Effects of oxygen availability on maximum aerobic performance in Mus musculus selected for basal metabolic rate or aerobic capacity

Maximum aerobic metabolic rate () determined during running on a treadmill is a standard tool to study maximum sustainable metabolic power production, which depends on the integrated function of a wide variety of organ systems (e.g. Swallow et al., 1998; Koch and Britton, 2001; Rezende et al., 2005). Since it sets the upper limit to sustainable activity, is relevant ecologically, evolutionarily and physiologically, as well as clinically. As cannot increase indefinitely in response to increasing demands for ATP production, it is important to learn what kind of constraints limit . Conceptually, could be set by the interactions between five sequential steps of O₂ transport and uptake: (1) ventilatory convection, (2) alveolar–capillary diffusion, (3) blood convection, (4) tissue capillary-to-cell diffusion and (5) mitochondrial capacity for O₂ uptake (Taylor and Weibel, 1981; Wagner, 1996). In principle, each of the sequential steps of O₂ delivery and uptake can be limiting and many earlier studies identified some of them as limiting, while often pointing to the existence of excess transport capacity at other steps. For example, several studies suggested that cardiac output (that is, blood convection) will limit (Bishop, 1997; Richardson et al., 1999a), and studies on human athletes suggested that their muscle tissue oxidative capacity is in excess of O₂ delivery capacity (e.g. Richardson et al., 1999b). Alternatively, it has been proposed that all steps in the O₂ cascade are optimized by natural selection, so their transport capacities match each other and none of them alone would set the limit for maximum organismal performance [‘symmorphosis’ (sensuTaylor and Weibel, 1981)].

Several decades of studies on have shown that the identification of limiting steps of O₂ delivery is not straightforward (Koteja, 1986; Henderson et al., 2002; Howlett et al., 2003; Seymour et al., 2004; Gonzales et al., 2006; Kirkton et al., 2009). A promising approach to identifying possible limits to is offered by artificial selection experiments on metabolic rates (Swallow et al., 2009) (for reviews, see Garland and Rose, 2009; Garland et al., 2010). Animals manipulated through artificial selection can be useful in two ways. First, one can consider artificial selection as a tool for ‘engineering’ animals that are close to the within-species boundaries of variation of traits related to metabolic rates. It is instructive to take advantage of that artificially increased variation and carry out comparative analysis in much the same way as at the interspecific level, but being freed from many confounding factors, such as between-species differences in anatomy, physiology, phylogeny, etc. Second, within-selection analyses can also be informative, because mechanisms limiting can be manifested as differences in within-selection responses to the applied metabolic stressor(s).

In recent years, much of the progress in understanding complex interactions between subsequent components of the O₂ cascade has been made by studying in artificially selected rodents subjected to hypoxic and/or hyperoxic conditions, in comparison to normoxic conditions (Howlett et al., 2003; Howlett et al., 2009; Rezende et al., 2006a; Rezende et al., 2006b). Metabolic responses to different partial pressures of O₂ (P_O2) can give insights into which of these steps is limiting.

We hypothesise that under moderately hypoxic conditions one can expect a reduction in (so-called central limitation) unless there exists an excess capacity for O₂ uptake and delivery at steps 1–3 that can offset the reduction of P_O2 relative to normoxia. The existence of limitations at steps 1–3 would be further supported by a significant increase of in response to hyperoxia as a result of facilitated pulmonary diffusion and/or blood convection. However, an increase in under hyperoxia would be only possible if the extra O₂ can be consumed at capillary-to-cell and mitochondrial level. Otherwise, the lack of such an increase would point to potential constraints at the tissue level (so-called peripheral limitation). The consistency (or the lack thereof) of the above responses across different artificial selection regimens applied within species can inform us about the ‘design constraint’ related to the symmorphosis principle.

