Mice selected for a high basal metabolic rate evolved larger guts but not more efficient mitochondria

Abstract

Intra-specific variation in both the basal metabolic rate (BMR) and mitochondrial efficiency (the amount of ATP produced per unit of oxygen consumed) has profound evolutionary and ecological consequences. However, the functional mechanisms responsible for this variation are not fully understood. Mitochondrial efficiency is negatively correlated with BMR at the interspecific level but it is positively correlated with performance capacity at the intra-specific level. This discrepancy is surprising, as theories explaining the evolution of endothermy assume a positive correlation between BMR and performance capacity. Here, we quantified mitochondrial oxidative phosphorylation activity and efficiency in two lines of laboratory mice divergently selected for either high (H-BMR) or low (L-BMR) levels of BMR. H-BMR mice had larger livers and kidneys (organs that are important predictors of BMR). H-BMR mice also showed higher oxidative phosphorylation activity in liver mitochondria but this difference can be hypothesized to be a direct effect of selection only if the heritability of this trait is low. However, mitochondrial efficiency in all studied organs did not differ between the two lines. We conclude that the rapid evolution of BMR can reflect changes in organ size rather than mitochondrial properties, and does not need to be accompanied obligatorily by changes in mitochondrial efficiency.

1. Background

Basal metabolic rate (BMR) is a key physiological and life-history parameter in endotherms [1,2]. Intra-specific variation in BMR reflects variations in several morphological and molecular traits (reviewed in [2]), including mitochondrial properties [3–7]. Most evolutionary and physiological studies investigating the variation in metabolic rate (including BMR) have assumed tacitly that (for the same diet type) the rate of oxygen consumption is directly proportional to the amount of obtained energy. However, it has been shown recently that there is wide inter- and intra-specific variation in mitochondrial efficiency (mitochondrial coupling, i.e. the amount of ATP produced per unit of consumed oxygen), which reflects variations in mitochondrial parameters such as the magnitude of proton leakage or the activity of the respiratory chain [8,9]. Since variation in mitochondrial efficiency affects the rate of both ATP synthesis and reactive oxygen species production, it can play an important role in the evolution of life histories and longevity [9–12].

BMR quantifies the unavoidable costs of body maintenance, and thus, one can expect the presence of strong selection for high mitochondrial efficiency. However, mitochondrial efficiency is negatively correlated with the body mass-specific metabolic rate at the interspecific level in mammals [13], although this may not hold for very small species [14]. This pattern presumably reflects a negative correlation between body mass and proton leakage (and therefore a negative correlation between the mass-specific metabolic rate and mitochondrial efficiency) observed in mammals and birds ([3,4] but see [7]). On the other hand, a higher mitochondrial efficiency can be related to better performance at the intra-specific level. For example, there is a positive correlation between mitochondrial efficiency and the growth rate [15] and locomotor performance [16,17]. In mice, there is a negative correlation between parental effort and the maximal rate of mitochondrial proton leakage, suggesting a positive correlation with mitochondrial coupling efficiency [18]. Similarly, in the pied flycatcher, the mitochondria became more coupled during chick feeding [19], whereas experimentally elevated mitochondrial uncoupling reduced the number of eggs laid in the zebra finch [20]. However, there are some exceptions, such as a negative link between mitochondrial coupling and total energy expenditure in mice [21] or a negative effect of training on mitochondrial coupling in humans [22].

Knowledge of the relationship between the BMR and mitochondrial efficiency at the intra-specific level is of utmost importance for understanding evolutionary changes in BMR. In particular, many models explaining the evolution of endothermy assume a positive correlation between the rate of basal energy expenditure and performance capacity [23–25]. However, the results presented above suggest that mitochondrial efficiency is negatively related to mass-specific BMR at the interspecific level but positively (although with exceptions) related to individual performance at the intra-specific level. This contrast is puzzling, as interspecific variation emerges from intra-specific variation. Moreover, this problem may be more general, as ectotherms also show both a negative correlation between proton leakage and body mass at the interspecific level [26] and a positive correlation between the coupling efficiency and growth performance at the intra-specific level [27].

