Antioxidant response to acute cold exposure and during recovery in juvenile Chinese soft-shelled turtles (Pelodiscus sinensis)

Intense temperature change, as occurs during acute cold exposure, disturbs an animal's oxygen consumption and energy metabolism, and can even cause mortality (Johnston and Bennett, 2008; Şahin and Gümüşlü, 2004). These adverse effects are believed to be related to oxidative stress – the imbalance between generation and elimination of reactive oxygen species (ROS) – in both endotherms and ectotherms. Among ectotherms, studies on North Sea eelpout Zoarces viviparous (7°C reduction maintained for 2 h), three spine stickleback Gasterosteus aculeatus (acclimated at 8°C for 1 week), and the common Indian toad Bufo melanostictus (about 24°C reduction maintained for 30 min) all showed increased oxidation of macromolecules in response to cold exposure and/or following recovery, suggesting elevated oxidative stress (Heise et al., 2006; Kammer et al., 2011; Sahoo and Kara, 2014). It was also reported that cold exposure significantly induced ROS production in spiny-tailed lizards Uromastyx aegyptius (Al-Johany and Haffor, 2007). However, different results were found in several freshwater turtle species. After enduring severe cold stress, and even whole body freezing, no signs of physiological damage was detected in turtle species (Baker et al., 2007; Storey, 2006). Our previous studies of the subtropical Chinese soft-shelled turtle Pelodiscus sinensis also indicated no increase in oxidative damage after severe cold exposure (a reduction from 28°C to 8°C in less than 5 min that was then maintained for 2–12 h) (Chen et al., 2015; Zhang et al., 2017b). All these results indicate a unique ROS balancing capacity, which may promote the excellent cold stress tolerance in freshwater turtles. To further understand the relationship between cold stress tolerance and ROS balancing capacity in freshwater turtles, it is necessary to understand the response of the antioxidant system, which works in clearing excess ROS, to acute temperature reduction. The antioxidant system in turtles has two components: the enzymatic antioxidant system and the non-enzymatic antioxidant system (Davies, 2000; Storey, 2006). The former includes classic antioxidant enzymes, such as superoxide dismutase (SOD), catalase (CAT) and glutathione peroxidase (GPx) (Willmore and Storey, 1997a). Gene expression of most antioxidant enzymes is regulated by activation of the nuclear factor erythroid 2-related factor 2 (Nrf2) antioxidant response element signaling pathway (Lushchak, 2014; Motohashi and Yamamoto, 2004). Upregulation of Nrf2 mRNA expression and its downstream antioxidant enzymes is considered to be an important measure for freeze tolerance of North American turtles during their winter hibernation (Krivoruchko and Storey, 2010a; Storey, 2006). Although not upregulated, Nrf2 also exhibited a possible regulatory role in antioxidant defense in Chinese soft-shelled turtle hatchlings during hibernation, wherein temperature never drops below freezing (Zhang et al., 2017a). Hence, our objective included investigating the enzymatic antioxidant system response and its regulation with regard to acute cold exposure at temperatures above 0°C.

The non-enzymatic antioxidant system includes many low molecular weight ROS scavengers and their related enzymes (Venditti et al., 2010). Among these, glutathione (GSH) and ascorbic acid (AA) are the best known (Rice et al., 2002; Willmore and Storey, 1997b), not only due to their exceptional ROS quenching efficiency and abundant tissue distribution, but also due to their direct responses to environmental stresses (Chen et al., 2015; Pérez-Pinzón and Rice, 1995; Rice, 2000; Zhang et al., 2017c). The circulation of these low molecular weight antioxidants depends on their regulating enzymes, for instance, GSH is a substrate of some antioxidant enzymes such as GPx or it can be oxidized to GSSG during decomposition of peroxides. Its de novo synthesis is mediated by the enzymes γ-glutamylcysteine synthetase and glutathione synthetase (GSS), whereas its oxidized form GSSG can be reduced to GSH by glutathione reductase (GR). In addition, the glutathione S-transferase (GST) family can conjugate GSH to various nucleophilic xenobiotics or oxidized cellular components to detoxify them (Halliwell and Gutteridge, 2015; Willmore and Storey, 1997b; Zhang et al., 2017c). Our previous studies found that the antioxidant-related enzymes of the AA system exhibited different responses to acute cold exposure and to long-term hibernation (Chen et al., 2015, 2018). However, how the regulation of GSH-related enzymes responds to these two stresses was not known. Thus, our objective here also includes investigating the glutathione (GSH) system response and regulation to acute cold exposure without freezing.