Our laboratory has been running two experiments on lines of mice: selection for high maximum metabolic rate elicited by swimming () and divergent selection on basal metabolic rate (BMR). These experiments provide us with a unique opportunity to examine interactions between aerobic limits elicited by different ambient partial pressure of O₂ and selection-generated differences in metabolism. We have already demonstrated that selection on resulted in correlated increases of both normoxic elicited by running () and heart mass, which we assume is a proxy of cardiac output (Bishop, 1997; Gębczynski and Konarzewski, 2009b). In both line types (selected and randomly bred), maximum metabolic rates elicited by cold exposure in a He–O₂ atmosphere (hereafter referred to as ) were significantly higher than (Gębczynski and Konarzewski, 2009b). When compared with mice selected for high BMR, mice selected for low BMR differed in spontaneous locomotor activity and heart mass, but did not differ with respect to in normoxia or (Książek et al., 2004; Brzęk et al., 2007; Gębczynski, 2008; Gębczynski and Konarzewski, 2009a). Moreover, in the BMR-selected lines, values of and were similar (unlike in the line type).

In the current study our objective was to use different levels of O₂ availability to help identify possible limits to elicited by forced exercise in mice from both selection experiments. We compared the responses to hypoxia and hyperoxia in mice of both line types in the context of the central and peripheral components of the O₂ cascade and their relevance to the idea of ‘symmorphotic’ design.

We used 12-week-old males of outbred Swiss-Webster laboratory mice (Mus musculus Linnaeus 1758) subjected to divergent non-replicated artificial selection toward high (H-BMR) and low (L-BMR) BMR corrected for body mass [for details see Książek et al. (Książek et al., 2004)]. Lines are maintained in animal facilities at the Institute of Biology, University of Bialystok, Poland. Briefly, depending on breeding success, in subsequent generations we maintain 30–35 families in each selected line type. Whenever possible, no less than three randomly chosen males and three females from each family were tested for BMR at 31–32°C, a temperature within the thermoneutral zone of our mice (M.K., unpublished data). Mice are fasted for 6 h before measurements. We defined BMR as the lowest 4 min readout during the last 2 h of the 3 h trial period, within which oxygen concentration does not change by more than 0.01%. Animals characterized by the highest (in H-BMR line type) or lowest (L-BMR line type) body-mass-corrected BMR are chosen as progenitors and always mated outside their families.

In the study described here we used 38 H-BMR males and 39 L-BMR males of generation F34. In H-BMR and L-BMR males the BMR averaged 71.0±2.3 and 44.5±2.1 ml O₂ h^–1, respectively, a 60% between-line-type difference. This difference is sufficiently large to be confidently attributed to the results of selection, rather than to genetic drift (Gębczynski and Konarzewski, 2009a). Body mass did not differ between line types (H-BMR line type: 36.4±0.7 g, L-BMR line type: 36.5±0.6 g).

We used 205 12-week-old males from the 12th generation. Body-mass-corrected differed considerably between selected and random bred (control) individuals entering the experiment (282.5±4.2 ml O₂ h^–1 and 253.2±4.4 ml O₂ h^–1, respectively; ANCOVA, F_3,6=23.1, P<0.01).

Briefly, we established eight genetically isolated Swiss-Webster laboratory mice lines and in each of them maintained 10–15 families. In four of the lines, mice were selected for high , and the other four were randomly bred (hereafter referred to as RB), control lines. To measure we used a vertically positioned cylindrical Plexiglas® metabolic chamber supplied with atmospheric air (700 ml min^–1). The chamber was partly filled with water maintained at 25±0.2°C, leaving a dry volume of 560 ml. Each mouse was placed just above the water level on a movable platform, and allowed 10 min for adaptation. The platform was then abruptly submerged to force the animal to swim. was defined as the highest body-mass-corrected oxygen consumption averaged over 2 min of a 5 min swim. Animals with the highest were chosen as progenitors of the selected lines. No fewer than three randomly chosen males and three females from each family were subjected to metabolic trials. Of these, no more than two (typically one) males and females were chosen as progenitors and mated outside their families. Exactly the same procedure was applied to the RB lines, except that the mated individuals were picked at random, but still mated outside their families.