The subject of our study was two lines of laboratory mice divergently selected towards either high (hereafter H-BMR line) or low (L-BMR line) levels of BMR. The difference in body mass-corrected BMR between these lines reached 50–80% in males (P. Brzęk, A. Gębczyński 2015–2021, personal observation). Approximately half of this difference can be explained by the larger relative size of the metabolically active internal organs in H-BMR mice [28,29]. Selection for BMR has affected several molecular and genetic traits [29–31], including fatty acyl composition [29], which is likely to be related to the mitochondrial coupling efficiency. Moreover, mice from the H-BMR line showed higher values of spontaneous locomotor activity [32], better lactation performance [33] and a faster pup growth rate [34] than the L-BMR line. Thus, these two lines are particularly suitable for testing the correlation between intra-specific variation in BMR and mitochondrial coupling efficiency.

Here, we compared mitochondrial coupling efficiency in the liver, kidney and skeletal muscles between the two lines. All of these organs have been proposed to be important predictors of BMR because of the high mass-specific metabolic rate in the liver and kidney [35] or the high total mass of the skeletal muscles [36,37]. We hypothesized that long-term, divergent selection for BMR was likely to modulate mitochondrial coupling efficiency because mitochondria are directly involved in oxygen metabolism, although we did not have a clear prediction about the direction of this modulation because of the previously observed contradictory patterns of the inter- and intra-specific links between the rate of metabolism and mitochondrial efficiency. Thus, we measured the oxidative phosphorylation activity (i.e. oxygen consumption and ATP synthesis) and efficiency (ATP/O ratio) in liver, kidney and skeletal muscle mitochondria.

2. Methods

(a) . Animals

All details of the selection procedures and animal maintenance are presented elsewhere [28,38]. Briefly, two lines of Swiss Webster mice were selected for either high (H-BMR) or low (L-BMR) levels of body mass-corrected BMR measured at three to four months old. The subjects in the present experiment were six-month-old males (eight from each line) from generation F57 of the selection experiment. All experiments were conducted in accordance with the guidelines of Polish law.

(b) . Quantification of mitochondrial respiration and mitochondrial oxidative phosphorylation efficiency

After BMR measurements (methods described in [28]), being a part of the selection procedure, the mice were sacrificed by cervical dislocation. Their liver, kidney and hindlimb skeletal muscle samples were rapidly dissected, weighed and subjected to biochemical assays. Detailed description of mitochondrial isolation and biochemical assays is presented in the electronic supplementary material. In brief, we quantified the following parameters of mitochondrial respiration using either complex I substrates (pyruvate/malate) or a combination of complexes I and II substrates (pyruvate/malate/succinate): basal non-phosphorylating respiration (basal-CI and basal-CI, II), phosphorylating respiration in the presence of 1 mM ADP (OXPHOS-CI and OXPHOS-CI, II) and maximal activity of electron transport system (ETS). We calculated the respiratory control ratio (RCR) as OXPHOS-CI, II (state 3)/basal-CI, II (state 4). In separate assays, we quantified mitochondrial oxidative phosphorylation efficiency by measuring the rate of oxygen consumption and ATP synthesis in mitochondria respiring on pyruvate/malate/succinate.

(c) . Statistics

Our study is based on the comparison of two selected lines. Interpretation of the results of unreplicated selection experiments cannot rely on conventional statistical significance; rather, one has to rule out the possibility that the observed difference between lines are due to genetic drift rather than direct effects of artificial selection [39,40]. Therefore, we analysed our results according to the guidelines suggested by Henderson ([39,40]; see also [41]). First, we calculated the within-line standard deviations (SDs) of analysed traits. Individual mice used in our experiment came from different families, and thus those SDs can be interpreted as phenotypic SDs, being a square root of the product of narrow-sense heritability of a given trait and its genetic SD [40]. We used obtained SDs to calculate standard deviation SD_x weighted across both selected lines as

$${\mathrm{SD}}_x\cong \sqrt{{\displaystyle \frac{{\mathrm{SD}}_{h\hspace{0.17em}}^2\hspace{0.17em}(n_h-1)+{\mathrm{SD}}_{l\hspace{0.17em}}^2\hspace{0.17em}(n_l-1)}{n_h+n_l-2}}},$$ 2.1

