Genetic variation in haemoglobin is associated with evolved changes in breathing in high-altitude deer mice

High-altitude natives are an exceptional model for understanding the genetic and physiological bases of evolutionary adaptation. Species that are broadly distributed across altitudes can provide powerful insight into the genetic basis of high-altitude adaptation, because it is possible to examine segregating variation for phenotypes that may contribute to hypoxia tolerance. Recent research has identified many genes that appear to have experienced selection in high-altitude taxa, including genes thought to be involved in O₂ transport, energy metabolism and hypoxia signalling (Simonson, 2015; Simonson et al., 2012; Storz and Cheviron, 2021). However, in most cases, the specific effects of these genetic variants on physiological function are poorly understood. Identifying these functional effects has the potential to uncover novel and adaptive physiological mechanisms, given the growing appreciation that protein variants can have auxiliary effects that are unrelated to the ‘canonical’ function of the protein in question (Marden, 2013a).

Evolved changes in haemoglobin (Hb) have contributed to hypoxia adaptation in many high-altitude mammals and birds (Storz, 2016). Hb is a tetramer containing two α- and two β-chain subunits, and its O₂-binding affinity is an important determinant of O₂ exchange at the lungs and peripheral tissues. Evolved increases in Hb–O₂ affinity have arisen in many high-altitude taxa, and are typically attributable to amino acid replacements in the α- and/or β-chain subunits that increase intrinsic O₂-affinity and/or reduce responsiveness to negative allosteric cofactors (e.g. 2,3-DPG in mammals) (Galen et al., 2015; Jendroszek et al., 2018; Natarajan et al., 2015a, 2016, 2018; Projecto-Garcia et al., 2013; Signore et al., 2019; Storz et al., 2010; Tufts et al., 2015; Zhu et al., 2018). An increased Hb–O₂ affinity is generally assumed to safeguard arterial O₂ saturation (Sa_O2) in hypoxia, but it remains possible that modifications of Hb function contribute to hypoxia tolerance via other physiological mechanisms that are not directly related to circulatory O₂ transport.

Hb is not generally thought to be directly involved in the regulation of breathing, but some evidence suggests that it might have a role. On the one hand, pharmacological manipulations that increase Hb–O₂ binding affinity do not generally appear to exert much influence on ventilatory responses to hypoxia in guinea pigs and rats (Birchard and Tenney, 1986; Rivera-Ch et al., 1994), nor do some naturally occurring mutant haemoglobins that have altered O₂ affinity (e.g. Andrew–Minneapolis mutation in the β-globin subunit of humans) (Hebbel et al., 1977). By contrast, affinity-altering mutations in Hbs of mice and sheep are associated with changes in ventilatory sensitivity to O₂ and/or CO₂ (Dawson and Evans, 1966; Izumizaki et al., 2003; Shirasawa et al., 2003). Furthermore, deoxygenated Hb can produce S-nitrosothiols that contribute to signalling the ventilatory response to hypoxia in rats (Lipton et al., 2001). These findings together suggest that Hb or its constituent α/β globin monomers may play an underappreciated role in the control of breathing, by a mechanism that is not directly associated with the role of Hb in circulatory O₂ transport.

