European Respiratory Society


Patients with acute respiratory distress syndrome and high-altitude pulmonary oedema build up excess lung fluid, which leads to alveolar hypoxia. In patients with acute respiratory distress syndrome and hypoxia, there is a decrease in oedema fluid clearance, due in part to the downregulation of plasma membrane sodium–potassium adenosine triphosphatase (Na,K-ATPase).

In alveolar epithelial cells, acute hypoxia promotes Na,K-ATPase endocytosis from the plasma membrane to intracellular compartments, resulting in inhibition of Na,K-ATPase activity. Exposure to prolonged hypoxia leads to degradation of plasma membrane Na,K-ATPase.

The downregulation of plasma membrane Na,K-ATPase reduces adenosine triphosphate demand, as part of a survival mechanism of cellular adaptation to hypoxia. Hypoxia has also been shown to disassemble and degrade the keratin intermediate filament network, a fundamental component of the cell cytoskeleton, affecting epithelial barrier function.

Accordingly, better understanding of the mechanisms regulating cellular adaptation to hypoxia may lead to the development of novel therapeutic strategies for acute respiratory distress syndrome and high-altitude pulmonary oedema patients.


Edited by N. Weissmann

Number 2 in this Series

The alveolar epithelium contributes to the maintenance of surface tension, basic host defence properties, gas exchange and oedema clearance 1. It is normally well oxygenated since oxygen is exchanged across the alveolocapillary membrane. However, under a number of conditions, the alveolar epithelium is exposed to low oxygen levels (hypoxia). For example, during ascent to high altitude, the partial pressure of oxygen drops due to the decline in barometric pressure, which can contribute to high-altitude pulmonary oedema (HAPE) 2. In turn, HAPE exaggerates alveolar hypoxia as a consequence of alveolar flooding 3. In addition, patients with acute respiratory distress syndrome or congestive heart failure develop pulmonary oedema, resulting in impaired oxygen transfer from the airspaces into the pulmonary circulation 1, 4, 5. Patients who cannot clear oedema efficiently have worse outcomes, suggesting that hypoxia plays a deleterious role in alveolar epithelial function 1, 6. The mechanisms contributing to alveolar epithelial dysfunction during hypoxia are not completely understood. Recent advances on this topic are reviewed herein.


In the human lungs, the alveolar epithelium is populated by squamous alveolar type (AT)I and cuboidal ATII cells. There are similar numbers of ATI and ATII cells; however, ATI cells are elongated, with long thin cytoplasmic extensions, and cover 95% of the surface area 7, 8. ATII cells, which produce, secrete and recycle surfactant, cover the remaining 5% of the surface area and are thought to differentiate into ATI cells. Recent studies have demonstrated that both ATI and ATII cells express sodium–potassium adenosine triphosphatase (Na,K-ATPase) and the amiloride-sensitive epithelial sodium channel (ENaC), and thus contribute to active Na+ transport and alveolar fluid resorption 710.

Interactions between adjacent epithelial cells through domains such as tight junctions and adherens junctions provide a physical barrier between the alveolar airspaces and the interstitium (fig. 1). Tight junctions surround the cells like rubber O-rings, preventing large molecules from crossing the epithelial layer 11. The relative impermeability of the alveolar epithelium to paracellular solute diffusion is predominantly regulated by tight junctions 1214. Tight junctions consist of integral membrane proteins: occludins, claudins and junctional adhesion molecules 15, 16. Cytoplasmic plaque proteins transduce signals between tight junctions and cytoplasmic signalling molecules or the actin cytoskeleton 16, 17. In addition, tight junctions divide the epithelial plasma membrane into the apical and basolateral domains 15, 16. Ion transporters and other membrane proteins are asymmetrically distributed in these two domains 15, 16.