where ${\mathrm{SD}}_h$ and ${\mathrm{SD}}_l$ indicate phenotypic standard deviations of the analysed trait for the H-BMR and the L-BMR lines, respectively, whereas $n_h$ and $n_l$ represent the numbers of families used in the selection experiment. Then, we expressed the magnitude of separation of the lines d with the studied trait as the multiples of SD_x:

$$d=\left|{\displaystyle \frac{{\overline{x}}_{\mathrm{high}}-{\overline{x}}_{\mathrm{low}}}{{\mathrm{SD}}_x}}\right|,$$ 2.2

where ${\overline{x}}_{\mathrm{high}}\hspace{0.17em}$ and ${\overline{x}}_{\mathrm{low}}$ indicate mean values found in lines selected for high and low BMR, respectively. Finally, we used a modified equation 16 from Henderson [40] to calculate the 95% CI for the analysed trait:

$$95\text{\%}\hspace{0.17em}\mathrm{CI}\cong 4\sqrt{(h_y^2\hspace{0.17em}F+1/n)}$$ 2.3

where $h_y^2$ is the narrow-sense heritability of studied trait, F is the level of inbreeding and n represents the number of families used in the selection experiment.

The Henderson approach can be interpreted as a test of the null hypothesis that an analysed trait has not been affected by artificial selection. In such a situation, the expected difference between mean values of selected lines is zero, and 95% CI quantifies the magnitude of the difference between lines that could arise because of genetic drift. Thus, only differences in traits with d > 95% CI can be claimed to represent the genuine effect of selection, whereas for d < 95% CI one cannot exclude the possibility that observed difference reflects the effect of genetic drift only. We emphasize that whereas the number of families used in the selection experiment affects the magnitude of expected genetic drift and thus 95% CI (see equation (2.3)), the number of animals in which the analysed trait was quantified does not affect d (since sample size occurs both in the numerator and denominator of equation (2.1)). Thus, the higher number of assayed animals does not increase the statistical power of the Henderson approach (although it can offer more accurate estimates of mean values and standard deviation of the analysed trait).

The coefficient of inbreeding (F) for generation 57 was 0.314 (calculated from equation (3.5) from [42]). The number of families maintained in the selection experiment has been gradually increased over time (from 80 to 190 in generation F57), and we assumed conservatively n = 80 when calculating 95% CI (although values of 95% CI calculated for n = 80 and n = 190 were similar; P. Brzęk 2022, personal observation). We are not aware of any published estimates of h² of mitochondrial oxidative phosphorylation or mitochondrial efficiency; moreover, these traits are the result of mitonuclear genetic interactions that can result in complex inheritance patterns [43]. Thus, we present figure plotting calculated 95% CI as a function of possible h² of the analysed trait (figure 1). We note that, to claim that the observed trait has been affected by selection, d must be at least 1.10 if h² of that trait is 0.2, and at least 0.84 if h² is 0.1. We did not make any assumptions about h² of BMR and the size of internal organs because our conclusions about the effect of selection on these traits were independent on their h² (see §3). Even though the Henderson approach is the most appropriate statistical method to analyse the results of an unreplicated selection experiment, in the electronic supplementary material we present results of conventional t-tests, ANOVA and ANCOVA for comparative purposes.

Figure 1.

Explanation of the Henderson approach applied in the present study. The magnitude of the observed separation of the H-BMR and L-BMR lines with respect to sample studied traits (d) along with the upper boundary of the difference between lines that could arise because of genetic drift (95% CI, solid line). Values of 95% CI are plotted as a function of potential values of narrow-sense heritability (h²) of analysed traits. Unlike the d values for BMR, liver mass and kidney mass, the d value for liver mitochondrial efficiency always falls below 95% CI and thus cannot be ascribed to the effect of selection. On the other hand, the d values for OXPHOS-CI and maximum ATP synthesis rate in liver can be interpreted as the effect of artificial selection rather than genetic drift only when the h² of those traits is low.

