Animals were collected inside a private area, Fazenda São Miguel (5°41′27″S, 36°26′19″W), near the municipality of Angicos, State of Rio Grande do Norte, Brazil. The farm has been inactive for a long time and is situated inside the domain of the Caatinga, a biome restricted to Northeastern Brazil. High average temperatures and prominent rainy seasons (usually in January–April) followed by dry seasons characterize this biome (Figure 1). Other peculiarity of the Caatinga is the unpredictability of the rain regime, which is evidenced by its highly variable monthly rainfall levels (Figure 1). During the dry season, some anuran species, such as Rhinella granulosa and R. jimi, remain active foraging in humid microhabitats (Madelaire et al., 2017), whereas other species, such as Pleurodema diplolister, Proceratophrys cristiceps, and Physalaemus spp., estivate buried under dry river beds until the arrival of the rainy season (Carvalho et al., 2010; Jared et al., 2019). Detailed information about the collection site, as well as the behavior and morphology of the estivating species, is available in a recent publication (Jared et al., 2019).

Adult specimens of P. diplolister and P. cristiceps were collected during four expeditions (2005–2012). Active animals were collected on the surface of the sandy soil during the rainy season. Estivating animals were collected during the dry season after excavating dry river beds and locating the animals buried in the soil. Once located and identified, whole animals were immediately frozen in liquid nitrogen to avoid any manipulation effects, and kept at −80°C until biochemical analyses. Active P. diplolister were collected in December 2005 and estivating P. diplolister were collected in August 2012. Procedures were approved by Universidade de São Paulo Animal Care and Use Committee (168/2012), under permits from IBAMA (02010.003380/04-82 and 14836-1) and SisGen (A0F0B20).

Monthly rainfall totals for 2005–2012 were obtained from a station of the National Meteorological Institute (INMET) located at 6°6′36″S and 36°48′36″W, the closest station to the collection site.

To measure the activity of metabolic and antioxidant enzymes, the whole muscle mass surrounding the right femur of each animal was homogenized as previously described (Moreira et al., 2020). Briefly, the muscle tissue was mechanically homogenized in 50mM Tris (pH 7.4), 1mM EDTA, 1mM phenylmethylsulfonyl fluoride and 1:500 (v/v) protease inhibitor cocktail (P8340, Sigma-Aldrich, St. Louis, MO, United States). Then, the homogenate was centrifuged at 10,000×g for 15min at 4°C. The supernatants were transferred to new tubes, kept on ice and used for the determination of enzymatic activity in the same day of the homogenization.

The activities of citrate synthase (CS), isocitric dehydrogenase (ICDH), malic enzyme (ME), pyruvate kinase (PK), lactic dehydrogenase (LDH), creatine kinase (CK), glutathione transferase (GST), total superoxide dismutase (tSOD), catalase, glutathione peroxidase using H₂O₂ as substrate (GPX), and glutathione peroxidase using cumene hydroperoxide as substrate (tGPX) were measured using kinetic spectrophotometric assays as previous described (Moreira et al., 2020 and references therein).

Enzymatic activity was normalized to the concentration of protein in the supernatants, which was measured using coomassie brilliant blue G-250 (Bradford, 1976), as previously described (Moreira et al., 2020).

To measure the levels of glutathione (reduced and disulfide forms) and protein carbonyl groups, the whole muscle mass surrounding the left femur of each animal was homogenized as previously described (Moreira et al., 2020). Briefly, the muscle tissue was mechanically homogenized in 10% (w/v) trichloroacetic acid (TCA), aliquoted in three tubes, and centrifuged for 10,000×g for 10min at 4°C. The supernatants were assessed immediately to measure levels of total glutathione (tGSH) and disulfide glutathione (GSSG) using the enzymatic recycling method (Rahman et al., 2006), as previous described (Moreira et al., 2020). Reduced glutathione (GSH) levels were calculated using the formula: GSH=tGSH−2×GSSG. Two pellets were stored at −80°C until the determination of protein carbonyl concentration using 2,4-dinitrophenylhydrazine in a colorimetric assay (Fields and Dixon, 1971), as previously described (Moreira et al., 2020).

