The adaptive benefit of evolved increases in hemoglobin-O₂ affinity is contingent on tissue O₂ diffusing capacity in high-altitude deer mice

Abstract

Background

Complex organismal traits are often the result of multiple interacting genes and sub-organismal phenotypes, but how these interactions shape the evolutionary trajectories of adaptive traits is poorly understood. We examined how functional interactions between cardiorespiratory traits contribute to adaptive increases in the capacity for aerobic thermogenesis (maximal O₂ consumption, V̇O₂max, during acute cold exposure) in high-altitude deer mice (Peromyscus maniculatus). We crossed highland and lowland deer mice to produce F₂ inter-population hybrids, which expressed genetically based variation in hemoglobin (Hb) O₂ affinity on a mixed genetic background. We then combined physiological experiments and mathematical modeling of the O₂ transport pathway to examine the links between cardiorespiratory traits and V̇O₂max.

Results

Physiological experiments revealed that increases in Hb-O₂ affinity of red blood cells improved blood oxygenation in hypoxia but were not associated with an enhancement in V̇O₂max. Sensitivity analyses performed using mathematical modeling showed that the influence of Hb-O₂ affinity on V̇O₂max in hypoxia was contingent on the capacity for O₂ diffusion in active tissues.

Conclusions

These results suggest that increases in Hb-O₂ affinity would only have adaptive value in hypoxic conditions if concurrent with or preceded by increases in tissue O₂ diffusing capacity. In high-altitude deer mice, the adaptive benefit of increasing Hb-O₂ affinity is contingent on the capacity to extract O₂ from the blood, which helps resolve controversies about the general role of hemoglobin function in hypoxia tolerance.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-021-01059-4.

Background

A long-standing goal of evolutionary biology is to understand how the functional integration of traits influences patterns of phenotypic change and adaptation [1]. Complex physiological phenotypes often represent an emergent property of functional interactions among different tissues and organ systems, which in turn may be developmentally interrelated and genetically correlated. The functional, developmental, and genetic interdependence of traits may facilitate environmental adaptation if semi-autonomous components of a complex phenotype respond synergistically to selection. Alternatively, functional integration and genetic correlations among components of a trait can limit and channel pathways of phenotypic evolution [2, 3]. Evolutionary questions about phenotypic integration and adaptation can be addressed most profitably by examining well-defined traits with well-characterized functions and well-documented associations with fitness under natural conditions.

The capacity for aerobic thermogenesis in small mammals at high altitude is a complex performance trait that is well suited to experimental studies of how patterns of phenotypic integration affect the process of adaptation. At high altitude, cold temperatures challenge the ability of endotherms to maintain body temperature and activity, which is especially difficult in smaller animals that have high surface area to volume ratios. Unsurprisingly, aerobic thermogenesis (quantified as maximal O₂ consumption, V̇O₂max, during acute cold exposure) in hypoxia is under strong directional selection in some small mammals at high altitude [4], which have evolved higher thermogenic V̇O₂max [5–9]. Thermogenic V̇O₂max is supported by the integrated function of the O₂ transport pathway, the conceptual steps (ventilation, pulmonary diffusion, circulation, tissue diffusion, and mitochondrial O₂ utilization) involved in transporting O₂ from inspired air to thermogenic tissues where O₂ is used by mitochondria to support oxidative phosphorylation [10, 11]. Therefore, studies of thermogenic V̇O₂max in high-altitude natives are ideal for understanding the mechanisms underlying the adaptive evolution of complex traits.

Evolved increases in the O₂ affinity of hemoglobin (Hb) are pervasive in high-altitude taxa and have become classic examples of biochemical adaptation [12]. However, the nature of the direct adaptive benefit conferred by increases in Hb-O₂ affinity in highland species is controversial. Many highland taxa have evolved increases in Hb-O₂ affinity independently, and in many cases, the molecular mechanisms underlying these changes in protein function are documented in detail [12–18]. These increases in Hb-O₂ affinity are often presumed to safeguard arterial O₂ saturation in hypoxia and thus help improve tissue O₂ delivery and aerobic capacity [5, 6, 19–26], although this has rarely been tested. Nonetheless, the relationship between Hb-O₂ affinity and V̇O₂max in hypoxia remains contentious [12, 25, 27–29]. Theoretical modeling of the O₂ transport pathway in humans suggests that increases in Hb-O₂ affinity do not increase aerobic capacity in hypoxia on their own [30], because the advantage of increasing Hb-O₂ affinity may be offset by a trade-off in O₂ offloading at tissues [11, 31, 32]. A recent study in humans with rare genetic Hb variants found that increases in Hb-O₂ affinity attenuated the hypoxia-induced decline in aerobic capacity, but subjects with high Hb-O₂ affinity also had compensatory polycythemia [33]. Considering the strong functional integration of Hb within the O₂ transport pathway, the advantages of increasing Hb-O₂ affinity in high-altitude taxa may be contingent on the evolution of other cardiorespiratory traits, but this has not been experimentally investigated.

