Phenotypic plasticity and genetic adaptation to high-altitude hypoxia in vertebrates

High-altitude environments provide fertile ground for investigating mechanisms of physiological adaptation. For air-breathing vertebrates that live at high altitude, hypobaric hypoxia is an unremitting environmental stressor that cannot be mitigated by behavioral avoidance. High-altitude environments are also characterized by low ambient temperatures relative to lowland environments at similar latitudes, and therefore present endothermic animals with the additional physiological challenge of sustaining metabolic heat production in spite of the reduced availability of O₂ for aerobic power generation.

In animal species that inhabit a broad range of elevational zones, physiological challenges associated with reduced O₂ availability and reduced ambient temperature will tend to increase in severity as a function of altitude. Altitudinal gradients – like latitudinal gradients – are also characterized by relatively uniform and consistent patterns of change in a number of other environmental variables such as net primary productivity. One key difference is that ecotonal transitions across altitudinal gradients occur more rapidly over a given distance than those associated with latitudinal gradients. The spatially fine-grained environmental variation across altitudinal gradients has important implications for the relative roles of genotypic specialization and phenotypic plasticity in physiological adaptation.

When environmental heterogeneity is fine grained, individuals are forced to tolerate a broad range of environmental conditions over the course of their lifetimes. Conversely, when environmental heterogeneity is coarse grained, individuals will experience a more limited range of conditions. An organism's perceived grain of environmental heterogeneity is determined by its life history and dispersal capability. For a set of taxa that are co-distributed across the same altitudinal gradient, spatial variation in environmental conditions would be perceived as fine grained by highly mobile organisms such as birds or bats, and would be perceived as coarse grained by rodents and other organisms that have more restricted dispersal capabilities (Levins, 1968; van Tienderen, 1991; van Tienderen, 1997). Spatially fine-grained environmental heterogeneity is especially conducive to the evolution of phenotypic plasticity, as an increased acclimatization capacity enables organisms to track changes in habitat-specific trait optima. If no single phenotype confers highest fitness across the full range of environments, theory predicts that dispersal among spatially separated subpopulations will generally promote the evolution of phenotypic plasticity over genotypic specialization (Scheiner, 1993; Scheiner, 1998; de Jong and Behera, 2002; Sultan and Spencer, 2002). This theoretical prediction is relevant to animal species that are continuously distributed across steep altitudinal gradients. With regard to the temporal component of environmental heterogeneity, theory predicts that fluctuations that occur on a within-generation timescale will be especially conducive to the evolution of reversible plasticity (Moran, 1992). This would be relevant, for example, to birds that migrate over high mountain passes and therefore have to cope with marked changes in ambient temperature and O₂ availability on a daily or seasonal basis.

Environment-dependent variation in the phenotypic expression of a given genotype can be described as a reaction norm, and genotypic variation in reaction norms is referred to as a genotype × environment interaction (Fry, 1992; Via et al., 1995). A given reaction norm can be considered adaptive if the phenotype expressed in each environmental state is in the same direction as the environment-specific optimum favored by selection (Conover and Schultz, 1995; van Tienderen, 1991; van Tienderen, 1997; Ghalambor et al., 2007). During the colonization of a new environment, an adaptive reaction norm would move the mean phenotype closer to the new trait optimum (Price et al., 2003; Ghalambor et al., 2007). By contrast, a maladaptive reaction norm would move the mean phenotype further away from the new optimum, and would therefore require selection on genetically based trait variation to counteract the environmentally induced changes. In cases in which the trait optimum is unchanged in the newly colonized environment, a maladaptive reaction norm would require compensatory genetic changes to restore the ancestral phenotype (Grether, 2005). Thus, maladaptive plasticity can result in a form of cryptic adaptive evolution in which phenotypic similarity between ancestral and derived populations is attributable to the effects of divergent selection that offset environmentally induced trait differences.

Much of our understanding of the acclimatization response to hypoxia derives from studies of animal species that are native to lowland environments. These studies can reveal whether the hypoxia-induced physiological response tends to move the population mean phenotype closer to the fitness optimum of the nonnative high-altitude environment. The nature of the ancestral acclimatization response can therefore indicate whether phenotypic plasticity can be expected to facilitate or impede adaptation to high altitude. Here, we review general mechanisms of physiological acclimatization and genetic adaptation to high-altitude hypoxia in birds and mammals. Rather than reviewing these mechanisms in detail, we evaluate whether the acclimatization response to environmental hypoxia can be regarded generally as a mechanism of adaptive phenotypic plasticity, or whether it represents a misdirected response that works at cross purposes with genetically based physiological adaptation.

Under hypoxic conditions, many animals are able to compensate for a reduced O₂ supply by suppressing total metabolism, thereby reducing O₂ demand (Ramirez et al., 2007). In many endotherms that are native to high-altitude environments, metabolic suppression is not a viable option for coping with reduced O₂ availability owing to the exigencies of aerobic thermogenesis and/or aerobic exercise. In birds, for example, O₂ consumption increases 15- to 20-fold during flight (Ward et al., 2002). In such cases, matching O₂ supply with an undiminished or even increased O₂ demand requires an enhanced flux capacity of the O₂ transport pathway.

There are several convective and diffusive steps in the O₂ cascade from inspired air to the tissue mitochondria where physiological adjustments can preserve the rate of O₂ flux in spite of hypoxia, thereby ensuring an uninterrupted supply of O₂ to the energy-producing machinery of the cells (Fig. 1). These steps include ventilatory convection, diffusion across the blood–gas interface, circulatory convection, diffusion across the blood–tissue interface (including facilitated diffusion by myoglobin), and O₂ utilization by the tissue mitochondria.