3. Results

Analysed mice from the H-BMR line had higher BMR that their counter partners from the L-BMR line (means ± s.e. for H-BMR and L-BMR mice, respectively: 72.6 ± 2.99 ml O₂ h⁻¹ and 37.8 ± 1.68 ml O₂ h⁻¹, d = 5.06). Similarly, H-BMR mice had larger liver and kidney size (least-square means corrected for body mass ± s.e. for H-BMR and L-BMR, respectively: liver mass—2.05 ± 0.04 g and 1.47 ± 0.04 g, d = 4.88; kidney mass—0.63 ± 0.016 g and 0.43 ± 0.016 g, d = 4.41). Values of d for all these traits were large enough to claim that they represent the result of artificial selection rather than genetic drift, independently on the potential h² of analysed trait (figure 1). Mice from the H-BMR line also had higher values of mitochondrial respiratory parameters in liver, though d values were much lower than in the case of liver mass (figure 2b). The highest value of d was found for the OXPHOS-CI (though it was even 1.31 for the OXPHOS-CI, II activity when one outlying point was removed from the analysis). With the exception of the basal-CI activity, all these traits could be claimed to be affected by selection only if their h² is not higher than 0.1–0.2 (compare figure 1 and figure 2b). Estimates of the magnitude of separation d for mitochondrial respiratory parameters in kidneys and skeletal muscles did not exceed 0.65 (with the exception of the basal-CI in kidneys; figure 2a,c) and thus they are unlikely to represent the effect of selection unless their h² is almost zero (in practice, lower than 0.05; compare figure 1). RCR did not differ between L-BMR and H-BMR mice (means ± s.e. for L-BMR and H-BMR mice: skeletal muscles—10.2 ± 1.79, 10.4 ± 1.15, d = 0.04; liver: 6.2 ± 0.47, 6.6 ± 0.71, d = 0.26; kidney—6.7 ± 0.881, 5.5 ± 0.53, d = 0.61).

Figure 2.

(a) Mitochondrial respiratory parameters in skeletal muscles, (b) liver and (c) kidney. Means ± s.e. and values of the magnitude of separation d of the H-BMR and the L-BMR lines are shown.

The lines depicting the relationship between oxygen consumption and the rate of ATP synthesis in both selected lines were fully superimposed and values of the magnitude of separation d for mitochondrial efficiency never exceeded 0.25 (figure 3), indicating that mice from each line synthesized the same amount of ATP per unit of consumed oxygen (i.e. artificial selection for BMR did not affect coupling efficiency). However, we found that the d value for the maximum ATP synthesis rate in liver mitochondria was 0.97 (figure 3b, although it decreased to 0.81 when one outlying point was deleted). Thus, under the same conditions, liver mitochondria from H-BMR mice produced more ATP (with the same coupling efficiency) than liver mitochondria from L-BMR mice; however, the h² of the maximum ATP synthesis rate cannot be higher than 0.15 to claim that this difference reflects the effect of selection (figure 1).

Figure 3.

Linear relationship between the rates of ATP synthesis and oxygen consumption in (a) skeletal muscles, (b) liver and (c) kidney. Means ± s.e. are shown. For each pair of points, numbers indicate values of the magnitude of separation d of the H-BMR and the L-BMR lines for oxygen consumption rate, ATP synthesis rate and mitochondrial efficiency, respectively.

4. Discussion

We found that even though divergent selection for BMR resulted in an almost twofold relative between-line difference in BMR in studied individuals, it did not affect mitochondrial efficiency in the most important internal organs. We emphasize that the studied mice came from two lines that have been subjected to almost 60 generations of continuous, divergent artificial selection. Nevertheless, the effect of selection on mitochondrial efficiency was virtually non-existent, irrespectively of the h² of this parameter (figure 1). The lack of any difference in mitochondrial efficiency between H-BMR and L-BMR mice is in line with the observation that all parameters quantifying mitochondrial activity in particular organs were either unaffected by selection or changed in unison (figure 2). On the other hand, we found that variables quantifying mitochondrial respiratory parameters in liver as well as the maximum ATP synthesis rate in this organ were higher in mice from the H-BMR line (figures 2b and 3b). However, these differences may be interpreted as the result of selection only if h² of analysed traits is low (no higher than 0.1–0.2; compare figures 1 and 2b). This is not unlikely, as traits related to tissue oxidative capacity were reported to possess rather low h² (lower than 0.25; [44]). However, even though the described selection experiment could potentially affect traits related to the mitochondrial oxidative phosphorylation activity in liver, the magnitude of between-line separation with respect to these parameters was much lower than we found for the relative size of liver and kidney, whereas the effect of selection on organ size was almost as strong as on BMR, i.e. the primary target of our selection (figure 1). Thus, divergent artificial selection for BMR significantly affected the relative size of the liver and kidneys (pattern observed also in earlier generations; [28,29]), but its effect on the intensity of mitochondrial respiration was markedly weaker (at best) and virtually non-existent for the efficiency of mitochondrial respiration.