The Chinese soft-shelled turtle P. sinensis is a commercially important species with remarkable medicinal and economical value in Southeast Asia (Gong et al., 2011; Niu et al., 1999). Here, we explored the regulation of the antioxidant defense system in this species in response to acute cold exposure. Juvenile turtles were acclimated at 28°C for 3 weeks, and then they were exposed to acute cold conditions (8°C) for 12 h, followed by a 24 h recovery period at 28°C. We measured changes in mRNA transcripts for Nrf2 and downstream antioxidant genes under its control (SOD, CAT and GPx) and the corresponding activities of these enzymes in the brain, liver and kidney. We also assessed changes in mRNA transcript levels of GSH-related genes (GSS, GR and GST isoforms) and their relative enzyme activities.

The experiments were conducted in accordance with the standards of the Ethic and Animal Welfare Committee (EAWC) of Beijing Normal University (approval no. CLS-EAW-2013-004).

Turtles (n=24; mass: 90.3±2.6 g) were purchased from a turtle hatchery facility (Yutian county, Hebei province, PRC). The optimum growth temperature for Chinese soft-shelled turtle ranges from 25 to 32°C (Chen et al., 2015) and all turtles were reared at 28±0.5°C in a recirculating-water tank system with free access to basking platforms and diving water (patent no.: ZL 201310189957.0) to acclimate for 3 weeks. The photoperiod was maintained at 12h light:12 h dark and the turtles were fed daily with commercial standard diets (Hebei Haitai Tech. Ltd, Shijiazhuang, PRC).

Considering the lowest temperature in the habitat during hibernation – the coldest period in P. sinensi’s life cycle – is usually ∼4°C, we chose 8°C as the acute stress temperature, along with a slow cooling process (Chen et al., 2015; Zhang et al., 2017a). Our pre-experiment found that cerebral antioxidant gene expression did not change during 15 days of cold stress (Fig. S1). However, significant changes in total antioxidant capacity was observed in the brain and liver of P. sinensis after 12 h of cold stress (Zhang et al., 2017b). Therefore, a duration of 12 h was selected for measuring the antioxidant response to acute cold stress.

At the end of the acclimation, turtles were transferred into four plastic tanks (140×60×70 cm) at a density of six individuals per tank and fasted for 48 h with water temperature maintained at 28±0.5°C. After randomly sampling the control group (n=8) from these three tanks, the water temperature was quickly dropped to 8°C in less than 5 min and maintained for 12 h with cooling machines. At the end of this treatment, eight turtles, defined as the ‘cold exposure group’, were sampled immediately. Then the water temperature was raised to 28°C in less than 5 min and maintained for 24 h, allowing the turtles to recover. The remaining eight turtles, defined as the ‘recovery group’, were then sampled (Chen et al., 2015). The brain, liver and kidney were quickly excised and frozen in liquid nitrogen. All samples were then transferred to −80°C for storage.

The protocol was the same as our previous studies (Chen et al., 2015; Zhang et al., 2017a). Briefly, total RNA was extracted from the brain, liver and kidney (50–100 mg) using a method combining Trizol reagent (Takara, Japan) and the Nucleospin RNA II kit (Macherey-Nagel, Germany) according to the product manual (Munang'Andu et al., 2013). After RNA extraction, electrophoresis and a NanoDrop 2000 spectrophotometer (Thermo, USA) were employed to determine the quality and quantity of the isolated RNA, respectively (A₂₆₀/A₂₈₀>1.9; 0.2–0.8 μg μl⁻¹). The PrimerScript II 1st strand cDNA synthesis kit (Takara, Japan) was used for reverse transcription. 2 μg total RNA was used as the template for each 30 μl reaction. The acquired cDNA was then diluted six times and stored at −20°C for use in real-time PCR.