The effectiveness of O₂ loading during high-altitude hypoxia is strongly influenced by plasticity in the pulmonary system, which comprises the first two steps in the O₂ transport pathway (Fig. 1). The rate of O₂ transport in the air compartment of the lungs can be described as the product of ventilation (V) and the inspired–alveolar difference in O₂ partial pressure (P_O2) (i.e. P_IO2–P_AO2). High rates of ventilation therefore help maintain a high alveolar P_O2, the pressure head for diffusion across the blood–gas interface. The rate of O₂ transport across the blood–gas interface can be described as the product of pulmonary O₂ diffusion capacity (D_L) and the alveolar–blood P_O2 difference (P_AO2–Pb_O2). The decrement in Pb_O2 is therefore mitigated by an enhanced pulmonary O₂ diffusion capacity, which is determined by lung morphology and pulmonary perfusion.

The most important factors affecting ventilation at high altitude are the partial pressures of O₂ and CO₂ and the pH of arterial blood (Scott and Milsom, 2009). Breathing is stimulated when a decline in arterial P_O2 (Pa_O2; hypoxemia) is sensed by chemoreceptors in the carotid bodies. However, this hypoxic ventilatory response increases respiratory CO₂ loss, causing a secondary hypocapnia (low CO₂) and alkalosis (high pH) in the blood. Because CO₂ and H⁺ normally stimulate breathing via central and peripheral chemoreceptors, secondary hypocapnia and alkalosis restrain the ventilatory response. The net effect is an increase in ventilation that is dependent on the severity of hypoxia, but the magnitude of this increase is less than if blood CO₂ levels had remained constant (Fig. 2) (Scott and Milsom, 2009). The magnitude of the hypoxic ventilatory response is also dependent on the duration of hypoxia (Powell et al., 1998). Compared with the rapid response that occurs within minutes of the onset of hypoxia, several days of acclimatization can further increase ventilation owing to plasticity in the O₂ sensing function of the carotid body chemoreceptors and in central integration sites within the neural feedback loop that regulates breathing (Black and Tenney, 1980; Powell, 2007). Over many months at high altitude, this hypoxic ventilatory response can be gradually attenuated by hypoxic desensitization (Powell et al., 1998; Brutsaert, 2007). It is difficult to ascribe adaptive significance to these various responses owing to the multiple functions of breathing. Increases in breathing during the acute response and early phases of acclimatization clearly improve O₂ uptake during hypoxia. Conversely, hypoxic desensitization might help to maintain blood CO₂/pH homeostasis, reduce respiratory water loss, or reduce the metabolic cost of breathing at the expense of diminished O₂ uptake (Powell, 2007).

The morphological capacity for O₂ diffusion across the lungs is generally increased in lowlanders after acclimatization to high-altitude hypoxia. Development or prolonged residence at high altitude can increase pulmonary diffusion capacity by increasing the surface area available for diffusion and by reducing the thickness of the pulmonary blood–gas interface (Hsia et al., 2005; Hsia et al., 2007; Ravikumar et al., 2009). Although relatively little is known about how O₂ levels control lung morphogenesis during development or during adult life stages (Metzger et al., 2008), these induced changes in lung morphology clearly help improve O₂ transport during hypoxia.

In lowland species, the acclimatization response to hypoxia generally involves a maladaptive change in pulmonary perfusion, the physiological component of pulmonary O₂ diffusion. Hypoxia directly stimulates the endothelial and smooth muscle cells in pulmonary blood vessels, causing vasoconstriction throughout the lungs and an increase in pulmonary blood pressure that can persist for prolonged durations at high altitude (Reeves and Grover, 1975; Gurney, 2002). This global vasoconstriction impairs O₂ diffusion because it can divert blood flow away from the gas exchange surface to pulmonary shunt vessels (Lovering et al., 2008), and the resultant pulmonary hypertension can cause fluid leakage into the air spaces, which in turn causes a thickening of the O₂ diffusion barrier (Maggiorini et al., 2001; Eldridge et al., 2006). Hypoxic pulmonary hypertension can also overburden the right ventricle of the heart and can contribute to pathophysiological conditions, such as chronic mountain sickness (Monge and León-Velarde, 1991; León-Velarde et al., 2010).

Some features of the acclimatization response to hypoxia can improve pulmonary O₂ transport at high altitude, whereas other features can be counterproductive. Populations at high altitude could therefore benefit from enhancing the facilitatory mechanisms or suppressing the inhibitory ones. Some highland species or populations are characterized by a heightened acute hypoxic ventilatory response, a more effective breathing pattern, a maintained acclimatization response, and/or a prevention of hypoxic desensitization (Black and Tenney, 1980; Beall et al., 1997; Wu and Kayser, 2006; Brutsaert, 2007; Scott and Milsom, 2007; Pichon et al., 2009). The underlying chemoreceptive mechanisms are not necessarily convergent at the cellular level. For example, the enhanced ventilatory response to acute hypoxia in the high-altitude bar-headed goose (Anser indicus) is partly due to an insensitivity to hypocapnia (Fig. 2) (Scott and Milsom, 2007; Scott and Milsom, 2009), whereas responses of high-altitude humans generally involve changes in O₂ sensitivity (Brutsaert, 2007). Other highland species or populations exhibit a blunted ventilatory response to acute hypoxia (Blake and Banchero, 1985; Brutsaert et al., 2005), which would be maladaptive at high altitudes without compensatory changes at other steps in the O₂ transport pathway that maintain peripheral O₂ supply. As might also be the case for hypoxic desensitization, this blunted response might help maintain blood CO₂/pH homeostasis or it might reduce the costs of breathing (Hochachka, 1998; Powell, 2007).