After weaning, animals were housed in same-sex and same-family groups of up to five individuals per cage at 23°C. They were maintained on a 12 h:12 h light:dark cycle and had unlimited access to murine chow (Labofeed H, FPP, Kcynia, Poland) and water.

Upon completion of measurements of BMR and as part of the routine selection protocols, randomly selected males (up to three individuals from each family within line type, not qualified as progenitors) were randomly assigned to three P_O2 trials and subject to measurements of carried out using a motorized treadmill enclosed in a metabolic chamber (700 ml in volume) that was fitted to the above-described respirometry system. We elected to carry out measurements of sequentially because of logistical constraints – first, within selection on BMR, then within selection on . This was a compromise between statistical and logistical demands, which allowed us to minimize the overall duration of measurements, and therefore to minimize the confounding effect of aging on measured traits (P. Brzęk, A. Książek, A. Dobrezyn and M.K., unpublished data), albeit at the expense of the desirable random block design of the experiment.

We used a measurement protocol similar to earlier published studies (Swallow et al., 1998; Rezende et al., 2006a), with an enclosed treadmill functioning as a metabolic chamber. Outside atmospheric air, or mixtures of 70% nitrogen/30% oxygen or 86% nitrogen/14% oxygen, was dried and forced through a copper coil submerged along with the treadmill in a water bath to equalize and control temperature (23±1.0°C). The flow rate was 700 ml min^–1 (mass flow meter, Sierra Instruments, Monterey, CA, USA). Animals were individually placed in the chamber while the treadmill was stopped and the system was sealed and submerged in the water bath. ‘Resting’ oxygen consumption was recorded for 2 min. The treadmill was then started at an initial speed of 0.7 km h^–1. Mice were induced to run by a mild electric current (200 V, 0.5–1.5 mA) through a horizontal grid of six 2 mm bars spaced 5 mm apart at the end of the moving belt. Speed was increased every 2 min by 0.5 km h^–1. Trials were ended when the mouse failed to keep pace with the treadmill, on average after 7.9 min of running (from 6.7 min in hypoxia to 8.7 min in hyperoxia). Oxygen concentration was recorded every 0.5 s by a computer. We used the highest oxygen consumption averaged over 1 min as .

Each mouse was measured only once, in hypoxia, normoxia or hyperoxia. Three animals failed to run on the treadmill and were subsequently removed from the experiment (two individuals and one individual from the BMR and selection experiments, respectively). For this reason within-line-type sample sizes varied between experimental groups (see Table 1).

We carried out measurements changing the P_O2 every half a day, whereas the P_O2 of the very first set of measurements on the very first morning was selected at random. The next P_O2 value was randomly drawn from the two remaining options. The next day, we started with measurements at the same P_O2 as on the previous afternoon and continued with the third (yet not applied) P_O2, and so on. This sequence assured an equal distribution of P_O2 values between morning and afternoon measurements.

All metabolic trials were carried out between 08:00 and 20:00 h. Metabolic data were collected and analysed with Sable System (Sable System, Salt Lake City, UT, USA) DATACAN V software. We calculated oxygen consumption rates using equation 4a of Withers (Withers, 1977), and attempted to correct instantaneous values of O₂ consumption for the chamber washout time by applying a Z transformation [Bartholomew et al. (Bartholomew et al., 1981) implemented in DATACAN V software]. However, as the magnitude of the correction was less than 1%, and caused an increase in variance, we elected not to apply the correction in our final analyses. We used the highest oxygen consumption averaged over 1 min as .

