Lung Circulation

Abstract

The circulation of the lung is unique both in volume and function. For example, it is the only organ with two circulations: the pulmonary circulation, the main function of which is gas exchange, and the bronchial circulation, a systemic vascular supply that provides oxygenated blood to the walls of the conducting airways, pulmonary arteries and veins. The pulmonary circulation accommodates the entire cardiac output, maintaining high blood flow at low intravascular arterial pressure. As compared with the systemic circulation, pulmonary arteries have thinner walls with much less vascular smooth muscle and a relative lack of basal tone. Factors controlling pulmonary blood flow include vascular structure, gravity, mechanical effects of breathing, and the influence of neural and humoral factors. Pulmonary vascular tone is also altered by hypoxia, which causes pulmonary vasoconstriction. If the hypoxic stimulus persists for a prolonged period, contraction is accompanied by remodeling of the vasculature, resulting in pulmonary hypertension. In addition, genetic and environmental factors can also confer susceptibility to development of pulmonary hypertension. Under normal conditions, the endothelium forms a tight barrier, actively regulating interstitial fluid homeostasis. Infection and inflammation compromise normal barrier homeostasis, resulting in increased permeability and edema formation. This article focuses on reviewing the basics of the lung circulation (pulmonary and bronchial), normal development and transition at birth and vasoregulation. Mechanisms contributing to pathological conditions in the pulmonary circulation, in particular when barrier function is disrupted and during development of pulmonary hypertension, will also be discussed.

Introduction

The pulmonary vasculature is unique, both in volume and function. During fetal life, the pulmonary circulation is a low flow, high-resistance circuit. With transition to postnatal life, the pulmonary vasculature dilates, accommodating the entire cardiac output (CO), with high blood flow maintained at low intravascular pulmonary arterial pressure (P_PA). As compared with the systemic circulation, pulmonary arteries have thinner walls with much less vascular smooth muscle and a relative lack of basal tone, likely a function of high production of endogenous vasodilators and low production of vasoconstrictors, resulting in a normal pulmonary vascular resistance (PVR) that is approximately one-tenth that of the systemic circulation. In the adult lung, factors controlling pulmonary blood flow include vascular structure, gravity, mechanical effects of breathing, and the influence of neural and humoral factors.

While autoregulation is a well-recognized feature of most systemic vascular beds, this phenomenon is absent in the adult pulmonary circulation. The pulmonary circulation also differs functionally from the systemic in that the pulmonary arteries carry mixed venous blood. From the pulmonary arteries, deoxygenated blood is channeled through the alveolar/capillary units where a large component of gas exchange occurs, and returned to the left heart by the pulmonary veins for distribution to the systemic circulation. The pulmonary vascular pressor response to hypoxia is also unique, as systemic arteries dilate as oxygen concentrations decrease. The acute pressor response occurs immediately and is rapidly reversible with return to normoxic conditions. If the hypoxic stimulus is maintained over a prolonged period, as can occur with various chronic lung diseases, contraction is accompanied by remodeling of the vasculature, resulting in elevated PVR and pulmonary hypertension, a potentially devastating disease. In addition to hypoxia, other genetic and environmental factors can also confer susceptibility to development of pulmonary hypertension even in the absence of a hypoxic stimulus.

In addition to gas exchange, the pulmonary vasculature also serves to filter the blood, removing microemboli and participates in the metabolic regulation of a variety of vasoactive hormones. The endothelium forms a tight barrier, actively regulating paracellular extravasation of proteins, solutes and fluids to control interstitial fluid homeostasis. Infection and inflammation compromise normal barrier homeostasis, resulting in increased permeability, fluid, protein and cellular extravasation, edema formation, and ultimately acute respiratory distress syndrome (ARDS).

The lung also has a systemic vascular supply, the bronchial circulation, which provides oxygenated blood from the systemic circulation to the walls of the conducting airways, pulmonary arteries and veins. The historical concepts (233) and anatomy (635) of the pulmonary circulation, comparative physiology (686), effects of exercise (445) and high altitude (233), the mechanics (172) and distribution of blood flow (217, 220), and mechanisms matching ventilation and perfusion (161,172,220) have been covered extensively in previous reviews. In this article, we will focus on reviewing the basics of the lung circulation (pulmonary and bronchial), normal development and transition at birth, and vasoregulation. We will also discuss the mechanisms contributing to pathological conditions in the pulmonary circulation, in particular when barrier function is disrupted and during development of pulmonary hypertension.

Anatomy—A Brief Overview

Any discussion of the pulmonary circulation requires at least a rudimentary understanding of the anatomy of the vasculature. A thorough review detailing the structure of the pulmonary vessels, including wall composition, has been published recently (635); herein we provide a brief overview to orient the reader.

In general, normal embryogenesis gives rise to a pulmonary circulation with an arterial network that closely parallels the airways, a capillary network at the level of the alveoli, and an irregular venous system that drains to the left heart. In the postnatal lung, blood exits the right ventricle into the main pulmonary artery, the diameter of which is similar to that of the aorta (<3 cm) (440). During gestation and immediately following birth, the wall thickness of the pulmonary artery is nearly identical to that of the aorta, but, postnatally, the elastic tissue gradually diminishes resulting in a pulmonary artery wall that is much thinner than the aorta [reviewed in (202)].

The main pulmonary artery divides into the right and left main branches, with the caliber of each branch approximately half that of the main pulmonary artery (440). The right and left branches further divide to supply each lobe before entering the lung. Within the lung, each lobar artery subdivides into rather irregular branches corresponding to the bronchial tree. The close proximity of the pulmonary arteries and airways under-scores the relationship between ventilation and perfusion that defines the normal function of the lung. The large pulmonary arteries (>2000 μm o.d.) are classified as elastic, with the media comprised primarily of elastic fibers and some smooth muscle (523,635). As vascular diameter decreases, these elastic arteries gradually give rise to vessels with increased smooth muscle content (523). In general, arteries between 2000 and 150 μm o.d. can be considered muscular arteries, but are still more thin-walled than systemic arteries of the same diameter (155, 523, 635). The pulmonary arterial wall also comprises collagen and advential fibroblasts, a longitudinal elastic lamina, which allows for expansion during inspiration and a thin intima composed of endothelial cells (ECs).

The small arterioles have a nonuniform smooth muscle cell (SMC) layer, giving way to the small nonmuscular preacinar arterioles, which are located proximal to terminal bronchi (264, 418, 523, 524). At the alveoli, the terminal arterioles break into a network of pulmonary capillaries within the alveolar walls. Although it had long been surmised that blood passed from the arterial to venous circulation via small vessels not visible to the eye or with traditional light microscopy [reviewed in (687)], the use of electron microscopy in the early 1950s allowed the first visualization of the pulmonary capillaries (375). The capillaries have a very thin wall consisting of a single layer of ECs (210,687) and, at an estimated 126 m² in the adult human (210), contain most of the surface area of the pulmonary vasculature. It should be noted that pulmonary capillaries are not uniform in thickness (267, 635, 678). For example, the gas-exchange portion of the intra-acinar alveolar capillaries wall can thin to essentially just a bit of cytoplasm between the two sections of plasma membrane, resulting in a thickness of 20 to 30 nm (635,678) with the cell contents and nucleus shifting to the “thick” side of the septum.

Gas exchange between the alveolar gas and blood takes place largely within the pulmonary capillary bed after which the blood flows into venules, which are almost indistinguishable in gross structure from arterioles (635). It should be noted, however, that oxygen uptake has been proposed to also occur across the walls of larger arteries (118, 297, 586), a notion recently confirmed by elegant in vivo imaging techniques (611). While each small arteriole supplies a specific unit of respiratory tissue, the venules drain several portions of the lung. Venules do not follow the bronchiole tree and unite to form the pulmonary veins, which conduct the oxygenated blood into the left ventricle.

Anatomically, vessels can be categorized as extrapulmonary or intrapulmonary. Extrapulmonary vessels are outside the lung, in the mediastinum, and the pressure surrounding these vessels is related to pleural pressures. The intrapulmonary vessels, including the capillaries, can be further subdivided anatomically and functionally into extra-alveolar or alveolar. Alveolar vessels are comprised of capillaries that are adjacent to alveoli, while extra-alveolar vessels are larger and more proximal. These two types of vessels also have different perivascular pressures due to the differences in the tissues that surround them, and are exposed to alveolar pressure (P_ALV) or pressure exerted by lung tissue connections. Both of these forces increase with lung inflation; however, they act in opposite directions, with P_ALV directed inward and tissue pressure directed outward. Thus, the alveolar vessels are surrounded by P_ALV, due to localization within the septal wall, whereas the extra-alveolar vessels are surrounded by septa and are typically contained in a sheath of connective tissue. These vessels include a subset of the pulmonary capillaries, called the corner vessels, which are located in the alveolar parenchyma at the alveolar corners. Detailed morphologic measurements of the alveolar vessel density, radius, path length from artery to vein, and other measurements have been made and are discussed in detail elsewhere (635).

Fetal and Neonatal Pulmonary Circulation

Vascular morphogenesis

In humans, during fetal development the pulmonary circulation begins to form during the embryonic stage (0–7 weeks) from the truncus arteriosus (332), which originates from the ventricles and subsequently divides into the ascending aorta and pulmonary trunk. The arteries and veins continue to develop during the pseudoglandular stage (7–17 weeks), following the branching patterns of the airways (248, 261). By the 17th week of gestation, development of the pulmonary circulation down to the preacinar pulmonary arteries is complete (202,261,262) (Fig. 1).

Figure 1.

Timeline of vascular development. Adapted, with permission, from Hislop and Pierce (265).

During organogenesis, vascularization can occur via vasculogenesis or angiogenesis, or a combination of the two. Immunohistochemical analysis of fetal lung specimens to characterize the temporal and spatial expression of endothelial markers suggested that intrapulmonary arteries in humans arise primarily via vasculogenesis, or the development of new blood vessels from endothelial progenitor cells (248). In these studies, endothelial tubes containing blood cells could be identified in the mesenchyme surrounding the airways and the foregut (248). In mice, however, a combination of vascular casting and electron microscopy was used to determine that angiogenesis, or sprouting of new vessels from existing ones, was a more prominent process (137). Examination of lung sections from mice expressing the bacterial lacZ gene under control of an endothelial promoter also revealed a major role for angiogenesis (475), although in this case, instead of deriving from more proximal vessels, sprouts were proposed to arise from the capillary network and remodel (i.e., gain smooth muscle) to form the more proximal vessels. Possible explanations for the apparent contradictions in reported results between studies include differences in the techniques (i.e., casting versus immunohistochemical visualization of structures) and terminology employed. Nonetheless, the exact contribution of vasculogenesis and angiogenesis to proximal vessel development remains unresolved and a topic of ongoing investigation.

As the pulmonary arteries and veins develop, the primitive vessels acquire a smooth muscle layer and extracellular matrix proteins are deposited to form a basement membrane that both separates different layers within the vascular wall and maintains the wall structure. In humans, the SMCs surrounding the arteries appear to originate from three sites (248). First, between 38 to 98 days gestation, pulmonary arterial SMCs (PASMCs) derive from the bronchial smooth muscle in adjacent airways. During this time period, PASMCs also appear to derive from the mesenchyme surrounding the arteries. Later in gestation (98–140 days gestation), PASMCs coexpressed smooth muscle specific α-actin and endothelial markers, suggesting endothelial to SMC transdifferentiation. In contrast, venous SMCs derive from the mesenchyme (247).