The ancestral acclimatization response for the morphological component of pulmonary O₂ diffusion will facilitate pulmonary O₂ transport at high altitude. However, pulmonary O₂ diffusion capacity can be enhanced further in some high-altitude adapted groups that have larger lungs than lowlanders (Brutsaert et al., 1999; Wu and Kayser, 2006; Brutsaert, 2007; Scott et al., 2010). This is particularly true in birds because their lungs are rigid and are ventilated by air sacs. The larger lungs of the high-altitude bar-headed goose (Scott et al., 2010), for example, should afford a greater surface area for diffusion and thus reduce the P_O2 difference between parabronchial gas and arterial blood. By contrast, the ancestral acclimatization response for the physiological component of pulmonary O₂ diffusion will impede pulmonary O₂ transport at high altitudes. Not surprisingly, the pulmonary vasoconstriction response is blunted or lost altogether in species that are adapted to high altitudes (Faraci et al., 1984a; Durmowicz et al., 1993; Groves et al., 1993; Ge et al., 1998; Sakai et al., 2003), which could represent an example of genetic compensation.

Under conditions of hypoxic hypoxemia, when the P_O2 of arterial blood (Pa_O2) is reduced, transport of O₂ by blood must meet a dual challenge: (1) it must maintain an adequate delivery of O₂ to the peripheral tissues; and (2) it must maintain an adequate P_O2 at the vascular supply source to permit O₂ diffusion to the tissue mitochondria.

At a given Pa_O2, O₂ delivery can be enhanced by increasing the total cardiac output, , and by increasing the blood O₂ capacitance coefficient, βb_O2 (Fig. 3). The latter parameter is defined as the ratio (Ca_O2–Cv_O2)/(Pa_O2–Pv_O2), where Ca_O2–Cv_O2 is the arterial–venous difference in O₂ concentration and Pa_O2–Pv_O2 is the arterial–venous difference in P_O2. With regard to the second challenge mentioned above, the lower critical P_O2 at the vascular supply source for tissue oxygenation can be expressed as , where is the rate of O₂ consumption by the tissues and the product is the specific blood O₂ conductance (Dejours et al., 1970; Bouverot, 1985). Because Pa_O2 is determined by ventilation and O₂ equilibration at the blood–gas interface, this equation shows that an increase in the specific blood O₂ conductance minimizes the decline in Pv_O2 under hypoxia, thereby maintaining an adequate pressure head for O₂ diffusion to the tissue mitochondria.

Under severe hypoxia, an increased blood–O₂ affinity will tend to maximize βb_O2 because the arterial–venous P_O2 difference spans the steepest portion of the sigmoidal O₂-equilibrium curve (Fig. 3). This can be visualized as an increase in the slope of the line connecting the arterial and venous points on the curve. The resultant increase in the specific blood O₂ conductance helps meet the dual challenge mentioned above: it minimizes the inevitable P_O2 decrement in the tissue capillaries while preserving a constant Ca_O2–Cv_O2 difference and, hence, a constant net O₂ flux. Likewise, an increased hemoglobin (Hb) concentration increases Ca_O2, thereby increasing blood O₂ conductance if Pa_O2, and all remain constant. With excessive polycythemia, however, potential advantages of an increased Hb concentration for O₂ carrying capacity might be more than offset by a corresponding reduction in .

When lowland natives ascend to high altitude, the acclimatization response to hypoxia generally entails correlated increases in hematocrit (Hct) and Hb concentration, but the adaptive significance of this erythropoeitic response is questionable (Bullard, 1972; Monge and Whittenbury, 1976; Winslow and Monge, 1987; Monge and Leon-Velarde, 1991). Under normoxia, a moderately increased Hct augments arterial O₂ content and therefore enhances aerobic capacity (Ekblom and Hermansen, 1968; Kanstrup and Ekblom, 1984; Ekblom and Berglund, 1991). However, the highest attainable Hct is not necessarily associated with the highest attainable aerobic power output (Crowell et al., 1959; Crowell and Smith, 1967; Villafuerte et al., 2004; Schuler et al., 2010). This is because the associated increase in blood viscosity produces a higher peripheral vascular resistance that might compromise , thereby reducing V_O2,max (Guyton and Richardson, 1961; Connes et al., 2006). It is reasonable to expect that a moderately increased Hct could confer a physiological advantage under hypoxia because a greater quantity of O₂ can be carried at partial saturations, but experimental results do not support this (McGrath and Weil, 1978; Winslow et al., 1985). In humans at high altitude, evidence suggests that the optimal Hb concentration might actually be quite close to the typical sea level value (Villafuerte et al., 2004).

In humans living at high altitude, hypoxia-induced polycythemia often vastly exceeds the optimal Hct. For example, in Andean natives living at altitudes of >3000 m, Hct and Hb concentrations can reach values of up to 0.83 and 270 g l^–1, respectively (Bouverot, 1985). This excessive polycythemia limits exercise performance, and pulmonary function, and appears to play a key role in the pathogenesis of chronic mountain sickness (Winslow et al., 1985; Monge and León-Velarde, 1991; Rivera-Ch et al., 2007). Richards (Richards, 1960) provided a vivid description of the effects of hypoxia-induced polycythemia in patients suffering from chronic pulmonary disease: “The increased red cell mass produces increased blood volume, which overfills the heart; and this organ, already under strain working against a restricted pulmonary vascular bed, goes into congestive heart failure, which still further increases blood volume. At the same time the added pulmonary vascular engorgement further aggravates the [hypoxia], thus setting up a second vicious cycle.”

In humans, the potential fitness consequences of excessive polycythemia seem especially clear cut when considering the effects on pregnancy outcomes at high altitude (Moore et al., 2004; Gonzales et al., 2009; Julian et al., 2009a; Julian et al., 2009b). Hospital-based studies have documented that the risk of adverse perinatal outcomes is increased with extreme maternal Hcts (Moore et al., 2001a; Gonzales et al., 2009) (Fig. 4). During pregnancy at high altitude, elevated Hct and the associated increase in blood viscosity contribute to an increased risk of stillbirth, preterm birth, and reduced birth weight, all of which appear to stem from a reduced uterine artery blood flow and, consequently, a reduced rate of O₂ delivery to the uteroplacental circulation (Moore et al., 1982; Moore et al., 2001a; Moore et al., 2001b; Zamudio et al., 1995; Wilson et al., 2007; Julian et al., 2008; Julian et al., 2009b). For this reason, acclimatized lowlanders at high altitude generally experience especially high rates of adverse pregnancy outcomes. In humans living at high altitude, the available evidence suggests that hypoxia-induced polycythemia represents an example of maladaptive plasticity.