The lack of a more pronounced effect of selection on mitochondrial function and efficiency can seem unexpected since H-BMR and L-BMR mice differ not only in BMR but also in the intensity of long-term, sustained activities such as spontaneous locomotor activity [32,38], parental investment [33] and growth rate [34]. However, both lines differ little or none in maximum aerobic capacity, as demonstrated by running or swimming in 25°C water ([28,38]; P. Brzęk 2010, personal observation). We hypothesize that mitochondrial properties are more likely to be modulated rapidly when selection acts on maximum energy requirements (i.e. the maximum metabolic rate) or on the capacity to address ‘extreme’ challenges such as fasting [45,46] or hypoxic/anoxic conditions. BMR and long-term sustained metabolic rate represent submaximal levels of aerobic metabolism [47] and can be related more to the delivery rate of oxygen and nutrients to cells than to mitochondrial properties. Accordingly, our selection treatment affected the size of organs such as the liver, kidneys, heart and intestine [28,29,33,34,48], which are related to food processing and oxygen/nutrient delivery, but not mitochondrial properties. Here, we must add an important caveat that even though the activity of mitochondria from the H-BMR and L-BMR lines is similar under the same conditions in vitro, we cannot exclude that both lines can differ in their intensity of mitochondrial activity in vivo. For example, if some tissues are more active in H-BMR mice, their higher cellular energy needs can result in higher mitochondrial activity. This is particularly likely because H-BMR mice have smaller erythrocytes and thus presumably a higher capacity for oxygen delivery [30]. However, this mechanism can affect the intensity of mitochondrial respiration but not its efficiency.

We concluded that a very large (almost twofold) difference in BMR can arise rapidly (from an evolutionary point of view) without changes in coupling efficiency at the level of isolated mitochondria in the liver, kidney and skeletal muscles. Although we cannot exclude that both lines differ in mitochondrial activity in vivo (see above) or in the number/density of mitochondria per tissue (compare [14,37,49]), it seems that under ‘normal’ conditions, both lines have similar mitochondria. Finally, the combined results of our earlier studies ([28,29]) and the present experiment show that divergent selection for BMR affected the relative size of internal organs such as the liver and kidney. Particularly relevant here is a recent study that found that the evolution of a very high BMR in an extremely small rodent, the African pygmy mouse (Mus mattheyi), was not accompanied by any changes in mitochondrial metabolism and efficiency that could be expected from the patterns observed in larger rodent species [14]. Conversely, pygmy mice have a relatively larger liver size and presumably a higher liver mitochondrial content [14]. Thus, both Böel et al. [14] and the present study suggest that the evolution of high BMR in small endotherms can proceed through changes in relative organ size rather than changes in mitochondrial activity and efficiency as suggested by wider interspecific comparisons [13]. Strikingly, we came to similar conclusion in our previous study of the fatty acid composition of cell membranes in H-BMR and L-BMR lines [29]. All these results lend experimental support to studies linking intra-specific variation of BMR to the size of internal organs (e.g. [35]), which were founded on the untested assumption of a lack of variation in molecular parameters, including mitochondrial efficiency. Therefore, the rapid evolutionary increase of BMR does not need to be obligatorily accompanied by potential life-history costs/benefits resulting from changes in mitochondrial efficiency (described in [9–12]).

Ethics

All experiments were conducted in accordance with the guidelines of Polish law.

Data accessibility

The data are provided in the electronic supplementary material [50].

Authors' contributions

P.B.: conceptualization, formal analysis, funding acquisition, investigation, methodology, resources, visualization, writing—original draft, writing—review and editing; D.R.: conceptualization, formal analysis, investigation, methodology, resources, visualization, writing—review and editing; M.K.: conceptualization, formal analysis, funding acquisition, investigation, methodology, resources, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

The authors declare that they have no competing interests.

Funding

This study was supported by the National Science Centre, Poland, grant nos. 2014/15/B/NZ8/00244 to P.B. and 2017/27/B/NZ8/02242 to M.K.