High-altitude deer mice have evolved an increased Hb–O₂ affinity relative to lowland conspecifics (Snyder et al., 1982; Storz et al., 2010), and this modification of protein function is associated with increased Sa_O2 in hypoxia (Ivy and Scott, 2017; Ivy et al., 2020; Tate et al., 2017; Wearing et al., 2021). Experimental studies have identified the mutations in the α- and β-chain subunits that are responsible for population differences in Hb–O₂ affinity (Jensen et al., 2016; Natarajan et al., 2013, 2015b; Storz et al., 2009, 2010) and population-genetic analyses on sequence variation in the underlying genes provided evidence that the altitudinal patterning of the Hb polymorphism is attributable to a history of spatially varying selection that favours different allelic variants in different elevational zones (Storz and Kelly, 2008; Storz et al., 2012). High-altitude deer mice have also experienced strong directional selection for increased aerobic capacity for thermogenesis (Hayes and O'Connor, 1999), which has led to evolved increases in maximal rates of O₂ consumption (V̇_O2,max) in hypoxia (Cheviron et al., 2013, 2014b; Tate et al., 2017, 2020). Studies of inter-population hybrids of deer mice from high and low altitudes, in which effects of individual genotypes and phenotypes can be assessed on a mixed genetic background, have been used to examine the contribution of Hb–O₂ affinity and other respiratory traits to this evolved increase in V̇_O2,max (Chappell and Snyder, 1984; Wearing et al., 2021). In addition, high-altitude deer mice have evolved an enhanced hypoxic ventilatory response and a deeper breathing pattern (larger tidal volumes but lower breathing frequencies at a given level of total ventilation) under routine conditions, both of which should increase alveolar ventilation and be more effective for gas exchange in hypoxia (Ivy and Scott, 2017, 2018; Ivy et al., 2020). However, the potential contribution of genetic variation in Hb to these evolved changes in control of breathing has yet to be examined.

Here, we report an investigation of whether high-altitude Hb variants contribute to the evolved changes in the control of breathing we have previously reported in high-altitude populations of deer mice (Ivy and Scott, 2017, 2018; Ivy et al., 2020). This was achieved using an F₂ intercross breeding design to isolate the effects of allelic variants of α- and β-globins against an admixed genetic background. We then examined whether the physiological effects of variation in Hb genes resulted from changes in Hb–O₂ affinity, using efaproxiral (a synthetic drug that acts as a negative allosteric regulator of Hb–O₂ binding) to acutely reduce Hb–O₂ affinity in vivo (Donnelly et al., 2006; Khandelwal et al., 1993).

Wild adult deer mice were live-trapped at high altitude on the summit of Mount Evans (Clear Creek County, CO, USA, at 39°35′18″N, 105°38′38″W; 4350 m above sea level) (Peromyscus maniculatus rufinus) 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) (Peromyscus maniculatus nebrascensis) and were transported to the University of Montana (elevation 978 m) or to McMaster University (elevation 50 m). The wild mice transported to Montana 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 were used for full-sibling matings to produce 4 families of second-generation hybrid progeny (F₂). These F₂ hybrids (N=26) were raised to adulthood, 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 for subsequent experiments (see below). Each F₂ hybrid was genotyped (described below) to determine the sequence of its α- and β-globin haplotypes, resulting in the 5 distinct combinations of highland and lowland haplotypes of α- and β-globin that were studied here: N=5 α^HHβ^HH, N=5 α^HHβ^HL, N=7 α^HHβ^LL, N=4 α^LLβ^HH and N=5 α^LLβ^HL. These F₂ hybrid mice were also used for a distinct study on aerobic capacity (Wearing et al., 2021). The wild mice transported to McMaster were bred in captivity to produce first-generation (G₁) progeny within each population. All mice were held in a standard holding environment (∼24–25°C, 12 h:12 h light:dark photoperiod) under normal atmospheric conditions before experiments, and were provided with unlimited access to water and standard mouse chow. All animal protocols were approved by institutional animal research ethics boards.

Adult isoforms of tetrameric Hb from 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) (Hoffmann et al., 2008; Natarajan et al., 2015b). We used a reverse-transcriptase PCR (RT-PCR) approach to obtain sequence data for all four of the adult-expressed α- and β-globin transcripts in the full panel of mice (Natarajan et al., 2015b; Storz et al., 2010). The RNeasy Plus Mini Kit (Qiagen, Valencia, CA, USA) was used to extract total RNA from red blood cells. We then amplified globin transcripts 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 conditions were as follows: 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, we used the same primer pair for HBA-T1 and HBA-T2 (forward: CTGATTCTCACAGACTCAGGAAG, reverse: CCAAGAGGTACAGGTGCGAG). For the β-globin transcripts, we used the same RT-PCR primer pair for HBB-T1 and HBB-T2 (forward: GACTTGCAACCTCAGAAACAGAC, reverse: GACCAAAGGCCTTCATCATTT). We cloned and sequenced the RT-PCR products using TOPO^® TA Cloning Kit (Life Technologies, Carlsbad, CA, USA), and we sequenced at least six clones per gene in order to recover both alleles from the paralogues. Full-length cDNAs of all expressed HBA and HBB genes were thereby sequenced at 6-fold coverage, and the haplotype phase of all variable sites was determined experimentally.