Fig. 1—

Schematic diagram of the alveolar epithelium. Tight (in orange) and adherens junctions (in red) on adjacent epithelial cells provide a restrictive barrier in order to maintain selective permeability. The asymmetrically distributed epithelial sodium channel (in green) and sodium–potassium adenosine triphosphatase (in light blue) confers the vectorial transport of sodium ions, which is crucial for alveolar fluid resorption. ↕: paracellular transport.

Alveolar fluid resorption is accomplished through active Na+ transport across alveolar epithelium 18, 19. As depicted in figure 1, Na+ is taken in on the apical surface of alveolar epithelial cells, primarily through ENaC. Subsequently, Na+ is actively extruded through the basolateral surface into the lung interstitium by the Na,K-ATPase, generating a transepithelial osmotic gradient. Water then follows the osmotic gradient into the interstitial space and pulmonary circulation, leading to the resorption of alveolar fluid 18, 19.

The apical surface of epithelial cells expresses the ENaC, which is a major Na+ transporter and widely distributed in the lung, kidney and colon 2023. It comprises three subunits (α, β and γ) and usually exists as a tetramer made up of two α-, one β- and one γ- subunit or a much larger complex containing three of each subunit 24, 25. Maximal transport requires the presence of all three subunits. ENaC-mediated ion transport is voltage-dependent, activated by calcium ions (at high doses) and inhibited by 1 μM amiloride 26. α-ENaC knockout mice die due to pulmonary oedema immediately after birth 27, whereas β- or γ-ENaC knockout mice show a compromised rate of fluid clearance 28, 29. The ENaC inhibitor amiloride does not completely inhibit Na+ transport in alveolar epithelial cells, suggesting the presence of amiloride-insensitive pathways that contribute to fluid resorption. Among these are cyclic nucleotide-gated cation channels, the Na+–glucose transporter and other co-transporters (e.g. Na+–amino acid) 25.

The Na,K-ATPase resides in the basolateral membrane of the cells. It utilises the energy released from adenosine triphosphate (ATP) hydrolysis to pump 3Na+ out of cells in exchange for 2K+ into cells, generating Na+ and K+ gradients across the plasma membrane 30. Na,K-ATPase is a heterodimer of α- and β-subunits, both of which are necessary for its activity 31. It is believed that the α- and β- subunits are synthesised independently and then assembled into a dimer in the endoplasmic reticulum and delivered to the plasma membrane 31, 32. The α-subunit is a transmembrane protein that catalyses ATP hydrolysis and contains the binding sites for Na+, K+ and the inhibitor ouabain 30. Four α-subunit isoforms, α1– α4, have been described 31. These isoforms are highly conserved, each containing a >77% identical primary amino acid sequence. The α1-isoform is found in most tissues, whereas the other isoforms are tissue-specific 30. Mice with deletion of either the α1- or α2-isoform do not survive, suggesting a fundamental role of the α1- or α2-isoform in organ development 33, 34. Loss-of-function mutations in the α2-subunit are associated with familial hemiplegic migraine type 2 3537, and missense mutation of the α3-subunit causes dystonia-Parkinsonism 38. The β-subunit has four isoforms and contains three glycosylation sites. It controls heterodimer assembly and insertion into the plasma membrane 30.


During hypoxia, cells respond to this stress through adaptive mechanisms 39, 40. One response is to increase the level of expression of genes responsible for angiogenesis, in order to provide more efficient blood flow, and of genes involved in glycolytic pathways 39. This regulation of gene expression during hypoxia is carried out by turning on the master transcription factor, hypoxia-inducible factor (HIF), a dimer consisting of the subunits HIF-α and -β 41. HIF-α is a short-lived protein, whereas HIF-β is constitutively expressed 42. Under normoxic conditions, oxygen- and iron-dependent prolyl hydroxylase hydroxylates HIF at prolines 402 and 564, two highly conserved amino acids within the oxygen-dependent degradation domain of HIF-α 43. In turn, von Hippel–Lindau protein (pVHL) recognises and binds to the prolyl-hydroxylation sites, targeting HIF-α for ubiquitination and eventual degradation in the proteasome 42, 4446. Under hypoxic conditions, prolyl hydroxylase activity is inhibited; therefore, HIF-α is stabilised, since it is neither hydroxylated nor degraded in the proteasome. After stabilisation, HIF translocates into the nucleus and activates downstream genes, such as vascular endothelial growth factor, erythropoietin, glucose transporter 1 and enzymes involved in the glycolytic cascade, to improve delivery of oxygen and glucose to cells 41, 42.