Immediately after measurements of , -selected and RB mice were killed by cervical dislocation. Blood for measurements was taken directly from the heart. To determine red blood cell (RBC) concentrations, 5 μl of blood was diluted with 1 ml of Hayem reagent solution and the cells were counted under a microscope using a Bürker counting chamber (Carl Roth, Karlsruhe, Germany). Haemoglobin concentration was estimated, in duplicate, by adding 20 μl of blood to a test tube containing 5 ml of Drabkin's reagent and the mixture was incubated for 30 min at room temperature in the light. Absorbance was then measured at 540 nm using a UV-1201V spectrophotometer (Shimadzu Corp., Kyoto, Japan). Concentrations were determined from calibration curves obtained with standard haemoglobin solutions diluted in Drabkin's reagent at the following concentrations: 0.0, 5.0, 10.0, 15.0 and 20.0 g 100 ml^–1.

of mice selected on was analysed with a mixed model extension of a general linear model (proc MIXED, version 9.1, SAS Institute, Cary, NC, USA, 1996) with line type (selected vs random bred) and partial oxygen pressure (hypoxia, normoxia and hyperoxia) as fixed factors, line (replication) nested within line type and family affiliation (nested within line) as random factors. Body mass measured immediately prior to the trial was used as a covariate. The effects of P_O2, line type and P_O2 × line-type interaction were tested by means of an F-test against variation of a random P_O2 × line-within-line-type interaction as the error term. We also applied a similarly structured statistical model (but without P_O2 as a fixed factor, body mass as covariate and line nested within line type as the error term) to analyse between-line-type differences in the number of RBCs and haemoglobin concentration.

Our selection on BMR was not replicated. To directly compare responses of mice from the BMR-selected lines with the -selected and RB lines we pooled for all line types and then analysed the between-line-type variation by means of the proc MIXED procedure with the model structure as described above (which we refer to as the ‘common’ model). Because the line type × P_O2 interaction was statistically significant in a common model, we carried out separate analyses for all line types within each level of P_O2 (with a random effect of line nested within line type as an error term).

To further elucidate the effect of P_O2 we separately compared of mice from our two artificial selection regimes with of four RB lines. We also analysed of mice divergently selected on BMR (H-BMR vs L-BMR) by means of the mixed model as described above for selection on , with family affiliation (nested within line type) as the error term.

Because mice used in this experiment were drawn from the pool of animals not qualified as progenitors, they cannot be considered as a random unbiased sample. For all analyses, assumptions of parametric tests were met (Sokal and Rohlf, 1995). All P-values shown are from two-tailed tests.

Mean body mass did not vary between lines and was strongly affected by P_O2 in all line types (Table 1).

A mixed ‘common’ model with line type (H-BMR, L-BMR, and RB) revealed a weak line type × P_O2 interaction (F_6,12=2.89, P=0.05, Fig. 1). To pinpoint the source of this interaction we repeated the analysis of within each level of P_O2. These analyses did not detect an effect of line type at 14% and 21% O₂ (F_3,6=1.68, P=0.27 and F_3,6=3.39, P=0.09, respectively), but revealed a significant line-type effect at 30% O₂ (F_3,6=6.58, P=0.02). A Tukey's test indicated that this effect was due to significant differences between H-BMR and randomly selected line types (P=0.05).

A mixed model of ANCOVA of with line type (H-BMR, L-BMR and RB) revealed a significant effect of P_O2 (F_2,6=191.6, P<0.001) and line type (F_2,6=19.1, P=0.002) with a nonsignificant line type × P_O2 interaction (F_4,6=3.7, P=0.07). differed significantly between all three levels of P_O2 (pair-wise Tukey's test, P=0.01). The Tukey's test also revealed significant differences between H-BMR and the other two line types (P=0.05).

