High basal metabolic rate does not elevate oxidative stress during reproduction in laboratory mice

The evolution of high basal metabolic rate (BMR) and endothermy are hypothesized to result from selection for intensive parental care (Farmer, 2000; Koteja, 2000). The ability to maintain a high, sustained level of energy expenditure and locomotor activity could allow for more efficient feeding, guarding or brooding of offspring, which in turn can decrease juvenile mortality and thus increase parental fitness (Kozłowski, 1992). However, one of the key assumptions of life history theory is that intense parental effort should lead to higher costs of reproduction, revealed by a lower survival of parents, or their reduced future reproductive success (Roff, 1992; Stearns, 1992). Thus, even if higher BMR of parents enables better survival of their offspring, it may confer no evolutionary advantage if it simultaneously incurs an increase of the physiological costs of reproduction and thereby parental mortality. It is unclear, however, to what extent elevated BMR is associated with such costs, most notably with increased oxidative stress (OS) hypothesized to represent a significant reproductive cost at the molecular level (Speakman, 2008; Monaghan et al., 2009).

OS occurs when there is an imbalance between the production of reactive oxygen species (ROS) and the capacity of antioxidant mechanisms to control their damaging effects (Monaghan et al., 2009). ROS are primarily byproducts of normal metabolic processes that can cause damage to lipids, proteins and DNA when not quenched by antioxidant mechanisms [e.g. enzymes such as catalase and superoxide dismutase, or non-enzymatic antioxidants such as glutathione (Monaghan et al., 2009; Pamplona and Costantini, 2011)]. Reproduction represents the period when animals may be particularly prone to OS, because elevated energy expenditure in reproduction can potentially increase the rate of ROS production and/or reduce investment in the antioxidative systems (Speakman, 2008; Monaghan et al., 2009). However, the relationship between BMR and the magnitude of oxidative damage is not obvious. First, higher metabolic rates do not universally elevate the rate of ROS production (Barja, 2007). For example, higher mitochondrial uncoupling may increase the rate of oxygen consumption but simultaneously decrease the ROS production (Speakman et al., 2004). Second, higher BMR may allow for more effective antioxidative mechanisms (Speakman et al., 2002). Finally, previous studies (reviewed in Stier et al., 2012; Metcalfe and Monaghan, 2013; Speakman and Garratt, 2014) on the relationship between OS and reproduction have produced ambiguous results. For example, laboratory experiments on small rodents reported both an increase (Stier et al., 2012) and a decrease (Garratt et al., 2011; Ołdakowski et al., 2012; Garratt et al., 2013) in oxidative damage during reproduction. In conclusion, although variation in BMR is likely to affect both the magnitude of OS and its changes during reproduction, the direction of the relationship between these parameters is difficult to predict. This issue is particularly important because variation in individual parental quality may mediate the importance of OS as the cost of reproduction (Metcalfe and Monaghan, 2013). Moreover, if BMR is an important predictor of parental quality, then variation in BMR (both within and between experiments) may be one of the factors responsible for the inconclusive results of previous studies of the changes of OS during reproduction.

In the present paper, we explore the presumed links between BMR and OS elicited by reproduction. We used laboratory mice from two line types with BMR manipulated by means of artificial selection for high (H-BMR) and low (L-BMR) body-mass-corrected BMR (Książek et al., 2004). The relative between-line-type difference in BMR reaches 40–50% (see Results). Such a considerable difference makes testing correlations between BMR and life history parameters on an intra-specific level possible. Although selection is aimed at BMR measured before reproduction, it resulted in a significantly higher parental investment during lactation in H-BMR females, enabling faster growth of pups (Sadowska et al., 2013). Thus, selection for BMR affected parental effort, and both line types are therefore particularly suited as models for studies on the association between OS, BMR and reproduction (see Metcalfe and Monaghan, 2013). Recently, it was shown that selection for high maximum aerobic metabolic rate in bank voles did

not affect the level of oxidative damage, though it elevated BMR (Ołdakowski et al., 2012). However, the difference in BMR between mice from H-BMR and L-BMR line types is approximately three times higher than between control and selected line types of bank voles (see Ołdakowski et al., 2012), and thus is more likely to reveal the effect of BMR on OS. Moreover, OS in females of bank vole was measured after weaning of their litters (Ołdakowski et al., 2012), whereas in the present experiment we assayed females at the peak lactation, when energetic costs of reproduction are highest.