F₂ hybrids were subjected to a series of measurements during adulthood (1–1.5 years old), both before and after exposure to chronic hypoxia. Acute hypoxia responses and haematology were first measured in mice held in normoxia. Four days later, the mice were moved into specifically designed hypobaric chambers that have been previously described (Ivy and Scott, 2017; Lui et al., 2015; McClelland et al., 1998) and were thus acclimated to chronic hypoxia (barometric pressure of 60 kPa, simulating the pressure at an elevation of 4300 m; O₂ pressure ∼12.5 kPa). During this time, mice were temporarily returned to normobaric conditions twice per week for <20 min for cage cleaning. Acute hypoxia responses were measured again after 6–8 weeks in chronic hypoxia. Mice were finally euthanized after a full 8 weeks of chronic hypoxia acclimation with an overdose of isoflurane followed by cervical dislocation, and blood was collected for the second set of haematology measurements.

Acute hypoxia responses were measured in unrestrained mice using barometric plethysmography, respirometry and pulse oximetry techniques that we have used in previous studies (Ivy and Scott, 2017, 2018; Ivy et al., 2020). Mice were placed in a whole-body plethysmography chamber (530 ml) that was supplied with hyperoxic air (30 kPa O₂, balance N₂) at 600 ml min⁻¹. Mice were given 20–40 min to adjust to the chamber until relaxed and stable breathing and metabolism were observed. Mice were then maintained for an additional 20 min at 30 kPa O₂, after which they were exposed to 20 min stepwise reductions in inspired O₂ pressure (P_O2) of 21, 16, 12, 10 and 8 kPa. Dry incurrent gases were mixed using precision flow meters (Sierra Instruments, Monterey, CA, USA) and a mass flow controller (MFC-4, Sable Systems, Las Vegas, NV, USA), such that the desired P_O2 was delivered to the chamber at a constant flow rate of 600 ml min⁻¹. At the end of this protocol, mice were removed from the chamber and returned to their home cage in the appropriate acclimation condition.

Breathing (total ventilation, breathing frequency and tidal volume), rates of O₂ consumption (V̇_O2), body temperature (T_b), heart rate (f_H) and Sa_O2 were determined during the last 10 min at each P_O2 as follows. Incurrent and excurrent air flows were subsampled at 200 ml min⁻¹; incurrent air was continuously measured for O₂ fraction (FC-10, Sable Systems), and excurrent air was analysed for water vapour (RH-300, Sable Systems), dried with pre-baked Drierite, and analysed for O₂ and CO₂ fraction (FC-10 and CA-10, Sable Systems). These data were used to calculate V̇_O2, expressed in volumes at standard temperature and pressure (STP), using established equations (Lighton, 2008). Chamber temperature was continuously recorded with a thermocouple (TC-2000, Sable Systems). Breathing frequency, tidal volume and total ventilation were measured using whole-body plethysmography as previously described (Ivy and Scott, 2017, 2018) and reported volumes are expressed at body temperature and pressure saturated (BTPS). Air convection requirement (ACR) is the quotient of total ventilation and V̇_O2. All the above data were acquired using PowerLab 16/32 and Labchart 8 Pro software (ADInstruments, Colorado Springs, CO, USA). T_b was measured using thermosensitive passive transponders (micro LifeChips with Bio-therm technology, Destron Fearing, Dallas, TX, USA), which were implanted subdermally on the left side of the abdomen close to the leg ∼2 weeks before normoxic measurements were conducted, along with a hand-held scanner from the same manufacturer. Sa_O2 and f_H were measured using MouseOx Plus pulse oximeter collar sensors and data acquisition system (Starr Life Sciences, Oakmont, PA, USA). This was enabled by removing fur around the neck ∼2 days before experiments.