Fine-tuned adjustments in blood–O₂ affinity can play a key role in matching O₂ supply and O₂ demand by shifting the steepest portion of the O₂-equilibrium curve within the range of in vivo P_O2s (Bouverot, 1985; Samaja et al., 2003). These adjustments could be mediated by changes in the intrinsic Hb–O₂ affinity, changes in the sensitivity of Hb to allosteric cofactors that modulate Hb–O₂ affinity, and/or changes in the concentration of allosteric cofactors within the erythrocyte (Nikinmaa, 2001; Weber and Fago, 2004; Weber, 2007; Storz and Moriyama, 2008). The latter mechanism involves regulatory changes in the chemical milieu within the erythrocyte and does not require structural modifications of the Hb protein itself. In the enucleated erythrocytes of most mammals, the most potent allosteric cofactor is 2,3-diphosphoglycerate (DPG), a metabolite of glycolysis [for exceptions, see Bunn (Bunn, 1980)]. In the nucleated erythrocytes of birds, the most potent allosteric cofactor is typically inositol pentaphosphate (InsP₅), a metabolite of oxidative phosphorylation. Ectothermic vertebrates typically make use of other nucleotide triphosphates (Nikinmaa, 2001). These various allosteric cofactors generally reduce Hb–O₂ affinity by preferentially binding and stabilizing the deoxygenated, ‘tense-state’ conformation of the Hb tetramer. Other cofactors such as H⁺ and Cl^– ions exert similar allosteric effects, thereby facilitating O₂ unloading in the tissue capillaries (Riggs, 1988; Nikinmaa, 2001; Mäirbaurl, 1994; Jensen, 2004; Weber and Fago, 2004).

In mammals that are native to lowland environments, the acclimatization response to hypoxia typically involves a DPG-induced reduction in Hb–O₂ affinity (a rightward shift in the O₂-equilibrium curve), which might be mediated primarily by an increase in the erythrocytic concentration of DPG (Lenfant et al., 1968; Lenfant et al., 1971; Mäirbaurl et al., 1993). In humans, for example, a rapid increase in the intracellular DPG concentration occurs within 24 h of ascent to high altitude (>4500 m), and the resultant decrease in Hb–O₂ affinity is partially offset by the respiratory alkalosis induced by increased ventilation (Lenfant et al., 1968; Lenfant et al., 1971; Weiskopf and Severinghaus, 1972).

Modulating the erythrocytic concentration of allosteric cofactors can be considered to be a mechanism of phenotypic plasticity because the same genotype (manifest as Hb structure) can express a range of different phenotypes (blood–O₂ affinities) depending on the prevailing circumstances. The adaptive significance of the DPG-induced decrease in Hb–O₂ affinity at high altitude has been the subject of much debate. Under conditions of moderate hypoxia, a reduction in Hb–O₂ affinity can enhance levels of tissue oxygenation by maximizing the capillary–tissue P_O2 difference. However, under conditions of severe hypoxia when Pa_O2 is substantially reduced, a leftward shift of the O₂-equilibrium curve is required to allow O₂ loading and unloading over the steepest portion of the curve (Turek et al., 1973; Turek et al., 1978a; Turek et al., 1978b; Samaja et al., 1986; Samaja et al., 2003). Even at low altitude, a left-shifted curve might be advantageous when the rate of O₂ transfer across the blood–gas interface is diffusion limited, as is the case during intense exercise (Bencowitz et al., 1982). In rats, a pharmacologically increased Hb–O₂ affinity was shown to greatly enhance metabolic performance and survival under conditions of extreme hypoxia (Eaton et al., 1974; Turek et al., 1978a; Turek et al., 1978b), and human subjects with mutant Hbs that exhibit increased O₂ affinity (Hb Andrews-Minneapolis) maintain normal arterial O₂ saturation and undiminished aerobic capacity at high altitude (3100 m) relative to subjects with wild-type Hb (Hebbel et al., 1978). It therefore appears that the DPG response of lowlanders might be maladaptive at high altitudes.

In contrast to the hematological changes that are typically associated with the acclimatization response to hypoxia in lowland mammals, genetically based changes in Hb structure that increase intrinsic O₂ affinity or that suppress sensitivity to allosteric cofactors are generally thought to make more important contributions to hypoxia tolerance in species that are high-altitude natives (Bunn, 1980; Monge and León-Velarde, 1991; Storz, 2007; Weber, 2007; Storz and Moriyama, 2008). This is because an increased Hb–O₂ affinity promotes pulmonary O₂ loading at low P_O2, and can thus amplify the effects of pulmonary adaptations to increase Pa_O2. However, an increased Hb–O₂ affinity also hinders O₂ unloading in the tissue capillaries. Other modulators of Hb–O₂ binding might be important in this regard, such as the Bohr effect (reduction in Hb–O₂ affinity at low pH). This trade-off in Hb function between normoxic and hypoxic environments can give rise to trade-offs in whole-organism aerobic performance (Hebbel et al., 1978; Chappell and Snyder, 1984; Chappell et al., 1988).