We sought to determine the effects of evolved increases in Hb-O₂ affinity in high-altitude deer mice (Peromyscus maniculatus) on thermogenic V̇O₂max in hypoxia and to examine whether the adaptive benefit of changes in Hb-O₂ affinity is contingent on other cardiorespiratory changes. Deer mice have the broadest altitudinal range of any North American mammal [34], ranging from near sea level to montane environments up to approx. 4350 m above sea level [35], and high-altitude populations have evolved elevated thermogenic V̇O₂max in hypoxia in response to directional selection [4–9]. In conjunction with a higher V̇O₂max in chronic hypoxia, high-altitude deer mice also exhibit higher pulmonary O₂ extraction, arterial O₂ saturation, cardiac output, and tissue O₂ extraction than their lowland counterparts [9, 36]. The latter is associated with several evolved changes in skeletal muscle phenotype and mitochondrial function [37–41]. Highlanders have also evolved a higher Hb-O₂ affinity as a result of amino acid replacements in duplicated genes that encode the α- and β-chain subunits of the α₂β₂ Hb tetramer [5, 6, 15, 20, 21, 23, 24, 26, 34, 42]. This evolved increase in Hb-O₂ affinity in highlanders is not complemented by an enhanced Bohr effect to augment O₂ unloading [43]. We initially hypothesized that these evolved increases in Hb-O₂ affinity would be responsible for higher thermogenic capacity in highland deer mice, compared to their lowland conspecifics. To investigate the effect of genetically based changes in Hb-O₂ affinity on whole-animal performance in hypoxia, we created F₂ hybrids between high- and low-altitude deer mice (F₂ intercross breeding design) to randomize associations between allelic globin variants, and we then examined the effects of α- and β-globin variants on red blood cell P₅₀ (the O₂ pressure, PO₂, at which Hb is 50% saturated), arterial O₂ saturation, thermogenic V̇O₂max, and other physiological traits on an admixed genetic background (see Additional file 1: Fig. S1 for a graphical overview of the experimental design). We performed physiological measurements before and after chronic exposure to hypoxia to test for the effects of Hb genotype on trait-specific acclimation responses. We then used our empirical data in an in silico model of the O₂ transport pathway to examine the interactive effects of Hb-O₂ affinity and the O₂ diffusing capacity of tissues (D_TO₂) on V̇O₂max. Our results suggest that increases in Hb-O₂ affinity only contribute to the adaptive enhancement of thermogenic V̇O₂max in hypoxia if accompanied by a corresponding increase in D_TO₂ to augment tissue O₂ extraction.

Results

We measured thermogenic V̇O₂max, arterial O₂ saturation, and other cardiorespiratory traits in vivo during acute exposure to cold heliox in normoxia (21% O₂) and hypoxia (12% O₂) in male and female F₂ hybrid mice (mean ± SEM of body mass before hypoxia acclimation, 23.8 ± 0.9 g; see Additional file 2: Fig. S2 for body masses for each genotype before and after hypoxia acclimation) that possessed a diverse array of different α- and β-globin genotypes. We also performed in vitro measurements of red blood cell P₅₀ using erythrocyte suspensions from the same set of mice. The F₂ hybrids were generated by crossing wild mice from populations at high and low altitudes to produce F₁ inter-population hybrids, followed by full-sibling matings to create 4 families of F₂ hybrid progeny with admixed genetic backgrounds. Measurements of physiological phenotypes were made before and after a 6-week acclimation period to hypobaric hypoxia (12 kPa O₂, simulating ~ 4,300 m above sea level). In general, hypoxia acclimation was associated with increased V̇O₂max in hypoxia, along with increases in pulmonary ventilation, arterial O₂ saturation, heart rate, hematocrit (Hct), and blood Hb concentration ([Hb]), as well as increases in red blood cell P₅₀ (Fig. 1, Additional file 3: Tables S1 and S2). However, hypoxia acclimation did not affect V̇O₂max under normoxic conditions. Below, we describe the effects of Hb genotype on thermogenic V̇O₂max and hematological traits in mice acclimated to normoxia, and then we describe how Hb genotype affects acclimation responses to chronic hypoxia.

Fig. 1.

Acclimation responses to chronic hypoxia in all F₂ inter-population hybrid deer mice of all genotypes for measurements at V̇O₂max in normoxia (21 kPa O₂) and hypoxia (12 kPa O₂) and for measurements of hematology. Hct, hematocrit; [Hb], blood hemoglobin content; P₅₀, O₂ pressure at 50% saturation. *P < 0.05 between measurements in normoxia vs. hypoxia within an acclimation condition. ^†P < 0.05 vs. pre-acclimation value. Bars display mean ± SEM (n = 26) with individual data superimposed (circles)

Genetically based decreases in red blood cell P₅₀ improved arterial O₂ saturation in hypoxia

In normoxia-acclimated mice, there was a significant main effect of Hb genotype on red blood cell P₅₀ (P = 0.0048; Fig. 2a, Additional file 3: Table S3), which appeared to be largely attributable to the effects of α-globin variants. Mice possessing highland α-globin variants had a lower red blood cell P₅₀ compared to those possessing lowland variants, reflecting a higher affinity for O₂. In contrast, Hb genotype did not affect Hct P = 0.8339), [Hb] P = 0.9351), or the Hill coefficient (n) that quantifies the cooperativity of Hb-O₂ binding (P = 0.8053; Additional file 4: Fig. S3, Additional file 3: Table S3).

Fig. 2.