The primers for real-time PCR were designed using Primer 3 software (http://primer3.ut.ee/). All primers were verified by sequencing after the PCR products were assembled into cloning vectors. The real-time PCR efficiency was calculated in accordance with our previous study (Zhang et al., 2017a). Gene accession numbers, forward and reverse primer sequences and real-time PCR efficiencies are given in Table 1.

Real-time PCR was performed on an Applied Biosystems 7500 real-time PCR system with the following reaction conditions: 10 μl of 2× SYBR Green PCR Master Mix (Applied Biosystems), 3 μl diluted cDNA template, 0.25 μl each specific forward and reverse primers (20 μmol l⁻¹) and 6.5 μl Milli-Q H₂O. For internal control gene selection, a one-way ANOVA test was conducted on the expression levels of several common control genes under experimental conditions. We selected β-actin as the internal control gene for brain (P=0.73) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the internal control gene for liver (P=0.24) and kidney (P=0.14). The reason for applying different internal control genes was the varied expression status of these genes in different tissues across the treatments, i.e. β-actin did not remain constant in liver and kidney, whereas the expression of GAPDH was not constant in brain across the whole experiment (P<0.05). Relative mRNA expression was calculated based on data from triplicate runs using the comparative CT method (Schmittgen and Livak, 2008). Where multiple isoforms of a gene were assessed, mRNA levels of the different isoforms in each tissue were expressed relative to the control transcript levels of the isoform with the lowest abundance that was standardized to 1.0.

All assays were performed using Diagnostic Reagent Kits (Nanjing Jiancheng, PRC) except for selenium-dependent GPx (Se-GPx) activity, which was measured using a kit from Beyotime (PRC). Tissue homogenates were prepared by 9-fold dilution in ice-cold physiological saline (0.68% NaCl). All assays were performed at 25°C, a common temperature for measuring enzyme activity in reptiles (Galli et al., 2013; Sánchez-Hernández et al., 2004) and the protocols were the same as in our previous studies (Zhang et al., 2017a,c).

All data are presented as mean±s.e.m. and P<0.05 was used as the level of statistical significance. Data analysis was conducted using the SPSS statistical package (v11, SPSS Inc., USA). Outliers beyond the mean±2×s.d. were excluded. The data were first checked for normality and homogeneity of variance. If appropriate, the statistical significance between groups was assessed using one-way ANOVA followed by Duncan post hoc tests. Otherwise, the Kruskal–Wallis test followed by Mann–Whitney post hoc U-test was used (Dytham, 2011).

Nrf2 mRNA levels in kidney decreased significantly by 37% compared with controls in response to cold exposure (P=0.01) but were restored after recovery (Fig. 1). However, Nrf2 mRNA levels in the brain and liver did not change significantly across the experiment (brain: P=0.98; liver: P=0.48; Fig. 1).

Levels of Cu/ZnSOD and MnSOD mRNA in the brain remained unchanged over the entire experiment (Cu/ZnSOD: P=0.34; MnSOD: P=0.85), whereas both hepatic Cu/ZnSOD and MnSOD mRNA content increased significantly during recovery (Cu/ZnSOD, 1.8-fold: P=0.038; MnSOD, 1.5-fold: P=0.043; Fig. 2A). The Cu/ZnSOD mRNA level in kidney remained stable throughout the experiment (P=0.85), but MnSOD mRNA significantly reduced during cold exposure and following recovery (P=0.04; Fig. 2A).

The total SOD activity (U mg⁻¹ protein) in liver and kidney were both more than 10-fold higher than that in brain. Cerebral and nephric SOD activity remained stable during the entire experiment (brain: P=0.06; kidney: P=0.77); but hepatic SOD activity decreased significantly during the recovery period (P=0.041; Fig. 2B).

Hepatic CAT mRNA levels increased significantly (1.48-fold) compared with the control during recovery (P=0.04), whereas both cerebral and nephric CAT transcript levels remained stable over the entire experiment (brain: P=0.84; kidney: P=0.41; Fig. 3A).

CAT activity in liver (U mg⁻¹ protein) was more than 10-fold higher than in kidney, and more than 40-fold higher than in brain. The CAT activity in all three tissues did not change significantly across the experiment (brain: P=0.16; liver: P=0.63; kidney: P=0.10; Fig. 3B).