Within-selection ANCOVA with P_O2 and line type (L-BMR and H-BMR only) as fixed effects did not reveal a significant line type × P_O2 interaction (F_2,47=0.77, P=0.47), but detected strong effects of P_O2 (F_2,47=36.8, P<0.001) and line type (F_1,47=11.5, P=0.001), with significant differences between all three levels of P_O2 (pair-wise Tukey's test, P<0.001; Fig. 2).

was significantly affected by P_O2 (F_2,12=453.1, P<0.0001) with values significantly different between all three P_O2 levels (pair-wise Tukey's test, P<0.001; Fig. 3). was also affected by the line type (F_1,12=21.8, P=0.0005), with values consistently higher in lines selected for across all P_O2 levels (Fig. 3). The P_O2 × line type interaction was not significant (F_2,12=2.6, P=0.1).

Comparison of the number of RBCs of selected vs RB lines did not reveal statistically significant differences (F_1,6=1.6, P=0.22). Likewise, haemoglobin concentration did not differ significantly between line types (F_1,6=1.9, P=0.26).

Recall that can be considered to be set by interactions between five major processes of O₂ transport and uptake: (1) ventilator convection, (2) alveolar–capillary diffusion, (3) blood convection, (4) tissue capillary-to-cell diffusion (Taylor and Weibel, 1981; Wagner, 1996) and possibly (5) mitochondrial oxidative capacity (Weibel et al., 1991). What can our results tell us about possible limiting steps?

Maximum oxygen consumption of mice from both selection experiments was significantly affected by gas composition, and relative to normoxia, significantly decreased in response to hypoxia (Fig. 1). This pattern was also reported in rats selected for increased endurance capacity (Henderson et al., 2002; Howlett et al., 2003) and mice selected for high voluntary activity (Rezende et al., 2006a; Rezende et al., 2006b). The most straightforward interpretation of such an outcome is that decreased in hypoxia is due to limitations in O₂ delivery in steps 1–3, which is further supported by an increase of in hyperoxia. However, although in our experiment consistently increased in hyperoxia in both selected and randomly bred lines, there was no difference between normoxia and hyperoxia in the control lines of the selection on voluntary wheel running (Rezende et al., 2006a). Thus, responses to hypoxia were similar across different selection regimens, but differed in mice exposed to hyperoxic conditions. Interestingly, significant between-line differences at hyperoxia reported in our study suggest that peripheral oxidative capacity can substantially increase through selection on high BMR (Fig. 1).

To identify which of the specific steps in the oxygen transport cascade are limiting oxygen delivery it is worth noting that in a previous study we demonstrated that in mice selected for increased the level of was at least 25% higher that (25% for RB and 26.7% for selected lines) (Gębczynski and Konarzewski, 2009b). Likewise, it has been demonstrated that in lines of mice selected for high voluntary activity the elicited by cold exposure was approximately 32% higher than that on the treadmill, with a similar mean increase for both selected (31.6% increase) and RB (32.4% increase) lines (Rezende et al., 2005). In contrast, in mice selected for divergent BMR, elicited by cold (Książek et al., 2004) and noradrenaline injection (Gębczynski, 2008) did not differ from in normoxic conditions (Gębczynski and Konarzewski, 2009a) (this study).

The main difference between oxygen delivery in cold () and exercise is the tissues where the O₂ is used. In exercise most of the O₂ is consumed in skeletal muscle (Mathieu et al., 1981), whereas during cold exposure largely reflects the extraction of O₂ by brown adipose tissue (Zaninovich et al., 2003). That is, steps 1–3 are common to both exercise and thermogenesis, and yet, mice selected on (Gębczynski and Konarzewski, 2009b) and voluntary activity (Rezende et al., 2005) are capable of delivering more oxygen under cold exposure. In other rodent species, is also higher than [e.g. deer mice (Chappell, 1984; Hayes and Chappell, 1986); ground squirrels (Chappell and Bachman, 1995)]. This points to step 4 (tissue capillary-to-cell diffusion) as limiting in mice selected on and voluntary running (but not in mice selected on BMR, as their elicited by exercise and thermogenesis did not differ). This view is further supported by results of studies on rats selected for divergent low- and high-endurance exercise capacity (Henderson et al., 2002; Howlet et al., 2003; Howlet et al., 2009; Gonzales et al., 2006). These studies demonstrated that the first structural and functional responses to selection enabling greater O₂ utilization in the high capacity runner line type were muscle adaptations that improved oxygen extraction, with little difference in O₂ delivery capacity at the level of heart and lungs (Henderson et al., 2002; Howlett et al., 2003). In particular, at generation seven of selection (G7), increased muscle extraction was coupled with 33% between-line-type difference in skeletal muscle oxygen conductance (Howlett et al., 2003).