Similar to the situation for pulmonary O₂ diffusion, the rate of O₂ transport from blood to mitochondria in tissues can be described as the product of tissue O₂ diffusion capacity (D_T) and the blood–mitochondria P_O2 difference (Pb_O2–Pmit_O2). Under hypoxia, the decrement in P_O2 at mitochondria is mitigated by an enhanced O₂ diffusion capacity. O₂ diffusion at peripheral tissues has both morphological and physiological components. The morphological capacity for O₂ diffusion is determined by the density and distribution of capillaries in peripheral tissues, as well as by the cellular arrangement of mitochondria (Fig. 5). Regional tissue perfusion (capillary recruitment) represents the physiological component, which determines the extent to which the morphological capacity for O₂ diffusion is realized. Tissue O₂ transport will be optimized at high altitudes by full perfusion of abundant and homogeneously spaced capillaries, and by mitochondria that are positioned close to capillaries.

It is unclear whether the morphological capacity for O₂ diffusion is altered by high-altitude acclimatization in low-altitude species. In general, there is a strong relationship between capillarity and metabolic demand [e.g. mitochondrial abundance (Hepple, 2000)], but capillarity must increase relative to demand for tissue O₂ transport to be improved in hypoxia. The issue of whether high-altitude acclimatization increases capillarity has been contentiously debated, based on methodological and conceptual arguments related to both O₂ supply and metabolic demand (Mathieu-Costello, 2001). There are examples where capillarity in the muscle has increased during high-altitude acclimatization without any change in mitochondrial abundance (Mathieu-Costello and Agey, 1997), but it is generally thought that any changes in capillarity occur as a secondary consequence of alterations in metabolic demand or cell size (Lewis et al., 1999; Hepple, 2000; Hoppeler and Vogt, 2001). The arrangement of mitochondria within cells must trade-off between reducing the intracellular diffusion distance for O₂ and reducing that for intracellular metabolites, which will be optimized when mitochondria are situated in predominantly subsarcolemmal (i.e. clustered near the cell membrane) or intermyfibrillar (i.e. distributed throughout the cell) locations, respectively (Fig. 5B) (Kinsey et al., 2007). Acclimatization to high-altitude hypoxia does not appear to alter mitochondrial distribution in muscle, based on a study of chronic hypoxia exposure in rats (van Ekeren et al., 1992). This contrasts with the proliferation of subsarcolemmal mitochondria that is induced when hypoxia is experienced only during short daily bouts of exercise (Schmutz et al., 2010). Although comprehensive support remains to be collected, low-altitude species appear to exhibit negligible plasticity in the morphological capacity for O₂ diffusion during high-altitude acclimatization.

Lowlanders at high altitude might sacrifice perfusion of muscle and some other peripheral tissues. Regional tissue perfusion during hypoxia is regulated by local paracrine signals that promote vasodilation (e.g. nitric oxide) and central signals that regulate vasoactivity in a tissue-specific manner (e.g. sympathetic neural activity). The former mechanism tends to increase blood flow and capillary recruitment, particularly to metabolically active tissues (Heinonen et al., 2010). By contrast, the latter mechanism can promote a redistribution of blood flow away from peripheral tissues during hypoxia in favor of hypoxia-sensitive tissues, such as the heart and brain (Hochachka, 1985). This reduction in blood flow to peripheral tissues might help animals tolerate short periods of O₂ lack, but it is not overcome after prolonged acclimatization to high altitude (Kamitomo et al., 1993; Hansen and Sander, 2003). Peripheral O₂ supply can remain impaired even when pulmonary and hematological adjustments restore arterial oxygenation to normoxic levels (Calbet et al., 2003), possibly because of the complex integration of feedforward and feedback influences from the central nervous system, the heart and the peripheral tissues (Noakes et al., 2004). The redistribution of blood flow might therefore represent a maladaptive response to high-altitude hypoxia.

In principle, tissue O₂ transport could be improved at high altitude by enhanced morphological capacities for O₂ diffusion. This has occurred in some high-altitude species, such as the bar-headed goose, which has expanded capillarity in the flight muscle and heart, as well as mitochondria that are redistributed closer to the more abundant capillaries (Fig. 5) (Scott et al., 2009a; Scott et al., 2010). These unique traits are canalized, occurring without any prior exposure to high altitude. The benefit of these specializations is probably restricted to tissues that are active during flight, because running exercise is not well sustained in bar-headed geese during hypoxia (Fedde et al., 1989). However, this phenotype does not appear to be a universal evolutionary response to high altitude, because muscle capillarity is not altered in some high-altitude mammals (Mathieu-Costello, 1989; Brutsaert, 2008). Tissue O₂ transport could also be improved by restricting the hypoxia-induced redistribution of blood flow. This hypothesis has not been thoroughly investigated, and the minimal support for this possibility is confounded by differences in arterial O₂ loading (Faraci et al., 1984b; Faraci et al., 1985).

Tissue O₂ utilization is altered in various ways by high-altitude acclimatization in low-altitude species. A reduction in whole-tissue metabolic capacity (i.e. maximal O₂ demand) occurs in muscle and other tissues during chronic hypoxia through reductions in cell size and mitochondrial abundance (Hoppeler and Vogt, 2001; Mathieu-Costello, 2001). Reductions in mitochondrial capacity lead to a shift in fuel preference in favour of carbohydrate oxidation at a given rate of ATP turnover (McClelland et al., 1998), which reduces the requirement for O₂ by 12–14% (Essop, 2007; Calbet et al., 2009). Chronic hypoxia might in some cases improve the efficiency of mitochondrial O₂ utilization (i.e. ADP/O₂ consumed) (Zungu et al., 2008) but, in general, mitochondrial function is sustained or even impaired under such conditions (Essop, 2007). The use of anaerobic metabolism in the muscle to synthesize ATP during exercise generally declines after high-altitude acclimatization, possibly because of an improved coordination between O₂ supply, glycolytic flux and cellular ATP demand (Hochachka et al., 2002). By contrast, chronic hypoxia has no effect on some traits that could have a profound influence on mitochondrial metabolism during O₂ lack, such as the sensitivity of mitochondrial respiration to low O₂ (mitochondrial O₂ kinetics) (Costa et al., 1997) and the activity of the creatine kinase shuttle (Daneshrad et al., 2001). The general result is that some metabolic traits respond to hypoxia to help match cellular O₂ supply and demand, whereas others exhibit negligible plasticity.