Variation in red blood cell O₂ affinity and arterial O₂ saturation associated with hemoglobin (Hb) genotype in F₂ inter-population hybrid deer mice acclimated to normoxia. a Red blood cell P₅₀ (O₂ pressure at 50% saturation). b Arterial O₂ saturation at V̇O₂max measured in normoxia (21 kPa O₂) and hypoxia (12 kPa O₂). Bars display mean ± SEM (n = 3-8) with individual data superimposed (circles). Different α- and β-globin genotypes are shown as superscripts with “^L” representing the lowland haplotype and “^H” representing the highland haplotype. *P < 0.05, hypoxia vs. normoxia within a genotype. P < 0.05 between genotypes for values not sharing a letter. c Linear regression of arterial O₂ saturation in hypoxia and red blood cell P₅₀ for individual data (P = 0.0103, R² = 0.2441; dotted line represents 95% confidence interval). Symbol colors reflect Hb genotype, as shown in a and b

Arterial O₂ saturation varied in association with red blood cell P₅₀ in hypoxia, but not in normoxia. There were significant main effects of both Hb genotype (P = 0.0189) and inspired PO₂ (P < 0.0001) on arterial O₂ saturation at V̇O₂max, with mice exhibiting reduced saturation in hypoxia (Fig. 2b, Additional file 3: Table S3). However, the effect of inspired PO₂ on arterial O₂ saturation was influenced by genotype (genotype x PO₂ interaction, P = 0.0389), as mice with the highland α-globin genotype exhibited a smaller reduction in arterial O₂ saturation under hypoxia compared to those with the lowland genotype. Consequently, mice with highland α-globin maintained 9–14% higher arterial O₂ saturation on average than those with lowland α-globin at hypoxic V̇O₂max. Higher red blood cell O₂ affinity was associated with higher arterial O₂ saturation in hypoxia, as indicated by a significant negative relationship between arterial O₂ saturation and red blood cell P₅₀ (P = 0.0103, R² = 0.2441; Fig. 2c).

Genetically based variation in red blood cell P₅₀ and arterial O₂ saturation had no effect on thermogenic V̇O₂max in hypoxia

V̇O₂max was significantly reduced in hypoxia compared to normoxia by ~ 24% on average (P < 0.0001; Fig. 3a, Additional file 3: Table S3). However, although Hb genotype had a significant main effect on V̇O₂max (P = 0.0416), V̇O₂max in hypoxia did not follow the pattern of variation seen for arterial O₂ saturation. As such, hypoxic V̇O₂max was not correlated with arterial O₂ saturation in hypoxia (Fig. 3b). Instead, the observed variation in V̇O₂max appeared to be associated with variation in heart rate, which was also significantly affected by inspired PO₂ (P < 0.0001), though the effect of genotype was only marginally significant (P = 0.0545; Additional file 5: Fig. S4A, Additional file 3: Table S3). Total ventilation, tidal volume, and breathing frequency were unaffected by the Hb genotype (Additional file 5: Fig. S4, Additional file 3: Table S3).

Fig. 3.

Variation in V̇O₂max was unrelated to variation in arterial O₂ saturation in F₂ inter-population hybrid deer mice acclimated to normoxia. a V̇O₂max measured in normoxia and hypoxia. See Fig. 2 for the details on hemoglobin (Hb) genotypes and symbols. b There was no correlation between hypoxic V̇O₂max and arterial O₂ saturation in hypoxia (P = 0.8103) across individuals (mean ± SEM for each genotype are shown as error bars)

Hb genotype influenced the acclimation responses of red blood cell P₅₀ and arterial O₂ saturation to chronic hypoxia

There were main effects of hypoxia acclimation that tended to increase both red blood cell P₅₀ (P = 0.0002) and arterial O₂ saturation measured at V̇O₂max in hypoxia (P = 0.0005), but the acclimation response appeared to differ between genotypes (Fig. 4, Additional file 3: Table S2). Mice with the lowland α-globin variant exhibited no plasticity in red blood cell P₅₀ in response to hypoxia acclimation, whereas mice with highland α-globin increased red blood cell P₅₀ to values that were comparable to mice with lowland α-globin. Conversely, mice with lowland α-globin showed much greater plasticity in arterial O₂ saturation in hypoxia following hypoxia acclimation, with all individuals increasing saturation (on average by ~ 13% saturation units). Mice with highland α-globin showed little to no change in saturation after hypoxia acclimation. Hypoxia acclimation increased Hct (P < 0.0001) and [Hb] (P < 0.0001; Fig. 1), but neither these traits nor the Hill coefficient was influenced by the Hb genotype (Additional file 4: Fig. S3, Additional file 3: Table S2).

Fig. 4.