The pulmonary capillaries appear to arise separately from the arteries and veins (137), becoming connected to the more proximal vasculature during the pseudoglandular stage (136, 248). As the fetus enters the canalicular stage, the capillaries increase in number, a process thought to involve a combination of angiogenesis, vasculogenesis and intussesception, or formation of pillars within existing capillaries to increase the surface area (84,136,137,475).

At the time of birth, the branching pattern of the pulmonary circulation is similar to that observed in the adult (Fig. 2) (265), although the size of the vessels increases with lung growth. Since the majority of alveolarization of the human lung continues after birth, the postnatal capillary bed continues to expand via angiogenesis and intussesception, with the surface area for gas exchange increasing approximately 20-fold from birth to adulthood (83).

Figure 2.

Arteriogram of a human lung obtained postmortem from a fetus (39 weeks gestation) showing distal vascularization. The arteries were injected with barium sulphate (mag × 0.75). Reprinted, with permission, from (265).

Factors regulating vascular development

Growth factors

While histological studies provided early details regarding vascular morphogenesis, more recent studies utilizing transgenic animals have yielded important clues as to the molecular aspects of vascular development. It is now recognized that a number of factors contribute to the development of the pulmonary vasculature, including transcription factors, growth factors and the extracellular matrix. One of the most widely studied angiogenic factors, vascular endothelial growth factor (VEGF), plays a central role in vascular development through receptor/ligand interactions with both VEGFR1 (Flt-1) and VEGFR2 (KDR/Flk-1) receptors. Mice with global heterozygosity for the Vegf allele exhibit defective vascularization and embryonic lethality (89). Similarly, mice with global knock-out of VEGFR2 are embryonic lethal with poor vascular formation (562) while VEGFR1-deficient mice die in utero due to lack of structural organization of vessel walls (178). Thus, both VEGF receptor subtypes are required for vascular development, with VEGFR2 and VEGFR1 controlling angiogenesis and vascular organization, respectively.

Early in development, expression of both VEGF and VEGFR2 are observed in the rat lung, with VEGF localized to epithelium (52,209), and VEGFR2 found primarily in the mesenchyme (52, 209). Moreover, transgenic mice with conditional inactivation of the Vegf gene in lung epithelium exhibited almost complete absence of pulmonary capillaries (707). These genetic studies coupled with observations on the pattern of VEGF/VEGF receptor localization suggested an important interaction between tissues in regulating vascular development.

In addition to VEGF, the glycoprotein growth factors, angiopoietins (Ang 1, 2, 3, and 4), also play a major role in vascular development. These proteins signal through their major receptor, Tie-2, to modulate survival and migration of ECs, vascular remodeling and vascular integrity (127, 373). Indeed, loss of either Ang 1 (604) or Tie-2 receptors (545) is embryonic lethal due to substantial microvascular defects leading to hemorrhage and edema. The immature and improperly organized vessels in these mice suggest that the normal role of Ang 1/Tie-2 is to facilitate branching, sprouting and remodeling. In mice, Ang 1 is produced by lung mesenchyme and smooth muscle, whereas Tie-2 expression is restricted to endothelium (545, 553). The expression of Ang 1 and Tie-2 in mouse lung is measureable early on (E9.5) and increases throughout gestation (116, 236), although the spatial expression was not determined in these studies.

The Wnt family of proteins bind to receptors of the Frizzled family on the cell surface, primarily signaling through a canonical pathway involving β-catenin (352). Wnt proteins are clearly involved in vascular development (115). With respect to the lung, the isoform Wnt7b has been shown to influence maturation of vessels. Mice with null alleles for the Wnt7b gene died perinatally from respiratory failure, with lungs that were poorly inflated and hemorrhagic (576). While the endothelium appeared to develop normally, the large pulmonary vessels were dilated, exhibited defective SMC layers and herniation of the endothelial lining. These observations suggest that Wnt7b may regulate the development and/or recruitment of the mural support cells during maturation of the lung vessels.

The transforming growth factor-β (TGF-β) superfamily consists of over 40 secreted cytokines that signal through serine/threonine kinase receptors. TGF-β family members, including TGF-β1 and bone morphogenetic proteins (BMPs), as well as their receptors, are expressed in the developing lungs (204, 289, 422, 491, 613, 734), but their roles in lung vascular development have not been completely defined. TGF-β signaling is important in the development of many tissues, including the vasculature (142), with complete loss of TGF-β1 resulting in 50% prenatal lethality associated with inadequate capillary tube formation and impaired endothelial differentiation. Loss of TGF-β type-I receptors also results in embryonic lethality due to vascular defects (344). Mice deficient in Smad1 or Smad5, the main signaling intermediaries downstream of the BMP receptors, die midgestation due to vascular abnormalities including enlarged blood vessels with insufficient surrounding vascular SMCs (712).

Other potential factors that are known to be involved in vascular morphogenesis include notch, platelet-derived growth factor, angiostatic proteins, ephrins, insulin-like growth factors, and endothelial monocyte activating polypeptides [reviewed in (202, 485, 592)]. While there is ample evidence to suggest that these factors could contribute to lung vascular development, confirmation of their exact roles has yet to be reported.

Transcription factors

The regulation of many of the aforementioned pathways is coordinated by transcription factors. For example, Forkhead box (Fox) proteins are a family of transcription factors that play diverse roles in development due to their control of regulation of cell cycle progression, cell survival, expression of differentiated genes, and cell metabolism (88). Mice with complete loss of the Foxf1 transcription factor died midgestation (396), whereas heterozygous mice developed further but showed a variety of foregut developmental defects, including malformed lungs, impaired lung vascular development and perinatal lethality (310, 395). In the lungs of these mice, distal lung capillary density was reduced, and tight junctions between endothelial and distal lung epithelial cells were disrupted, leading to hemorrhage (310). Foxf1 regulates a variety of genes and is expressed in the mesenchyme during lung branching morphogenesis [reviewed in (392)]. Although the exact Foxf1 targets that modulate lung vascular formation have not been identified, lungs from Foxf1^−/− mice exhibited reduced VEGF expression (310), suggesting a potential mechanism by which Foxf1 may regulate pulmonary vascular development.

Hundreds of genes are controlled by activation of hypoxia-inducible factors (HIFs), oxygen-sensitive transcription factors. HIFs are highly conserved transcription factors that are tightly regulated by oxygen availability. The first family member identified (560), HIF-1, is found in all nucleated cells and exists as a heterodimer, consisting of HIF-1α and HIF-1β subunits. HIF-1β is ubiquitously expressed, whereas HIF-1α is found at very low levels under normoxic conditions. When oxygen levels are normal, HIF-1α protein is ubiquitinated and subjected to proteasomal degradation. This process is interrupted during hypoxia [reviewed in (505)], allowing for rapid accumulation of HIF-1α protein and activation of the transcriptional complex. Thus, HIF-1α confers sensitivity and specificity for hypoxic induction of HIF-1 transcriptional activity. Subsequently, HIF-2α was identified (156). Like HIF-1α, HIF-2α is regulated in a similar manner by oxygen and dimerizes with HIF-1β, but HIF-2α exhibits a much more restricted pattern of expression. During fetal development, the lung of the normal embryo is hypoxic; thus, it is not surprising that HIFs would be important contributors to lung vascular development. HIF mRNA and protein levels are quite high in the fetal lung (23, 232, 234), with the entire HIF system in place as early as 8 weeks of gestation in the human (232). Analysis of the spatial pattern of expression in human lungs during initial development of the pulmonary vasculature revealed high levels of HIF-1α protein localized predominantly in branching epithelium (232). HIF-2α protein was also present in epithelium, as well as in mesenchymal structures (232). The different patterns of expression suggest the paralogs may serve distinct and specific functions in epithelial and vascular morphogenesis, a possibility supported by studies in HIF-1α and HIF-2α loss-of-function models. In mice, global deletion of HIF-1α or HIF-1β results in severe cardiovascular malformations and embryonic lethality at day 10 (117, 295, 331, 399). Loss of HIF-2α, on the other hand, leads to death in approximately 50% of embryos, with the viable offspring exhibiting impaired lung development, reduced production of surfactant, postnatal respiratory distress, and neonatal lethality (80, 117). Interestingly, animals heterozygous for either HIF-1α or HIF-2α develop normally and survive to adulthood, with outwardly normal lung vasculature and function under normoxic conditions. In late development, fetal lung levels of HIF-1α and HIF-2α protein are normally high, with a rapid decrease in HIF-1α protein levels upon delivery (23,234), and in utero depletion of HIF-1α using antisense oligonucleotides reduced lung branching morphogenesis and vascularization (645). Similarly, a decline in HIF-1α and HIF-2α protein levels in the lungs of mechanically ventilated preterm animals is associated with vascular and alveolar hypoplasia, neonatal respiratory distress, and bronchopulmonary dysplasia (23,234). Taken together, these studies point to a central role for HIFs in pulmonary vascular morphogenesis.

Blood flow in the fetal lung

During gestation, the placenta serves as the primary gas-exchange organ for the fetus; thus, the fetal lung receives modest blood flow. Rather than traversing the pulmonary circulation, oxygenated blood crosses from the right atrium to the left atrium through the foramen ovale. The majority of blood pumped out by the right ventricle returns to the aorta via the ductus arteriosus, a wide muscular vessel connecting the pulmonary arterial trunk to the descending aorta (325, 519). In the human, blood flow to the lung increases from 10% to 15% of the combined CO at 20 week gestation (421, 519) to 25% at 30 weeks (519). Between 30 and 38 week gestation, fetal lung blood flow decreases to 21% of CO (519). Slightly lower values have been reported in fetal lamb (538), with the lung receiving <10% of the CO. As the fetal lamb matures from midgestation to near term, P_PA and aortic blood pressure increase in parallel from 30 to 50 mm Hg, while pulmonary blood flow increases approximately 40-fold and PVR decreases from 6 mm Hg/mL*min to 0.3 mm Hg/mL*min (539). Similar increases in pressure were measured in the right and left ventricles of the human fetus (305). In both the lamb and human, it is likely that the early increase in blood flow at midgestation is due to expansion of the pulmonary vascular bed since reactivity of the pulmonary circulation during this period is largely unaffected by increasing maternal oxygenation (518). Nearer term, both the pulmonary arteries and veins actively respond to vasoactive factors (86,93,128,351,735), suggesting the development of vasoreactivity during this period and that at this stage, high PVR is maintained by a combination of the thick SMC layer surrounding the vessels and significant release of vasoconstrictor substances. Interestingly, evidence in lambs also suggests that the fetal pulmonary circulation possesses a myogenic response that contributes to high PVR by opposing vasodilatory stimuli (5,47,603).

Postnatal vasodilation

In utero, the oxygen tension in pulmonary arterial blood is approximately 18 mmHg and oxygen saturation 50% (202). In this low oxygen fetal environment, the high levels of endogenous vasoconstrictors, high myogenic tone and low basal release of vasodilators such as nitric oxide (NO) and prostacyclin (PGI₂) contribute to the high PVR. Toward the end of gestation, increasing oxygen tension from 24 to 46 mmHg increases pulmonary blood flow by 10-fold in lambs (432) and maternal hyperoxygenation reduces PVR in the human fetus (518, 519) and lamb (329), indicating that oxygen concentration is a major factor regulating vasomotor tone in the mature fetus. Similar results have been reported in primates (46).