Blood was collected into heparinized capillary tubes for haematology, sampled from the facial vein under light anaesthesia for mice acclimated to normoxia (∼130 µl), or by severing the jugular vein for mice that were euthanized and sampled after acclimation to chronic hypoxia (∼400 µl). Blood Hb content was measured using Drabkin's reagent according to the manufacturer's instructions (Sigma-Aldrich, Oakville, ON, Canada). Haematocrit was measured by spinning the blood in the capillary tubes at 12,700 g for 5 min. Oxygen dissociation curves were generated at 37°C for all mice using a Hemox Analyzer (TCS Scientific, New Hope, PA, USA) using 10 µl of whole blood in 5 ml of buffer (50 mmol l⁻¹ Hepes, 10 mmol l⁻¹ EDTA, 100 mmol l⁻¹ KCl, 0.1% bovine serum albumin and 0.2% antifoaming agent, pH 7.4; TCS Scientific). Hb–O₂ affinity (P₅₀, the P_O2 at which Hb is 50% saturated with O₂) was calculated using Hemox Analytic Software (TCS Scientific). These measurements of blood Hb content, haematocrit and P₅₀ have been previously published (Wearing et al., 2021) but are reported again here to provide insight into the measurements of acute hypoxia responses.

The larger blood samples collected after hypoxia acclimation were also used for measurements of the nitric oxide (NO) metabolites nitrite, S-nitrosothiols (SNO) and iron-nitrosyl and N-nitrosamine derivatives (FeNO+NNO) in plasma and red blood cells. This was accomplished by reductive chemiluminescence using a Sievers Nitric Oxide Analyzer (NOA model 280i, Boulder, CO, USA) and previously described protocols (Hansen and Jensen, 2010; Yang et al., 2003). Blood was sampled in dim light conditions, and spun at 16,000 g for 2 min to separate plasma from red blood cells, which were then quickly frozen in liquid N₂ and stored at −80°C. Frozen samples (100 µl) were thawed and immediately incubated at room temperature for 5–10 min in the dark with a SNO-stabilizing solution (900 µl) described elsewhere (Yang et al., 2003) and then centrifuged (10,000 g for 2 min). The supernatant of each sample was injected into the NOA purge vessel in serial aliquots (300 µl, peak A) and after 2 min incubation with sulfanilamide (270 µl, peak B) and sulfanilamide and HgCl₂ (270 µl, peak C) to obtain values for nitrite, SNO and FeNO+NNO from the three peaks, as previously described (Yang et al., 2003). These measurements were only conducted for a subset of mice because of an unforeseen technical issue that resulted in the loss of some samples during storage.

Captive G₁ populations of deer mice from high and low altitude, held in standard holding conditions in normoxia, were used to assess the acute effects of manipulating Hb–O₂ binding. Mice were placed in the plethysmography chamber and exposed to normoxic conditions (21 kPa O₂) for 40 min to make baseline measurements. Mice were then removed and given an intraperitoneal injection of either saline or efaproxiral sodium at a volume of 20 ml kg⁻¹ body mass (Fisher Scientific, Whitby, ON, Canada). Efaproxiral was prepared in sterile saline (0.9% NaCl solution) on the day of experiments and was administered at a dose of 200 mg kg⁻¹ body mass. Mice were then returned to the chamber, and measurements were made for 50 min in normoxia and 20 min in hypoxia (12 kPa O₂). Breathing, V̇_O2, T_b, f_H and Sa_O2 were measured in the last 10 min of each exposure as described above. Every individual underwent both saline and efaproxiral injections, conducted in random order and separated by 1 week, and the efaproxiral dose used was determined in preliminary tests to have persistent effects on Sa_O2 for the duration of the experiment.

