Regulation of sarcoendoplasmic reticulum Ca²⁺-ATPase (SERCA) in turtle muscle and liver during acute exposure to anoxia

Several species of turtles have a superb ability for prolonged survival without oxygen. The freshwater red-eared slider, Trachemys scripta elegans, can live for as long as 3 months submerged in cold deoxygenated water, a feature that supports winter hibernation under water in ice-locked lakes (Jackson, 2004; Ultsch, 2006). Multiple biochemical adaptations contribute to anoxia tolerance, the most important of these being a profound anoxia-induced depression of resting metabolic rate (RMR) (Storey and Storey, 2004; Storey and Storey, 2007). In turtles, metabolic depression lowers RMR to just 10–20% of the corresponding aerobic rate at the same temperature (Herbert and Jackson, 1985). Depression of RMR includes both coordinated global suppression of all ATP-producing and ATP-consuming cell functions as well as a reprioritization of ATP use to sustain vital activities (e.g. ion pumping to sustain membrane potential difference) while shutting down optional activities such as protein synthesis and the cell cycle (Storey and Storey, 2007). For example, studies with turtle hepatocytes showed a net 94% decrease in ATP turnover under anoxia but a strong proportional increase in the fractional use of ATP by Na⁺/K⁺-ATPase (62% of total ATP turnover under anoxia versus 28% in normoxia) (Hochachka et al., 1996). Calcium pumping via the sarcoendoplasmic reticulum (SER) Ca²⁺-ATPase (SERCA) is a major ATP expenditure of cells, and we propose that, like Na⁺/K⁺-ATPase, SERCA must be differentially regulated under anoxic conditions in turtle tissues.

Calcium is a dynamic signaling molecule that triggers a variety of responses across all cell types (Shimoda et al., 2006). Cytoplasmic calcium concentration is tightly controlled, and usually maintained at low levels between 40 and 100 nmol l^–1 (Collins and Thomas, 2001). Influx of Ca²⁺ into the cytoplasm is gated by calcium channels, whereas calcium is actively transported out of the cytoplasm via plasma membrane Ca²⁺-ATPase (PMCA) and SERCA. In myocytes, depolarization triggers calcium release from SER stores and cytoplasmic Ca²⁺ functions in cross-bridge formation and contraction (Endoh, 2006). Cytoplasmic Ca²⁺ is then pumped, at the cost of one ATP per two Ca²⁺ ions moved, into the SER lumen by SERCA to re-establish resting conditions (MacLennan et al., 1997). SERCA activity is most prominent in muscle, accounting for 10–25% of basal ATP turnover (Clausen et al., 1991; Simonides et al., 2001), but SERCA is ubiquitous and consumes varying amounts of ATP in cells depending on cell type and physiological state.

In anoxia-intolerant species such as mammals, anoxia triggers an immediate and drastic increase in intracellular calcium due to ATP depletion. For example, cytoplasmic [Ca²⁺] rises by more than 10-fold in mammalian neurons within minutes when oxygen is depleted (Bickler et al., 2001) and multiple events triggered by Ca²⁺ are responsible for many of the injuries to mammalian cells caused by anoxia or ischemia (Hochachka, 1986). By contrast, turtle neurons show a much lower rise in cytoplasmic Ca²⁺ of ∼2-fold within the first few hours of anoxia exposure, before stabilizing at this level over subsequent weeks of anaerobiosis (Bickler, 1998). Extracellular calcium levels also rise in anoxic turtles, due to calcium released from shell and bone matrices to provide buffering of the byproducts of anaerobic fermentation (Jackson, 2002; Jackson, 2004). This mobilization of calcium results in increased [Ca²⁺] in plasma and cerebrospinal fluid over the duration of anoxia exposure (Cserr et al., 1998; Reese et al., 2001).

The present study examined the regulation of SERCA in response to anoxic submergence in T. s. elegans tissues. A stable decrease in activity occurred in several tissues and significant differences in SERCA kinetic and stability properties were identified when aerobic versus anoxic forms were analyzed in skeletal muscle and liver. The molecular basis of changes in SERCA properties between the two states was traced to reversible phosphorylation of the enzyme.