After birth, the lung becomes the organ of gas exchange. As breathing commences, pulmonary blood flow increases, reaching total CO in the human infant (519). During the transition from fetal to neonatal life P_PA gradually decreases, and within hours approaches 50% of systemic pressure (158,540). As the systemic vascular resistance and pressure become greater than P_PA, the foramen ovale closes. The ductus arteriosus begins to close within the first few hours after birth (436), closing completely by 2 days after birth (280). In the normal infant, P_PA reaches adult levels within the first 2 weeks of life (158,540). During maturation, the myogenic response is also lost.

The rapid postnatal reduction in P_PA can be attributed to a reduction in PVR, achieved primarily via marked vasodilation. Initiated by ventilation, oxygenation, and shear stress due to increasing blood flow, a variety of vasorelaxant mediators are produced, including NO and PGI₂, in conjunction with falling levels of vasoconstrictor agents such as endothelin-1 (ET-1) (202, 257, 701). The initial, immediate changes in PVR that occur with the onset of breathing are followed by postnatal vascular reorganization and remodeling that occur rapidly after birth and continue over the first few months of life. The factors controlling the changes in vasomotor tone that accompany the normal transition of the pulmonary circulation at birth, including mechanical factors and vasoactive agents, have recently been reviewed in detail (202).

The Adult Pulmonary Circulation

To effectively distribute CO to the vascular beds of various organs that possess unique requirements with respect to blood flow and, in humans, are separated by vertical distances up to 180 cm, the systemic circulation requires a high-pressure gradient. Thus, the vascular resistance of the systemic vessels is precisely controlled to maintain adequate pressure. While baroreceptors monitor pressure and utilize neural pathways to regulate intravascular pressure during changes in posture, physical activity, temperature, and blood volume, it is well recognized that the systemic vasculature also exhibits substantial autoregulation. This is in stark contrast to the adult pulmonary circulation, which is required to accommodate the total CO and, thus, cannot actively adjust its total blood flow. Instead, the adult pulmonary circulation is a low-resistance, highly compliant vascular bed whose properties are greatly influenced by external forces such as P_ALV, CO, and posture.

As noted earlier in this review, the pulmonary circulation can be broadly divided into extra-alveolar, alveolar, arterial and venular segments. In addition to serving as a conduit for blood and, in the case of the microvasculature, gas exchange, each segment of the pulmonary vasculature is unique in its responses to hemodynamic stress, and is subject to a specific set of luminal and perivascular pressures due to the anatomic differences between the segments. These differences in turn produce differential responses to mechanical events such as lung inflation. Given the existing regional inhomogeneities in structure, mechanical forces, ventilation, and gas exchange, the design of the cardiorespiratory apparatus provides for local differences in blood flow and compensatory mechanisms for rearrangement that match blood flow to alveolar ventilation insofar as is possible.

General aspects of pulmonary hemodynamics

As might be expected, the pattern of blood flow in the pulmonary circulation varies with vessel caliber. In earlier studies, flow in the pulmonary artery was noted to be borderline turbulent with flow velocities approximating 18 cm/s (174). Newer methods of blood flow imaging, using magnetic resonance imaging, demonstrated similar average flow velocities (22 cm/s) at rest in supine healthy volunteers, but complex blood flow patterns in these large vessels (26). Average and peak flow velocities increased with exercise, although the magnitude of these changes may be position dependent. Exercise is also associated with increased flow and capillary recruitment, but the effect of exercise on hemodynamic parameters of the pulmonary circulation also depends on exercise intensity and degree of hypoxia [as reviewed in (445)].

Effects of posture on hemodynamics

Starting in the 1960s, heterogeneity of lung perfusion has been investigated in a variety of species and using multiple methods. It is now widely recognized that pulmonary flow distribution is affected by numerous factors including local/regional changes in flow and other nongravitational factors, such as genetic determination of pulmonary vascular branching and changes in alveolar capillary perfusion that are independent of alveolar and arterial pressures (the mechanisms of which are yet unclear), as discussed in detail by Glenny et al. (220). Briefly, in the upright lung, the net result of forces that increase flow (gravity) and decrease flow (regional areas of shunting, areas of high P_ALV) is an overall increase in blood flow in the bases compared to the apex, a finding that has been shown using radiotracers as well as multiple newer methods, including single photon emission tomography and positron emission tomography (278). Regardless of the vertical effects of gravity in the prone versus supine position, the presence of isogravimetric gradients suggest that factors unrelated to gravity likely also play a role in perfusion heterogeneity (15, 217, 244, 306). While earlier studies focused on changes in perfusion between standing and supine subjects (688), there has been renewed interest in the effects of the different types of horizontal posture (supine vs. prone) on ventilation and perfusion due to the clinical benefit of prone positioning reported in patients with ARDS (275); however, results of the published studies have been somewhat contradictory. Multiple animal studies have shown changes in distribution of flow between supine and prone position in mechanically ventilated animals, even in the absence of additional parenchymal injury (12, 438). In contrast, use of PET imaging in healthy, nonsmoking adults on no ventilatory support revealed flow gradients favoring dependent regions in both supine and prone positions, but did not detect appreciable changes in flow between positions (444). Similar findings were noted in healthy nonsmokers not on mechanical ventilation, where regions of interest examined on MRI images also showed similar gradients in both positions (602). Interestingly, SPECT imaging (with microaggregate albumin as tracer) on healthy adults during application of continuous positive airway pressure (CPAP) showed more uniform perfusion in the prone position but only when CPAP was not used, suggesting that the presence of positive ventilation further complicates existing postural perfusion gradients (439). The finding of more uniform perfusion in the prone position in ventilated and nonventilated healthy humans was also corroborated by other studies (256, 463). In addition to the aforementioned reports, several other publications have also shown opposing results. These studies and potential reasons underlying the conflicting data has been discussed by Hughes et al. (277).

The studies described above were performed in healthy adults; however, the possibility exists that the effects of prone position on lung perfusion may differ in the diseased lung. Positional effects are of most interest in ARDS, where positive pressure and changes in CO may exert more significant changes in perfusion compared to healthy lungs (245,254). For example, in studies where CT scans were performed in patients with ARDS, prone position had salutary effects on alveolar recruitment and improved stress and strain, which may lead to more homogeneous perfusion (205, 490). Positional studies in animal models of lung injury also showed benefits to prone positioning (144, 341, 530). In dogs undergoing oleic acid injury, more uniform distribution of perfusion was noted in the prone position before injury and redistribution of perfusion away from dependent regions in prone but not supine position following injury (693). These results are consistent with studies in multiple animal models of injury where, species differences between animals studied and between quadrupeds and humans notwithstanding, beneficial alterations in perfusion in the prone position have been reported (661, 686). Taken as a whole, it would appear that whether or not vertical gradients of flow exist and are changed in the prone position independent of redistribution of lung parenchyma and changes in oxygenation in healthy humans, in humans with diseased lungs, aeration, and compliance improve with prone positioning (238,490,530). Thus, clinical improvement observed with prone position in pathologic states such as ARDS may reflect increased lung perfusion distribution due to changes in tissue distribution (495) and/or mechanics.

Effects of respiration on blood flow

Owing to low intravascular pressure, surrounding pressures exert a substantial effect on pulmonary vessel caliber. As noted earlier, alveolar vessels are highly influenced by P_ALV since changes in P_ALV produce similar (but not equivalent) changes in alveolar vessel transmural pressures (the difference between the pressure inside and outside the vessel). Thus, these vessels undergo compression at higher P_ALV. Blood flow control in alveolar vessels can be independent of changes in upstream pressures, and may be regulated by local factors (672). In addition to the effects of P_ALV, the connection of these vessels to lung parenchyma subjects them to substantial tissue forces.

Extra-alveolar vessels are, by definition, not affected by changes in P_ALV; however, changes in lung inflation also influence transmural pressures in these vessels. For instance, with inflation, extra-alveolar vessels dilate. This interdependence between lung volume and extra-alveolar vessel diameter is related to the effects of inflation on the perivascular interstitium surrounding these vessels (48), which is composed of loose parenchymal tissue, collagenous fibers and lymph vessels. Inflation deforms the surrounding tissue to produce changes in transmural pressures that distend extra-alveolar vessels independent of luminal vascular pressures (340). This effect is minimal at functional residual capacity, but increases with increasing lung volumes. Though individual responses of alveolar and extra-alveolar vessels to changes in lung volume are different, the net effect of lung inflation on PVR is predictable (Fig. 3). Resistance is lowest at normal breathing and increases with both increased and decreased lung volume due to the differential effects of lung volume on the diameters of alveolar and extra-alveolar vessels.

Figure 3.

Graph illustrating the relationship between PVR and volume. Adapted, with permission, from (581).

Since perfusion of alveolar vessels is dependent on P_PA, pulmonary venous pressure (P_V) and P_ALV, the lung is described as being divided to three major functional zones based on these pressures (Fig. 4). In Zone 1, P_ALV exceeds P_PA and, as a consequence, alveolar vessels are collapsed. In this situation, the V/Q ratio approaches infinity. The behavior of the lung vasculature under varying venous and alveolar pressures has been described using the principles of starling resistors (described below). However, alternate models have also been proposed to explain zonal blood flow (172); for instance, Fung et al. (191) have modeled the lung vasculature using a sheet-flow framework, where blood flow is envisioned to occur in between two alveolar sheets that are subject to elastic deformation due to changes in alveolar pressures.

Figure 4.

Original three-zone model proposed by West et al., (688) to illustrate the regional heterogeneity in blood flow. Within each of the three zones, the behavior of the blood vessels is different, based on the relative magnitudes of pulmonary arterial, alveolar and venous pressures (P_a, P_A, and P_v, respectively). In zone 1, P_A is greater than P_a, occluding collapsible vessels and preventing flow. In zone 2, P_a is greater than P_A, which exceeds P_v, such that blood flow is dictated by the P_a-P_A pressure gradient. In zone 3, both P_a and P_v exceed P_A, and vessels are held open, allowing blood flow based on the P_a-P_v pressure gradient. Reproduced, with permission, from West et al. (688).

In the upright lung, gravitational forces cause increases in P_PA such that when P_PA exceeds P_ALV, flow is determined by the pressure gradient between P_PA and P_ALV. This condition is described as Zone 2, and is characterized by P_ALV exceeding P_V, but remaining below P_PA. Thus, in Zone 2, blood flow is determined by the difference between P_PA and P_ALV, rather than the arterial-venous pressure difference (as seen in Zone 3). This hemodynamic situation, in which flow is independent of downstream pressure, has been likened to a vascular waterfall or Starling resistor. Since flow is independent of the difference between arterial and venous pressure in this zone, lower downstream pressures (i.e., increasing the height of the waterfall) have no effect on flow. This phenomenon was illustrated by Permutt et al. (494), who showed that, when flow is held constant, P_PA falls linearly with P_ALV at higher alveolar pressures, but at lower P_ALV, P_PA becomes invariant to changes in P_ALV. This behavior would not be expected if P_PA and P_ALV were the only factors that influenced flow. Cessation of flow at an arterial-alveolar gradient >0 suggests an additional source of resistance (371, 494). One possibility is that when P_ALV drops below the critical closing pressure of a segment of the pulmonary vasculature, critical closure, or vessel collapse, occurs. Thus, critical closure is defined as a state of zero flow despite a positive driving pressure (P_PA − P_ALV).