However, further selection on low- and high-endurance exercise capacity resulted in substantial differences in pulmonary function (Kirkton et al., 2009) and cardiac output (Gonzales et al., 2006). Howlett et al. therefore hypothesized that central factors (cardiovascular and/or pulmonary changes) could also become limiting during (Howlett et al., 2009). In our mice, divergently selected on BMR, the probable role of lungs as the central limitation in oxygen delivery is supported by the lack of differences between elicited by cold, noradrenaline injection and exercise (Książek et al., 2004; Gębczynski, 2008; Gębczynski and Konarzewski, 2009a). Pulmonary limits are particularly well documented in thoroughbred racehorses, in which maximal exercise results in hypercapnia, hypoxemia and, sometimes fatal, exercise-induced pulmonary haemorrhaging (West and Mathieu-Costello, 1995). However, studies on humans indicate that their ventilatory capacity is largely excessive and fully exploited only in the most elite endurance athletes (Lindstedt and Conley, 2001). Likewise, results of interspecific comparisons led Weibel and coworkers (Weibel et al., 1991) to the conclusion that the capacity of mammalian lung for oxygen uptake exceeds the normal demand and that they have a ‘reserve capacity’. The excess capacity of lungs suggested by interspecific comparisons is difficult to reconcile with our results on mice divergently selected for BMR (and the above mentioned studies on rats), which indicate that limitations on ventilator capacity are context dependent.

Another possible limiting step of the oxygen delivery cascade is cardiac output (Mitchell et al., 1958; Liguzinski and Korzeniewski, 2007). Heart size/mass has often been used as a predictor of cardiac output and ultimately (Bishop, 1997; Richardson et al., 1999a; Chen et al., 2001; Hussain et al., 2001). All else being equal, heart mass should be closely correlated with stroke volume (Karas et al., 1987; Bishop, 1997) and therefore, can be considered as a significant predictor of the heart's pumping capacity. Moreover, the highly significant positive correlation between of mice selected for high voluntary activity and ventricle mass at different oxygen concentrations (Rezende et al., 2006b) suggest that cardiac output is indeed an important factor in determining individual differences in .

However, the possibility of cardiac output acting as a limiting step does not square with our earlier studies of divergent selection on BMR that resulted in a significantly increased heart mass in the H-BMR line type (Książek et al., 2004; Gębczynski, 2008; Gębczynski and Konarzewski, 2009b), but this did not provide any compensation compared with the L-BMR line type and RB lines under hypoxic conditions (Fig. 1). To put this finding in an even broader context, we plotted literature data on heart masses vs body masses of mice subjected to different selection regimens (Fig. 4), which confirms an exceptionally high heart mass of H-BMR mice. The question then rises of why selection on high BMR resulted in increased cardiac output, which after all, may indicate central limitation on . We suggest that the increased heart mass of H-BMR mice is part of an indirect response to selection resulting in enlarged metabolically active internal organs (liver, small intestine and kidneys in addition to the heart), whose metabolism accounts for a significant proportion of BMR (Konarzewski and Diamond, 1995; Książek et al., 2004; Gębczynski and Konarzewski, 2009a). However, an increase of mass of those organs did not result in a correlated increase in elicited by cold or exercise (Gębczynski and Konarzewski, 2009a). Thus, it is unsurprising that in this study we did not observe any compensation in of H-BMR mice in hypoxia.