Linear mixed-effects models were used in experiments with F₂ hybrids to test for effects of α- and β-globin genotype, acclimation environment and inspired P_O2. They were also used in the efaproxiral experiments to test for effects of efaproxiral, mouse population and inspired P_O2. We initially tested for the random effects of sex and family, but they did not near statistical significance (P>0.10) and were therefore removed from the final models reported here. The full results of the linear mixed-effects models are given in Tables S1–S3), and the salient findings are reported in the Results. Holm–Šidák post-tests were used as appropriate. Statistical analysis was conducted using the lme4 package in R (v. 3.6.0) (Bates et al., 2015) with a significance level of P<0.05. Values are reported as means±s.e.m.

When F₂ hybrids were considered all together, both acute and chronic hypoxia affected breathing, metabolism and Sa_O2 (Fig. 1). Adult mice were subjected to acute step-wise decreases in inspired partial pressure of O₂ (P_O2) both before and after chronic hypoxia acclimation (6–8 weeks at ∼12 kPa O₂). Total ventilation increased in response to acute hypoxia as a result of increases in breathing frequency that offset smaller declines in tidal volume. Chronic hypoxia augmented these increases in total ventilation, particularly in response to severe acute hypoxia (acclimation environment×inspired P_O2, P<0.001), which arose from significant increases in breathing frequency (environment×P_O2, P<0.001). Chronic hypoxia also had significant main effects on several other phenotypes, attenuating declines in Sa_O2 (P<0.001) and f_H (P<0.001) in response to acute stepwise hypoxia, and increasing O₂ consumption rate (P=0.009) and T_b (P<0.001) (Fig. 1). Chronic hypoxia also increased haematocrit, whole-blood Hb content and Hb P₅₀ (P<0.001 for all environment effects) (Table 1).

Several of these cardiorespiratory and metabolic phenotypes were associated with α-globin and/or β-globin genotype (Tables S1, S2). Below, we discuss the statistically significant differences associated with genotype, but we include the full suite of measurements for each genotype in Figs S1–S3. There was a strong effect of α-globin genotype (P=0.012), but not β-globin genotype (P=0.711), on Sa_O2. Sa_O2 in severe hypoxia was higher in the highland (H) α-globin genotypes compared with the lowland (L) α-globin genotypes in measurements among normoxia-acclimated mice (Fig. 2A). However, although chronic hypoxia tended to reduce the decline in Sa_O2 across genotypes (Fig. 1A), this effect was greater in mice with the α^LL genotypes, such that the difference in Sa_O2 between genotypes was abolished after hypoxia acclimation (Fig. 2B). This variation in Sa_O2 appeared to be associated with variation in Hb–O₂ affinity measured in intact red blood cells, for which there was also an effect of α-globin genotype (α-globin effect, P<0.001; α-globin×environment, P=0.035) but not β-globin genotype (Table S2). In particular, α^HH mice exhibited significantly higher Hb–O₂ affinity (i.e. lower P₅₀) than α^LL mice before exposure to chronic hypoxia, but Hb–O₂ affinity decreased in response to chronic hypoxia only in mice with the α^HH genotype, such that α-globin genotypes were similar after hypoxia acclimation (Fig. 2C). Measurements of NO metabolites (i.e. nitrite, S-nitrosothiols, iron-nitrosyl and N-nitrosamine derivatives) were made after hypoxia acclimation, and α-globin genotype affected plasma nitrite concentration but had no effect on plasma concentrations of other NO metabolites (Table 2).

There was a strong association of α-globin genotype with variation in breathing pattern, both before and after exposure to chronic hypoxia (Fig. 3). In measurements both before and after hypoxia acclimation, mice with the α^HH genotype breathed using significantly deeper breaths (α-globin effect, P<0.001) but at a slower frequency (α-globin effect, P<0.001) than mice with the α^LL genotype, with no significant differences between α-globin genotypes in total ventilation. These differences in breathing pattern persisted across a range of inspired O₂ levels, from hyperoxia at 30 kPa O₂ (when Hb was fully saturated with O₂) to severe hypoxia.