Due to the nonrigid nature of blood vessels, critical closure occurs in both the systemic and pulmonary circulation when-ever the outflow pressure in a vessel is less than the pressure at which the vessel collapses (critical pressure) (172). In addition to Zone 2 hemodynamics, this concept also explains why venous return (VR) in the vena cava fails to continue increasing when right atrial pressure falls below a certain thresh-old (381). In the sheet-flow model, Fung et al. (192) have described this phenomenon as a “sluicing gate.” The exact critical closing pressure of a vessel depends on both factors intrinsic to the vessel itself as well as perivascular pressures. Since the latter changes in the lung based on factors such as lung volume and local edema, the exact critical closing pressure for any particular segment of the vascular tree is hard to determine. Based on isolated, perfused lung experiments, both the extra-alveolar (79, 371, 494) and alveolar (230) vessels have been implicated as sites of critical closure, and the exact site likely depending on lung volume. Moreover, each site of critical closure within Zone 2 may be determined by local factors controlling interstitial pressures and, thus, the measured critical closing pressure likely represents an average. Lastly, changes in resistance and flow in larger, upstream vessels may contribute to the lack of decrease in arterial pressures at low alveolar pressures, unrelated to whether or not critical closure occurs. In sum, Zone 2 flow is determined not only by P_ALV but also by Starling resistor behavior both at the level of the alveolus and possibly other segments of the vasculature, producing variable capillary opening and closing that depends on P_ALV and interstitial pressures.

In Zone 3, both P_PA and P_V are higher than P_ALV and, thus, in theory flow is completely independent of P_ALV. Additionally, regional heterogeneity in perfusion exists even in isogravimetric planes, implying that additional factors related to vascular structure and/or the capillary endothelium may play a role in blood flow in the lung microvasculature (220). For example, the most dependent part of the upright lung would be expected to have high flow due to gravitational effects; however, this area has paradoxically decreased flow relative to Zone 3. This area, known as Zone 4, is thought to experience decreased flow due to increased PVR in the extra-alveolar vessels as a consequence of mechanical forces, such that the effective precapillary vascular pressures are diminished. This point can be further illustrated in experiments using fluorescent microspheres to measure blood flow distribution in baboons studied in the upright, supine, prone, and head-down postures (218). In all positions, vertical gradients of perfusion were observed, although multiple stepwise linear regression analysis revealed that both gravity and geometry of the vascular tree influenced regional blood flow. Further complexity is added by heterogeneity in blood flow even between neighboring groups of alveoli, as shown by Baumgartner et al. (39). Consideration of these factors has led to the proposal of a refined model of lung zones (217, 219) whereby all three zones can exist within the same horizontal (isogravimetric) plane, but the numbers of different zones within each plane shifts with increasing hydrostatic pressure down the lung, from predominantly zones 1 and 2 at the apex to all zone 3 conditions in the dependent lung regions.

Pressure-flow relationships

The pulmonary vasculature exhibits high distensibility and low resistance. Propagation of flow in the pulmonary circulation is dependent on a myriad of factors including blood viscosity, vascular compliance, and transmural pressures. Additionally, nonlinear flow patterns originate as a result of vascular branching that occurs with each subsequent generation of pulmonary artery bifurcations. Though both the systemic and the pulmonary circulation feature a system of branching vessels downstream of a ventricular pump, they have different patterns of reflection, as evidenced by differences in pressure and flow curves between these two circuits (643). Like the systemic circuit, flow is pulsatile throughout the pulmonary circuit, remaining pulsatile, though with signal attenuation, to approximately the midcapillary region. In a mathematical model of wave propagation, the drop in pressure pulse appears to be greatest at the entry to the microcirculation (694). The venular segment of the pulmonary circulation exhibits flow patterns that are more dependent on left atrial pressures than changes in P_PA (516). Together, the pulmonary arterial, capillary, and venular segments serve to maintain synchrony between right ventricular output and left ventricular filling. Though often described in relation to resistance (see below), it should be noted that the pressure-flow relationships have been modeled as a function of vessel distensibility rather than resistance (357), especially when describing pressure-flow dependence on perfusate viscosity.

The pulmonary arterial tree maintains flow by adjusting resistance and capacitance (in opposite directions) at any given P_PA. According to this concept of the time constant (resistance times capacitance), the pulmonary arterial tree tends to be self-adjusting. Consequently, agents that decrease the caliber of the resistance vessels (small arteries and arterioles) in the lungs simultaneously stiffen the larger pulmonary arteries (capacitance vessels). The total PVR of the pulmonary circulation is dispersed across the various types of vessels, with each type (large arteries, smaller arteries, microcirculation, and veins) contributing to a portion of the total PVR. Given that PVR is dependent on various mechanical and hemodynamic factors, it is not surprising that the portion contributed by each subsegment of the vasculature, as well as total PVR, changes with conditions such as hypoxia, volume loading, and positive pressure ventilation (74).

Pressure and flow in the pulmonary arteries

P_PA is pulsatile and decreases slightly with inspiration. By comparison, P_PA is significantly smaller in magnitude than aortic pressure, displaying a rapid rise to a rounded peak during systole, a brisk small, and gradual decrease in pressure during diastole. End diastolic pressure in the pulmonary artery is approximately 7 to 12 mmHg; P_PA rises to approximately 20 to 30 mmHg during systole. PVR is calculated by measuring the drop in pressures across the pulmonary circulation and dividing by pulmonary blood flow. Pulmonary capillary wedge pressure (P_CWP), measured with a Swan-Ganz catheter introduced into the right-sided vessels and advanced via the pulmonary artery into the pulmonary circulation, is often used as a measure of P_V to make this calculation. In the normal human lung, PVR is approximately 50 to 100 dynes*s/cm⁵ (174, 445). Calculation of PVR is based on several assumptions including homogeneity of the intraluminal fluid, rigid vessels, and laminar flow; however, blood is a nonhomogeneous fluid, the pulmonary vessels are not rigid and flow is often turbulent or nonlaminar. Furthermore, increases in P_PA decrease PVR by way of recruitment, a process that essentially alters the circuit by adding additional sources of capacitance and resistance. Lung volumes affect PVR in a parabolic fashion. At volumes less than FRC, extra-alveolar and alveolar vessel resistance increase and decrease, respectively, contributing to a net increase in PVR and consequently, reduced flow. At volumes higher than FRC, the opposite occurs; alveolar vessel resistance increases due to increasing alveolar pressures, and extra-alveolar vessel resistance decreases, with the same net effect of increasing PVR. In summary, in addition to the limitations associated with interpreting isolated PVR measurements, interpreting changes in measured PVR in the lung is confounded by several factors, including: (i) functional zone of the vessels surrounding the catheter; (ii) lung volume; (iii) position; and (iv) changes in P_PA. For example, if the vessels surrounding the site of P_V measurement (i.e., the vessel in which the Swan-Ganz catheter resides) do not display Zone 3 behavior, the measured P_V may actually reflect P_ALV instead of true P_V, thus artificially lowering calculated PVR. Moreover, there is a preexisting gradient of decreasing vascular resistance from top to bottom in the vertical lung and since resistance and arterial pressures are correlated, rises in P_PA are sufficient to cause large decreases in PVR due to recruitment alone.

Pressure and flow in the microcirculation

Obtaining precise values of pressures in the microcirculation has proven difficult due to technical challenges. Studies utilizing isogravimetric, isolated perfused lobes have suggested that the majority of PVR occurs at the precapillary (i.e., arterial) level (481). However, micropuncture measurements performed in subpleural microvessels suggest a significant contribution to total PVR in this segment, with marked drops in pressure occurring just proximal (i.e., <50 μm diameter vessels) to the pulmonary capillaries (54). Exact measurements of pressure gradients across the microcirculation are also confounded by the effects of inflation, the choice of experimental preparation (i.e., isolated and perfused lung), differences between subpleural and deeper vessels, and lack of anatomic precision regarding the location of the vessels studied in a given experimental preparation. PVR measurements are further complicated by capillary recruitment, where the number of vessels with blood flow is subject to change based on ventilatory factors and subject position. In sum, precapillary resistance plays a large role in total PVR, but venous and microcirculatory sources of resistance exist as well.

Pulmonary capillary wedge pressure

As discussed above, P_CWP, an estimate of left atrial pressure, can be recorded by introducing a Swan-Ganz catheter through the great veins and advancing the catheter through the right ventricle, pulmonary artery and, ultimately, “wedging” the catheter, with a balloon, in a small precapillary vessel. Assuming Zone 3 conditions, there is an open stream of blood between the tip of the catheter and the left atrium and thus, the pressure recording is said to reflect left atrial pressures. This method of measuring right sided pulmonary vascular pressures and left atrial pressures remains in clinical use, though the utility of P_CWP to meaningfully alter clinical care is not clear (517). Multiple efforts to ensure proper position of the catheter are often taken, including analysis of the characteristic wedge tracing, measurement of oxygen saturation from the distal port of the catheter (which, if positioned properly, should reflected pulmonary venous, oxygenated blood) and comparison of the P_CWP to pulmonary arterial mean and diastolic pressures. However, despite these efforts, the catheter may still provide incorrect information due to the dynamic nature of zones in the lung. The catheter may be wedged in a Zone 1 or Zone 2 area and thus may not accurately reflect P_V. Other considerations such as the presence of parenchymal disease, such as fibrosis or emphysema, and presence of embolic disease may further confound the reliability of these measurements. Finally, two additional factors must be taken into consideration when analyzing pulmonary vascular pressures: transmural pressures and the effects of mechanical ventilation.

Intravascular pressure measurements solely provide information regarding luminal pressures. Since transmural pressure (and not luminal pressure) governs filling and distension, interpreting measurements of intravascular pressure relies largely on the assumption that perivascular pressures are constant between measurements, which may not be the case. While pleural pressure can be measured using a variety of techniques including esophageal balloons, extrapolation of pressures on the surface of the pulmonary vessels from pleural pressure is confounded by several factors including known vertical gradients in pleural pressure in the upright lung and variable transmission of pleural pressure to the various segments of the pulmonary vasculature.

Mechanical ventilation, especially with high positive end expiratory pressure (PEEP) in the setting of hypoxic respiratory failure, has multiple effects on both left sided CO as well as VR. The exact mechanisms behind decreased VR with application of PEEP in both healthy subjects and ill patients have been critically reviewed elsewhere (164, 381). PEEP affects both the mean systemic pressure within the venous system as well as right atrial pressure, and also changes the nature of flow limitation in the vena cava. In a manner similar to Zone 2 of the lung, application of PEEP alters the critical closing pressure of the vena cava, thus preventing further drops in right atrial pressure from affecting VR and producing the “flat” portion of the VR curve (Fig. 5). Transmission of elevations of airway pressure to other thoracic structures (such as the pericardium or the great veins) is variable and depends on a host of anatomic and physiologic factors, such as lung compliance, heart size, and volume status. In general, mechanical ventilation decreases VR and left ventricular afterload in patients with heart failure; however, it should be noted that precise calculations of the effects of mechanical ventilation and/or PEEP on transmural pressures in various segments of the pulmonary circulation are technically challenging and likely vary based on a number of patient-specific factors.

Figure 5.

Effect of PEEP and no, or zero, end expiratory pressure on VR and CO. (A) Effect of PEEP on CO and VR if PEEP has no effects on venous return flow limitation. Points A and B represent effects of PEEP without changes in mean systemic pressure (Pms). Point C represents effect of PEEP with Pms compensation. (B) Effect of PEEP on CO and venous return if increased VR flow limitation (FL2) occurs with higher PEEP. Reprinted, with permission, from Luecke and Pelosi (381).