In the present experiment, we predicted that: (1) if higher BMR increases the rate of ROS production and oxidative damage, the oxidative damage should be higher in non-reproducing H-BMR than L-BMR females. Moreover, elevated energy expenditures during reproduction are likely to further increase OS, particularly in H-BMR line type. (2) Alternatively, if higher BMR decreases the rate of ROS production (e.g. via more uncoupled mitochondria), enables more effective antioxidant mechanisms or is related to lower susceptibility to ROS-related damage, then non-reproducing H-BMR females should have lower oxidative damage than non-reproducing L-BMR females. Under such a scenario (positive correlation between BMR and antioxidative defence or ROS resistance and/or negative correlation between BMR and ROS production), elevated energy expenditures during reproduction may not affect or may even reduce oxidative damage. To test these hypotheses, we measured two parameters quantifying oxidative damage to different types of molecules (Monaghan et al., 2009) in reproducing and non-reproducing female mice of both line types: the level of malondialdehyde (MDA; the product of lipid peroxidation) in liver, kidneys and heart, and the concentration of 8-oxo-2′-deoxyguanosine (8-oxodG; the product of repair of ROS-mediated damage of guanosine) in blood serum as markers of oxidative damage to lipids and DNA, respectively. We also measured the activity of catalase in liver and kidneys, an enzyme that represents an important component of antioxidative defence (Pamplona and Costantini, 2011).

Body-mass-corrected BMR of reproducing females differed consistently between line types both before (least-square means ± s.e.m.; H-BMR: 65.11±0.88 ml O₂ h⁻¹, L-BMR: 43.28±0.83 ml O₂ h⁻¹) and after first reproduction (H-BMR: 71.24±1.27 ml O₂ h⁻¹, L-BMR: 54.53±1.16 ml O₂ h⁻¹). A significant interaction between line type and order of measurement reveals that the effect of first reproduction on BMR was line-type-specific (Table 1). Indeed, BMR increased between measurements in the L-BMR (P=0.0002) but not in the H-BMR line type (P=0.96). However, the standardized between-line type differences (d) were higher than their upper limit of 95% confidence interval (ddrift) for BMR measured both before (d=5.11 versus ddrift=0.71) and after (d=2.70 versus ddrift=0.71) first reproduction, suggesting that the difference in BMR arose as a result of selection rather than genetic drift and was still highly significant in females after first reproduction.

Among non-reproducing females, all parameters quantifying oxidative damage were higher in mice from the L-BMR line type, and the magnitude of most differences between line types exceeded the values expected under the effect of genetic drift (Table 2, Fig. 1). Oxidative damage to lipids was unaffected by reproduction (Table 2, Fig. 1). There was a significant interaction between line type affiliation and reproductive status for the level of 8-oxodG in blood serum (Table 2, Fig. 1D): non-reproducing L-BMR females had higher level of 8-oxodG than H-BMR females (P=0.0004), but this difference disappeared in reproducing females (P=0.86).

There was also a significant interaction between reproductive status and line type for the activity of catalase in liver (Table 2, Fig. 2A). A Tukey's test showed that the activity of catalase did not differ between non-reproducing females from L-BMR and H-BMR line types (P>0.99). Although it was elevated during reproduction in both line types, this increase was stronger in L-BMR than in H-BMR mice (Tukey's test, P<0.0001 and P=0.014, respectively), resulting in a significant difference between reproducing females from both line types (P=0.016). Catalase activity in the kidneys was not affected by the line type, and significantly decreased during reproduction in both line types (Table 2, Fig. 2B).