This research was approved by the Carleton University Animal Care Committee. Adult male red-eared slider turtles (T. s. elegans Schoepff) of ∼20–25 cm shell length were acclimated in large tanks filled with running dechlorinated water (5–7°C) and fed ad libitum on a diet of trout pellets, lettuce and egg shells for ∼6 weeks prior to experimentation. Control (aerobic) turtles were sampled from this condition. For anoxia exposure, some turtles were transferred to 40 l tanks (three animals per tank) filled with deoxygenated water (previously bubbled with nitrogen gas for several hours) at 5–7°C. A wire mesh placed 20 cm below the water surface prevented the animals from surfacing. Nitrogen gas was continuously bubbled for the duration of the anoxic treatment (1, 5 or 20 h). Animals were killed by decapitation; tissues were immediately removed and frozen in liquid nitrogen and then stored at –80°C until use.

The mouse anti-chicken IgG serum SERCA antibody (CaF2-5D2) was purchased from the Developmental Studies Hybridoma Bank (DSHB, University of Iowa, Iowa City, IA, USA), and HRP-conjugated goat anti-mouse secondary antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Okadaic acid was purchased from Calbiochem (La Jolla, CA, USA). All other chemicals were obtained from Sigma Corporation (St Louis, MO, USA).

Frozen tissue samples were homogenized 1:5 (w:v) using a Polytron PT1000 homogenizer (Brinkmann Instruments, Rexdale, ON, Canada) in pre-chilled (4°C) extract buffer containing 25 mmol l^–1 imidazole pH 7.2, 10% (v:v) glycerol, 0.2% (w:v) sodium deoxycholate, 250 mmol l^–1 sucrose and 10 mmol l^–1 2-mercaptoethanol, with the following additions included to inhibit endogenous kinases and phosphatases: 2 mmol l^–1 EDTA, 5 mmol l^–1 spermine and 25 mmol l^–1 β-glycerol phosphate (βGP). A few crystals of phenylmethylsulfonyl fluoride (PMSF), 1:1000 (v:v) of Sigma Protease Inhibitor Cocktail (P8340) and 1:1000 (v:v) of Sigma Phosphatase Inhibitor Cocktail 1 (P2850) were added just prior to homogenization. Homogenates were centrifuged at 10,000 g for 20 min at 4°C. Both the pellet [insoluble protein, resuspended in an equal volume of extract buffer containing 1% (w:v) sodium deoxycholate] and supernatant (containing soluble protein) fractions were desalted by low speed centrifugation through small columns of Sephadex G25 (300 μl of supernatant per 3 ml of G25) equilibrated in extract buffer to remove endogenous ions, small molecules and free phosphate, and assayed immediately.

General serine/threonine and tyrosine phosphatase inhibitors (NaF and Na₃VO₄) were omitted from the homogenizing buffer because they reduced recoverable SERCA activity, and alternative reagents were used (1 mmol l^–1 EDTA, 5 mmol l^–1 spermine, 25 mmol l^–1 βGP, and Sigma Phosphatase Inhibitor Cocktail 1 to inhibit endogenous kinases and phosphatases). Different detergents (deoxycholate, Brij-35, CHAPS) were tried during homogenization to solubilize turtle SERCA and 0.2% sodium deoxycholate gave the highest recoverable activity in the soluble fraction with negligible activity remaining in the pellet. Thus, the data reported are for SERCA in soluble extracts homogenized in the presence of 0.2% sodium deoxycholate.

SERCA activity was determined from assays in the absence versus presence of 50 nmol l^–1 thapsigargin (TG), a specific inhibitor of SERCA (Lytton et al., 1992). The difference in activity was attributed to SERCA. Initial experiments determined that 50 nmol l^–1 TG was sufficient to abolish 100% SERCA activity in crude extracts. Assays were initiated by adding 25 μl of desalted extract to paired reaction tubes (±50 nmol l^–1 TG) in 250 μl final reaction volumes containing 25 mmol l^–1 imidazole pH 7.0, 2 mmol l^–1 MgATP, 400 μmol l^–1 CaCl₂, 50 mmol l^–1 KCl, 5 mmol l^–1 MgCl₂ and 2 μmol l^–1 calcium ionophore. The reaction proceeded for 20 min at 22°C. Blank reaction tubes contained no calcium and 5 mmol l^–1 EGTA. Thus, T. s. elegans SERCA was defined as calcium-dependent, TG-sensitive, ATPase activity. Initial experiments validated that enzyme activity was linear with respect to time and extract volume, and optimized assay conditions with respect to pH, ATP and Ca²⁺concentration. SERCA activity was assayed over a range of MgATP (0–250 μmol l^–1) and Ca²⁺ (0–5 μmol l^–1) concentrations to determine kinetic constants. SERCA activity was assayed spectrophotometrically by measuring the amount of inorganic phosphate produced using a malachite green/ammonium molybdate dye reagent; this red dye reagent turns green when it forms a complex with inorganic phosphate, and intensity is quantified (OD₅₉₅). The reagent was prepared as previously described (MacDonald and Storey, 1999). SERCA assays were stopped by removing 50 μl aliquots of the assay mix and mixing these into tubes containing 750 μl of ddH₂O and 200 μl of the dye reagent. After thorough mixing, color was allowed to develop for ∼10 min followed by measurement of OD₅₉₅ using an MR5000 microplate reader (Dynatech Laboratories, Chantilly, VA, USA) and Biolinx 2.0 software. Phosphate quantity was determined from a standard curve prepared with known amounts of KH₂PO₄.