Depending on the species in question, ancestral acclimatization responses in tissue O₂ utilization could either facilitate or impede high-altitude adaptation. The reduction in metabolic capacity might be beneficial in species that usually sustain sub-maximal rates of aerobic metabolism. Mitochondrial abundance is constitutively reduced in the muscle of Tibetan humans compared with that of native lowlanders, and is not affected by environmental O₂ availability (Kayser et al., 1991; Kayser et al., 1996). This might represent an example of genetic assimilation (Pigliucci et al., 2006), because the phenotypic plasticity exhibited by lowlanders has become constitutively expressed in Tibetans. Conversely, reductions in metabolic capacity might be maladaptive in high-altitude birds, which need to sustain near-maximal rates of aerobic metabolism during flight. In the bar-headed goose, for example, aerobic capacity of the flight muscle is expanded by an increased proportion of oxidative fibers, whereas the inherent respiratory capacities, O₂ kinetics, and phosphorylation efficiencies of their mitochondria are unaltered (Scott et al., 2009a; Scott et al., 2009b). Human populations at high altitude have also evolved an enhanced preference for carbohydrate oxidation in the muscle and heart, along with a decreased propensity for anaerobic lactate production during exercise (Hochachka et al., 1991; Hochachka, 1998). Unlike the situation in lowlanders, these phenotypes exhibit very little plasticity in highland natives during acclimatization to hypoxia and might therefore represent another case of genetic assimilation. In addition, some traits could be enhanced in high-altitude species that are not plastic in lowlanders, such the creatine kinase shuttle (Scott et al., 2009b), whereas other traits remain unaltered, such as mitochondrial phosphorylation efficiency and O₂ kinetics (Scott et al., 2009a).

Another important aspect of O₂ utilization at high altitudes that has important consequences for cellular function, but which is not specifically related to matching O₂ supply and demand, is the production of reactive O₂ species (ROS). ROS production can increase at high altitudes if O₂ limitation to cytochrome c oxidase (COX), the enzyme that catalyzes the terminal reduction of O₂ in oxidative phosphorylation, causes a shift in mitochondrial redox towards a more reduced state (Aon et al., 2010). Although endogenous antioxidant systems exist to help prevent ROS from causing oxidative damage, the acclimatization response to hypoxia in lowland natives often involves a downregulation of antioxidant capacity (Dosek et al., 2007). With diminished protection, oxidative damage to various cellular components increases at high altitudes, as indicated by increased rates of lipid peroxidation and DNA strand breaks (Currie, 1999; Dosek et al., 2007).

Mechanisms to reduce ROS production or to protect against oxidative damage would be beneficial at high altitude. One possible means of reducing ROS production is to maintain a more oxidized electron transport chain (Brown et al., 2009). This could be the general strategy employed by bar-headed geese, as reflected by an alteration in COX enzyme kinetics (higher substrate affinity) that might allow the electron transport chain to operate at a lower redox state with less reduced-cytochrome c (Scott et al., 2010). This might be caused by a single amino acid substitution in subunit 3 of the protein that alters inter-subunit interactions (Scott et al., 2010). The importance of reducing ROS production in preventing oxidative damage is emphasized by studies comparing high- and low-altitude human populations: oxidative damage at high altitudes is generally less in highlanders even though most (but not all) antioxidant enzyme levels are reduced (Gelfi et al., 2004; Sinha et al., 2009a; Sinha et al., 2009b; Sinha et al., 2010). It is presently unclear how these evolutionary changes relate to the ancestral acclimatization response, but they should nonetheless be important for improving cell function at high altitudes.

Bullard (Bullard, 1972) and Monge and León-Velarde (Monge and León-Velarde, 1991) described a physiological syndrome characteristic of birds and mammals that are genetically adapted to high-altitude hypoxia. These animals are typically characterized by an elevated Hb–O₂ affinity, a normal or slightly increased Hct, thin-walled pulmonary vessels that respond to hypoxemia with moderate hypertension, and increased muscle capillary density. Some of these features are shared with burrowing mammals that are adapted to the chronic hypoxia of subterranean burrow systems (Hall, 1966; Jelkmann et al., 1981; Campbell et al., 2010). By contrast, in lowland-adapted animals, acclimatization to hypoxia typically involves increased erythropoeitic activity, decreased blood–O₂ affinity, pulmonary arterial hypertension, and numerous other responses that can potentially impair O₂ transport. These physiological responses can overburden the right side of the arterial circulation and can contribute to pathophysiological conditions such as chronic mountain sickness, suggesting that at least some features of the typical acclimatization response to hypoxia are maladaptive. Moreover, evidence suggests that many responses to hypoxia, such as polycythemia and reduced Hb–O₂ affinity, do not simply represent pathological overshoots of the phenotypic optimum; instead, the ancestral norm of reaction is actually pointed in the wrong direction.

