine:ceramide cholinephosphotransferase 1 and Sphingomyelin synthase related), as well as potential lipid-binding proteins such as lipocalin (supplementary material Table S1). The KEGG steroid biosynthesis (KO00100) and glycerophospholipid metabolism (KO00564) pathways were well represented in genes weakly induced by cold acclimation (Fig. 2F), as were a number of additional lipid metabolism and modification pathways (supplementary material Table S2).
Membrane fatty acid composition
Twenty-three fatty acids were identified, with six present in every sample (16:0, 18:0, 18:1n9c, 18:1n7, 20:5n3 and 22:6n3), and 11 present in more than 50% of samples (Table 2). There was no effect of sex on membrane fatty acid composition (P>0.5), so the sexes were pooled in the analysis. The number of double bonds (8°C, 2.03±0.228 double bonds per molecule; 18°C, 1.95±0.183 double bonds per molecule; t5=0.256, P=0.81) and ratio of saturated:unsaturated fatty acids (8°C, 0.84±0.226; 18°C, 0.88±0.722; t5=0.11, P=0.91) did not differ significantly between crabs acclimated to 8 or 18°C at the thermal acclimation endpoint (10–11 weeks; Fig. 3). The best model of acclimation effects on saturated:unsaturated fatty acids and the number of double bonds per molecule included only time (saturated:unsaturated, F1,33=6.24, P=0.018; double bonds, F1,33=6.52, P=0.015), reflecting the decreased ratio after 24 h of acclimation, but the lack of influence of acclimation temperature on membrane composition (P>0.15 for both temperature and temperature × time in both cases).
The intertidal zone porcelain crab P. cinctipes can experience sub-zero temperatures in nature (Stillman and Tagmount, 2009), and has a cardiac CTmin that is within 0.5°C of the freezing point of seawater when cold acclimated (Stillman, 2003; Stillman, 2004). In this study, we acclimated P. cinctipes to 8 and 18°C, temperatures that reflect variation in the thermal mean in the natural environment (Stillman and Tagmount, 2009), and that are known to elicit near-maximal variation in cardiac thermal performance (CTmax and CTmin) in this species (Stillman, 2003; Stillman, 2004). We investigated whether P. cinctipes is freeze tolerant, how the first 24 h of thermal acclimation affects tolerance to supercooling, changes in cardiac transcriptome profiles that occur during the first 24 h of thermal acclimation, and membrane phospholipid composition of cardiac tissues from warm- and cold-acclimated crabs. Each area of our study is discussed separately, below.
Here we show that although P. cinctipes can tolerate low temperatures in an unfrozen state, it is killed by internal ice formation. This limited cold tolerance is likely sufficient to tolerate the brief sub-zero temperatures experienced during wintertime low tides (Stillman and Tagmount, 2009), but is in contrast to reports of freeze tolerance in a range of other intertidal invertebrates, including tidepool copepods (McAllen and Block, 1997), barnacles (Crisp et al., 1977), bivalves (Compton et al., 2007; Williams, 1970) and gastropods (Hawes et al., 2010; Sinclair et al., 2004). Freeze tolerance has not been demonstrated in any species of crab, and the lack of freeze tolerance of P. cinctipes may impact the northern distribution limit of this species. Petrolisthes cinctipes is distributed from Point Conception, California, to the northern tip of Vancouver Island, Canada (Haig, 1960). A subtidal congener, P. eriomerus, is found further north into northern British Columbia and Alaska, but is likely never exposed to sub-freezing temperatures. Distribution limits in fiddler crabs across sharp biogeographic boundaries have been demonstrated to be due to effects of cold temperatures on larvae rather than adults (Sanford et al., 2006). Some species of fiddler crab live in locations where surface temperatures are below freezing, but the crabs avoid freezing temperatures by remaining in burrows during winter months (Sanford et al., 2006). Intertidal zone porcelain crabs live on rocky shores and do not burrow, and thus cannot avoid exposure to freezing temperatures.
Cold acclimation for as little as 6 h enhanced cold tolerance in P. cinctipes (Fig. 1), and the enhanced tolerance was likely initiated by processes other than transcriptional regulation, as there were few changes in gene expression (in cardiac tissue) observed until 12 h of acclimation (Fig. 2). This rapid induction of enhanced cold tolerance is significant for intertidal zone animals, where extreme cold exposure is both unpredictable and rapid due to the interaction of weather and tidal exposure. The rapidity by which cold tolerance is increased may be in contrast to the slow and predictable decrease in temperature that occurs during seasonal induction of cold and freeze tolerance that is observed in terrestrial animals that survive freezing over winter (Loomis, 1995). Rapid changes in thermal tolerances are well documented in insects and other terrestrial arthropods (Lee and Denlinger, 2010), but the mechanisms of this rapid cold-hardening are not well understood, so it is not possible to speculate about whether those mechanisms are shared by intertidal crustaceans.