Regulation of Pulmonary Vascular Tone

Although a low resistance, low-tone circulation under normal conditions, several factors influence vascular SMC tone under a variety of physiologic and pathologic conditions, including neural and circulating factors and oxygen concentration. These potent regulators of both pulmonary vasomotor responses and vascular caliber exert a substantial influence on pulmonary blood flow. If the changes in vasomotor tone induced by these factors are not uniform, significant redistribution of blood flow can occur. While the following discussion focuses on major factors involved in vasomotor regulation, it should be noted that in some cases, factors elicit differential responses depending on the level of initial tone, as discussed in (172).

Methods used to measure vascular tone

Various technical approaches have been utilized to examine vasomotor responses in the lung. Previous reviews have explained in detail some of the caveats of the methods used for detecting changes in vasomotor tone (172). Briefly, measurements in the lungs of intact animals are complicated by potential changes in heart rate, CO, and regional redistribution of pressure/flow. Interpretation of data from intact animals must also consider potential contributions from changes in neural and systemic inputs. In an attempt to remove these confounding factors, many investigators have turned to isolated, perfused lung/lobe preparations, where nervous influences are removed and pulmonary blood flow and left atrial pressure can be precisely regulated. Even more reduced experimental conditions, consisting of isolated pulmonary artery/vein preparations, are used to eliminate potentially confounding effects of extravascular influences, such as circulating mediators, and allow evaluation of influences localized to the vessel wall, such as the endothelium. In these preparations, vasomotor tone has been measured in vascular rings or strips mounted on wire myographs for isometric force measurement or in cannulated vessel segments where vascular diameter can be directly measured under conditions of constant transmural pressure.

Vasodilators

A number of humoral factors participate in active regulation of pulmonary vascular tone (36,129,172). Among the factors that induce pulmonary vasodilation, NO, adenosine, atrial natriuretic factor (ANP) and the eicosanoid, PGI₂, are the most studied. Some of these factors are endogenous, derived from the vascular endothelium, while others are produced by circulating cells or in other vascular beds. In this section, we will discuss these factors, and their role in regulating tone under normal and pathologic conditions.

Nitric oxide

NO is produced primarily by the endothelium and is perhaps the most widely studied of the pulmonary vasodilators. NO is generated by oxidation of L-arginine to L-citrulline in a reaction that requires molecular oxygen and is catalyzed by the enzyme NO synthase (NOS) (473). Three NOS isoforms have been identified, with endothelial (eNOS) and neuronal (nNOS) isoforms being Ca²⁺ dependent and constitutively expressed. In contrast, the inducible form (iNOS) is Ca²⁺ independent and is expressed in response to cytokines and other stimuli (471, 472). In ECs, constitutive NO synthesis occurs via the activity of eNOS, whereas NO levels can be enhanced by activation of iNOS. Given its short half-life, NO is not suited for action as a circulating factor and instead, once released by ECs, quickly diffuses to the underlying smooth muscle and promotes relaxation via activating soluble guanylate cyclase, leading to generation of cGMP and cGMP-dependent decreases in intracellular calcium concentration ([Ca²⁺]_i) and/or myofilament Ca²⁺ sensitivity that can limit or reverse ongoing contraction (90).

Numerous studies have utilized NOS antagonists to investigate the role of NO in maintaining low normal pulmonary vascular tone. In the neonate, studies clearly demonstrate a critical role for NO as a modulator of PVR (143, 228, 455). In the adult lung, however, contradictory results have been reported. While NO antagonists increased normoxic pulmonary vasomotor tone in humans (64), and rats (35,92,139), suggesting that NO was exerting a vasodilatory influence, in other studies these interventions had minimal effect on baseline pulmonary vascular tone (35, 157, 163, 249). These data suggest that the contribution of NO to maintenance of low pulmonary vasomotor tone during normoxia depends on the presence or absence of contractile influences and/or relaxing influences other than NO, which in turn may depend on species, preparation, experimental conditions, and other factors.

As a highly reactive gas with short biological half-life, NO was initially thought of as functioning primarily as an autocrine/paracrine signaling molecule, at most diffusing the short distance from the endothelium where it was produced to the underlying smooth muscle. What has only been appreciated more recently is the potential for NO to be stored as nitrite. With a half-life of approximately 50 min (498), nitrite is relatively stable compared to NO. Originally, nitrite was viewed simply as a byproduct of NO metabolism, useful mostly as a biomarker for NO production. However, it is now recognized that nitrite can readily be reduced to NO, thus forming an in vivo physiological reservoir from which NO can be recycled independent of NOS (119, 738). Experiments where exogenous nitrite was administered showed that nitrite had the ability to dilate the systemic vasculature (193,288,324,345) and the pulmonary circulation under conditions where tone was increased (281, 739). Whether NO derived from nitrite contributes to the maintenance of low pulmonary vasomotor tone is unclear at this point, but is a likely possibility, especially under conditions where NO bioavailability is compromised.

Bradykinin

Bradykinin, a peptide product of the renin-angiotensin system, dilates isolated pulmonary vascular rings in many species. In the intact pulmonary vascular bed, however, whether bradykinin is a vasodilator may depend in part on the species tested and/or the level of preexisting tone. Under basal conditions, bradykinin minimally constricted the pulmonary vascular bed of the rabbit (250), but reduced P_PA in isolated perfused dog lung (350). Vasodilation became evident, however, when the vasculature was precontracted (183,359).

Bradykinin exerts its dilatory influence by stimulating endothelial NO, and in some cases PGI₂, release (21, 287, 474). Indeed, in vitro, removal of the endothelium results in a loss of dilation in response to bradykinin, with vasoconstriction observed in some cases due to activation of receptors on the smooth muscle.

Adenosine

Adenosine, a metabolite of AMP or S-adenosylhomocysteine, dilates most vascular beds. In the lung of various species, including humans, adenosine causes relaxation of pulmonary arteries, via binding to its cell surface receptors and transducing signals through G-protein coupled adenylyl cyclase (189,229,251,330,415,453,454,486,525,535). In some cases, adenosine has also been reported to elicit pulmonary vasoconstriction (56, 77, 103, 535). The exact reason for the discrepancy between studies is unclear, but may depend on species, concentration, or initial vasomotor tone. Consistent with this possibility, adenosine contracted isolated pulmonary vessel preparations at resting tone (364,695), but induced relaxation when vessels were precontracted (153,364,403).

Adenosine infusion increases pulmonary blood flow in healthy humans (255), suggesting that under normal conditions, adenosine exerts a vasodilatory action. Although infusion of adenosine receptor inhibitors had no effect on the high PVR in near-term fetal lamb (329), the effects of adenosine receptor inhibitors in the adult human lung have not been tested and whether endogenous adenosine contributes to maintenance of low basal tone is unknown.

Atrial natriuretic peptide

ANP is produced from cardiomyocytes primarily in response to stretch of the right heart, such that with an increase in P_PA, circulating ANP levels are elevated (24). ANP is produced and stored in cardiomyocytes as an inactive precursor, pro-ANP, and upon secretion is cleaved to the active 28-amino acid peptide (α-ANP) (708). Given the source, the pulmonary circulation is the first vascular bed to see ANP; not surprisingly, ANP plasma levels are 30% greater in the pulmonary than systemic circulation. Additionally, pulmonary arteries exhibit greater sensitivity to the vasodilatory effects of ANP than arteries from other vascular beds. Vasodilation is mediated by ANP-A and ANP-B receptors, which are coupled to guanylate cyclase (461). These receptors have been localized in the lung (4, 573, 639) and pulmonary vessels exhibit a high density of ANP-binding sites (16). Exogenous application of ANP causes relaxation of precontracted isolated pulmonary vessels (298,356) and vasodilation in the perfused pulmonary vascular bed under basal conditions (108, 302) that was enhanced when the lung was precontracted (108). In COPD patients, ANP reduces pulmonary vascular pressure and resistance (8), suggesting that in vivo ANP vasodilates the pulmonary circulation, especially under conditions of increased tone.

Eicosanoids

Under basal conditions, when tone is low, infusion of arachidonic acid constricts the perfused pulmonary vascular bed (559, 692) but causes relaxation when vascular tone is elevated (211). Arachidonic acid is the precursor of several vasoactive mediators, including the cyclooxygenase product, PGI₂. Released from endothelium, PGI₂ stimulates adenylate cyclase and increases production of cAMP in the underlying smooth muscle, vasodilating both the pulmonary and systemic circulations (282). The role of PGI₂ in maintaining low vasomotor tone is incompletely understood. Several investigators found no effect of the cyclooxygenase inhibitors, indomethacin or meclofenamate, on baseline P_PA and/or PVR in either intact animals or isolated, perfused lungs (94, 487, 496, 522, 680). In contrast, other investigators observed increases in P_PA with administration of these inhibitors (307, 522, 618, 659), suggesting basal endogenous production of a vasodilating prostanoid, likely PGI₂. The reason for these differences is unclear, but may be related to concentrations of drugs used or the duration of exposure to drug before measurements were made, since most of the measurements showing no effect were made shortly (<30 min) after drug administration. Along these lines, long-term administration of cyclooxygenase inhibitors (3 weeks) was reported to increase P_PA (417). Overall, these data suggest that, at least under certain conditions, vasodilating cyclooxygenase products may contribute to low pulmonary vasomotor tone.

In some cases, endothelium-dependent relaxations induced by acetylcholine and other agents were not completely abolished with inhibition of NO and PGI₂, indicating the presence of additional endothelium-dependent vasodilation pathways (67, 102, 162, 328). Since these vasodilatory effects occurred in association with smooth muscle hyperpolarization, it was postulated that release of an endothelium-derived hyperpolarizing factor (EDHF) was responsible. The exact identity of EDHF is still unclear, but one of the most likely possibilities is a cytochrome P-450 product of arachidonic acid metabolism, perhaps an epoxyeicosatrienoic (EET) acid (68, 175, 177, 206). In both the systemic and pulmonary circulations, cytochrome P-450 metabolites can elicit either dilation or contraction, although in the systemic circulation, EETs mediate relaxation while 20-HETE is a constrictor (176). Remarkably, the opposite appears true in the pulmonary vascular bed, with infusion of 11,12-EET increasing P_PA in isolated perfused mouse lungs (320) and 20-HETE causing dilation in human PAs (59). Although EDHF appeared to play a role in vasomotor responses of isolated pulmonary arteries or lungs (198, 732), whether it contributes to maintenance of low basal tone is unclear.

Vasoconstrictors

Pulmonary vasoconstriction is caused by a variety of factors, including serotonin, ET-1, angiotensin II (ANG II), and prostaglandins, some of which are derived from the vascular endothelium. Circulating cells are also an important source of vasoconstrictors that can act on the pulmonary circulation, including serotonin, a primary product of activated platelets.