Cells can also adapt to hypoxia by maintaining ATP homeostasis 40. During hypoxia, insufficient oxygen limits ATP production through mitochondrial oxidative phosphorylation, and thus cellular ATP levels decrease 47, 48. In order to maintain ATP homeostasis, cells can either increase ATP production via anaerobic glycolysis or decrease ATP demands via inhibition of ATP-consuming enzymes, such as Na,K-ATPase and the protein synthesis machinery 40. Na,K-ATPase activity may account for ∼40% of ATP consumption in cells, dependant upon cell type 40, 49, 50. It has been reported that hypoxia decreases ATP demand by reducing the amount of plasma membrane Na,K-ATPase 51, 52.


Hypoxia has been shown to impair alveolar fluid clearance by inhibiting transepithelial active Na+ transport 53, 54. It has been demonstrated that exposure of rodents to hypoxia results in a significant decrease in alveolar fluid resorption, which is associated with a decrease in ENaC, as well as Na,K-ATPase, activity 5456. In cultured alveolar epithelial cells, hypoxia-mediated downregulation of Na,K-ATPase is time and oxygen concentration-dependent 52, 57, 58. Short-term exposure to hypoxia decreases Na,K-ATPase activity and its protein abundance at the plasma membrane without significant changes in its total amount, suggesting that endocytosis of Na,K-ATPase occurs during hypoxia 52.

The endocytosis of Na,K-ATPase during hypoxia appears to be mediated by mitochondrial reactive oxygen species (ROS), since ROS scavengers prevented the hypoxia-induced downregulation of Na,K-ATPase 52. Moreover, treatment with hydrogen peroxide is sufficient to cause both Na,K-ATPase endocytosis from plasma membrane and a decrease in Na,K-ATPase activity 52. The source of ROS in this process was further assessed in mitochondrial-DNA-depleted (ρ0) A549 cells 59. These cells lack a competent electron transport chain and are thus incapable of generating ROS during hypoxia 60. In ρ0 A549 cells, hypoxia failed to induce Na,K-ATPase endocytosis and decrease Na,K-ATPase activity. Furthermore, in rodents exposed to hypoxia, overexpression of manganese superoxide dismutase inhibited mitochondrial ROS production and blocked the decrease in Na,K-ATPase abundance and alveolar fluid resorption 54. Together, these studies suggest a role for mitochondrial ROS in the hypoxia-mediated endocytosis of Na,K-ATPase.

During hypoxia, mitochondrial ROS activate protein kinase Cζ, leading to Na, K-ATPase α1-subunit phosphorylation at serine 18 52. In addition, plasma membrane Na,K-ATPase was ubiquitinated; however, mutation of the four lysines surrounding serine 18 to arginine prevented Na,K-ATPase ubiquitination, implying that these are the ubiquitination sites 61. Mutation of serine 18 to alanine prevented ubiquitination and endocytosis of the Na,K-ATPase α1-subunit, suggesting that serine 18 phosphorylation is a prerequisite for these processes (fig. 2) 52, 61.

Fig. 2—

Hypoxia induces endocytosis and degradation of plasma membrane sodium–potassium adenosine triphosphatase (Na,K-ATPase; in purple). In alveolar epithelial cells exposed to hypoxia, generation of mitochondrial reactive oxygen species activates protein kinase Cζ, which phosphorylates (in red) plasma membrane Na,K-ATPase, leading to its ubiquitination (in light blue). This series of events triggers Na,K-ATPase endocytosis via clathrin (in yellow)-coated vesicles. Na,K-ATPase-containing endosomal vesicles merge to form the late endosome (LE) and Na,K-ATPase is degraded.