Soluble protein content was quantified using the Coomassie Blue dye binding method and the BioRad prepared reagent (Hercules, CA, USA) with a standard curve of bovine serum albumin. SERCA activity was reported as mU mg^–1 soluble protein (1 mU=1 nmol phosphate released per min); soluble protein content of extracts did not change between aerobic and anoxic conditions.

For studies characterizing the Michaelis constant (K_m) for MgATP and maximal velocity (V_max) as a function of temperature, assay reaction tubes were incubated in water baths at the appropriate temperature (4–68°C) for ∼10 min before initiating the reaction. Assays were stopped and phosphate production determined as previously described. For V_maxversus temperature studies, activation energy (E_a) values were determined (in kJ mol^–1) from the linear portion of Arrhenius plots.

To assess the effects of phosphorylation state on enzyme activity, soluble tissue extracts were prepared in extract buffer and desalted on G25 Sephadex equilibrated in extract buffer (without phosphatase and kinase inhibitors) as previously described. Then aliquots were incubated in vitro at 22°C for 8 h under conditions that promoted either protein phosphorylation or protein dephosphorylation (Ramnanan and Storey, 2008). Desalted tissue extracts were incubated 1:3 (v:v) with 25 mmol l^–1 imidazole pH 7.2, 10% (v:v) glycerol, 250 mmol l^–1 sucrose and 10 mmol l^–1 2-mercaptoethanol with additions as follows.

Control incubations: 5 mmol l^–1 spermine, 25 mmol l^–1 βGP, 2 mmol l^–1 EDTA, and 1:1000 (v:v) Sigma Phosphatase Inhibitor Cocktail 1 (P2850) to inhibit all kinase and phosphatase activities.

Stimulation of endogenous kinase activities: 5 mmol l^–1 MgATP, 5 mmol l^–1 spermine, 25 mmol l^–1 βGP, 1:1000 P2850, and one of the following: (i) 1 mmol l^–1 cAMP to stimulate protein kinase A (PKA); (ii) 1 mmol l^–1 cGMP to stimulate protein kinase G (PKG); (iii) 1.3 mmol l^–1 CaCl₂ + 7 μg ml^–1 phorbol myristate acetate to stimulate protein kinase C (PKC); (iv) 1 U calmodulin activity per incubation tube + 1.3 mmol l^–1 CaCl₂ to stimulate calcium/calmodulin-dependent protein kinase (CAMK); or (v) 20 mmol l^–1 AMP to stimulate AMP-activated protein kinase (AMPK).

Stimulation of endogenous protein phosphatase (PP) activities: (i) for total PP activity: 5 mmol l^–1 CaCl₂ and 5 mmol l^–1 MgCl₂; (ii) for total PP minus PP2A: 5 mmol l^–1 CaCl₂ and 5 mmol l^–1 MgCl₂ + 10 nmol l^–1 okadaic acid (abolishes 100% of PP2A activity); (iii) for PP1/PP2A activity: 2 mmol l^–1 EGTA + 10 nmol l^–1 cypermethrin + 2 mmol l^–1 EDTA; (iv) for total PP minus PP1/PP2A activity: 5 mmol l^–1 CaCl₂ and 5 mmol l^–1 MgCl₂ + 1 μmol l^–1 okadaic acid (abolishes 100% of PP1 and PP2A activity); (v) for PP2B activity: 5 mmol l^–1 CaCl₂ + 1 U of calmodulin activity per tube + 2 mmol l^–1 EDTA + 1 μmol l^–1 okadaic acid; (vi) for PP2C activity: 5 mmol l^–1 MgCl₂ + 1 μmol l^–1 okadaic acid + 10 nmol l^–1 cypermethrin; or (vii) to promote full dephosphorylation: 1 U commercial calf intestinal alkaline phosphatase (AP) per incubation tube + 5 mmol l^–1 CaCl₂ and 5 mmol l^–1 MgCl₂.