Transcript levels of three isoforms of Se-GPx were assessed. The mRNA level of Se-GPx1, Se-GPx3 and Se-GPx4 in brain (Se-GPx1: P=0.31; Se-GPx3: P=0.74; Se-GPx4: P=0.41) and liver (Se-GPx1: P=0.71; Se-GPx3: P=0.85; Se-GPx4: P=0.84) as well as Se-GPx1 and Se-GPx3 in kidney (Se-GPx1: P=0.11; Se-GPx3: P=0.31; Fig. 4A) remained stable over the entire experiment. However, Se-GPx4 mRNA in kidney decreased significantly by about 25% during cold exposure and recovery (P=0.028; Fig. 4A).

Se-GPx activity (U g⁻¹ protein) was the highest in kidney among the three tissues tested (Fig. 4B). Cerebral GPx activity remained constant over the entire experimental course (P=0.10). However, hepatic GPx activity decreased significantly during the recovery period (P=0.02) whereas nephric GPx activity decreased significantly during cold exposure and remained at this low level in recovery (P<0.001; Fig. 4B).

Acute cold exposure and recovery did not affect the mRNA expression of either GSS_X1 or GSS_X2 in liver and kidney (liver: GSS_X1, P=0.39; GSS_X2, P=0.59; kidney: GSS_X1, P=0.44, GSS_X2, P=0.83). In brain, GSS_X1 transcript levels remained stable across the entire experiment (P=0.65), but GSS_X2 mRNA increased significantly (1.8-fold) during recovery relative to control (P=0.03; Fig. 5).

GR mRNA levels did not change in any of the three tissues over the entire experiment (brain: P=0.13; liver: P=0.32; kidney: P=0.44; Fig. 6A). GR activity (U g⁻¹ protein) in liver was 6.8-fold and 25.7-fold higher than in brain and kidney, respectively (Fig. 6B). Hepatic and nephric GR activity remained constant across the experimental course (liver: P=0.17; kidney: P=0.19), but cerebral GR activity increased significantly during cold exposure and then returned to control levels in recovery (P=0.006; Fig. 6B).

All GST isoforms in brain showed stable mRNA expression levels over the entire experiment (GSTZ1_X2: P=0.13; GSTZ1_X1: P=0.61; GST2: P=0.91; GSTC: P=0.67; GST1: P=0.59; GSTK1: P=0.70; GSTP1: P=0.63; GST3: P=0.42). The same was true for the transcript abundance of 7 isoforms of GSTs in both liver (GSTZ1_X2: P=0.26; GSTZ1_X1: P=0.51; GSTC: P=0.14; GST1: P=0.26; GSTK1: P=0.88; GSTP1: P=0.65; GST3: P=0.23) and kidney (GSTZ2_X1: P=0.15; GST2: P=0.98; GSTC: P=0.78; GST1: P=0.053; GSTK1: P=0.18; GSTP1: P=0.61; GST3: P=0.083). However, there were two exceptions. Hepatic GST2 mRNA levels increased significantly during recovery (1.7-fold; P=0.008), whereas nephric GSTZ1_X1 mRNA showed a small but significant decrease during recovery (P=0.045; Fig. 7A).

Hepatic and nephric total GST activities were both more than 2-fold higher than that in the brain of juvenile P. sinensis (Fig. 7B). The GST activity in all three tissues did not change significantly throughout the entire experiment (brain: P=0.30; liver: P=0.57; kidney: P=0.10; Fig. 7B).

This is the first study to examine the response and regulation of the major antioxidant systems (both the classic enzymatic antioxidant system and the GSH system) in response to acute cold exposure in a Southeast Asian turtle. It differs from the previous studies in evaluating cold exposure (above 0°C) versus prior work that evaluated the effects of whole body freezing at subzero temperatures on antioxidant defenses by North American freeze-tolerant turtles (Baker et al., 2007; Krivoruchko and Storey, 2010a; Storey, 2006).