The effects of hypoxia acclimation on red blood cell O₂ affinity and arterial O₂ saturation differed between hemoglobin (Hb) genotypes in F₂ inter-population hybrid deer mice, but the effects of hypoxia acclimation on V̇O₂max did not. a Red blood cell P₅₀ (O₂ pressure at which Hb is 50% saturated) and b arterial O₂ saturation at V̇O₂max in hypoxia, measured before and after a 6-week acclimation to hypobaric hypoxia (12 kPa O₂). ^†P < 0.05 vs. pre-acclimation value within a genotype. P < 0.05 between genotypes within an acclimation condition for values not sharing a letter. c The change in hypoxic V̇O₂max plotted against the change in arterial O₂ saturation in hypoxia in individuals in response to hypoxia acclimation (mean ± SEM for each genotype are shown as error bars). See Fig. 2 for other details on Hb genotypes

V̇O₂max in hypoxia increased after hypoxia acclimation (P = 0.0013), but this response was not influenced by Hb genotype (P = 0.1764; Additional file 6: Fig. S5, Additional file 3: Table S2). The magnitude of change in hypoxic V̇O₂max following hypoxia acclimation was not associated with the magnitude of change in arterial O₂ saturation (Fig. 4c). Hypoxia acclimation also increased heart rate (P = 0.0031), total ventilation, (P < 0.0001), tidal volume (P = 0.0005), and breathing frequency (P < 0.0001) measured at V̇O₂max in hypoxia, but none of these traits was affected by Hb genotype (Additional file 7: Fig. S6, Additional file 3: Table S2). Normoxic V̇O₂max was not affected by hypoxia acclimation or Hb genotype, nor were the measurements of heart rate, total ventilation, tidal volume, or breathing frequency at normoxic V̇O₂max affected by Hb genotype (Additional file 8: Fig. S7, Additional file 3: Table S4). However, there was a main effect of genotype on arterial O₂ saturation measured at V̇O₂max in normoxia (P = 0.0291) that appeared to result from slightly lower saturation values in mice with characteristic lowland α- and β-globin genotypes (α^LLβ^LH; Additional file 8: Fig. S7B, Additional file 3: Table S4).

Sensitivity analysis suggested that effects of Hb-O₂ affinity on V̇O₂max in hypoxia are contingent on tissue O₂ diffusing capacity (D_TO₂)

We examined the interactive effects of Hb-O₂ affinity and D_TO₂ on V̇O₂max in hypoxia using a mathematical model of O₂ flux through the O₂ transport pathway. We generated the initial solutions of the model using empirical data collected for deer mice and then performed a sensitivity analysis to determine the effects of increasing D_TO₂ on V̇O₂max at each of the red blood cell P₅₀ values for mice with characteristic highland (α^HHβ^HH) and lowland (α^LLβ^LH) Hb genotypes. Increasing D_TO₂ by 50% increased V̇O₂max, but the effect was greater with the P₅₀ of the high-affinity α^HHβ^HH genotype (11.8%) than with the lower affinity α^LLβ^LH genotype (8.5%; Fig. 5a). The effect of P₅₀ was accentuated when D_TO₂ was increased above 41%, when venous PO₂ (and thus venous O₂ saturation) fell to zero at the higher P₅₀ (Fig. 5b). These results indicate that an increase in Hb-O₂ affinity only contributes to an enhancement of V̇O₂max in hypoxia if it is paired with an increase in D_TO₂ in thermogenic tissues (i.e., skeletal muscle and/or brown adipose tissue).

Fig. 5.

Effects of increasing tissue O₂ diffusing capacity (D_TO₂) on hypoxic V̇O₂max and O₂ partial pressures (PO₂) using mathematical modeling of the O₂ transport pathway. a Relative changes in hypoxic V̇O₂max and b changes in alveolar (P_AO₂), arterial (P_aO₂), and venous (P_vO₂) PO₂ in response to relative increases in D_TO₂. The effects were modeled using the mean (bold lines) ± SEM (fine lines) values of red blood cell P₅₀ for mice with hemoglobin genotypes that were most characteristic of lowlanders (α^LLβ^LH) and highlanders (α^HHβ^HH)

Discussion

Our study provides evidence that the adaptive benefit of increasing Hb-O₂ affinity is contingent on the capacity of active tissues to extract O₂ from the blood. In agreement with previous studies [5, 6, 20, 21, 24, 26, 43], our data from F₂ inter-population hybrids demonstrate that Hb variants from high-altitude deer mice confer a higher Hb-O₂ affinity than Hb from lowland conspecifics and that this evolved increase in affinity augments arterial O₂ saturation in hypoxia by 9–14%. However, these genetically based changes alone did not augment V̇O₂max (i.e., aerobic performance) in hypoxia. Modeling of the O₂ transport pathway revealed that increases in Hb-O₂ affinity would only be expected to enhance V̇O₂max in hypoxia if O₂ diffusing capacity was increased to augment tissue O₂ extraction. Importantly, recent evidence suggests that high-altitude mice have evolved a highly aerobic skeletal muscle phenotype with an enhanced capacity for O₂ diffusion [37–41]. In particular, the gastrocnemius muscle of highland deer mice has greater capillary density and a redistribution of mitochondria to a subsarcolemmal location that is closer to capillaries, each of which would increase O₂ diffusing capacity. Our results therefore suggest that increases in both Hb-O₂ affinity and tissue O₂ diffusing capacity likely contributed to the adaptive increases in V̇O₂max in high-altitude deer mice. These findings suggest the testable hypothesis that other hypoxia-adapted, high-altitude vertebrates that have evolved derived increases in Hb-O₂ affinity will also have evolved increases in tissue capillarity and/or other changes that augment O₂ diffusing capacity.

The genetically based differences in Hb function led to predictable differences in arterial O₂ saturation during acute and chronic hypoxia. Amino acid variation in Hb genes is not always associated with changes in O₂-binding properties [16, 44], and even in cases where it has been possible to document causal effects of specific mutations on Hb function [13, 15, 23–26, 34, 42, 45–48], the in vivo effects on blood oxygenation have rarely been examined. Our study suggests that it is critically important to examine how genetic changes in proximal biochemical phenotypes affect higher-level physiological phenotypes (e.g., arterial O₂ saturation and V̇O₂max in hypoxia) to fully understand their potential adaptive significance.