Following incubation, extracts were desalted using low speed centrifugation through small columns of G25 Sephadex equilibrated in extract buffer and SERCA activity was assayed immediately.

Kinetic analysis to determine K_m, V_max, E_a and I₅₀ (inhibition constant for 50% inhibition) values used a Kinetics computer program (Brooks, 1992). Kinetic constants were calculated using the concentration of free ions during assay conditions as determined by the Bound and Determined (BAD) computer program (Brooks and Storey, 1992).

Isolation, electrophoresis, blotting and immunodetection of SERCA protein were performed essentially as described previously (Ramnanan and Storey, 2008) with the following minor modifications. Aliquots containing 35 μg of soluble protein were subjected to electrophoresis and proteins were subsequently wet-transferred to PVDF membranes. Membranes were blocked with 5% non-fat dried milk in Tris-buffered saline containing Tween-20 (TBST: 10 mmol l^–1 Tris-base pH 7.0, 150 mmol l^–1 NaCl, 0.1% v:v Tween 20) for 1 h at room temperature. Membranes were incubated overnight at 4°C with mouse serum anti-SERCA (CaF2-5D2) diluted 1:1000 v:v in TBST. After 3×5 min washes with TBST, membranes were then incubated with HRP-conjugated goat anti-mouse secondary antibody (diluted 1:2000 v:v in TBST) for 1 h at room temperature. Following 3×5 min washes in TBST, immunodetection was performed using Western Lightning Chemiluminescence Plus reagents (Perkin Elmer, Boston, MA, USA) following the manufacturer's protocols, and imaged and quantified using the Syngene ChemiGenius bio-imaging system (Frederick, MD, USA) and associated GeneTools software (v3.00.02). Samples were normalized for slight variations in protein loading as described previously (Ramnanan and Storey, 2006; Ramnanan and Storey, 2008).

We assessed the structural stability of SERCA in muscle and liver by treating extracts with increasing concentrations of denaturant (urea) followed by pulse proteolysis incubation with thermolysin, as described previously (Ramnanan and Storey, 2006). Tissue extracts were prepared as described above with the exception that protease inhibitors were omitted during homogenization. Aliquots of soluble crude extract were incubated for 24 h in urea solutions (0–3.4 mol l^–1 final concentration) prior to treatment with 10 mg ml^–1 thermolysin (stock prepared with 2.5 mol l^–1 NaCl and 10 mmol l^–1 CaCl₂). Thermolysin activity was stopped after 10 min by the addition of 20 μl of 50 mmol l^–1 EDTA (pH 8.0). SERCA protein content was assessed by western blotting and parameters of protein unfolding (C_m) were calculated.

Statistical analysis was performed using Graph-Pad Prism 5.0. All data are expressed as means ± s.e.m. For comparision of two groups, Student's t-test was used. For comparison of more than two groups, ANOVA was performed followed by post hoc Tukey's Multiple Comparison to determine significance between groups.

Total SERCA protein content and maximal activity (V_max) were assessed in several tissues of normoxic and anoxic turtles (Fig. 1). Immunoblotting detected a single band corresponding to the typical size of vertebrate SERCA (110 kDa). After 20 h of anoxic submergence, SERCA protein levels remained unchanged in skeletal muscle, liver and heart, but decreased by 36% and 24% in kidney and brain, respectively (Fig. 1A,B). SERCA V_max decreased by 46% in muscle, 47% in liver and 35% in heart of anoxic turtles, but did not change significantly in other tissues (Fig. 1C). Subsequent experiments focused on characterizing SERCA in skeletal muscle and liver.