Our results showed that the mRNA expression of the Nrf2 transcription factor and its downstream antioxidant enzyme genes were generally unchanged in response to cold exposure (20°C reduction and maintained for 12 h) or recovery (acclimation temperature and maintained for 24 h) in P. sinensis. The brain showed no change in transcript levels of any of the antioxidant genes measured (Nrf2, Cu/ZnSOD, MnSOD, CAT or GPx). However, nephric Nrf2, MnSOD and GPx4 mRNA levels decreased significantly with acute cold exposure (Fig. 1, Fig. 2A and Fig. 4A), whereas hepatic Cu/ZnSOD, MnSOD and CAT levels were upregulated during recovery (Fig. 2A and Fig. 3A). Overall, then, the turtles exhibited a ‘conservative’ transcriptional response to cold exposure by antioxidant genes. By contrast, empirical studies of freeze-tolerant North American turtles revealed that despite a global depression of transcription and translation, Nrf2 and its downstream antioxidant enzyme genes were among a small group of genes that were upregulated in response to freezing (Krivoruchko and Storey, 2010a; Storey, 2006). Combining our results with the evidence of the studies on North American turtles, it appears that activation of the Nrf2 pathway depends on the extent of oxidative stress in tissues. Turtles cannot regulate their metabolic rate independently of ambient temperature. Their metabolic rate decreases with cold exposure, possibly resulting in reduced ROS production and decreased oxidative stress (Hermes-Lima and Zenteno-Savın, 2002). In this case, activation of the Nrf2 pathway to upregulate antioxidant defenses in the cold may be unnecessary and this may account for the stable or reduced Nrf2 mRNA transcript levels in our results. However, freezing can lead to physical damage, cell dehydration and whole body ischemia that can cause severe tissue damage and oxidative stress (Dinkelacker et al., 2005; Storey, 2006; Voituron et al., 2002), resulting in activation of Nrf2 and its downstream antioxidant enzymes when temperatures were reduced below 0°C. Furthermore, it has been reported that the ROS did not increase during acute cold exposure in P. sinensis (Zhang et al., 2017b), indicating a stable balance between ROS production and antioxidant defense, which indicates that cold exposure for 12 h did not produce oxidative stress sufficient to trigger Nrf2 pathway activation of antioxidant defenses.

The mRNA levels of Nrf2 and its downstream antioxidant enzyme genes either remained stable in brain and liver or showed a synchronized changing pattern in kidney in acute cold exposure compared with the control, suggesting a potential regulatory role of Nrf2. However, the changing patterns of these mRNAs differed dramatically during recovery, suggesting that the transcriptional regulation of antioxidant enzyme genes might not be mediated by Nrf2 mRNA expression changes in the recovery period. One explanation is that the activation of the Nrf2 pathway can be achieved by the accumulation of Nrf2 protein in the nucleus, which is more likely a result of mechanisms that decrease the rate of its degradation (Itoh et al., 2004; Nguyen et al., 2009). Another possible explanation could be the involvement of other transcription factors as regulators, such as NF-κB and AP-1 (Krivoruchko and Storey, 2010b; Lushchak, 2011).

Compared with the enzymatic antioxidant system, the mRNA expression of enzymes in the GSH system exhibited a more conservative response: genes for all three enzyme tested (GSS, GR, GST) maintained a constant transcription level (compared with controls) under acute cold exposure (Fig. 5, Fig. 6A and Fig. 7A). Furthermore, in the recovery period, only cerebral GSS_X2 and hepatic GST2 showed a significant increase in transcript levels whereas nephric GSTZ1_X1 mRNA showed a significant decrease (Fig. 5 and Fig. 7A). In combination, these results allow us to conclude that regulation of the antioxidant system in response to acute cold exposure in P. sinensis may not lie mainly at the transcriptional level, which is supported by the observation that the transcription status of all antioxidant enzyme genes was stable or reduced under cold exposure. None of the antioxidant enzyme genes showed increased mRNA levels under cold exposure, and only 5 out of 18 tested genes showed significant increases mRNA expression during recovery, with maximum expression increases of less than 2-fold. Furthermore, there was a mismatch between the changing patterns of mRNA levels and the activities of the antioxidant enzymes. For example, our results showed that hepatic Cu/ZnSOD and MnSOD mRNA levels increased significantly during recovery whereas total SOD activity decreased significantly in recovery (Fig. 2). Similarly, cerebral GR mRNA did not change throughout the experiment, but its enzyme activity increased under cold exposure and was then restored in recovery (Fig. 6). The only match between mRNA levels and enzyme activity was found for GPx in kidney, where both parameters decreased significantly during acute cold exposure and following recovery (Fig. 4). Given that an enzyme's catalytic activity is the dominant factor in the ROS detoxification process, the above evidence suggests that up- or downregulation of antioxidant enzymes at the transcriptional level might not necessarily lead to increased or decreased capacity of the animal's antioxidant defenses. Nevertheless, it is also possible that a time lag between mRNA expression and enzyme protein synthesis might exist across the experimental time course (Zhang et al., 2017a).