The third pellet was used to measure protein levels and normalize the concentrations of glutathione and protein carbonyl groups. Protein concentration in the pellets was determined using coomassie brilliant blue G-250 (Bradford, 1976) after resuspending the pellets in 100mM NaOH, as previously described (Moreira et al., 2020).

First, we analyzed the biochemical data for P. diplolister as previously (Moreira et al., 2020). Briefly, data had their distribution tested using the Shapiro–Wilk normality test. We used unpaired t tests (two-tailed) for data with a normal distribution and Mann–Whitney tests (two-tailed) for data with a distribution other than normal to compare active with estivating P. diplolister. Then, we compiled the original data generated in the present study with those already published in a similar study (Moreira et al., 2020) and conducted a two-way ANOVA to analyze the effects of species and metabolic state on each biochemical variable. Finally, we also analyzed the compiled data using a principal component analysis (PCA) in Minitab 19 (Minitab, State College, PA, United States). Considering the strong (R²=0.9960) and significant (Pearson correlation, p<0.0001) correlation between tGSH and GSH levels among all animals, we excluded tGSH levels in the PCA to avoid distortions of the outcome; all other parameters were included. Normality tests, t tests, Mann–Whitney tests, Pearson correlation and two-way ANOVA tests were conducted using GraphPad Prism 8 (GraphPad, San Diego, CA, United States). Graphs were created using GraphPad Prism 8. For all analyses values of p below 0.05 were considered to indicate statistically significant differences/effects.

The activities of CS and ICDH, enzymes of oxidative metabolism, were lower in the muscle of estivating frogs than those in active frogs. Compared with active animals, CS activity was 35.8% lower (Figure 2A) and ICDH activity was 47.9% lower (Figure 2B) in the muscle of estivating P. diplolister. The activity of muscular pyruvate kinase (PK), which catalyzes a key step in glycolysis, did not differ between active and estivating P. diplolister (Figure 2C). The activities of enzymes involved in anaerobic ATP production from phosphocreatine (CK), NADPH production (ME) and anaerobic metabolism (LDH) did not differ between active and estivating animals (Figures 2D–F).

There was no difference between active and estivating animals in terms of reduced glutathione (GSH) and total glutathione (tGSH=GSH+2×GSSG) concentration in skeletal muscle (Figures 3A and B). Similarly, GST, an enzyme that uses GSH to detoxify electrophilic molecules, was unaffected by estivation in P. diplolister (Figure 3C). Estivating P. diplolister had 65.5% less disulfide glutathione (GSSG) than active P. diplolister, which resulted in a 72.4% decrease in GSSG/tGSH ratio in the muscle of estivating animals compared with that of active animals (Figures 4A and B). The degree of oxidative damage to muscular proteins assessed as protein carbonyl levels was unaffected by estivation in P. diplolister (Figure 4C).

Total superoxide dismutase (tSOD) activity did not differ between estivating and active animals (Figure 5A). Estivating P. diplolister had higher activity of peroxide-detoxifying enzymes than that in the muscle of active P. diplolister. The activities of catalase, GPX, and tGPX were 74.4, 72.7, and 74.1% higher, respectively, in the muscle of estivating animals than in those of active animals (Figures 5B–D).

To assess the effect of species (P. diplolister or P. cristiceps) and metabolic state (active or estivating) on the biochemical biomarkers measured in the skeletal muscle of active and estivating P. diplolister (present study) and active and estivating P. cristiceps collected in the same site (Moreira et al., 2020), we assembled the data (Table 1) and conducted a two-way ANOVA (Table 2). Simple main effects analyses revealed that species had an effect on the activities of CS, ICDH, ME, CK, GST, catalase, GPX and tGPX, as well as on GSSG/tGSH ratio (Table 2). In the case of metabolic state, simple main effects analyses showed that estivation had an effect on the activities of CS, ICDH, catalase, GPX and tGPX, as well as on the levels of tGSH, GSH, GSSG, and GSSG/tGSH (Table 2). Moreover, there was an interaction effect between species and metabolic state on GSSG and GSSG/tGSH levels (Table 2).