Genetic variation in Hb altered the acclimation response to chronic hypoxia, as highland α-globin genotypes were associated with increased plasticity in Hb-O₂ affinity of red blood cells. This variation was likely a result of differences in sensitivity to 2,3-diphosphoglycerate (2,3-DPG), an allosteric modulator of Hb-O₂ affinity. Concentrations of 2,3-DPG in erythrocytes are known to increase in response to chronic hypoxia, which tends to reduce red blood cell Hb-O₂ affinity [49–53]. Previous studies have shown that Hb from high-altitude deer mice is more sensitive to 2,3-DPG in the presence of Cl^- than Hb from low-altitude mice [24, 26]. Therefore, in the current study, if red blood cell concentrations of 2,3-DPG were comparable across genotypes, differences in 2,3-DPG sensitivity could explain the differences in plasticity of red blood cell Hb-O₂ affinity. This mechanism may also explain why genotypes differed in the magnitude of plasticity in arterial O₂ saturation in response to chronic hypoxia. Several physiological adjustments contribute to increasing arterial O₂ saturation after hypoxia acclimation, including increases in total ventilation (Fig. 1) and adjustments in lung function to augment pulmonary O₂ diffusion [36, 54], and these effects could potentially be counteracted by reductions in red blood cell Hb-O₂ affinity. Such reductions in affinity did not occur in mice with lowland α-globin, such that they experienced greater improvements in arterial O₂ saturation after hypoxia acclimation.

Our results indicate that the adaptive benefit of increasing Hb-O₂ affinity is contingent on the O₂ diffusing capacity of active tissues. Our study provides empirical evidence that genetically based increases in Hb-O₂ affinity and arterial O₂ saturation alone are not sufficient to improve aerobic capacity in hypoxia. We also demonstrate that the adaptive benefit of increasing Hb-O₂ affinity is contingent on having a tissue O₂ conductance (D_TO₂) that is sufficiently high to take advantage of the greater arterial O₂ saturation and extract more O₂ from the blood. The relationship between Hb-O₂ affinity and V̇O₂max in hypoxia is a contentious topic [12, 25, 28], with different empirical studies and theoretical models providing contradictory results [5, 27, 29, 30, 33, 55]. In fact, previous investigation in deer mice has shown that mice possessing highland α-globin alleles with higher Hb-O₂ affinity did have higher V̇O₂max in hypoxia than mice with lowland α-globin haplotypes [5]. However, in this previous study (in which genotyping was based on protein electrophoresis), different α-globin alleles were backcrossed into a highland genetic background [5], unlike the current study in which alternative allelic variants were randomized against an admixed highland/lowland background. As discussed above, highland deer mice appear to have evolved a higher capacity for O₂ diffusion and utilization in the skeletal muscles than their lowland conspecifics, comparable to some differences between high-altitude and low-altitude human populations [56]. It is therefore possible that the highland mice used in this previous study [5] had a higher D_TO₂ than the F₂ inter-population hybrids used in our present study, which would explain the observed differences in the relative influence of Hb genotype on V̇O₂max. Indeed, our modeling shows that the adaptive benefits of increasing Hb-O₂ affinity are critically dependent on D_TO₂. Together, our findings suggest adaptive increases in V̇O₂max in high-altitude deer mice may have been facilitated by evolved increases in D_TO₂, which were required in order for increases in Hb-O₂ affinity to confer an adaptive benefit at high-altitude.

Conclusions

Complex organismal traits are often the result of multiple interacting genes and phenotypes, but the role of these interactions in shaping adaptive traits is poorly understood. Our findings demonstrate that adaptive increases in thermogenic capacity result from a functional interaction between blood hemoglobin and active tissues, in which the adaptive benefit of increasing hemoglobin O₂ affinity is contingent on the capacity for O₂ diffusion from the blood. This helps reconcile controversy about the general role of hemoglobin in hypoxia tolerance and provides insight into the physiological mechanisms of high-altitude adaptation.

Methods

Animals

Wild deer mice (Peromyscus maniculatus) were live-trapped at high altitude on the summit of Mount Evans (Clear Creed County, CO, USA, at 39° 35′ 18″ N, 105° 38′ 38″ W; 4350 m above sea level) and at low altitude on the Great Plains (Nine Mile Prairie, Lancaster County, NE, USA, at 40° 52′ 12″ N, 96° 48′ 20.3″ W; 430 m above sea level), and were transported to the University of Montana (elevation 978 m). The wild mice were used to produce one family of first-generation inter-population hybrids (F₁), created by crossing a highland male and a lowland female. These F₁ hybrids were raised to maturity and used for full-sibling matings to produce 4 families of male and female second-generation hybrid progeny (F₂). These F₂ hybrids (n = 26) were raised to adulthood (1–1.5 years old), a small volume of blood was obtained for genotyping (sampled from the facial vein and then stored at −80 °C), and mice were then transported to McMaster University (near sea level) for subsequent experiments (see below). Prior to experimentation, all mice were kept in standard holding conditions (24–25 °C, 12:12-h light-dark photoperiod) under normal atmospheric conditions, with unlimited access to water and standard mouse chow. All animal protocols were approved by the institutional animal research ethics boards.