Differences in BMR between line types of studied mice significantly affected all examined markers of oxidative damage; however, reproduction did not affect oxidative damage to lipids in internal organs, and the concentration of 8-oxodG in blood serum was reduced only in females with low BMR (Table 2, Fig. 1). For most traits quantifying oxidative damage, the magnitudes of phenotypic differences between line types for non-reproducing females were large enough to attribute them to selection on BMR, rather than genetic drift (Table 2). This finding agrees with the results of other studies suggesting that the resistance to OS may have a significant genetic component [e.g. Kim et al. (Kim et al., 2010) and studies cited therein]. Interestingly, in non-reproducing females, the level of oxidative damage was higher in L-BMR mice, whereas earlier studies reported positive inter- and intra-specific correlation between the concentration of 8-oxodG and basal and/or standard metabolic rate (Foksinski et al., 2004; Topp et al., 2008).

The results of the present study do not allow us to pinpoint the exact mechanisms underlying the effect of between-line-type variation in BMR on oxidative damage. However, the magnitude of oxidative damage reflects the balance between the rate of ROS generation and neutralization, susceptibility to ROS, and the efficiency of repair mechanisms (Monaghan et al., 2009). Between-line type differences in BMR did not affect the activity of catalase in liver and kidneys (Fig. 2), and we have demonstrated previously that males of both line types did not differ with respect to antioxidative capacity of blood serum (Brzęk et al., 2012). Thus, lower oxidative damage in the H-BMR mice probably cannot be attributed to their enhanced antioxidative defences, at least those assayed here and previously (Brzęk et al., 2012). Alternatively, mice with high BMR may have lower rate of ROS production because of higher mitochondrial uncoupling (Speakman et al., 2004). Finally, L-BMR mice may be more susceptible to OS because of a higher proportion of ROS-susceptible polyunsaturated fatty acids in their cell membrane lipids (Brzęk et al., 2007). The products of lipid peroxidation may induce DNA damage (Evans and Cooke, 2006; Hulbert et al., 2007), and these mechanisms may explain the higher concentration of both MDA and 8-oxodG in L-BMR mice.

Our study is not the first that has reported no change or a reduction of oxidative damage in internal organs of reproducing rodents (Garratt et al., 2011; Ołdakowski et al., 2012; Garratt et al., 2013). The simplest explanation for the lack of increase in OS during reproduction is an improvement of antioxidative mechanisms. Indeed, we found a significant increase in the activity of catalase in the liver during reproduction (Fig. 2A), which agrees with observations of elevated level of glutathione (Garratt et al., 2011) and the activity of superoxide dismutase (Garratt et al., 2013) in the same organ. Moreover, the increase of catalase activity was significantly higher in L-BMR females. Such a pattern may indicate that elevated energy metabolism during reproduction triggered higher upregulation of ROS-neutralizing mechanisms in this line type, presumably to counterbalance a higher propensity to oxidative damage incurred by their more ROS-susceptible cell membrane lipids (Brzęk et al., 2007). Unexpectedly, the activities of catalase in the kidneys were significantly reduced during reproduction, indicating that other antioxidative mechanisms must be responsible for the lack of increase in oxidative damage in this organ (compare Fig. 1B and Fig. 2B). This result also suggests that the effect of reproduction on antioxidative defences can vary even between vital organs, such as the liver and kidneys [similarly, the effect of reproduction on the activity of superoxide dismutase in Brandt's vole and Mongolian gerbil is tissue dependent (Yang et al., 2013; Xu et al., 2014)].