In humans, the acclimatization response to severe hypoxia includes an increase in Hct (mediated by increased production of erythropoeitin) and a decrease in Hb–O₂ affinity (mediated by increased red cell concentrations of DPG and a reduced intracellular pH). For reasons explained above, both of these hematological adjustments are counterproductive under conditions of severe hypoxia. What accounts for this off-target response? One possible explanation is that the increased Hct and decreased Hb–O₂ affinity are part of a misdirected response to environmental hypoxia that originally evolved as a response to anemia (Hebbel et al., 1978; Storz, 2010). Environmental hypoxia and anemia both result in reduced levels of tissue oxygenation, but they have different root causes. In the case of anemia, reduced tissue oxygenation is caused by a curtailed oxygen transport capacity of the blood, and can therefore be rectified by increasing the Hct (which increases Ca_O2) and by decreasing Hb–O₂ affinity [which, under normoxia, enhances O₂ unloading without compromising O₂ uptake in the lungs (Brauner and Wang, 1997)]. In the case of environmental hypoxia, by contrast, reduced tissue oxygenation has an external cause: the reduced P_O2 of inspired air, which in turn leads to a reduced O₂ saturation of arterial blood. Under these circumstances, O₂ delivery might be further compromised by an elevated Hb concentration (due to the associated increase in blood viscosity) and a decreased Hb–O₂ affinity (due to reduced pulmonary O₂ loading that leads to a further diminution of arterial O₂ saturation). In humans and other lowland mammals that experience hypoxic stress at high altitude, the joint increase in Hb concentration and DPG/Hb ratio provide examples of maladaptive phenotypic plasticity, as the induced physiological changes are in the opposite direction of the phenotypic optimum under severe hypoxia.

Pulmonary vasoconstriction is another example of a misdirected response to environmental hypoxia, which probably evolved in terrestrial vertebrates to help match perfusion to ventilation. At low altitude, heterogeneity occurs throughout the lungs and can lead to some regions being slightly hypoxic relative to others. Pulmonary vasoconstriction in these isolated regions is beneficial because it causes preferential shunting of blood to the most oxygenated areas of the lungs. However, as described above, this is maladaptive when hypoxia occurs globally throughout the lungs, causing pulmonary hypertension, a diversion of blood flow away from the gas exchange surface to pulmonary shunt vessels (Lovering et al., 2008), and a thickening of the O₂ diffusion barrier due to fluid leakage into the air spaces (Maggiorini et al., 2001; Eldridge et al., 2006). A blunted pulmonary pressor response should therefore be advantageous for O₂ uptake in hypoxia and this benefit should outweigh the potential disruption to ventilation–perfusion matching.

In cases in which the acclimatization response to hypoxia is maladaptive, selection will favor an attenuation of the induced phenotypic change. This can give rise to a pattern of countergradient variation (Conover and Schultz, 1995; Grether, 2005) caused by selection on genetically based trait variation that offsets the environmentally induced changes. The blunted erythropoietic and pulmonary vasoconstriction responses to hypoxia in numerous high-altitude species and populations provide possible examples of this phenomenon.

Recent genomic surveys of nucleotide polymorphism in Tibetan highlanders revealed strong evidence for a history of positive selection at EPAS1 and other genes in the hypoxia-inducible factor (HIF) oxygen-signaling pathway (Beall et al., 2010; Bigham et al., 2009; Bigham et al., 2010; Simonson et al., 2010; Yi et al., 2010). The HIF family of transcription factors regulates systemic and cellular oxygen homeostasis by coordinating the transcriptional response to hypoxia. EPAS1 (also known as HIF2α) encodes the oxygen-sensitive α subunit of the heterodimeric HIF2 transcription factor and plays an especially important role in regulating erythropoeisis (Rankin et al., 2007). The studies by Beall et al. (Beall et al., 2010) and Yi et al. (Yi et al., 2010) revealed a striking signal of positive selection at EPAS1, and documented that noncoding nucleotide variants in and around the EPAS1 gene are strongly associated with a reduced Hb concentration in Tibetan highlanders. Previous studies demonstrated that Tibetan highlanders are characterized by low Hb concentrations relative to acclimatized lowlanders, as well as to Andean highlanders that are resident at comparable altitudes (Beall and Reichsman, 1984; Winslow et al., 1989; Beall et al., 1990; Beall et al., 1998; Garruto et al., 2003; Wu et al., 2005; Beall, 2007). Remarkably, Tibetans living at elevations of up to 4000 m present a hematological profile similar to what would be expected at sea level. Evidence suggests that this same blunted erythropoeitic response would be beneficial in other high-altitude human populations as well. According to Villafuerte et al. (Villafuerte et al., 2004): “...the overall symptomatic and physical improvement that comes from reducing Hct to sea level values while at high altitude suggest that Andean humans would be better suited for life at high altitude if they could maintain a [Hb] within the sea level range”. It is possible that Andean highlanders have not evolved a similar mechanism for attenuating the erythropoietic response to hypoxia because of their shorter history of residence at high altitude (Beall et al., 1998; Moore, 2001; Beall, 2006; Beall, 2007).

The Hb concentration of Tibetan highlanders does not fall outside the ancestral range of variation characteristic of closely related lowlanders, such as the Han Chinese. Instead, selection appears to have favored a blunted erythropoietic response such that Hb concentration at high altitude is maintained at the sea-level status quo. This represents a potential example of genetic compensation (Grether, 2005) in which a maladaptive acclimatization response requires selection on genetically based trait variation to restore the ancestral phenotype. Although the mechanism has yet to be uncovered, it is possible that noncoding changes that alter the expression of EPAS1 and other HIF-related genes have recalibrated the homeostatic set point for hypoxia-induced erythropoeisis in Tibetans. It remains to be seen whether Hb concentration represents the direct phenotypic target of selection in Tibetans, or whether changes in Hb concentration simply represent an ancillary effect of selection on some other physiological trait that is altered by regulatory changes in the HIF cascade (Zhuang et al., 1993; Moore et al., 2006).