Endothelin-1

Discovered in 1988, ET-1 is a peptide secreted by primarily ECs (709). Of the three isoforms that have been identified (ET-1, −2, and −3), ET-1 is the most widely expressed, and thus, most studied. ET-1 is a 21-amino acid peptide that causes profound pulmonary vasoconstriction in every species tested to date [reviewed in (606)], and is perhaps the most potent endogenous vasoconstrictor in the lung. ET-1 binds to two membrane bound receptors (ET_A and ET_B) (629), both of which are expressed in PASMCs (124,188), and mediate contraction, proliferation and migration. In contrast, ECs express only ET_B receptors, which are thought to act as a “sink” for circulating ET-1. While ET-1 can elicit NO production and transient dilation when EC receptors are activated, the majority of ET-1 is secreted basolaterally; thus, in vivo PASMCs are likely to be the main target of ET-1 with the main action of ET-1 being vasoconstriction. At concentrations as low as 10⁻¹⁰ M, ET-1 constricted isolated pulmonary arteries through activation of ET_A or ET_B receptors on PASMC, while infusion of ET-1 caused long-lasting increases in vascular resistance in isolated perfused lungs [reviewed in (570, 606)]. ET-1 secretion from the pulmonary endothelium has been shown to increase in response to shear stress and hypoxia [reviewed in (563, 570, 606)]. Binding of PASMC ET receptors initiates a complicated signaling pathway (570, 629), involving inhibition of K⁺ channel expression and activity, increased [Ca²⁺]_i and activation of NHE1 and Rho kinase (ROCK) (570,642,691).

Acute infusion of an ET_A receptor antagonist causes pulmonary vasodilation in fetal sheep, suggesting that basal ET-1 activity contributes to high in utero PVR (293, 700). Similar results were observed in intact adult pigs (268, 683), but not dogs (624), perhaps reflecting a species-dependent difference in either endogenous ET-1 synthesis or participation in regulating basal tone.

Arachidonic acid metabolites

Arachidonic acid, which is readily taken up in the pulmonary circulation, is metabolized into a number of vasoactive eicosanoids, including PGE₂, PGF_2α, thromboxane, and leukotrienes, all of which diffuse to the smooth muscle and cause contraction. Thromboxane, or thromboxane mimetics, constrict isolated perfused lungs (166, 169, 515) as well as isolated pulmonary arteries (114, 166, 299, 360, 443). Synthesized via the lipoxygenase pathway, leukotrienes C₄, D₄, and E₄ caused pulmonary vasoconstriction in a variety of species, including humans [reviewed in (36, 658)]. Arachidonic acid can also be metabolized by cytochrome P-450, which produces midchain hydroxyeicosatetraenoic (HETE) and cis-EET acids in pulmonary arterial ECs and PASMCs (731, 737), the latter of which can be further metabolized by soluble epoxide hydrolase, forming dihydroxy derivatives. In the precontracted pulmonary vasculature, HETE and EET can cause vasodilation (59, 296, 599, 722) but primarily mediate vasoconstriction under unstimulated conditions (190,372,706,736). Since administration of arachidonic acid constricts the normal perfused pulmonary vascular bed (559, 692), it is likely that under normal conditions of low tone, the balance of arachidonic acid metabolism is shifted toward a vasoconstrictor effect.

Serotonin (5-HT)

Pulmonary neuroendocrine cells secrete various vasoactive substances, including 5-HT (304, 346). A 5-HT transporter protein [serotonin transporter (SERT)] in plasma membranes allows circulating 5-HT to be taken up by cells, including pulmonary ECs and platelets, where it is stored and subsequently released during aggregation. At baseline, storage in platelets and the presence of reuptake mechanisms ensure low circulating 5-HT levels, but plasma levels can be elevated by selective 5-HT reuptake inhibitors (SSRIs). Application of exogenous 5-HT constricts the pulmonary vasculature, although its potency varies greatly among species [reviewed in (36,606)]. When tone is low, contraction to 5-HT is mediated primarily by 5-HT2A receptors; with elevated tone, a portion of 5-HT-induced contraction can be mediated by 5-HT1 receptors (388). In the intact animal, infusion of the 5-HT2A receptor antagonist, ketanserin, reduced PVR in intact cats (75), fetal sheep (135) and in patients with stable angina (270), but had no effect on PVR in isolated perfused dog lung (34). Similar contradictory results were obtained with application of SSRIs. Infusion of the SSRI, sertraline, increased PVR in fetal sheep (135), whereas the SSRI, dexfenfluramine, had had no effect on PVR in porcine lungs and had limited effect in human lungs, with vasoconstriction only noted at high concentrations (258). Whether these situations reflect differences between intact animals and isolated perfused lung preparations, or conditions where tone and/or endogenous 5-HT levels were elevated, is unclear.

Angiotensin II

Circulating angiotensin I, a product of the renin-angiotensin system, is converted into the vasoactive peptide, angiotensin II (ANG II), in the lung by means of endothelium-bound angiotensin converting enzyme. In point of fact, the lung endothelium is the primary site of metabolism of angiotensin I, with approximately 60% to 80% of plasma angiotensin I converted to ANG II in a single pass through the pulmonary circulation. Exogenous ANG II causes pulmonary vasoconstriction in most species, both in isolated vessel and intact lung preparations (37,70,87,160,185,222,404,408,462,506,534, 614, 682). Some studies suggest that basal activation of the renin-angiotensin system and production of ANG II exerts a slight vasoconstrictor effect on the pulmonary circulation (222, 462), although other studies failed to show an effect of ANG II blockade at baseline (246,276,323,335,506).

Summary

In addition to the vasodilators and vasoconstrictors described in the preceding sections, a number of other factors have been shown to exert regulatory influences on tone in the pulmonary vasculature, including histamine, platelet-activating factor, vasopressin, vasoactive intestinal peptide, and substance P. The actions of these agents on the pulmonary circulation have been reviewed in detail elsewhere (36, 129) and interested readers are encouraged to consult these previous reviews for further information regarding pulmonary vasoregulation by these factors.

The preponderance of evidence suggests that although most of these factors can produce changes in vasomotor tone, with the exception of NO and PGI₂, none appear to exert a major regulatory influence on basal pulmonary vascular tone in the normal lung. However, under pathological conditions modulation of tone by a variety of factors may become more pronounced. For example, bradykinin induces a moderate fall in P_PA in patients with hypoxic pulmonary hypertension (62). Similarly, adenosine acts as a pulmonary vasodilator in pigs following infarction (589), in patients with pulmonary hypertension (252, 431) and in cardiac surgical patients (189). Moreover, vasoconstrictors can modulate tone under pathological conditions, such as in pulmonary hypertension induced by anorexic agents, where serotonin uptake is impaired, and excessive production of ET-1 is a common feature of pulmonary hypertension.

Nervous control

The pulmonary circulation is richly innervated by the autonomic nervous system, with both sympathetic and parasympathetic connections (36, 173, 337, 529). In most organs, arterioles exhibit the highest level of sympathetic innervation (561); in the lung the opposite is true, with sympathetic noradrenergic innervation density highest in the extrapulmonary vessels (96, 171, 243). The degree to which sympathetic innervation penetrates the vascular tree varies with species [reviewed in (36)]. Rat and mouse intrapulmonary arteries do not appear to be innervated (96,171); however, in most animals, including humans, sympathetic axons extend down to the level of small arterioles (96, 171, 243). Intrapulmonary veins are also innervated (96,309,338), although to a lesser extent than the arterial network.

Given evidence indicating that pulmonary arteries are abundantly innervated, it is perhaps surprising that there appears to be minimal nervous control of basal vascular caliber in the pulmonary circulation, with adrenergic antagonists causing a slight, if any, increase in PVR in anesthetized and conscious dogs and sheep (312,398,442). While these results suggest that there may be a mild degree of basal sympathetically driven dilatory influence on tone under normal physiological conditions, nervous stimulation under experimental conditions clearly modulates tone. In general, increased sympathetic activity leads to release of catecholamines (e.g., dopamine, norepinephrine, epinephrine, and neuropeptide Y) that cause vasoconstriction and an increase in PVR (82,338). Indeed, in the intact lung, sympathetic nerve stimulation caused an increase in PVR that could be inhibited by chemical denervation or adrenergic blockade (283,308,309). Similarly, stimulation of sympathetic nerves constricted resistance-sized pulmonary arteries, which was reversed by an α-adrenergic blocker (574). This point is further illustrated in numerous studies where infusion of α-adrenergic agonists caused pulmonary vasoconstriction in intact animals with controlled pulmonary perfusion (32, 499, 536, 580). Interestingly, infusion of β-adrenergic agonists dilated the pulmonary vasculature (500,580), suggesting β-adrenergics mediate vasodilation in response to circulating catecholamines (65,285) and locally released norepinephrine (284). Taken together, control of the pulmonary circulation by the sympathetic nervous system may vary with stimulus and/or ultimately may be a coordinated response including both α- and β-adrenergic components.

Pulmonary arteries contain fewer cholinergic than adrenergic nerve fibers, although the distribution is similar, with highest density at the extrapulmonary arteries (152, 243). Parasympathetic stimulation causes release of acetylcholine and vasoactive intestinal polypeptide (11,96,237,407), which mediate vascular dilation and act to decrease PVR. However, cholinergic blockade had no effect on basal PVR (442). The lung also contains nonadrenergic, noncholinergic (NANC) nerves that can release vasoactive intestinal peptide, calcitonin-gene related peptide, ATP, substance P and NO to mediate vasodilation (9,227,241,362,363,404,557). The presence of nNOS has been demonstrated in nerve fibers in and around the walls of pulmonary arteries (168,241), consistent with NO being the primary neurotransmitter responsible for NANC-induced pulmonary vasodilation (227,241,362,363). However, NOS blockade did not fully inhibit NANC-mediated dilation (241), suggesting that the other factors may also participate in the response. While the existence and activity of NANC nerves have been demonstrated in vitro [reviewed in (36)], regulation of tone in vivo by these nerves has not been demonstrated.

Thus, the majority of evidence suggests little in the way of contribution from the nervous system in control of resting vasomotor tone in the lung; however, stimulation of adrenergic nerves may modulate PVR and blood flow during exercise and cold exposure and may increase in regulatory contribution during pathological states, particularly during pulmonary edema and embolism.

Hypoxia

One of the unique aspects of the pulmonary circulation is the hypoxic pressor response. Unlike the systemic circulation, where hypoxia causes vasodilation to increase blood flow and oxygen delivery to tissues, reductions in alveolar and arterial oxygen content cause profound constriction in the pulmonary vasculature. This phenomenon was suggested as early as 1852 (51), when the first measurements of P_PA were reported and showed that stopping ventilation increased P_PA. Later studies by von Euler and Lillestrand in 1946 (652) confirmed this early finding, and provided the first detailed characterization of hypoxic pulmonary vasoconstriction (HPV).

While systemic hypoxic vasodilation plays an obvious role in working to meet the metabolic demands of the tissue, the exact reason why the pulmonary circulation contracts in response to hypoxia remains a matter of some debate. HPV is most commonly believed to be either a residual feature from fetal development, where the hypoxic environment maintains a closed pulmonary circulation until birth, or a mechanism to match ventilation and perfusion during periods of localized hypoxia, by diverting blood flow from poorly oxygenated portions of the lung. What is not in debate is the response itself. Numerous studies have explored the effects of hypoxia on the pulmonary circulation in a variety of species [reviewed in (606)]. Uniformly, these studies have described an increase in PVR that occurs within 1 to 2 min after a reduction in alveolar oxygen tension that reaches a maximal response in 5 min. This immediate contraction is maintained for the duration of the hypoxic exposure, but varies in intensity by level of hypoxia, sex, and species (606). Upon reoxygenation, HPV rapidly reverses, with P_PA typically returning to normal levels within minutes.

Although the large pulmonary arteries are capable of responding to hypoxia, it is now generally accepted that the main site of increased resistance during HPV is in the small, muscular arterioles. Postcapillary hypoxic venoconstriction may also occur, although the magnitude of the contribution of these vessels to the increase in total PVR is uncertain.