Each mouse was genotyped for the determination of α- and β-globin haplotypes. Tetrameric hemoglobin isoforms of adult P. maniculatus incorporate α-chain subunits that are encoded by two tandem gene duplicates, HBA-T1 and HBA-T2 (separated by 5.0 kb on chromosome 8), and β-chain subunits that are encoded by two other tandem duplicates, HBB-T1 and HBB-T2 (separated by 16.2 kb on Chromosome 1) [34, 57, 58]. A reverse-transcriptase PCR (RT-PCR) approach was used to obtain sequence data for all four of the adult-expressed α- and β-globin transcripts [26, 34]. Total RNA was extracted from red blood cells using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA, USA). Globin transcripts were then amplified from 1 μg of extracted RNA using the One-Step RT-PCR system with Platinum Taq DNA polymerase High Fidelity (Invitrogen, Carlsbad, CA, USA). PCR cycling was performed with 1 cycle at 50 °C for 30 min; 1 cycle at 95 °C for 15 min, 34 cycles at 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 1 min; and then a final extension cycle at 72 °C for 3 min. For the α-globin transcripts, the same primer pair was used for HBA-T1 and HBA-T2 (forward: CTGATTCTCACAGACTCAGGAAG, reverse: CCAAGAGGTACAGGTGCGAG). For the β-globin transcripts, the same RT-PCR primer pair was used for HBB-T1 and HBB-T2 (forward: GACTTGCAACCTCAGAAACAGAC, reverse: GACCAAAGGCCTTCATCATTT). Gel-purified RT-PCR products were then cloned into pCR4-TOPO vector using the TOPO TA cloning kit (Invitrogen), and automated DNA sequencing of cloned PCR products was performed using Big Dye chemistry (ABI 3730 capillary sequencer; Applied Biosystems, Foster City, CA, USA). For each mouse, we sequenced 6 clones containing products of HBA-specific RT-PCR and 6 clones containing products of HBB-specific RT-PCR. Thus, full-length inserts representing cDNAs of all expressed HBA and HBB genes were sequenced at 6-fold coverage, and the haplotype phase of all variable sites was determined experimentally. We thus identified 5 distinct combinations of highland (H)- and lowland (L)-associated α- and β-globin haplotypes: n = 5 α^LLβ^LH, 2 males, 3 females; n = 3 α^LLβ^HH, 1 male, 2 females; n = 8 α^HHβ^LL, 5 males, 3 females; n = 5 α^HHβ^LH, 5 males; and n = 5 α^HHβ^HH, 5 males.

Acclimation treatments

Physiological measurements (see below) were taken for each mouse both before (mean ± SEM body mass, 23.8 ± 0.9 g) and after (23.7 ± 0.8 g) a 6-week acclimation to hypobaric hypoxia (approx. 12 kPa PO₂), approximating the O₂ levels experienced by highland deer mice living at 4350 m above sea level in the wild (see Additional file 2: Fig. S2 for body masses for each genotype before and after hypoxia acclimation). This was achieved by placing mice into custom-made hypobaric chambers inside which barometric pressure was maintained at 60 kPa using a vacuum pump, as previously described [37, 59]. Mice were removed from the chambers for < 20 min twice per week for cage cleaning.

Respirometry, plethysmography, and pulse oximetry

We combined open-flow respirometry, plethysmography, and pulse oximetry to simultaneously measure aerobic capacity for thermogenesis (thermogenic V̇O₂max), pulmonary ventilation, arterial O₂ saturation, and heart rate during acute cold (−5 °C) exposure in heliox [9, 36]. We used established methods of measuring thermogenic V̇O₂max that have been shown to elicit values of V̇O₂max that equal or exceed those measured during exercise V̇O₂max in deer mice [22, 60, 61]. These measurements were performed twice before mice were acclimated to hypoxia: once in normoxic heliox (21% O₂, 79% He) and once in hypoxic heliox (12% O₂, 88% He) in random order. After hypoxia acclimation, both normoxic and hypoxic V̇O₂max trials were repeated. V̇O₂max was measured inside a 530-ml plethysmography chamber that has been previously described in detail [62] and was kept in a regulated freezer to maintain the internal chamber temperature at −5 °C (measured with a PT-6 thermocouple, Physitemp). Two days prior to initial trials, the neck fur of each mouse was removed with Nair™ hair-removal product to facilitate pulse oximetry. Immediately before each trial, each mouse was weighed, fitted with a MouseOx pulse oximetry neck collar (Starr Life Sciences, Oakmont, PA, USA), and placed inside the chamber for 10 min of continuous recording. Incurrent gas flowed through the animal chamber at a rate of 1500 ml min⁻¹ (controlled by an MFC-2 mass flow controller, Sable Systems, Las Vegas, NV, USA) and was cooled before entering the plethysmograph by passing through copper coils that were also placed in the freezer. Excurrent gas was subsampled at a rate of 200 ml min⁻¹ and dried with pre-baked Drierite before passing through O₂ and CO₂ analyzers to determine the fractional concentrations of each gas (FoxBox Respirometry System, Sable Systems). Breathing-induced changes in flow across a pneumotachograph in the chamber lid were measured from pressure oscillations in the animal chamber relative to an identical reference chamber using a differential pressure transducer (Validyne DP45; Cancoppas, Mississauga, ON, Canada) and carrier demodulator (Validyne CD15, Cancoppas), and signals were volume-calibrated before each trial with 300-μl injections using a gas-tight syringe. The core body temperature (T_b) of each mouse was obtained (RET-3-ISO; Physitemp, Clifton, NJ, USA) immediately after being removed from the plethysmograph, and then again at room temperature exactly 24 h afterwards, allowing us to estimate T_b at V̇O₂max for use in tidal volume calculations (see below) by assuming that T_b dropped linearly throughout the trial. All data were acquired and recorded using a PowerLab 8/32 and LabChart 8 Pro Software (ADInstruments, Colorado Springs, CO, USA), with the exception of pulse oximetry data, which were obtained using Starr Life Sciences acquisition hardware and software.