β-Globin genotype was associated with variation in the hypoxic ventilatory response, as reflected by a significant interaction effect between β-globin genotype and inspired P_O2 on total ventilation (P=0.009). The association of β-globin genotype with total ventilation was particularly evident among normoxia-acclimated mice that were homozygous for highland α-globin, with no significant associations of β-globin genotype with V̇_O2 (Fig. 4; Table S1). Among these normoxia-acclimated mice, those that were homozygous for highland β-globin had higher total ventilation than both heterozygotes and lowland homozygotes at 12 kPa O₂, and higher total ventilation than heterozygotes in more severe levels of acute hypoxia (Fig. 4A). However, these differences between β-globin genotypes disappeared after hypoxia acclimation (Fig. 4B), potentially because the effects of hypoxia acclimation were greater in heterozygotes and lowland homozygotes.

Variation in T_b was associated with both α-globin (α-globin effect, P=0.007) and β-globin (β-globin effect, P=0.037) genotypes (Fig. S3, Table S1). T_b in normoxia was similar across genotypes, ∼36–38°C on average. Both the magnitude of T_b depression and the P_O2 at which T_b depression occurred varied among genotypes, but the magnitude of T_b depression tended to be reduced after hypoxia acclimation for all genotypes. However, neither α-globin nor β-globin genotype had significant associations with O₂ consumption rate before or after hypoxia acclimation (Fig. S3, Table S1).

We next sought to examine whether the association of globin genotype with respiratory phenotypes stems from variation in Hb–O₂ affinity. We used efaproxiral – a synthetic drug that acts as a negative allosteric regulator of Hb–O₂ binding – to reduce Hb–O₂ affinity acutely in captive-bred deer mice from high- and low-altitude populations in vivo. This treatment was expected to manifest as a reduction in Sa_O2 in acute hypoxia. Indeed, efaproxiral reduced Sa_O2 in high-altitude deer mice in hypoxia (Fig. 5A) and in low-altitude deer mice in both normoxia and hypoxia (Fig. 5B) compared with saline controls (P<0.001 for treatment effect and treatment×P_O2 interaction; Table S3). The magnitude of the effect of efaproxiral on Sa_O2 differed between populations (population×treatment, P=0.002; population×treatment×P_O2, P=0.048), driven by larger effects in lowlanders than in highlanders. However, efaproxiral had no consistent effects on breathing frequency, tidal volume or total ventilation in each population (no significant treatment or treatment×P_O2 effects) (Fig. 5). Efaproxiral also had no post-injection effects on oxygen consumption rate, air convection requirement or T_b (Table 3). However, efaproxiral did affect f_H, as reflected by a significant treatment×P_O2 interaction (P=0.005) that was driven primarily by increased f_H in lowlanders after efaproxiral injection (population×treatment, P=0.012; Table 3). Therefore, our treatment was successful in reducing Sa_O2 and leading to potential compensatory changes in f_H but it had no effect on breathing, in stark contrast to the differences observed between globin genotypes.

Evolved changes in Hb–O₂ affinity have contributed to hypoxia adaptation in numerous high-altitude mammals and birds and is often assumed to confer a physiological benefit by safeguarding Sa_O2 in hypoxia (Storz, 2016). However, the possibility that adaptive modifications of Hb function might contribute to other physiological processes that are not directly related to circulatory O₂ transport has been largely unexplored. Here, we show that allelic variants of the α-globin and β-globin genes in high-altitude deer mice are associated with changes in breathing that augment alveolar ventilation in hypoxia. These effects could not be recapitulated by acute changes in Sa_O2 or Hb–O₂ affinity using pharmacological manipulations. Our results suggest that allelic variation in Hb genes is associated with multiple respiratory phenotypes, and may contribute to environmental adaptation via physiological mechanisms that are not commonly ascribed to this protein.