The exact mechanisms by which the pulmonary vasculature senses a drop in oxygen tension and transduces this signal to vasoconstriction are still an area of active investigation. Studies using reduced preparations, including isolated vessels and single cells, clearly established that hypoxia had a direct effect on PASMCs, although it is also evident that the maximal contractile response requires a coordinated response between ECs and SMCs(1, 606). For example, it is well documented that ECs provide a priming or modulating influence through the release of vasoactive agents [reviewed in (1, 606)]. In addition, ECs communicate with SMCs via myoendothelial gap junctions, which allow transmission of electrical current and small signaling molecules between cells and facilitate HPV (326, 433). Indeed, myoendothelial gap junction signaling contributes to HPV via Rho kinase-dependent increases in Ca²⁺ sensitization of the contractile apparatus (326). More recently, studies revealed a fundamental role for the endothelium as part of the initiating sensor of HPV, with capillary ECs exhibiting depolarization in response to alveolar hypoxia that is propagated upstream via endothelial connexin 40 gap junctions, eventually leading to production of EET acids (669). Thus, the in vivo response to hypoxia likely results from a coordinated response including sensing by, and communication between, both the ECs and SMCs.

Considerable effort has gone into delineating the SMC response to hypoxia, sometimes with confusing and/or contradictory results [reviewed in (606)]. For example, substantial evidence suggests that reactive oxygen species (ROS) appear to be key mediators of HPV; however, the source, direction of change during hypoxia and downstream targets have varied across laboratories [reviewed in (606)]. Initial reports indicated that ROS production decreased during hypoxia, leading to inhibition of redox-sensitive K⁺ channels, depolarization and activation of voltage-gated Ca²⁺ channels (19, 20, 419) (redox hypothesis; Fig. 6A). Later studies, however, suggested that ROS formation in fact increased during hypoxia (347, 361, 520, 665, 674, 675, 677). While perhaps counterintuitive, given the fact that oxygen is a substrate for ROS production, it was suggested that despite a fall in oxygen concentrations, ROS production from mitochondrial complex III was enhanced (ROS hypothesis; Fig. 6B), leading to increased [Ca²⁺]_i due to both Ca²⁺ release and influx. Interestingly, Ca²⁺ release has also been shown to inhibit K⁺ channels during hypoxia (502), suggesting potential links between these two proposed hypotheses.

Figure 6.

Proposed mechanisms by which hypoxia caused pulmonary vasoconstriction. Reproduced, with permission, from Sylvester et al. (606).

The exact reasons underlying the differences in results remains to be fully explained, although the recent development of ratiometric ROS sensors that could be targeted to specific compartments within cells has provided potential insights. Using these probes, it was found that there were substantial differences in ROS production during hypoxia that varied by cellular compartment; ROS levels increased in the cytosol and mitochondrial intermembrane space, but decreased in the mitochondrial matrix (676). Thus, significant differences in results could be due to the compartment within the cell in which measurements were obtained. Overall, it would seem that the majority of data now supports that, in PASMCs, hypoxia leads to production of mitochondrial ROS (98, 674), localized areas in which ROS are reduced (19, 676), calcium release from the sarcoplasmic reticulum (458,663,705), opening of calcium permeable nonselective cation channels (NSCCs) (458,664,679), inhibition of voltage-gated potassium channels (19, 503, 728, 729), membrane depolarization (503, 729), and subsequent activation of Ca²⁺ influx through voltage-gated Ca²⁺ channels (409,664).

Although the exact mechanisms by which hypoxia activates Ca²⁺ signaling are still being investigated, recent data demonstrated a role for sphingolipid signaling that might link ROS with Ca²⁺ responses and HPV (610). Oxidant stress activates neutral sphingomyelinase, releasing ceramide, which meditates both the recruitment of the NSCC, TRPC6, to the plasma membrane and activation of TRPC6 via PLC. The sphingolipid product, sphingolipid-1-phosphate (S1P), concomitantly increases Rho kinase activation, combining with Ca²⁺ influx to cause PASMC contraction.

Increased production of vasoconstricting (i.e., ET-1), and/or reduced production of vasodilating (i.e., NO), factors by the neighboring ECs augment PASMC responses to hypoxia (606), indicating that while the main contractile mechanics underpinning HPV reside in PASMCs, the endothelium certainly plays a modulatory role. Other factors that modulate HPV include prostaglandins, angiotensin II, serotonin and leukotrienes, although all of these have been ruled out as mediators of the response [reviewed in (606)]. The autonomic nervous system has also been ruled out as a mediator, as the response can be observed in isolated, perfused lungs that lack nervous input (123, 367, 368) and most experiments in intact animals suggest that sympathetic activity has no effect on HPV [reviewed in (606)]. Indeed, in some cases, blockade of β-adrenergic receptors enhanced HPV (32, 76, 367, 499), suggesting that while sympathetic activity does not mediate HPV, it could modulate the response under certain conditions. In contrast, parasympathetic nerves do not appear to modulate HPV (349,351).

The fact that the pulmonary circulation contracts almost immediately upon initiation of hypoxia would also appear to rule out a role for transcriptional regulation in the acute HPV response. However, HPV is enhanced in individuals with Chuvash polycythemia, a rare genetic condition where germline hypomorphic mutations in the gene encoding von-Hippel Lindau protein—a major component of the HIF degradation pathway—result in augmented HIF expression under nonhypoxic conditions (585), suggesting a potential role for HIFs in regulating acute pulmonary vasomotor responses to hypoxia. Similarly, studies of humans with HIF-2α gain-of-function mutations also exhibit enhanced HPV (179). Although it is unlikely that acute changes in gene transcription underlie the immediate pulmonary vascular response to hypoxia, these findings support an intriguing possibility that HIF may modulate the basal expression of targets believed to be directly involved in the contractile response (i.e., K⁺ or Ca²⁺ channels) or circulating factors known to be required for enhancement or priming of PASMCs (i.e., ET-1).

Pulmonary Barrier Function and Edema

The pulmonary endothelium is a contiguous structure, one of the main functions of which is to protect the alveolus from flooding and injury due to flux of electrolytes and other small molecules, proteins, and fluid from the vascular compartment. The thinness of the pulmonary endothelium and minimal distance between the capillary wall and alveolar epithelium are ideal for gas exchange, but also predispose the lungs to edema formation. In recent years, the barrier protective function of the pulmonary vasculature has been investigated intensely [reviewed in (53, 60, 327, 476)]. Recent studies from several laboratories suggest that the pulmonary microvascular endothelium is much less permeable than that of capillaries in other organs and less permeable than the pulmonary macrovasculature (134, 477, 480, 601). Phenotypic variation between the ECs that comprise the intima of these vascular segments likely contributes to the differences in their responses to various barrier-disrupting stimuli. Understanding the cellular mechanisms that control endothelial barrier function has been the focus of many recent studies of the pulmonary circulation, especially given that increases in microvascular and macrovascular endothelial permeability have been identified as key processes in the development of diseases like ARDS.

Vascular phenotypes: Microvascular and microvascular endothelium

Since the pulmonary vascular endothelium plays a key role in the pathogenesis of various lung diseases, recent investigations into the molecular biology of populations of lung ECs led to the classification of vessels based on their endothelial phenotype, as either macrovascular or microvascular. Like older classification systems, this system is also based on vessel diameter size; however, it includes additional functional and morphologic characteristics that differentiate these two cell types. While differences between alveolar and extra-alveolar vessels have been studied in the context of mechanical forces such as inspiration or changes in vascular pressure, the differences between the vascular endothelial phenotypes have been studied largely in the context of their gene and protein expression, cytoskeletal function and responses to barrier disrupting stimuli.

Differences between the various segments of the pulmonary endothelium begin in utero, where it has been shown that the microvascular ECs develop from a pool of progenitor cells distinct from arterial and venular ECs (208). All ECs in the lung share some features that are common to ECs throughout the body, such as size, fine control of cell adhesion and a basic array of cell surface markers such as von Willebrand factor and PECAM-1 (208). However, it is logical that differences in local milleu dictate the genotypic and phenotypic variation seen within the pulmonary endothelium. For instance, macrovascular ECs are exposed to more pulsatile flow and are not subjected to nearby alveolar forces, unlike microvascular ECs. Consequently, these two cell types exhibit unique morphologic characteristics, distinct genomic and proteomic profiles and differential responses to a variety of injurious stimuli (635). In general, microvascular ECs form a tighter barrier and are more restrictive to fluid efflux through paracellular routes, and when barrier disruptive stimuli such as pathogens, alterations in intracellular Ca²⁺ or oxidative stress are applied, macrovascular and microvascular cells respond differently to these injuries (13,109). For example, in experiments utilizing various Ca²⁺ channel agonists in isolated, perfused lung preparations, increased permeability in microvascular endothelium can be induced with agonism of the Ca²⁺ channel, TRPV4, but not depletion of Ca²⁺ stores with thapsigargin, whereas macrovascular endothelial permeability can be increased by store depletion (104). In vitro, store depletion and activation of store-activated Ca²⁺ channels appears to be sufficient to induce paracellular leak in macrovascular, but not microvascular, ECs, suggesting differential expression of these channels as well as fundamental differences in barrier disrupting signal transduction (376). There are additional differences in virtually all domains of cell function, from cytoskeletal dynamics to proliferation and responses to flow (600). For instance, compared to macrovascular ECs, microvascular ECs express binding motifs for interactions with Griffonia simplicifolia lectin, but lack Weibel-Palade bodies (600). Microvascular ECs also express more nucleosome assembly protein-1, which has been implicated in their ability to proliferate faster than macrovascular ECs (111). In vitro, application of cyclic stretch produces more changes in cell-cell adhesions in macrovascular endothelium (7). Indeed, more global analysis of gene expression using microarrays has demonstrated distinct gene expression profiles between micro- and macrovascular ECs suggesting large differences in cell function and signaling between these two adjacent EC types (107). In vivo, there appears to be a distinct transition zone, occurring around vessels with outer diameters of approximately 25 μm, when microvascular ECs appear and macrovascular ECs disappear (635). The exact reasons underlying the observed functional differences between pulmonary micro- and macrovascular ECs are still being investigated, but may lie in differential expression of various ion channels, kinases and signaling molecules that participate in barrier regulation, as will be discussed later in this review.

Methods of measuring endothelial permeability

To compare across studies, a fundamental understanding of the multiple methods that have been developed to assess lung endothelial permeability in vitro and in vivo is helpful [reviewed in (727)]. In vivo measures of endothelial permeability measure leak of fluid into the interstitial space due to either increase in intravascular pressures or endothelial permeability (478). An important consideration is that the net efflux of fluid from the vascular space is a function of Starling forces, as well as properties of the luminal fluid, permeability state of the endothelium, flooding in the alveolus and lymphatic flow (401), and occurs in both the alveolar and extra-alveolar vessels [reviewed in (478)]. Methods of quantifying endothelial leak in vivo in intact animals include gravimetric methods (e.g., wet/dry ratios), light and electron microscopy, and visual quantification of dye extravasation from the vascular compartment (i.e., Evans Blue Dye). These methods are not without drawbacks, however, including arti-factual increases in perceived permeability due to changes in lung blood volume and dye binding to native proteins.