Breathing, heart rate, and arterial O₂ saturation were recorded at thermogenic V̇O₂max for each trial. O₂ consumption rate (V̇O₂) was calculated from gas concentration and flow measurements using established equations [63]. V̇O₂max was defined as the maximal V̇O₂ measured over a 10-s period during the 10-min trial. V̇O₂ usually increased to V̇O₂max within 5–6 min of entering the chamber and would then decline to less than 90% V̇O₂max, and all mice had depressed T_b by the end of each 10-min cold exposure. Tidal volume was calculated (in volumes at BTPS) using established equations for the barometric method in flow-through conditions [64, 65]. Total ventilation was calculated as the product of tidal volume and breathing frequency.

Hematology

Hematology was measured both before and after hypoxia acclimation. Blood samples were taken from the facial vein 3 days after V̇O₂max measurements. We measured Hb content using Drabkin’s reagent (according to the instructions from the manufacturer, Sigma-Aldrich) and hematocrit by spinning the blood in capillary tubes at 12,700g for 5 min. The O₂ affinity of intact erythrocytes was measured using 10 μl blood in 5 ml buffer containing 0.1 M Hepes, 0.05 M EDTA, 0.1 M NaCl, 0.1% bovine serum albumin, and 0.2% antifoaming agent at pH 7.4. Oxygen dissociation curves were generated at 37 °C using a Hemox Analyzer (TCS Scientific), and red blood cell P₅₀ and Hill coefficient (n) were calculated using the Hemox Analytic Software.

Statistics

We used linear mixed effects models to test for the effects of Hb genotype and acclimation condition using the lme4 [66] package in R (v.3.1.3, R Core Team, 2013). We carried out one set of models to examine the fixed effects of Hb genotype in normoxia-acclimated mice, in the absence of the effects of hypoxia acclimation and with inspired PO₂ as an additional fixed effect. We then carried out a second set of models including data from both before and after chronic hypoxia exposure to examine the effects of Hb genotype, hypoxia acclimation, and their interaction. We used a backwards model selection approach, in which initial models included sex, family, and individual subject as random factors, as well as body mass as a covariate. If these terms had P values above 0.1, they were removed by stepwise backward deletion (starting with the term with the highest P value), and the model was re-run until all terms in the model (with the exception of fixed factors and individual subject) had P values below 0.1. Family was thus included in only 6 of the models (see Additional file 3: Tables S1-S4), while the effects of sex were never significant and were removed from all models. Tukey’s HSD post hoc tests were performed to test for pairwise differences between genotypes within an acclimation/PO₂ treatment and between acclimation/PO₂ treatment groups within each genotype. Data are presented as individual values and as mean ± SEM, unless otherwise stated.

Modeling the O₂ transport pathway

Mathematical modeling of the O₂ transport pathway of deer mice was used to determine the interactive effects of blood-O₂ affinity and tissue O₂ diffusing capacity on V̇O₂max in hypoxia. This was done using established equations that have been used previously to build similar models [30, 67–71]. The Fick equation describes the diffusion of oxygen from the alveoli into the blood along capillaries in the lung:

$$\frac{d{\left[O_2\right]}_L}{\mathit{dt}}=\frac{D_L{O}_2}{t_L\dot{Q}}.\left(P_A{O}_2-P_L{O}_2\right)$$ 1

where [O₂]_L and t_L are the instantaneous O₂ content and transit time of the lung capillaries, D_LO₂ is the physiological O₂ diffusing capacity of the lungs, Q̇ is cardiac output, and P_AO₂ and P_LO₂ are the PO₂ in the alveoli and instantaneously along lung capillaries, respectively. P_LO₂ began at mixed venous PO₂ (P_vO₂), and the equation was then integrated over the length on the lung capillaries to determine arterial PO₂ (P_aO₂) using the Hill equation (Eq. 2) to relate [O₂] and PO₂ in the blood (dependent on blood O₂ affinity and hemoglobin content):

$$\left[O_2\right]=4\left[\mathit{Hb}\right].\frac{P{O_2}^n}{P{O_2}^n+{P_{50}}^n}$$ 2

where [Hb] is the hemoglobin content of the blood, P₅₀ is the PO₂ at which blood is 50% saturated with O₂, and n is the Hill coefficient that describes the cooperativity of blood-O₂ binding. The Fick equation also describes O₂ diffusion from the blood in tissue capillaries to the mitochondria:

$$\frac{d{\left[O_2\right]}_T}{\mathit{dt}}=\frac{-D_T{O}_2}{t_T\dot{Q}}.\left(P_T{O}_2-P_M{O}_2\right)$$ 3

where [O₂]_T and t_T are the instantaneous O₂ content and transit time of the tissue capillaries, D_TO₂ is the O₂ diffusing capacity of the tissues, and P_TO₂ and P_MO₂ are the PO₂ instantaneously along the tissue capillaries and at mitochondria, respectively. P_MO₂ was set to zero to facilitate modeling, but mitochondrial PO₂ is likely quite low and relatively close to zero at V̇O₂max [69]. In this case, P_TO₂ begins at P_aO₂, and the equation is integrated along the tissue capillaries to determine P_vO₂. Mass conservation then matches V̇O₂max measured from O₂ extraction at the lungs to that at tissues:

$${\dot{V}}_A.\left(F_I{O}_2-F_A{O}_2\right)=\dot{Q}.\left({\left[O_2\right]}_a-{\left[O_2\right]}_v\right)$$ 4

where V̇_A is the alveolar ventilation, F_IO₂ and F_AO₂ and the O₂ fractions of inspired and alveolar gas, respectively, and [O₂]_a and [O₂]_v are the arterial and venous oxygen content, respectively.

The above equations were solved using an iterative approach for the key unknown outcome variables, P_AO₂, P_aO₂, and P_vO₂, from which V̇O₂max was calculated. This was achieved using a F_IO₂ of 0.123, our empirical measurements of total ventilation at hypoxic V̇O₂max, red blood cell P₅₀ and n, and [Hb] in normoxia-acclimated α^LLβ^LH mice, a body mass-specific lung dead space volume of 6.4 μl g⁻¹ [72] to calculate alveolar ventilation from total ventilation, and cardiac output at hypoxic V̇O₂max from our previous measurements in low-altitude deer mice [36]. Values for D_LO₂ and D_TO₂ were chosen by trial and error to re-produce in vivo measurements of P_aO₂, P_vO₂, and V̇O₂max. All of the initial values used to solve the model are listed in Additional file 3: Table S5. The model was solved iteratively as follows. Using previously recorded in vivo P_vO₂ [36] and an initial estimate of P_AO₂ as a starting point, Eq. 1 was integrated to calculate a predicted value of P_aO₂. This P_aO₂ value was then used to calculate P_vO₂ by integrating Eq. 3. The two above steps were repeated until the P_aO₂ and P_vO₂ values became stable to within 0.05% (< 10 iterations). V̇O₂max was then calculated using both the left and right sides of Eq. 4. If the values did not agree to within 0.05%, P_AO₂ was adjusted, and the above steps were repeated until the left and right sides of Eq. 4 were equal to within 0.05%. Reaching a stable solution of the model generally took less than 10 iterations, and the final outcome was independent of the starting estimate of P_vO₂.

We conducted a sensitivity analysis of the effects of increasing D_TO₂ on V̇O₂max using the mean for the most ancestral “lowland” P₅₀ that was measured in α^LLβ^LH mice, and then again using the mean for the “highland” P₅₀ that was measured in α^HHβ^HH mice, with all other parameters in the model kept constant (including the potential effects of variation in blood pH on P₅₀), with the exception of P_AO₂, P_aO₂, and P_vO₂ (which were under the influence of the changes in D_TO₂ and P₅₀). We also used the mean + SEM and mean − SEM values of P₅₀ for each genotype in order to examine the influence of biological variation in P₅₀ within each genotype. V̇O₂max and D_TO₂ values are expressed here relative to the initial solution generated using the data here from α^LLβ^LH mice or from previous measurements in lowland deer mice (Additional file 3: Table S5), which we have termed the “ancestral” values. The calculations were carried out using spreadsheet software (Microsoft Excel), as in some previous models of the O₂ transport pathway [71], and we have included the spreadsheet template with values for the initial lowland P₅₀ model (Additional file 9: Dataset and Modeling).

Authors’ contributions

GRS, JFS, and ZAC conceived and designed the study. OHW, CMI, JPV, NG-P, SCC-S, and CN acquired the data. OHW analyzed the data. OHW and GRS interpreted the data and drafted the manuscript. All authors read and approved the final manuscript.

Funding

This research was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant to G.R.S. (RGPIN-2018-05707), the National Science Foundation grants to Z.A.C. (IOS-1354934, IOS-1634219, IOS-1755411, and OIA-1736249) and J.F.S. (IOS-1354390 and OIA-1736249), and the National Institutes of Health (NIH) grant to J.F.S. (HL087216). Salary support was provided to O.H.W. by the NSERC Vanier Canada Graduate Scholarship; to C.M.I. by the NSERC Postgraduate Scholarship and an Ontario Graduate Scholarship; to J.P.V. by the NIH National Heart, Lung and Blood Institute Research Service Award Fellowship (1F32HL136124-01); to S.C.C.-S by the NSF (DBI) Postdoctoral Fellowship in Biology Award (#1612283); and to G.R.S. by the Canada Research Chairs Program.

Availability of data and materials

The dataset supporting the conclusions of this article is included within the article and its additional files.

Declarations

Ethics approval and consent to participate

All animal protocols were approved by institutional animal research ethics boards.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.