α-Globin genotype had a strong association with breathing pattern, in which α^HH deer mice breathed deeper but less frequently, a change that likely augments alveolar ventilation without affecting total ventilation. These observed differences in breathing pattern between α-globin genotypes can completely account for previously documented differences that distinguish highland deer mice from lowland conspecifics and a closely related lowland congener (Peromyscus leucopus), which are observed in wild deer mice as well as deer mice raised for one or two generations in captivity in normoxia (Ivy and Scott, 2017, 2018; Ivy et al., 2020). This is nicely illustrated in plots of total ventilation against tidal volume (Fig. 6), which are useful for visualizing variation in breathing pattern independent of differences in total ventilation. The pronounced rightward shift in interpopulation hybrids with the α^HH genotype compared with the α^LL genotype (Fig. 6A), which reflects deeper but less frequent breaths at any given level of total ventilation, is nearly identical to the rightward shift in highlanders compared with lowlanders in previous measurements of deer mice acclimated to normoxia (Fig. 6B). The first-generation lab-raised mice used for the latter measurements are different from those used for breeding in the current study, but they do represent the physiology of the parents used to generate the F₂ interpopulation hybrids studied here. A similar deepening of breathing pattern has evolved in the high-altitude bar-headed goose (Scott and Milsom, 2007), which also possesses an α-globin variant that contributes to an evolved increase in Hb–O₂ affinity (Natarajan et al., 2018), suggesting that there may be an association between α-globin genotype and breathing pattern in other high-altitude taxa.

β-Globin genotype was associated with the hypoxic ventilatory response among normoxia-acclimated deer mice, with the α^HHβ^HH genotype exhibiting higher total ventilation than α^HHβ^HL and α^HHβ^LL genotypes at 12 kPa O₂. These findings mirror those in lab-strain mice possessing the Hb Presbyterian β-globin mutation, which is associated with a reduced Hb–O₂ affinity and an attenuated hypoxic ventilatory response (Izumizaki et al., 2003). Differences in β-globin genotype could thus, in addition to α-globin, contribute to the increases in total ventilation in highland deer mice compared with lowland deer mice and P. leucopus that we have observed among mice acclimated to normoxia (Ivy and Scott, 2017, 2018; Ivy et al., 2020). However, these associations with β-globin genotype were abolished after hypoxia acclimation, suggesting that β-globin may influence the effects of chronic hypoxia on control of breathing, a process termed ventilatory acclimatization to hypoxia (VAH). If so, β-globin may contribute to the attenuation of VAH that we have previously observed in highland deer mice (Ivy and Scott, 2017, 2018).

The association of globin variants with variation in breathing phenotypes could not be recapitulated by acute manipulations of Hb–O₂ affinity or Sa_O2. Treatment of deer mice from highland and lowland populations with efaproxiral to reduce Hb–O₂ affinity had no effect on total ventilation or breathing pattern in normoxia or hypoxia. There were differences in the magnitude of the effects of efaproxiral on Sa_O2 and f_H between populations, possibly because the normally lower Hb–O₂ affinity of lowlanders (Ivy et al., 2020) made them more susceptible to impairments in pulmonary O₂ loading upon further reduction in affinity, but efaproxiral had no effect on breathing in either population. Furthermore, the differences in breathing pattern between α^HH and α^LL deer mice persisted in hyperoxia (30 kPa O₂) when blood O₂ tension was well above that needed to fully saturate Hb with O₂. However, the effects of efaproxiral on Hb–O₂ affinity are temporary and we only measured the drug's acute effects, and it is possible chronic changes in Hb–O₂ affinity (akin to the differences between adult mice with highland versus lowland globin genotypes) are needed to induce variation in breathing phenotypes. Alternatively, our findings with efaproxiral treatment could also suggest that globin variants regulate breathing via mechanisms that are not directly associated with the role of Hb in circulatory O₂ transport. The globin family is ancient and carries out various cellular functions beyond O₂ transport (Fago et al., 2004; Kamga et al., 2012). The α/β monomers of Hb are expressed in various non-erythroid cells, including neurons and vascular endothelium (Biagioli et al., 2009; Newton et al., 2006; Richter et al., 2009; Schelshorn et al., 2009; Straub et al., 2012), so it is possible that α- and/or β-globins expressed in non-erythroid tissues could regulate ventilatory phenotypes.