Although reproduction did not change the magnitude of oxidative damage to lipids in internal organs, it significantly reduced the concentration of 8-oxodG in blood serum of the L-BMR line type (Fig. 1D). However, the concentration of 8-oxodG in blood serum is a general marker of OS at the whole-body level, and imbalance between ROS production and neutralization may differ between organs (Garratt et al., 2011; Garratt et al., 2012; Speakman and Garratt, 2014). One possibility is that elevated energy metabolism during reproduction triggered an enhanced upregulation of ROS-neutralizing mechanisms in the L-BMR line type, similar to the pattern observed for the catalase activity in the liver. Alternatively, whereas excretion rates of 8-oxodG in a steady state are largely unaffected by repair capacities (Loft et al., 2008) and thus mainly reflect the rate of DNA damage, one might hypothesize that reproduction reduced the activity of mechanisms excreting damaged DNA in the L-BMR line type. We cannot exclude such a scenario; however, we emphasize that it would mean that reproducing females from the L-BMR line type accumulated more mutagenic DNA lesions than females from the H-BMR line type, a pattern that still indicates that higher BMR does not result in relatively higher oxidative damage during reproduction.

To the best of our knowledge, this is the first analysis of the link between variation in BMR and changes in OS during reproduction. Our results have three important evolutionary implications. First, the observed patterns suggest that the evolution of high BMR and endothermy via selection for more effective parental care [as proposed in Koteja (Koteja, 2000)] does not necessarily elevate the costs of reproduction in terms of increased oxidative damage (as its value in reproducing H-BMR females never exceeded that observed in L-BMR females). In fact, our results suggest a negative association between energy expenditures and OS. An important caveat is that reproducing mice in our experiment had unlimited access to food, whereas the fitness effects of BMR are often context-dependent (Burton et al., 2011), and reproduction is more likely to increase OS when breeding individuals face limited food resources (Fletcher et al., 2013) or must cope with additional stresses (van de Crommenacker et al., 2012). However, reproducing females from the H-BMR line type cope better with lower ambient temperature than those from the L-BMR line type (Sadowska et al., 2013). Thus, it is unlikely that even under worse conditions, elevated OS in the H-BMR line type would counter-balance profits resulting from better parental care. Second, results for DNA damage and the activity of catalase in the liver show that the initial differences in BMR can affect how parameters related to OS change during reproduction, confirming that inter-individual variation may be important in studies of the link between OS and the cost of reproduction (Metcalfe and Monaghan, 2013). Thus, we recommend that – whenever possible – future studies on the association between the costs of reproduction and OS should take into account systematic variation in the level of basal energy expenditures (such as BMR) specific to the studied organisms. Finally, contrasting directions of changes of the catalase activity in the liver and kidneys during reproduction (without simultaneous changes in markers of OS in these organs) suggest the presence of significant variation between organs in the activation of antioxidative mechanisms during reproduction, a pattern also found in other recent studies (Yang et al., 2013; Speakman and Garratt, 2014; Xu et al., 2014).

Subjects in our experiment were female Swiss-Webster mice (Mus musculus Linnaeus 1758) from generation 36 of an artificial selection experiment for high and low body-mass-corrected BMR. The selection experiment and the BMR assays are described in detail elsewhere (Książek et al., 2004; Gębczyński and Konarzewski, 2009). Briefly, males and females characterized by the highest and lowest mass-corrected BMR measured at age 12–16 weeks were chosen as progenitors of the H-BMR and L-BMR line types, respectively. A similar procedure was repeated in subsequent offspring generations, yielding significant differentiation of the line types with respect to BMR, without simultaneous changes in body mass. Although the described selection experiment has no replication, between-line-type differences in BMR and several other traits have been shown several times to be large enough to claim that they represent a genuine change in frequencies of alleles directly related to BMR rather than genetic drift (Książek et al., 2004; Brzęk et al., 2007; Gębczyński and Konarzewski, 2009).

Throughout the course of the selection experiment, mice were maintained in a climatic chamber at an ambient temperature of 23°C under a 12 h:12 h light:dark cycle, and were offered water and food ad libitum (Murine Laboratory Chow, Labofeed H, Wytwórnia Pasz A. Morawski, Kcynia, Poland). The same conditions were applied during the present experiment.