SERCA V_max and K_m for MgATP and Ca²⁺ substrates were characterized as a function of temperature using enzyme extracts from both tissues, and similar trends were observed in liver (Fig. 2) and muscle (data not shown). When assayed at intermediate temperatures (15–25°C), the K_m for MgATP was significantly higher for SERCA from 20 h anoxic turtles as compared with the enzyme from liver of normoxic control turtles. Conversely, when assayed at low temperatures (4–10°C), K_m MgATP values for SERCA from 20 h anoxic turtles were significantly lower than those for the normoxic control enzyme (Fig. 2A). K_m for MgATP was generally stable (between 31 and 37 μmol l^–1 MgATP) in anoxic liver at all assay temperatures, but K_m MgATP values for SERCA from normoxic liver were generally elevated at low temperatures. No temperature-dependent or anoxia-induced effects were observed for calcium kinetics of the liver (Fig. 2B) or muscle enzyme. The effect of temperature on SERCA V_max was then analyzed and is shown as an Arrhenius plot (log V_maxversus inverse temperature in Kelvin) in Fig. 2C. This revealed a linear relationship from 4 to 25°C, with a break point apparent at 25°C (Fig. 2C). The calculated E_a from the linear portions of these plots was increased by 60% and 41% for muscle and liver SERCA, respectively, of turtles subjected to 20 h anoxic submergence (Fig. 2D).

Cellular pH decreases in tissues when turtles are anoxic because of metabolic acidosis associated with glycolytic ATP production and lactic acid accumulation. We examined SERCA activity as a function of pH and similar results were generally observed in muscle (Fig. 3) and liver (data not shown). Normoxic SERCA was more sensitive to assay pH than anoxic SERCA in muscle. Anoxic muscle SERCA was relatively stable (∼280–360 mU mg^–1 protein) between pH 6.6 and pH 8.0 (Fig. 3A). Conversely, normoxic SERCA showed a broad peak of activity that was greater than anoxic muscle SERCA activity between pH 7.0 and 7.4, but normoxic SERCA activity sharply declined on either side of that range, and was less than that of anoxic muscle SERCA at low pH (6.6–6.8). K_m MgATP of muscle SERCA was also assayed over a range of pH values (Fig. 3B). K_m MgATP of anoxic SERCA was significantly higher (i.e. lower affinity) than that of the normoxic enzyme at higher pH values (7.0, 7.4) but the situation was reversed at pH 6.6 when K_m of the anoxic enzyme decreased (i.e. affinity decreased).

To determine whether there was a difference in the structural stability of SERCA between normoxic and anoxic states, the liver and muscle enzymes from both conditions were incubated with varying concentrations of the denaturant urea, followed by pulse proteolysis with thermolysin to selectively degrade denatured (unfolded) protein. The remaining amount of native protein was then assessed by immunoblotting with quantification of the intensity of the 120 kDa band. As shown in representative western blots (Fig. 4A) and a plot of relative protein versus urea concentration (Fig. 4B), the population of SERCA in liver from anoxic animals was more stable after urea treatment/thermolysin-mediated proteolysis. The kinetic constant for unfolding (C_m; the urea concentration required to denature 50% of test protein) was 33% and 41% higher, respectively, for SERCA from muscle and liver of anoxic versus normoxic turtles (Fig. 4C).

Fig. 1 showed that SERCA activity was strongly suppressed after 20 h of anoxia exposure. To determine how quickly enzyme activity was shut down, turtles were sampled over a time course of anoxic submergence. In both tissues examined, SERCA activity decreased rapidly, by 39% and 32% in muscle and liver, respectively, within 1 h of anoxia exposure (Fig. 5). Subsequently, SERCA activity decreased further over time to reach levels that were 46% and 47% of the corresponding normoxic values in muscle and liver, respectively, after 20 h of anoxia treatment.

To determine whether the differences in SERCA enzyme kinetics during oxygen deprivation were due to anoxia-induced phosphorylation of the enzyme, tissue extracts were incubated under conditions that promoted the activity of specific endogenous protein kinases or protein phosphatases. Data are presented relative to SERCA activity recovered from control incubations that contained inhibitors of all kinases and phosphatases. In muscle, incubations promoting PKA or PKG activity reduced SERCA activity in extracts from normoxic turtles by 65% and 83%, respectively, to levels more typical of anoxia-treated animals (Fig. 6A). Incubation under conditions that stimulated PKG also decreased SERCA activity in extracts of anoxic muscle by ∼50%. Stimulation of PKC activity had no effect on SERCA activity. Incubations stimulating individual endogenous protein phosphatases failed to affect recoverable muscle SERCA activity (data not shown). However, incubation with a commercial heat-stable AP led to 56% and 246% increases in SERCA activity in extracts from normoxic and anoxic muscle, respectively (Fig. 6A). The changes in muscle SERCA activity were related to changes in substrate kinetics (Fig. 6B), where PKG stimulation increased K_m MgATP (reduced affinity) for both normoxic and anoxic SERCA, and incubation with AP decreased K_m MgATP (increased affinity) of the anoxic enzyme.