The association of α-globin genotype with variation in Sa_O2 was contingent upon acclimation environment, consistent with other recent findings (Wearing et al., 2021). Deer mice that were homozygous for highland α-globin (α^HH) maintained higher Sa_O2 in hypoxia and had higher Hb–O₂ affinity than lowland homozygous deer mice (α^LL) when comparisons were made among normoxia-acclimated deer mice, but these differences were abolished after hypoxia acclimation. This discrepancy could be explained by differences in sensitivity to 2,3-DPG, a key negative allosteric regulator of Hb–O₂ binding in mammalian erythrocytes that can increase in concentration after acclimation to the levels of chronic hypoxia used here (Lenfant et al., 1968; Snyder, 1982). Indeed, previous studies of O₂-binding properties of stripped Hb suggest that 2,3-DPG sensitivity is greater in high-altitude populations of deer mice when measured in the presence of physiologically relevant concentrations of Cl⁻ (Storz et al., 2010). Therefore, highland homozygotes could have been more sensitive to the increases in red cell 2,3-DPG concentration that may have occurred with hypoxia acclimation, and could have thus exhibited a more pronounced decrease in Hb–O₂ affinity and a less pronounced increase in Sa_O2 in hypoxia. This emphasizes the potential advantage of the evolved (genetically based) reduction in erythrocyte 2,3-DPG levels that has been observed in natural high-altitude populations of deer mice, which arose independent of changes in α-globin genotype (Snyder, 1982; Snyder et al., 1982). This evolved reduction in 2,3-DPG levels counteracts the effects of hypoxia acclimation on this trait, such that 2,3-DPG levels are elevated only slightly in high-altitude populations in the wild (Snyder, 1982). CO₂ and H⁺ are also key allosteric regulators of Hb–O₂ binding, and it is possible that differences in alveolar ventilation between genotypes affected blood CO₂/pH and thus influenced our in vivo measurements of Sa_O2. However, variation in blood CO₂/pH between genotypes would not have persisted in the buffer used for in vitro measurements.

The unexpected association between Hb genotype and control of breathing is especially intriguing in light of population genetic evidence for altitude-related selection on the α- and β-globin genes in deer mice (Storz and Kelly, 2008; Storz et al., 2009, 2012). The adaptive relevance of Hb–O₂ affinity is well established in high-altitude vertebrates, but the present findings force us to consider the possibility that allelic variation in Hb function may affect a broader diversity of physiological processes than previously assumed. Recent studies suggest that Hb functions not just as an O₂ carrier but also as an O₂ sensor and O₂-responsive transducer of NO vasoactivity in the microcirculation, thereby contributing to hypoxic vasodilation that helps match perfusion to tissue O₂ demand (Jensen, 2009; Storz, 2018; Zhang et al., 2015, 2016). Indeed, our results suggest that the α-globin variant in high-altitude deer mice affects the concentration of NO metabolites in the circulation (Table 2). The results reported here also suggest that the physiological effects of Hb may even transcend circulatory O₂ transport, with direct or indirect effects on the control of breathing. The next step is to identify and characterize the causal mechanism underlying the unexpected genotype–phenotype association.

Our findings contribute to a growing awareness that protein polymorphism can sometimes have phenotypic effects that are unrelated to the ‘canonical’ function of the protein in question. For example, genetic variation in enzymes of central metabolism can affect physiological phenotypes via mechanisms independent of pathway flux as a result of signalling functions of intermediary metabolites or non-enzymatic ‘moonlighting’ functions of the enzymes (Marden, 2013a,b). Such non-canonical effects of Hb variants may be common, based on previous observations that genetic variation in globins is not always associated with variation in Hb–O₂ binding (Barlow et al., 2017; Cheviron et al., 2014a; Nelson et al., 2019). The realization that there may be physiologically important auxiliary functions still waiting to be discovered in a protein as intensively studied as Hb highlights the importance of maintaining a wide field of vision when investigating causal connections between genotype and phenotype.

The authors would like to thank the animal care technicians and the members of the Scott and McClelland labs for help and support while completing this project.