Endocytosis of Na,K-ATPase has been reported to be clathrin-dependent 62. In alveolar epithelial cells, hypoxia-induced endocytosis of Na,K-ATPase requires the binding of adaptor protein 2 to the tyrosine-based motif (tyrosine 537) located in the Na,K-ATPase α1-subunit, leading to the incorporation of Na,K-ATPase into clathrin vesicles 63. Trafficking of clathrin vesicles requires the actin cytoskeleton in mammalian cells 64. Activation of Rho GTPase leads to rearrangement of the actin cytoskeleton, thus regulating trafficking of clathrin vesicles 64. Rho GTPases can activate two types of actin nucleators that directly stimulate actin polymerisation and also regulate cofilin to affect actin reorganisation via Rho-associated protein kinase (ROCK)/p21-activated kinase/LIM domain kinase signalling 64. More recently, it has been suggested that hypoxia-mediated endocytosis of Na,K-ATPase is dependent upon the activation of RhoA/ROCK signalling and actin stress fibre formation 65. During hypoxia, mitochondrial ROS activate the small GTPase RhoA, leading to the formation of actin stress fibres in alveolar epithelial cells 65, and dominant negative RhoA and ROCK inhibitor prevent the hypoxia-mediated Na,K-ATPase endocytosis 65.

Prolonged hypoxia leads to the mitochondrial ROS-dependent degradation of plasma membrane Na,K-ATPase, and both the proteasome and the lysosome are involved in this process 51, 54. Since the ubiquitination of plasma membrane Na,K-ATPase is required for its endocytosis 61, it is likely that Na,K-ATPase ubiquitination acts as a signal for its endocytosis, stimulating its merger with the proteasome and/or the lysosome for degradation. In contrast, hypoxia did not alter the half-life of total pool Na,K-ATPase, suggesting that degradation of total pool Na,K-ATPase is not affected by hypoxia 51. A recent study reported that loss of pVHL prevented hypoxia-mediated degradation of plasma membrane Na,K-ATPase 66. Although this process required the pVHL E3 ligase activity, HIF stabilisation was not required, indicating a role for pVHL in hypoxia-mediated Na,K-ATPase degradation independent of HIF 66. This study suggests that pVHL may play a dual function during hypoxia: 1) hypoxia inhibits pVHL–HIF interaction, resulting in stabilisation of HIF and upregulation of HIF-responsive genes for glycolytic ATP production and better oxygen delivery; and 2) pVHL facilitates plasma membrane Na,K-ATPase degradation to decrease ATP demands through an HIF-independent mechanism. Taken together, these data suggest that hypoxia increases mitochondrial ROS generation and induces endocytosis and degradation of plasma membrane Na,K-ATPase, resulting in the inhibition of Na,K-ATPase activity and impaired alveolar fluid clearance.


Keratin intermediate filaments (IFs) are the major cytoskeletal component of epithelial cells and play a crucial role in maintaining the structural integrity of cells 67. Although it was originally thought that the IF is a static structure, accumulating data suggest that keratin IFs can undergo rapid deformation/displacement in epithelial cells in response to stress, suggesting that the IF cytoskeleton transmits mechanical signals from the cell surface to all regions of the cytoplasm 68.

Keratin IFs are assembled as heteropolymers of type I and type II IF proteins. Keratins consist of a conserved central α-helical domain, a non-α-helical N-terminal head and a C-terminal tail domain containing all of the known phosphorylation sites 69. Lung alveolar epithelial cells primarily express keratins K8 and K18 in equal amounts 70, 71. Phosphorylation of K8 and K18 promotes their depolymerisation and redistribution in vitro 72.