With liver extracts, incubations promoting PKA and PKC led to 41% and 59% reductions in SERCA activity from normoxic animals, but did not affect recoverable SERCA activity in extracts from anoxic turtles (Fig. 6C). Incubations to stimulate PKG were ineffective in modifying liver SERCA activity (data not shown). Stimulation of endogenous ion-independent phosphatases (PP1/PP2A) led to an increase in recoverable SERCA activity of 40% and 100% for the normoxic and anoxic enzyme, respectively (Fig. 6C), but incubations that promoted ion-dependent phosphatases were ineffective (data not shown). This stimulatory effect was blocked in an incubation containing 10 nmol l^–1 okadaic acid, which inhibits endogenous PP2A but not PP1 (i.e. Total PP2A, Fig. 6C) and was also blocked in all other incubations that contained higher okadaic acid (1 μmol l^–1) which inhibited both PP2A and PP1. Hence, the stimulatory effect of endogenous phosphatases was localized to the action of PP2A. Again, changes in SERCA activity in liver were consistent with changes in K_m MgATP (Fig. 6D); incubations promoting PKA increased the K_m MgATP of normoxic SERCA to levels associated with anoxic SERCA, whereas promoting PP2A activity decreased K_m MgATP in anoxic extracts to levels similar to those observed for normoxic SERCA.

Pronounced depression of RMR is crucial for survival without oxygen by anoxia-tolerant species, and must be carefully regulated to achieve the overall suppression of metabolism while still fueling vital cellular processes. It stands to reason that ion-motive ATPases, due to their energetically demanding functions in maintaining the transmembrane ion gradients that are necessary for life, are critical loci for regulation during transitions to or from states of hypometabolism. For example, the overall metabolic rate of turtle hepatocytes was reduced by 95% in response to anoxia, but the fraction of ATP consumed by the ATP-dependent sodium–potassium pump (Na⁺/K⁺-ATPase) rose from 28% to 75% of overall ATP hydrolyzed (Hochachka et al., 1996). Indeed, it has been shown that Na⁺/K⁺-ATPase and SERCA are closely regulated and are strongly suppressed in hypometabolic states including hibernation, estivation and diapause (MacDonald and Storey, 1999; Hemmings and Storey, 2001; Malysheva et al., 2001; Ramnanan and Storey, 2006; Ramnanan and Storey, 2008; McMullen and Storey, 2008). Anoxia has been shown to suppress SERCA activity in neurons (Henrich and Buckler, 2008) and hypoxia decreased SERCA expression in hearts (Pei et al., 2003) of rats. If SERCA is sensitive to low oxygen in mammals, then it is likely to be a major target for energy savings in anoxia-tolerant species. We proposed that SERCA would be differentially regulated in response to anoxia exposure in tissues of T. s. elegans, a well-established model of vertebrate facultative anaerobiosis (Lutz, 1992; Hochachka and Lutz, 2001; Jackson, 2002; Storey and Storey, 2007), as part of the coordinated transition to the hypometabolic state.

A degree of ATP conservation would be expected in skeletal muscle during winter hibernation of turtles because of the overall infrequency of muscle movements. However, ATP consumption by ion-motive ATPases is a major part of RMR for all animals. In particular, it is well established that thyroid hormone increases RMR by stimulating SERCA expression and activity. The RMR of hypothyroid and hyperthyroid mammals is ∼70% and ∼150% of euthyroid values, respectively, and is associated with proportional changes in SERCA activity (Clausen et al., 1991; Simonides et al., 2001). Therefore, it is logical to propose that the anoxia-induced transition to a hypometabolic state must include strong suppression of SERCA activity as a key part of long-term anaerobiosis and the present study shows that this is indeed true for the best-studied vertebrate facultative anaerobe, the freshwater turtle T. s. elegans.

We characterized SERCA activity in crude extracts of T. s. elegans by using the SERCA-specific inhibitor TG to differentiate between SERCA and PMCA activities; TG binds SERCA (but not PMCA) and is widely used to study SERCA activity without the need for SERCA purification (Treiman et al., 1998). After 20 h of anoxia exposure, SERCA maximal activity decreased in muscle, liver and heart, without a change in SERCA total protein (as measured by immunoblotting). Conversely, SERCA protein levels decreased in brain and kidney of anoxic turtles without causing a change in SERCA activity (Fig. 1). Hence, SERCA activity in turtle organs was not strictly determined by the amount of SERCA protein.