Following BMR measurements, being a part of the selection procedure, females used in the present study were assigned randomly to reproducing and non-reproducing (control) treatments. Virgin, non-reproducing females (32 in the H-BMR and 30 in the L-BMR line type) were maintained in separate cages. Reproducing females (23 in the H-BMR and 26 in the L-BMR line type) were concurrently bred at 22 weeks of age (this reproduction was a part of our selection procedure). After weaning they were paired again with males from their respective line types and gave birth at 30 weeks of age (males were removed before parturition). Cages were bedded with sawdust and provided with paper towels for nest construction. Litter size did not differ between line types (P>0.05 for both reproductive attempts).

All females were killed on day 17 of the second lactation of reproducing dams by cervical dislocation, and their liver, kidneys and heart were dissected and immediately frozen in liquid nitrogen. Blood samples were taken and centrifuged to collect blood serum. All samples were stored at −80°C. Samples from control females were collected at the same age as in reproducing females.

Measurements of resting metabolic rate in reproducing females do not represent BMR (i.e. the primary target of artificial selection) because they include the cost of pregnancy or milk synthesis. Therefore, we did not quantify metabolic rate of females during reproduction. However, to estimate whether the between-line-type difference in BMR was still significant after the first reproduction, we measured BMR in reproducing females right after the weaning of their first litter and compared this with values found in the same individuals before the experiment [see Książek et al. (Książek et al., 2004) for a description of the BMR assays].

All experimental procedures were accepted by the by the Local Ethical Committee in Białystok (permission 12/2009).

MDA was measured by means of the NWK-MDA01 assay kit (Northwest Life Science Specialties LCC, Vancouver, WA, USA) according to the manufacturer's instructions (before spectrophotometric analyses, samples were extracted with butane-pyridine mixture v/v 15:1). Catalase activity was assayed according to the method of Aebi (Aebi, 1983). Protein content in the supernatant was determined by means of the Lowry method using a Sigma Aldrich TP0300 kit (Sigma-Aldrich, St Louis, MO, USA). Repeatability of all assays was r≥0.9 (with the exception for protein content in heart, where r=0.83). MDA concentration was expressed in nmol mg⁻¹ protein, and catalase activity was expressed [following recommendation by Aebi (Aebi, 1983)] as the rate constant of a first-order reaction (k) per milligram protein.

The concentration of 8-oxodG in blood serum was quantified with the Trevigen HT 8-oxodG ELISA kit (4370-096-K; Trevigen Inc., Gaithersburg, MD, USA; analyzed in duplicate, r>0.93), and expressed as nanograms of 8-oxodG per milliliter blood serum. Immunoassays may overestimate 8-oxodG content (Cooke et al., 2008; Cadet et al., 2011); still, they can be a reliable test (Cooke et al., 2006), particularly when the aim of the study is to compare the relative level of 8-oxodG in different groups (Cooke et al., 2008).

Sample sizes for different assays differed because of limitations in the quantity of sample available for analysis, but the number of non-analyzed animals never exceeded two per group.

Variables were analyzed with an ANOVA with line type, reproductive status (reproducing versus non-reproducing females) and the order of measurements (first and second, for changes in BMR between two measurements) as main factors, their respective interaction terms as main factors, and the family affiliation (nested within line type) as a random factor. Body mass was added as a covariate in analyses of BMR. Besides line type and reproductive status, all other terms were included in the final model only when significant (P<0.05). We subsequently tested between-group differences by means of a Tukey's post hoc test. The magnitude of oxidative damage in lipids in liver, the concentration of 8-oxodG, and BMR were log-transformed, and catalase activity in liver was exponentially transformed before analyses to improve homogeneity of variance. All analyses were carried using procedure MIXED in SAS software.

We thank M. Lewoc, B. Lewończuk, S. Płonowski, P. Selewestruk and U. Wysocka for technical assistance in carrying out our experiment and analysis of data, and G. Bartosz and B. Kozłowska-Szerenos for advice and help in biochemical analyses. K. D. Kohl and K. O'Hara helped to edit an earlier version of the paper.