The PCA with all biochemical variables, except for tGSH levels, identified two major components that effectively distinguished individual animals in terms of metabolic state (activity vs. estivation) and species (P. diplolister vs. P. cristiceps; Figure 6). Active P. diplolister clustered in the upper right quadrant, estivating P. diplolister in the lower right, active P. cristiceps in the upper left and estivating P. cristiceps in the lower left. The first component, which separated the species, was mainly a contribution of CS, ICDH, and GST activities (Table 3). On the other hand, the second component, which separated active animals from estivating animals regardless of the species, was mainly a contribution of GSSG/tGSH ratio, GPX activity, and tGPX activity (Table 3).

The decrease in the activity of two enzymes associated with oxidative metabolism (CS and ICDH) in the muscle of both P. diplolister and P. cristiceps (Moreira et al., 2020) is in line with the metabolic adaptations of estivation. The suppression of oxygen-dependent metabolic pathways and whole-animal oxygen consumption are hallmarks of estivation (Guppy and Withers, 1999). Indeed, resting metabolic rate of both P. diplolister and P. cristiceps decrease by ~50% in the dry season (i.e., during estivation) when compared with that measured during the rainy season (Pereira, 2016). In Australian frogs Cyclorana alboguttata, oxygen consumption by either the whole animal or saponin-permeabilized skeletal muscle fibers is strongly suppressed after 4 months of estivation (Reilly et al., 2014). Changes in CS activity during the estivation of C. alboguttata depends on specific muscle functions (i.e., in which specific muscle the measurement is made). For example, the activity of CS decreases in sartorius, iliofibularis, and rectus abdominus muscles, whereas it increases in gastrocnemius, of C. alboguttata estivating for 6months when compared with control animals (Mantle et al., 2010). Estivation is also part of the life history of lungfish, when whole-organism oxygen consumption declines (Fishman et al., 1992). Accordingly, muscle CS activity decreases as the African lungfish Protopterus dolloi enters the hypometabolic phase of estivation (Frick et al., 2008b). Overall, our findings with anurans collected directly in the field corroborate previous observations of decreased oxidative metabolism in other vertebrates induced to estivate in laboratory conditions.

With a reduction in oxidative metabolism, one might expect compensatory adjustments of oxygen-independent alternative pathways. Here, we found no changes in the activities of PK, CK, and LDH in the muscle of P. diplolister and no effect of metabolic state (i.e., estivation) on these enzymes. In P. cristiceps, a very similar response was observed, when muscular CK and LDH activities remained unchanged, whereas PK activity had a minor decrease, in estivating animals compared with the control group (Moreira et al., 2020). The responses of these enzymes in Caatinga frogs agree with previous observations in C. alboguttata, where, depending on muscle type, LDH activity is similar or lower in estivating frogs compared with that in the active group (Mantle et al., 2010). In the African lungfish, the capacity of the glycolytic pathway, measured as PK and hexokinase activities, is also maintained in skeletal muscle during estivation (Frick et al., 2008a). Moreover, LDH activity in the skeletal muscle of P. dolloi is not affected by estivation (Frick et al., 2008a). These findings indicate that, despite reduced oxygen consumption, the downregulation of ATP-expensive pathways maintains energetic supply/demand balance and oxygen availability is not a significant constraint during estivation for these animals. Indeed, the oxygen concentration inside the burrows of estivating frogs in the Caatinga does not fall below 20.7% down to 1.5m depth (Carvalho et al., 2010). However, the biochemical responses of these metabolic pathways differ in the North American spadefoot toad, whose estivation is associated with large increases in PK and CK activities in skeletal muscle (Cowan et al., 2000). Possible muscle functionality differences aside, S. couchii seems to differ from C. alboguttata, P. dolloi, and the Caatinga frogs in terms of biochemical adaptations associated with energy metabolism during estivation. Once again, our findings from field-collected animals corroborate previous observations of the lack of compensatory upregulation of oxygen-independent pathways in other vertebrates induced to estivate under laboratory conditions.