The cellular regulation of Hb–O₂ affinity in high-altitude deer mice provides another potential example of how the acclimatization response to hypoxia might work at cross purposes with genetically based adaptation. Comparisons between highland and lowland deer mice have revealed genetically based differences in baseline DPG/Hb ratios (Snyder, 1982) and Hb–O₂ affinity (Snyder et al., 1982; Snyder, 1985; Chappell and Snyder, 1984; Chappell et al., 1988; Storz et al., 2009; Storz et al., 2010). Adult deer mice captured at high altitude showed a 14% reduction in DPG/Hb ratios after 65 days acclimation to sea level, and a 23% reduction after ∼230 days. Common garden experiments under normoxic conditions revealed that the highland mice were characterized by lower baseline DPG/Hb ratios than lowland mice, and experimental crosses revealed that these differences have a genetic basis (Snyder, 1982). Moreover, comparisons of Hb function between highland and lowland mice revealed that genetically based differences in Hb–O₂ affinity are largely attributable to a reduced sensitivity to DPG and Cl^– ions in the Hbs of highland mice (Storz et al., 2009; Storz et al., 2010). The reduced baseline DPG/Hb ratios and the suppressed anion sensitivity both appear to be adaptive mechanisms for maintaining an elevated blood–O₂ affinity and, yet, high-altitude deer mice still exhibit a hypoxia-induced increase in DPG levels. Why hasn't selection abolished the seemingly maladaptive DPG response in high-altitude deer mice? Perhaps for the same reason that Andean humans have not yet evolved a blunted erythropoeitic response to hypoxia. According to Snyder (Snyder, 1982): “...[it might be] that the physiologically mediated DPG response evolved not in the context of high-altitude hypoxia, but in response to other sorts of hypoxic stress encountered by animals living at low altitudes. By that line of thought, the DPG response in high-altitude deer mice is an inextricable feature of the animal's physiological architecture that cannot be modified without pleiotropic repercussions”.

Cases in which induced physiological responses require compensation by genetically based changes suggest that the ancestral acclimatization response might act as a hindrance to high-altitude adaptation. Such cases also suggest the possibility that many phenotypic similarities between highland and lowland populations of the same species could be attributable to the effects of countergradient selection.

Some of the best prospects for identifying adaptive mechanisms of hypoxia tolerance, and for teasing apart the relative roles of genotypic specialization and phenotypic plasticity, are afforded by genetic and genomic studies of population-level variation in species that are continuously distributed across steep altitudinal gradients. Although much previous work on high-altitude adaptation has focused on comparisons between separate species that have different altitudinal range limits, comparisons between highland and lowland populations of the same species offer several important advantages for identifying genetic mechanisms of adaptation. First, the ability to conduct crosses is crucial for disentangling genetic and environmental sources of phenotypic variation. Second, mechanistic studies of population-level variation provide the greatest experimental power for assessing genotypic differences in fitness-related physiological performance (Garland and Adolph, 1991; Storz and Wheat, 2010). Many mechanisms of physiological adaptation involve modifications of regulatory networks, and newly developed genomic technologies provide a means of investigating the nature of these modifications. In particular, the simultaneous analysis of sequence variation and expression profiles for a set of genes in the same pathway provides a means of identifying both structural and regulatory mechanisms of adaptation.

At the molecular level, phenotypic plasticity is associated with changes in transcriptional regulation (Gracey et al., 2008). Recent common garden experiments that were designed to quantify the proportion of transcriptomic variation that is due to environmental and genetic effects have demonstrated tremendous plasticity of gene expression in species with broad altitudinal distributions (Appenzeller et al., 2006; Cheviron et al., 2008). In rufous-collared sparrows (Zonotrichia capensis), hundreds of genes that are involved in aerobic metabolism and oxidative stress responses are differentially expressed in the skeletal muscle of birds native to high (>4000 m) and low (<2000 m) altitude (Cheviron et al., 2008). However, none of these transcriptional differences persists after one week of sea-level acclimation. These results demonstrate considerable plasticity in the transcriptomic profiles of rufous-collared sparrows, and suggest that many genome-wide expression differences between highland and lowland individuals are environmentally induced.

One especially promising genomic approach to evaluate the relative contributions of phenotypic plasticity and genotypic specialization in high-altitude adaptation is whole transcriptome shotgun sequencing (WTSS) or RNA-seq (Wang et al., 2009). RNA-seq uses next-generation sequencing to simultaneously characterize transcript abundance and sequence polymorphism in thousands of expressed mRNAs. This type of population genomic approach therefore holds much promise for identifying novel molecular mechanisms of phenotypic plasticity and genetic adaptation, particularly for loci whose function and adaptive significance were previously unanticipated. After candidate loci for hypoxia adaptation have been identified, it is then possible to conduct follow-up experiments to measure the physiological effects of altered protein function (in the case of amino acid mutations) or protein expression (in the case of regulatory mutations) (Feder and Walser, 2005; Storz and Wheat, 2010).

Global climate change has already been shown to impact the physiological tolerances and range limits of many organisms, and both altitudinal and latitudinal range shifts can be expected to continue as warming becomes more severe (Deutsch et al., 2008; Moritz et al., 2008; Huey et al., 2009; Sinervo et al., 2010; Somero, 2010). Populations that are restricted to high altitudes could be at risk of increased competition as lowland species shift their altitudinal ranges towards cooler environments, but this will depend on the capacity of lowlanders to acclimatize and/or adapt to high-altitude hypoxia. In birds and mammals that are native to lowland environments, certain features of the acclimatization response to hypoxia might constrain adaptation to high-altitude environments. This is because some of the induced physiological changes (e.g. polycythemia, increased erythrocytic phosphate/Hb ratios, and pulmonary vasoconstriction) are in the opposite direction of the apparent phenotypic optimum under chronic hypoxia. Insights into the acclimatization response of lowland animals to high-altitude hypoxia can provide a basis for predicting how altitudinal range limits might shift in response to climate change. When lowland animals colonize high-altitude environments, adaptive phenotypic plasticity can mitigate the costs of selection, thereby enhancing prospects for population establishment and persistence (Ghalambor et al., 2007). By contrast, maladaptive plasticity has the opposite effect by increasing the cost of selection, thereby diminishing the prospects of successful colonization.