Lack of abrupt regulation at the transcriptional level does not mean that the turtle's antioxidant system is unresponsive to acute cold exposure, but could be due to existing powerful antioxidant defenses, which are sufficient to counteract cold-induced oxidative stress. Our previous studies have proven that no oxidative damage occurred in these three tissues of P. sinensis during both cold stress and recovery periods (Chen et al., 2015; Zhang et al., 2017b). Hence, there would be no need to activate extra transcriptional regulation to further strengthen the defense system. Indeed, several studies have reported that high constitutive antioxidant enzyme activities occur in freshwater turtles as a defense against environmental stress (Chen et al., 2015; Hermes-Lima et al., 2001; Hermes-Lima and Zenteno-Savın, 2002; Storey, 2006, 2007; Zhang et al., 2017a). In addition, the juveniles in our study exhibited extraordinary development in their antioxidant defenses: hepatic SOD activity increased more than 10-fold while hepatic GR activity increased more than 100-fold when compared with that in P. sinensis hatchlings (Fig. 2B and Fig. 6B) (Zhang et al., 2017a,c).

Tissue-specific antioxidant system responses to acute cold exposure were also evident in our study. Transcripts of most cerebral antioxidant genes remained stable over the entire treatment and the corresponding enzyme activities of most were also unchanged and were lower compared with other tissues. Our previous study found that ascorbic acid (AA) concentration in brain was more than 10-fold higher than that in other tissues, and acute cold exposure induced a >30% decrease in brain AA concentrations that did not occur in other tissues (Chen et al., 2015). Hence, it appears that low molecular weight antioxidants, and not antioxidant enzymes, may be the primary antioxidant force in turtle brain that responds to acute cold exposure. The best evidence for transcriptional regulation of antioxidant enzyme genes was found in liver, which also showed the highest activities of most antioxidant enzymes, suggesting the importance of an enzyme-dependent antioxidant defense strategy in liver. The kidney had intermediate responses at both transcriptional and enzyme activity levels compared with the brain and liver. The only antioxidant enzyme activity to decrease under acute cold exposure (GPx) was found in kidney. Since (1) the liver is the primary production center for GSH (Deneke and Fanburg, 1989; Limón-Pacheco and Gonsebatt, 2009), (2) blood transportation can be impaired during acute cold exposure in turtles (Chen et al., 2015; Galli and Richards, 2012) and (3) the nephritic GR level is relatively low, the reduction in GPX activity in the kidney might be due to an insufficient delivery of GSH, which serves as a necessary substrate for the GPx reaction and may be consumed during cold stress (Zhang et al., 2017c). In addition, nephritic metabolic depression during acute cold stress could easily contribute to the decreased GPx activity.

In conclusion, Nrf2 and its downstream antioxidant enzyme genes showed relatively synchronized changes in response to acute cold stress but not recovery treatment. Most genes of the enzymatic antioxidant and GSH systems exhibited conservative transcriptional regulation; the changing pattern of mRNA transcript levels differed from the corresponding enzyme activities in many cases, indicating that transcriptional control of the antioxidant system may not be a major factor in the response to acute cold exposure in P. sinensis. In general, high and stable constitutive antioxidant enzyme activities were evident in turtle tissues in the current study, but with some tissue-specific characteristics. The liver possessed the highest activity of most antioxidant enzymes, whereas the brain has the lowest enzyme activities in most cases. In combination with our previous findings, it appears that the antioxidant defense response to acute cold exposure was mainly achieved by low molecular weight antioxidants in the brain and antioxidant enzymes in the liver of P. sinensis.

We are grateful to Professor Zhen-dong Cao, Dr Zuo-bing Zhang, Dr Xiao-xuan Li and Lin Yuan for their help in this study.