Glutathione is a cofactor for antioxidant enzymes and an antioxidant itself. This tripeptide is present at millimolar intracellular concentrations in animal tissues, where it can be found in its reduced (GSH) and disulfide (GSSG) free forms. More important than their absolute levels, the ratio GSSG/tGSH is a reliable indicator of disturbances in redox balance (Flohé, 2013). For both P. cristiceps (Moreira et al., 2020) and P. diplolister, the skeletal muscle had a more reduced cellular environment months after the onset estivation (i.e., lower GSSG/tGSH values) compared with that of animals assessed during the active season. Muscular GSSG/tGSH ratio behaves differently in other estivating vertebrates. The spadefoot toad S. couchii experience no changes in GSSG/tGSH levels after 2months of estivation (Grundy and Storey, 1998). In the muscle of another African lungfish, Protopterus annectens, GSSG/tGSH increases sharply after 6months under estivation (Ong, 2016). The decrease in GSSG/tGSH observed in frogs estivating in the Caatinga can be explained by the expected decrease in the production of reactive oxygen species once the hypometabolic state is achieved and oxygen consumption stabilize at a slow rate (i.e., a slow electron flow through the mitochondrial electron transport chain). Indeed, the skeletal muscle of estivating C. alboguttata frogs produces less hydrogen peroxide when compared with active frogs (Reilly et al., 2014). The behavior of GSSG/tGSH in the muscle of freely estivating Caatinga frogs is consistent with a more reduced intracellular redox state that should favor muscle preservation during estivation, as observed for C. alboguttata (Hudson and Franklin, 2002; Young et al., 2013; Reilly et al., 2014).

In addition to GSSG/tGSH, we also assessed levels of oxidative damage to proteins in the muscle of Caatinga frogs. For both P. cristiceps (Moreira et al., 2020) and P. diplolister, protein carbonyl levels remained stable months after the onset of estivation. Protein carbonyl and malondialdehyde (a lipid peroxidation marker) levels are maintained in the gastrocnemius muscle of C. alboguttata frogs during estivation (Young et al., 2013). Likewise, the degree of oxidative damage to lipids in skeletal muscle (assessed as TBARS and hydroperoxide levels) does not differ between freshwater controls and 6-month estivating P. annectens (Ong, 2016). Protein carbonyl levels decrease by 45% in the muscle of P. annectens after 6months of estivation (Ong, 2016). On the other hand, the levels of two biomarkers of oxidative damage to lipids increase in the muscle of S. couchii estivating for 2months compared with its active control group (Grundy and Storey, 1998) and protein carbonyl content increases in the iliofibularis muscle, but not in gastrocnemius, from estivating frogs compared with control frogs (Young et al., 2013). Our findings indicate the lack of accumulation of oxidative damage to biomolecules in the skeletal muscle of freely estivating Caatinga frogs despite the monthslong estivation period spent burrowed.