Exposure to hypoxia caused a time-dependent disassembly of K8 and K18, which was associated with an increase in phosphorylation of K8 73. In alveolar epithelial cells, the hyperphosphorylation and disassembly of keratin during hypoxia was mediated by mitochondrial ROS, which activate PKC and phosphorylate keratin, leading to disassembly of the keratin IF network 73.

In rats exposed to hypoxia, there was a significant decrease in keratin in alveolar epithelial cells compared to normoxic rats, due to degradation of keratin IFs 73, 74. Keratin was degraded in alveolar epithelial cells exposed to hydrogen peroxide, suggesting that ROS are necessary for the degradation of keratin 74. The proteasome inhibitor MG132 (carbobenzoxy-l-leucyl-l-leucyl-l-leucinal) prevented the degradation of keratin IFs in alveolar epithelial cells exposed to hypoxia, suggesting that keratin degradation was mediated via the ubiquitin/proteasome pathway 74. These data suggest that keratin IF dynamics may play a role in adaptation to hypoxia in alveolar epithelial cells.


Since downregulation of Na,K-ATPase contributes to alveolar epithelial dysfunction, its upregulation may improve epithelial function during hypoxia or other pathological conditions. Indeed, overexpression of the Na,K-ATPase β1-subunit via viral and nonviral gene transfer resulted in an increase in alveolar fluid resorption in normal and injured rat lungs 7579. These studies suggest a potential therapeutic application of Na,K-ATPase gene transfer in the improvement of epithelial function and lung oedema clearance 80, 81.

Administration of catecholamines, such as dopamine, terbutaline and isoproterenol, has been shown to be effective in increasing Na,K-ATPase activity and alveolar fluid resorption. Parallel to this, treatment with dopamine and isoproterenol was associated with increases in Na,K-ATPase protein abundance at the cell basolateral membrane 82. This increase in protein abundance was mediated by exocytosis of Na,K-ATPase from endosomal compartments into the basolateral membrane 83. More importantly, the impaired alveolar function of injured lungs can be reversed by administration of dopamine, terbutaline and isoproterenol in animal models of hypoxia, hyperoxia, increased left atrial pressure or ventilation-induced lung injury 53, 54, 84, 85. The data from these animal model studies led to the development of a clinical trial, in which patients receiving salbutamol, a β-adrenergic agonist, showed a significant reduction in extravascular lung water 86. Moreover, inhalation of salmeterol has been shown to prevent HAPE in susceptible subjects 87. These findings support the hypothesis that β-adrenergic agonists accelerate the resolution of alveolar oedema and may improve survival.


Epithelial dysfunction, in acute respiratory distress syndrome and high-altitude pulmonary oedema patients, is associated with increased oedema formation, and results in a hypoxic environment. Hypoxia contributes to the impaired lung fluid clearance via the downregulation of sodium–potassium adenosine triphosphatase, which regulates active sodium ion transport and alveolar fluid clearance. Hypoxia also results in disassembly and degradation of the keratin intermediate filament network, which may aggravate the epithelial dysfunction. β-Adrenergic agonists have been demonstrated to increase active sodium ion transport, suggesting that it may form part of the therapeutic armamentarium for acute respiratory distress syndrome and high-altitude pulmonary oedema patients.

Support statement

This study was supported, in part, by grants from the National Institutes of Health (Bethesda, MD, USA; HL-071643 and HL-48129) and a Parker B. Francis Foundation (Dept of Environmental Health, Harvard School of Public Health, Boston, MA, USA) fellowship (to G. Zhou).

Statement of interest

None declared.


The authors are grateful to A. Kelly and L. Welch (Division of Pulmonary and Critical Care Medicine, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA) for their critical review of this manuscript.


  • Previous articles in this series: No. 1: Wagner PD. The biology of oxygen. Eur Respir J 2008; 31: 887–890.

  • Received November 19, 2007.
  • Accepted November 27, 2007.


View Abstract