Transcriptome response to thermal acclimation
In our study, the thermal acclimation temperatures chosen reflect minimal (8°C) and maximal (18°C) shifts in seasonal mean temperatures (Stillman and Tagmount, 2009). We observed shifts in cold tolerance within 6 h of acclimation (Fig. 1), but transcriptome responses required a longer period of thermal acclimation to manifest, as differences were not observed until after 12 h of thermal acclimation (Fig. 2). This observation leads us to conclude that during a single low-tide period (i.e. 6 h), exposure to extreme cold conditions are not likely to be adequate to induce the same effects as a constant temperature immersion. This conclusion is supported results from a previous study, in which transcriptome responses to cold shock (0°C) induced only slightly more cardiac tissue transcripts in winter-than in summer-acclimatized crabs (Stillman and Tagmount, 2009). In that study, there was between twofold and fourfold higher induction of approximately 20 transcripts in winter following cold stress (Stillman and Tagmount, 2009), the functions of which suggested increased oxidative metabolism (e.g. NADH dehdryogenase, isocitrate dehydrogenase and cytochrome c) and extracellular binding related to immune function (plasma kallikrein, peritrophin A, carcinin). We observed a much larger transcriptional response in the present study, with over 2000 transcripts exhibiting a twofold to fourfold increase in induction in cold-acclimated specimens (Fig. 2, Table 1), suggesting that the response to non-extreme cold is fundamentally different than the response to extreme cold in P. cinctipes.
Initial enhancement of cold tolerance occurred without concomitant changes in the cardiac transcriptome, potentially because of regulation of existing proteins through post-translational modification, e.g. phosphorylation (Dieni and Storey, 2010). The measurement of protein modification during the initial phase of thermal acclimation is beyond the scope of this study, but changes in gene expression that we observed present target candidates for future study. Maintenance of the cold-tolerant phenotype is presumably related to changes in expression that were induced starting at 12 h of acclimation to cold temperatures (Fig. 2E,F).
We found sugar transport genes to be upregulated in both warm- and cold-acclimated crabs, but warm-acclimated crabs induced different sugar transport genes than cold-acclimated crabs (Table 1, supplementary material Table S1). Cold-acclimated crabs induced a much greater diversity of genes for sugar transporters and translocation proteins (Table 1) than did warm-acclimated crabs. Genes induced in cold-acclimated cardiac tissues included many that would function in elevation of plasma glucose, including genes for glycogen degradation, production of glucose through gluconeogenesis, and glucose transporters to increase intracellular glucose (supplementary material Table S1). Glucose has been shown to be a cryoprotectant in terrestrial vertebrates including amphibians (Devireddy et al., 1999a; Devireddy et al., 1999b; Holden and Storey, 2000; Storey, 2004) and reptiles (Voituron et al., 2002), and also in invertebrates including mussels (Storey and Churchill, 1995) and insects (Muise and Storey, 2001), as an intermediate in the production of the cryoprotectant glycerol. Glycerol and genes encoding proteins involved with glycerol synthesis have been shown to increase during cold acclimation of smelt (Hall et al., 2011). There is debate about the role of hemolymph glucose concentrations in rapid cold-hardening of insects (MacMillan et al., 2009; Overgaard et al., 2007), but glucose is seldom used as a high-concentration cryoprotectant in insects (Storey, 1997), although it may be converted to glycerol or other, less reactive, cryoprotectants such as trehalose (Colson-Proch et al., 2009; Issartel et al., 2005). Further evidence that cold acclimation involves regulation of intracellular glucose can be taken from the fact that warm-acclimated cardiac tissues upregulated a different set of proteins involved with glucose signaling and transport. Warm-induced proteins included alpha-endosulfine, which stimulates insulin secretion through K(ATP) channel currents (Bataille et al., 1999), and insulysin, which degrades both insulin as well as glucagon (Shen et al., 2006); taken together, these changes indicate that warm acclimation is acting to sequester glucose and make glycogen. In contrast to the other glucose production or sequestration gene expression patterns observed in warm-acclimated crabs, we observed induction of glucose mobilization gene endoglucanase A, which is an enzyme that degrades internal beta-1,4-glucosidic bonds found in plant cellulose and lichen materials, and is involved in glucose mobilization (Liu et al., 2011). Overall, the picture that emerges is that 8°C-acclimated crabs may be increasing free glucose, transporting that glucose and modifying low-molecular-weight osmolytes or cryoprotectant concentrations to a much greater extent than 18°C- or 13°C-acclimated crabs, a hypothesis that could be tested in future studies aimed specifically at carbohydrate metabolism in this species.
Other functional groupings of genes whose expression was higher in cold-acclimated cardiac tissues suggested increases in nucleic acid binding and chromatin remodeling activity, and structural or cytoskeletal remodeling (Table 1). Increased expression of these genes could indicate temperature compensation to overcome Q10 effects of the cold by increasing the protein amount of the same genes, or specific changes within the cells such as modification of euchromatin or restructuring of the cytoskeleton in a temperature-specific fashion. Regulation of transcription varied with thermal acclimation of fish heart muscle (Castiho, 2009), and many genes involved with regulation of transcription and translation, most notably transcription factors and RNA modification proteins, were differentially regulated with seasonal acclimatization in P. cinctipes (Stillman and Tagmount, 2009).