Substrate affinity of liver SERCA for MgATP differed between normoxic and anoxic states, and at higher assay temperatures (15–25°C) the higher K_m MgATP of the anoxic enzyme would suggest a less active enzyme form under anoxic conditions (Fig. 2A). At low temperatures (4–10°C), however, the opposite was true which could be an advantage during winter hibernation in cold water. Notably, the K_m value of the anoxic enzyme form was actually stable across the temperature range whereas the K_m of normoxic SERCA changed 2-fold over the same temperature range; this feature demonstrates a substantial difference between the two enzyme forms.

The activity of anoxic SERCA showed good stability with respect to pH change, whereas normoxic SERCA activity declined sharply below pH 7.0. However, at a typical physiological pH (pH 7.0–7.4), SERCA from anoxia-exposed animals had a lower V_max (Fig. 3A) and decreased substrate affinity for MgATP (Fig. 3B) relative to normoxic SERCA, consistent with a less active form of the enzyme. At pH values <7.0, however, the situation was reversed and anoxic SERCA activity was higher and its K_m MgATP was lower than the normoxic form when assayed at pH 6.6. While this pH is subphysiological, physiological pH in turtles does gradually decrease over long (weeks to months) periods of anoxia. Thus, anoxic SERCA activity is conferred a degree of stability by an early (within 20 h of initial anoxia submergence) modification that may be beneficial for maintaining activity during longer bouts of anoxia exposure that include a substantial decline in the pH of body fluids.

The enhanced kinetic stability of SERCA with respect to temperature and pH in anoxic tissues also corresponded to increased structural and conformational stability. Thus, SERCA from liver of 20 h anoxic turtles was more resistant to urea denaturation and unfolding than was the normoxic form (Fig. 4). Thus, whereas the SERCA population in muscle and liver was dominated by the lower activity form of the enzyme after 20 h of anoxia exposure, this low activity conformation had increased stability in terms of substrate kinetics, susceptibility to urea denaturation, and ability to function efficiently (greater affinity for ATP) at the low temperatures and low pH values that T. s. elegans tissues would encounter during extended bouts of winter hibernation under water.

The Arrhenius relationship (Fig. 2C) was generally linear from the lowest temperature assayed (4°C) up to 22°C, which correlates well with the temperature range naturally encountered by turtles in their native habitat. A break in an Arrhenius plot is indicative of a temperature-dependent shift in a rate-limiting aspect of the enzymatic reaction, and can be a marker for a deleterious shift in the local, membrane lipid environment of an ATPase. Lipid membranes are well established as regulators of SERCA activity (Lee, 1998). Previous studies have established that Arrhenius breaks in SERCA plots are tightly correlated with environmental temperatures faced by the organism and this pattern holds true across the animal kingdom, including (i) rabbits where the break occurred at 20°C (Rubstov et al., 1986); (ii) cold water invertebrates like the deep sea scallop Placopecten magellanicus and lobster Homarus gammarus that show Arrhenius breaks at 10°C (Madeira et al., 1974; Kalobokis and Hardwicke, 1998); (iii) several species of fish where the break points were related to ambient living temperatures (Godiksen and Jessen, 2001; Godiksen and Jessen, 2002); and (iv) the desert snail Otala lactea where the break occurred above 30°C (Ramnanan and Storey, 2008). Calculated from the linear part of the relationship, the activation energy for turtle SERCA was increased in both tissues after 20 h of anoxia exposure, consistent with a less active form of the enzyme (Fig. 2D).

The transition to anoxia-induced hypometabolism in T. s. elegans has been shown to be associated with substantial changes in the phosphorylation state of enzymes and the activities of protein kinases/phosphatases (Greenway and Storey, 2000; Rider et al., 2009). Therefore, in light of previous studies of the regulation of ATP-dependent ion pumps in hypometabolic states (MacDonald and Storey, 1999; Ramnanan and Storey, 2006; Ramnanan and Storey, 2008), we proposed that the mechanism responsible for SERCA regulation in turtle tissues during anoxia exposure would be reversible protein phosphorylation.

Strong suppression (39% and 32% in muscle and liver) of SERCA activity was evident within 1 h of anoxia submergence (Fig. 5) and activities remained low after 20 h of anoxia exposure. Changes in SERCA activity occurred without a decrease in SERCA protein (Fig. 1) indicating that proteolysis was not a factor. This suggested that the stable suppression of SERCA activity in response to acute anoxia exposure was due to a post-translational modification of the protein, presumably by reversible protein phosphorylation. This is consistent with the observation that substantial changes in the phosphorylation state of proteins involved in signal transduction (Greenway and Storey, 2000) and protein synthesis (Rider et al., 2009) are also evident within hours of anoxia exposure in this model.