Regardless of the reduced intracellular redox state and the lack of oxidative damage (at the time the animals were collected), both P. cristiceps (Moreira et al., 2020) and P. diplolister had higher activities of peroxide-detoxifying enzymes during estivation than those during activity in the rainy season. In fact, the second principal component of PCA, which separated active from estivating individuals of both species, was mainly a contribution of three variables directly related to redox metabolism: tGPX, GPX, and GSSG/tGSH. Moreover, when data from both species were analyzed together, we detected an effect of metabolic state (i.e., estivation) on catalase, GPX and tGPX activities. The modulation of antioxidant enzymes during estivation has been studied in the skeletal muscle of a few vertebrate species. Similar to Caatinga frogs, GPX and tGPX activities increase and tSOD activity is maintained after 6months of estivation in the muscle of P. annectens compared with the control group (Ong, 2016). In the case of S. couchii, toads estivating for 2months have increased in SOD and tGPX activities in skeletal muscle compared with those of active toads; other four antioxidants had their activities maintained at control levels (Grundy and Storey, 1998). For the Australian burrowing frog, total antioxidant capacity of both water-soluble and membrane-bound fractions of skeletal muscle does not differ between active and 6-month-estivating frogs (Hudson et al., 2006). At the transcript level, the number of transcripts for catalase and glutathione peroxidase remains stable, whereas that for mitochondrial superoxide dismutase decrease, in skeletal muscle of estivating C. alboguttata compared with the levels measured in control frogs (Hudson et al., 2006). The activity of both mitochondrial and cytosolic SOD, however, is higher in the muscle of estivating frogs than in that of control frogs (Young et al., 2013). The antioxidant boost in the muscle of estivating frogs is further corroborated by the observation of the upregulation of several genes under the regulation of Nrf2 (Reilly et al., 2013), a transcription factor that controls the expression of antioxidants and other cytoprotective genes (Kaspar et al., 2009; Niture et al., 2014). The measurement of antioxidant levels in freely estivating P. diplolister confirmed the activation of the endogenous antioxidant machinery during estivation as observed in other vertebrate models of estivation.

The upregulation of endogenous antioxidants is a hallmark of the biochemical responses that allow animals to adapt and survive adverse environmental conditions. This biochemical phenomenon was first observed in snakes exposed to freezing or anoxia, and later confirmed in a few more species of vertebrates and invertebrates, being coined in 1998 as POS (Hermes-Lima and Storey, 1993; Hermes-Lima et al., 1998). Since then, the occurrence of POS in animals exposed to adverse ambient has been shown to be widespread in more than 80 animal species distributed in eight phyla (Hermes-Lima and Zenteno-Savín, 2002; Welker et al., 2013; Moreira et al., 2016, 2017). Despite recent advances in the field, studies reporting the POS phenotype have been mostly performed under artificial laboratory settings (Hermes-Lima et al., 2015; Giraud-Billoud et al., 2019) and only recently studies on the modulation of antioxidants in animals under ecophysiological relevant natural scenarios have been published. For example, mussels Brachidontes solisianus stimulate glutathione synthesis after 4h of air exposure (a condition that impairs oxygen uptake) in a rocky beach under natural settings (Moreira et al., 2021). In the case of Astyanax elachylepis, fish collected in a naturally hypoxic intermittent stream suffer more oxidative damage and have increased GPX activity in gills compared with fish living in a naturally normoxic perennial stream during the dry season (Ondei et al., 2020). More recently, the levels of redox metabolism biomarkers in naturally hibernating common Asian toads, Duttaphrynus melanostictus, at the end of the winter were compared with those in naturally active toads collected during summer (Patnaik and Sahoo, 2021). The redox balance in the brain and liver of hibernating toads shifts toward a more oxidized state (i.e., increased GSSG/GSH ratio; Patnaik and Sahoo, 2021). Such a disruption of the redox balance is associated with increased damage to proteins and lipids and the induction of catalase activity in both tissues of hibernating toads compared with active toads (Patnaik and Sahoo, 2021). Therefore, evidence from field-collected mussels, fish and anurans, including our previous (Moreira et al., 2020) and present study, indicates the occurrence of POS under ecologically relevant settings.