Selection for high also resulted in an increased heart mass, albeit not that conspicuous as in the case of H-BMR mice (Fig. 4) (Gębczynski, 2008; Gębczynski and Konarzewski, 2009b). Larger cardiac output delivered by the heavier hearts of mice selected on is in agreement with their high , but does not necessarily result in the compensation of in hypoxia (Fig. 1) if the latter is limited peripherally in this line type. However, the lack of compensation of of H-BMR mice in hypoxia may be due to pulmonary limits suggested by comparable elicited by cold, noradrenaline injection and (Książek et al., 2004; Gębczynski, 2008; Gębczynski and Konarzewski, 2009a).

Apart from the pumping capacity of the heart, the key step in oxygen transport is convection through blood, which depends on both cardiac output and blood oxygen capacity. Recently, Kolb et al. showed that administration of an erythropoietin (EPO) analogue significantly increased by ∼5% in mice selected on voluntary wheel running, with no dose × line type interaction (Kolb et al., 2010). This increase was presumably the result of higher blood haemoglobin content after EPO treatment. It is important to note in this context, that the line type selected on high BMR studied here is characterized by 30% smaller RBCs than low BMR mice. Nevertheless, both line types have a similar haematocrit (E. Bonda-Ostaszewska and S. Maciak, personal communication). Smaller and more numerous RBC should result in higher erythrocyte total surface area, which suggests more effective O₂ blood convection in H-BMR mice. Despite this, of H- and L-BMR line types did not differ. In contrast, of mice selected on high is consistently higher than that of RB mice (Fig. 1), despite the lack of between-line-type differences in RBC and haemoglobin concentration. Clearly, variation of blood parameters in mice of our two selections is not consistently associated with differences in .

If responses to artificial selection on traits related to are symmorphotic, one would expect effects of P_O2 on to be independent of selection regimens, and differences between line types would remain constant across different air compositions. This is because according to the principle of symmorphosis, all traits related to the should evolve in concert, so that none of them would become a limiting (or excessive) step in oxygen delivery and utilization cascades. In the case of our study, this would manifest itself as the lack of line type × P_O2 interaction and was rejected in a ‘common’ model. Overall, our results suggest that the effect of P_O2 on was selection-history-dependent only for hyperoxia (Fig. 1), as in mice selected for high voluntary wheel running (Rezende et al., 2005). Interestingly, unlike in our selection on , in which hyperoxia elicited increased in RB lines (Fig. 1), elicited by normoxia and hyperoxia in the RB lines of Rezende et al.'s (Rezende et al., 2005) study did not differ. These contrasting responses of RB lines are also not in agreement with the notion of common symmorphotic design, as one would expect that increased P_O2 would not affect if cellular oxidative capacity was symmorphotic with normoxic delivery rates.

In summary, the apparently ‘symmorphotic’ consistent response of mice of all three selections to hypoxia and normoxia contrasts with their responses to hyperoxia and higher elicited by cold or after administration of EPO (Kolb et al., 2010). Thus, even though whole-organism must be physiologically constrained, the limitations are clearly context dependent (and probably species dependent as well), which makes the experimental search for generalizations about a possible symmorphotic nature of oxygen transport mechanisms a questionable effort.

 
• BMR

  basal metabolic rate

 
• H-BMR, L-BMR

  mice selected for high and low BMR, respectively

 
• He–O₂

  helium–oxygen atmosphere

 
• P_O2

  oxygen partial pressure

 
• 

  maximum oxygen consumption

 
• ,

  elicited by swimming or running, respectively

We are greatly indebted to B. Lewonczuk, M. Lewoc and S. Plonowski for their skilful technical assistance in carrying out our experiments. We would like to thank Mark Chappell, Paweł Koteja and an anonymous reviewer for valuable comments.