Strong evidence for this mechanism in the regulation of turtle SERCA was obtained from in vitro incubations that stimulated the activities of specific endogenous kinases and phosphatases and measured the consequent effects on SERCA activity. Stimulation of PKA and PKG in muscle (Fig. 6A) and PKA and PKC in liver (Fig. 6C) significantly reduced SERCA activity compared with that in normoxic extracts, similar to the effects of in vivo anoxia exposure on the enzyme. PKG stimulation also further reduced SERCA in anoxic muscle extracts, whereas no kinases affected SERCA activity in anoxic liver, suggesting that the enzyme was already in a maximally phosphorylated and inhibited state. Thus, regulatory phosphorylation of SERCA seems to occur in a tissue-specific manner that includes differential responses to the three kinases evaluated. Furthermore, these kinases generally increased the K_m MgATP for SERCA in normoxic extracts to levels that were approximately equal to the K_m values of anoxic SERCA in muscle (Fig. 6B) and liver (Fig. 6D).

Incubation with AP strongly increased SERCA activity in both normoxic and anoxic muscle extracts (Fig. 6A), and decreased K_m MgATP in anoxic extracts (Fig. 6B), indicating that SERCA activity can be enhanced by dephosphorylation. In anoxic liver extracts, SERCA activity also increased in response to phosphatase action (Fig. 6C) and K_m MgATP decreased (Fig. 6D). Both of these responses in liver appeared to be specifically stimulated by PP2A, and these effects were blocked under any incubation condition that involved inhibition of PP2A with okadaic acid. These data suggest that regulatory phosphorylation mediates the anoxia-dependent changes in SERCA function in both turtle liver and muscle, under tissue-specific kinase and phosphatase action. The change in SERCA activity seems to be related to a change in substrate affinity for MgATP, and the in vivo dephosphorylation of liver SERCA appears to be mediated by PP2A.

Transiently phosphorylated catalytic intermediates have been established for SERCA (Ryan et al., 2004). While liver SERCA is not typically known to have sites that are targets for stable, covalently bound phosphate with regulatory properties, our findings do have a precedent in muscle (Hawkins et al., 1994). Additionally, the presence of covalently bound phosphate on bay scallop SERCA was shown to increase stability of both structure and enzymatic activity (Ryan et al., 2004), comparable to the results seen in anoxic turtle tissues at low pH and low temperature.

Thus, it is apparent that SERCA activity is suppressed in liver and muscle in T. s. elegans early (within 1 h) into anoxia exposure via stable alterations to enzyme activity and kinetic properties. Our data are consistent with the mechanism being protein phosphorylation with dephosphorylation of the anoxic enzyme reinstating properties that are like the normoxic enzyme. This post-translational mechanism of regulating SERCA also stabilizes enzyme structure and could help to protect/stabilize the enzyme during long-term anoxic submergence while also being quickly reversible when animals transition back to breathing air again. Thus, there is a conserved drive in vertebrates to protect SERCA protein content and/or activity. Regulatory phosphorylation allows for quick attenuation of enzyme function, as evident in anoxic turtles where SERCA was suppressed after 1 h of anoxia exposure. This fits in with the overall move to suppress RMR under anoxic conditions, as SERCA consumes a considerable percentage of cellular ATP even in tissues at rest.

Thanks to J. M. Storey for editorial comment on the manuscript.

 
• AMPK

  AMP-activated protein kinase

 
• AP

  alkaline phosphatase

 
• C_m

  constant for protein unfolding

 
• CAMK

  calcium/calmodulin-dependent protein kinase

 
• E_a

  Arrhenius activation energy

 
• I₅₀

  inhibition constant

 
• K_m

  Michaelis constant

 
• PKA

  protein kinase A

 
• PKC

  protein kinase C

 
• PKG

  protein kinase G

 
• PMCA

  plasma membrane Ca²⁺-ATPase

 
• PP

  protein phosphatase

 
• PP1

  type 1 protein phosphatase

 
• PP2

  type 2 protein phosphatase (A,B,C)

 
• RMR

  resting metabolic rate

 
• SER

  sarcoendoplasmic reticulum

 
• SERCA

  sarcoendoplasmic reticulum Ca²⁺-ATPase

 
• TG

  thapsigargin

 
• V_max

  maximal enzyme velocity