We have previously reported the upregulation of selected antioxidants in the muscle of P. cristiceps estivating under natural conditions (Moreira et al., 2020). Here, we measured the same biochemical variables in the muscle of another anuran species that share the same microhabitat in the Brazilian Caatinga during the dry seasons when they both estivate. Most biochemical changes observed in P. diplolister matched those observed in P. cristiceps: downregulation of CS, ICDH, and GSSG/tGSH levels; upregulation of catalase, GPX and tGPX activities; and the maintenance of ME, LDH, CK, GST, protein carbonyl, and tSOD levels in estivating vs. active frogs. Responses observed exclusively in P. cristiceps were the small decrease in PK activity and the increase in GSH concentration in estivating vs. active frogs (Moreira et al., 2020). Although GSSG/tGSH decreased in both Caatinga species, the underlying causes differed between P. diplolister and P. cristiceps. While in P. diplolister the reduction in GSSG/tGSH was driven by a decrease in GSSG concentration, in P. cristiceps the cause was an increase in GSH levels in estivating vs. active frogs (Moreira et al., 2020). This indicates that, despite the resource restriction in food and water in the Caatinga, P. cristiceps stimulate GSH synthesis during estivation, a response that is not shared with P. diplolister. Moreover, given the significant interaction (species×metabolic state) effects on GSSG and GSSG/tGSH, it appears that estivation affects glutathione redox balance in a species-specific manner.

In addition to the few different biochemical responses associated with the estivation phenotype, we also found significant effects of species on some biochemical variables. When comparing species, the skeletal muscle of P. diplolister appeared to have higher levels of aerobic metabolic enzymes than that of P. cristiceps. Accordingly, the first principal component of the PCA, which separated species into two distinct groups, was mainly a contribution of CS, ICDH, GST activities. These findings agree with the discrepancy in whole-animal oxygen consumption between P. diplolister (~0.25mlg⁻¹ h⁻¹) and P. cristiceps (~0.065mlg⁻¹ h⁻¹) collected during activity in the rainy season (Pereira, 2016). Noteworthy, estivating P. diplolister immediately resumes activity when disturbed, whereas estivating P. cristiceps takes minutes to show any sign of activity. Such a “more active/oxidative” characteristic of P. diplolister is associated with a more oxidized intracellular redox state in skeletal muscle (i.e., higher GSSG/tGSH ratio) and higher levels of antioxidant enzymes than those of P. cristiceps.

The ability to dig out the soil, associated with muscle conservation, is a vital point at the end of the estivation of the amphibians that are buried (Grundy and Storey, 1998). In this sense, there are metabolic differences associated with energy production and redox balance between P. diplolister and P. cristiceps in relation to the conservation of muscle tone. Although there is a significant decrease in the capacity of oxidative pathways in muscle in both species, P. diplolister tends to maintain higher levels of both anaerobic and aerobic metabolism than P. cristiceps does. This evidence may be relevant within the ecological relationships between these species because buried individuals of P. diplolister can react quickly to environmental signals of the first rains at the end of the dry season, having quick access to the reproduction areas which will not be used simultaneously by P. cristiceps, which represents an example of niche segregation by space (dos Santos Protázio et al., 2015; Jared et al., 2019; Lima et al., 2019).

The genus Pleurodema is endemic to South America, with several representatives in environments with low water availability and contrasting thermal extremes, such as P. brachyops in xerophytic forests in Colombia or P. bufoninum in the cold Argentine steppes (Faivovich et al., 2012). Within this context, the mechanisms associated with estivation in P. diplolister could be more associated with a phylogenetic trait than with a convergent evolution with the environmental conditions of the caatingas. Moreover, the high activity of metabolic enzymes in the skeletal muscle of P. diplolister (compared with those in P. cristiceps) can also be related to the rapid onset of reproductive activity on the arrival of the rainy season. Although the first rains form incipient water bodies, these are sufficient to allow the reproduction of P. diplolister (Hödl, 1992; Jared et al., 2019). Unlike this, P. cristiceps only comes out of estivation in the presence of heavy rain, since adults need greater availability of aquatic microhabitats for their reproduction and the deposition of eggs occurs in a place and vocalization in another (Jared et al., 2019). The results found for P. diplolister and P. cristiceps during estivation may be an example of how the richness of physiological mechanisms enhances the overlap of niches during the dry season, while during times of activity there are possibly ecological processes of competition.