The strongest induction of any microarray features was observed in warm-acclimated specimens (Fig. 2, cluster B). Genes represented by those microarray features were largely for those that may be involved in immune responses through antimicrobial activity (e.g. carcinin), the complement and coagulation cascade (e.g. complement C1q, kalikrein and von Willebrand factor), as well as binding at the extracellular surface (e.g. L-selectin and cell wall proteins) (supplementary material Table S1). However, in cardiac tissues of field-acclimatized P. cinctipes, some of the same genes were observed to have higher expression in specimens collected during winter, and from sites with lower average temperatures (Stillman and Tagmount, 2009), a result that is contradictory to findings in the present study. Genes involved in immune responses have been observed in other functional genomic studies of crustacean responses to thermal stress [e.g. shrimp (Cottin, 2010)], as well as in response to infection in shrimp (Robalino, 2007). Genes associated with the immune response are upregulated 6 h after cold exposure in Drosophila melanogaster (Zhang et al., 2011), but not in flies exposed for shorter time periods (Qin et al., 2005), so it is possible that the immune response observed in the field is a longer-term response. Nevertheless, the specific role of immune response genes in thermal responses vs responses to infection and/or hemolymph clotting requires further investigation in both crustaceans and insects.
Chilling injury in insects is commonly thought to accrue via membrane phase transitions, which can be obviated via remodeling of the membrane (MacMillan and Sinclair, 2011). For example, cold acclimation induced membrane phospholipid changes consistent with homeoviscous adaptation in the flesh fly Sarcophaga similis (Goto and Katagiri, 2011), the Arctic collembolan Megaphorura arctica (Purac et al., 2011) and bivalve molluscs (Pernet et al., 2007). The fatty acid composition of P. cinctipes hearts was broadly consistent with that of muscle tissue of other crustaceans (Cuculescu et al., 1995). We did not observe remodeling of cardiac tissue membrane phospholipids in P. cinctipes during the first 24 h of thermal acclimation, nor did we detect differences in membrane phospholipids following 10–11 weeks of thermal acclimation. Low sample sizes in some cases may have influenced our ability to detect phospholipid species changes, as sample sizes ranged from two to seven across acclimation groups, but as variances were similar in groups with large and small sample sizes (Table 2), so this is not likely to be a big issue. We examined pooled phospholipid classes, and in some cases there was a differential response to long-term acclimation in the fatty acids linked to phosphatidylethanolamine and phosphatidylcholine (Cuculescu et al., 1999), but in those cases there was nevertheless a consistent shift towards unsaturation. There may be tissue-specific regulation of membrane phospholipids in cardiac tissue that do not follow expectations of homeoviscous adaptation. Tissue-specific variation in the degree of homeoviscous adaptation has been shown previously (Cossins and Prosser, 1978; Crockett and Hazel, 1995), and may be related to the degree to which the membrane is influenced by thermal change, given other environmental characteristics (e.g. pH) of the internal and external milieu surrounding the membrane (Crockett and Hazel, 1995). Whether crab cardiac tissue alters membrane physical structure to compensate for thermal changes is unclear, but for animals living in an environment where temperature changes rapidly and unpredictably, such as the marine intertidal zone, modification of membrane composition to achieve homeoviscous adaptation may sometimes (Williams and Somero, 1996) but not always (Rais et al., 2010) be the case, and changes in viscosity are not always associated with shifts in phospholipid fatty acid composition (Lahdes et al., 2000).
Here we show that gene expression and cold tolerance of P. cinctipes changes after only 6 h of thermal acclimation, and that this is accompanied by changes in gene expression after 12 h. However, membrane phospholipid fatty acid composition does not change in a manner consistent with homeoviscous adaptation in the first 24 h. Transcriptomic analysis suggests that regulation of glucose, or an end-product derived from glucose (e.g. trehalose), plays a role in enhancement of cold tolerance in these crabs, but this enhancement does not impart tolerance of internal ice formation. Our results form the basis for multiple testable hypotheses regarding the cellular mechanisms regulating the initial phase of thermal acclimation; we suggest that these hypotheses can be tested in future comparative studies among porcelain crab species that vary in their responses to habitat temperature change.
Thanks to Eddy Price and Heath MacMillan for their advice on fatty acid determination, Chris Guglielmo for the use of his gas chromatograph, Mani Tagmount and Claudia Tomas-Miranda for their advice on microarray analysis, and two anonymous referees for their criticism of an earlier version of the manuscript.
This research was funded by the University of Western Ontario Academic Development Fund (B.J.S.), the Canadian Foundation for Innovation (B.J.S.), The Natural Sciences and Engineering Research Council of Canada (B.J.S., J.P.W.), the National Science Foundation (J.H.S.) and San Francisco State University (D.R.).