Isolated, perfused lung models offer the advantage of maintaining the integrity of the vasculature and its relationship to ventilated alveoli, while allowing for more precise measurements of filtration across the endothelium by manipulating arterial and venous pressures. First described in 1967 (194), measuring the endothelial filtration coefficient (K_f) using an isolated, perfused lung preparation remains the gold-standard tool for assessing pulmonary vascular permeability (479) particularly via the paracellular pathway (478). Assuming that further recruitment is not taking place (479), rises in K_f directly correlate to increased conductance of fluid across the endothelium. Accordingly, this method has been used in both wild-type and transgenic mice and rats in multiple models of lung injury. However, as explained in detail in a recent review (55), K_f measurements are prone to errors due to changes in pulmonary volume when the atrial pressure is artificially increased during the measurement process. Additional considerations include injury to the ECs due to the measurement procedure itself, and whether or not concomitant alveolar edema is present. Much work has been invested developing techniques to minimize these sources of error as much as possible, and detailed descriptions of specific recommendations for experimental procedures have been published (55,478). Once an appropriate isolated, perfused lung preparation is in place, in addition to K_f, further measures of permeability such as the reflection coefficient and permeability surface area product can be obtained as well (478). Techniques to measure flux occurring via vesicular and transcellular pathways have also been developed, and are discussed elsewhere (319,532,727).

The choice of a specific method to measure endothelial permeability may be dictated, to some degree, by the type of model in which lung injury is incited. Physiologically functional models where the cardiopulmonary circuit is left intact (e.g., survival ischemia-reperfusion models, endotoxin (lipopolysaccharide; LPS) or cecal-ligation puncture models of lung injury), while closest to native physiology, do not allow for enough manipulation of hemodynamic parameters to reliably make K_f measurements unless the animal is sacrificed afterward for use in the isolated, perfused lung preparation. Instead, for these models, a general dye-based method such as extravasation of Evans Blue Dye is often used (670). It should be noted, however, that limitations in this method related to binding of dye to albumin and other proteins (122) and changes in vascular tone and recruitment between animals (727) have been noted and may complicate interpretation of results.

Injury models in intact animals are particularly useful for determination and assessment of in vivo endothelial barrier function, but do not allow for detailed mechanistic exploration of the precise cellular mechanisms involved in injury. To combat this shortcoming, in vitro models of EC barrier permeability have been developed. The use of cultured EC monolayers allows for multiple modes of manipulation of pathways and evaluation of cell dynamics, albeit in a less physiologic setting, due to the absence of structures such as a proper glycocalyx, basement membrane, and surrounding extracellular matrix as well as the lack of tonic shear from luminal flow and changes in the endothelial proteome that occur due to in vitro handling (641). Methods for assessing permeability in cultured ECs include transwell assays, which determine permeability by measuring migration of a large weight molecule from an upper chamber to a lower chamber across a confluent layer of ECs (579) and electrical cell impedance sensing (ECIS), which utilizes the electrical properties of an EC monolayer, with the cells acting as capacitors, and pores that open when permeability increases acting as resistors, to measure total impedance (269). In the case of the latter method, increases in permeability will thus decrease measured ECIS resistance. One of the most exciting recent developments in the field is the newer “lung on a chip” in vitro model (279), which recapitulates many of the physiologic processes that occur in the native lung (Fig. 7). In this preparation, epithelial cells and ECs are appositionally seeded, separated by a porous synthetic membrane that has been coated with matrix components. The epithelial side is then exposed to air while the endothelial side is exposed to perfusate. Additional side chambers allow for simulation of cyclic stretch that occurs with normal tidal breathing. Thus, several of the limitations inherent in traditional static culture models such as lack of endothelial shear stress, alveolar stretch, alveolar-matrix, and matrix-endothelial interactions are overcome in this powerful new technique, providing the potential to combine more physiologically relevant parameters with the ability to manipulate individual cell populations in a coculture setting.

Figure 7.

The lung-on-a-chip microdevice. (A) Compartmentalized channels form an alveolar-capillary barrier on a thin, porous, flexible membrane. Physiological breathing movements are mimicked by applying a vacuum to the side chambers, causing mechanical stretching of the membrane that forms the alveolar-capillary barrier. (B) Cartoon illustrating the mechanical effects of inspiration in the living lung, where contraction of the diaphragm reduced pleural pressure (P_ip), resulting in distension of alveoli and physical stretching of the alveolar-capillary interface. (C) The three-layer system forming parallel microchannels with a porous membrane. Scale bar, 200 μm. (D) After the three layers are bonded together, the membrane layers in the side channels are removed, producing two side chambers to which vacuum is applied to cause mechanical stretching. Scale bar, 200 μm. (E) Actual images of a lung-on-a-chip device. Reproduced, with permission, from (279).

Mechanisms regulating endothelial permeability

Numerous investigators have invested considerable effort to understanding the complex mechanistic underpinnings of barrier homeostasis and dysregulation. Our understanding of the molecular mechanisms of endothelial permeability is complicated by the numerous agents that have been used to experimentally induce injury. Confounding matters further is the variable behavior of signaling pathways in vivo compared to in vitro cell culture systems, due to critical modifications in endothelial structure and/or environment (i.e., loss of glycocalyx, lack of interactions with underlying smooth muscle and matrix proteins, absence of flow, etc). As a consequence, agonism of several endothelial barrier disrupting pathways have produced differential effects on lung EC permeability in vivo compared to in vitro (641). Recent excellent reviews have discussed some of the controversies, and the signaling cascades underlying maintenance and loss of endothelial barrier function and various mechanisms by which fluid flux occurs in detail (146,327,413,646,651,727). Thus, in this section, we will highlight major pathways that have been implicated, focusing on some of the more recent findings.

Signals leading to initiation of barrier disruption

Multiple upstream mediators of intracellular events leading to cellular contraction, loss of cell-cell interactions and increased transcellular and paracellular flux have been identified [reviewed in (727)]. In particular, VEGF/VEGFR binding has been associated with a myriad of second messaging systems (727), including: (i) increased [Ca²⁺]_i; (ii) generation of inositol triphosphate (IP₃) and diacylglycerol, which activate Ca²⁺ release from intracellular stores and protein kinase C (PKC), respectively; and (iii) activation of Src and MAP family kinases, all of which have been implicated in endothelial permeability (as discussed later in this section). Human studies examining VEGF levels in patients with ARDS showed higher VEGF levels in plasma and lower levels in bronchoalveolar lavage fluid (626). When plasma from patients with ARDS was perfused atop confluent microvascular ECs, paracellular leak occurred that was attenuated with neutralization of plasma VEGF (626). Although animal models have shown injurious effects of VEGF on the lung vasculature, there is also evidence suggesting that VEGF may be protective (437). It is likely that VEGF plays a dual role in the pulmonary vasculature, either pathologic or protective depending on the stage and type of injury.

Clinically, sepsis is one of the major causes of ARDS. Bacterial toxins activate inflammatory cascades upon recognition by the family of toll-like receptors. As might be expected in this case, ligation of toll-like receptor 4 by LPS plays a key role in initiation of endothelial hyperpermeability in endotoxin-mediated barrier injury (130). In addition, other upstream receptors have been shown to be involved in barrier regulation; thrombin increased EC permeability in EC monolayers (315) and increased macrovascular (but not microvascular) EC permeability in isolated, perfused rat lungs (636).

In addition to direct effects on ECs via surface bound receptor signaling, LPS promotes barrier disruption via initiating derangements of extracellular endothelial components, including the glycocalyx (552,715). For example, degradation of the glycocalyx allows for exposure of previously hidden binding sites for inflammatory cells and other proinflammatory molecules on the EC surface (715). Moreover, the glycocalyx is connected to the cytoskeleton and focal adhesions, allowing for potential direct cross-talk between alterations in the extracellular environment and intracellular components known to modify cell shape. During inflammatory states, neutrophils traversing the lung exit the vasculature and cross the endothelial barrier at the capillary level (657). LPS-mediated loss of the glycocalyx resulted in neutrophil attachment to ECs, the first step in endothelial transmigration. It is well recognized that neutrophilic infiltration of the interstitium contributes to the pathogenesis of ARDS by generating local stimuli that are injurious to the endothelium (699), such as ROS (generated by neutrophils via NADPH oxidase pathways) and release of inflammatory factors stored in neutrophil granules, including various proteases (such as matrix metalloproteinase and elastase) which may injure the endothelium through indiscriminate proteolysis of nearby proteins. Proteases likely represent one in a series of neutrophil-dependent injurious stimuli that are released in ARDS; indeed, inhibition of neutrophil elastase alone has not been shown to be of clinical benefit in ARDS (294).

Finally, changes in blood flow during ischemia may also contribute to barrier dysregulation. It is now appreciated that the presence of pulsatile flow in the luminal side of ECs exerts tonic effects, and that changing flow across the endothelial layer initiates multiple secondary signaling pathways that affect the function of junctional proteins such as α-catenin and VE-cadherin, destabilizing cell-cell adhesions and promoting paracellular fluid flux (240).

Transduction of the barrier disruption signal

Regardless of the method by which the signal for barrier disruption is generated, it is now recognized that intracellular signaling converges on several key pathways. In particular, production of ROS (from NADPH oxidase or other sources), is a common event, whether generated in ECs or surrounding cells (i.e., neutrophils). ROS serve important signaling functions at baseline levels, with activation of many more pathways when ROS levels are increased (97). Indeed, recent evidence has implicated proteins that participate in endogenous ROS generation and regulation, such as NADPH oxidase and peroxiredoxin (170, 200) as playing critical roles in initiation of the intracellular signaling in acute lung injury and ARDS.

Another early signaling event activated by numerous barrier disrupting agents is increased [Ca²⁺]_i. The increase in [Ca²⁺]_i may occur as a direct consequence of receptor activation, or secondary to ROS (66). Both influx from extracellular sources, through TRP channels such as TRPC1, TRPC4, and TRPV4 (13, 631, 634), store operated channels such as Orai1 (200) and release from internal stores (via IP₃) lead to activation of various Ca²⁺-dependent systems. For instance, increases in [Ca²⁺]_i can directly activate components of the contractile apparatus [primarily through myosin light chain kinase (MLCK)]. [Ca²⁺]_i may also play an important role in linking membrane events to the cell-cell junction complex that controls paracellular permeability, with a rise in [Ca²⁺]_i ultimately inducing hyperpermeability via disruption of the junctional complex and/or via effects on cytoskeletal elements (13,200,634).

While much of the work examining the role of [Ca²⁺]_i in EC barrier function has been performed in vitro, in vivo Ca²⁺ dynamics in ECs include mechanisms such as wave propagation of rises of [Ca²⁺]_i from cell to cell in an oscillatory fashion (717) and spread of Ca²⁺ though gap junctions (483) that are not seen in vitro. The extent to which these mechanisms may also play a role in regulating endothelial permeability, under both basal and stimulated conditions, is still being investigated.

Elevations in [Ca²⁺]_i can also activate certain isoforms of PKC, which can lead to disassembly of junctional components as well as other barrier disrupting effects through activation of MLCK (631). In addition, Ca²⁺-dependent activation of PKC may serve as a starting point for activation of several downstream second messenger systems, including MAP kinases, focal adhesion kinase, and small GTPases (727), several of which have been implicated as part of a complex system of second messengers that effect hyperpermeability in ECs. For example, Rho-GTPase members of the Ras family of proteins associate with the actin cytoskeleton in ECs and are involved in thrombin-induced changes in endothelial permeability. Within this family, RhoA diminishes barrier function through its effects on myosin light chains (146) while Rac1 improves barrier function by stabilizing adherens junctions (61, 147, 327). Along these lines, silencing or pharmacological inhibition of the downstream effector molecule, ROCK1, prevented F-actin organization induced by in vivo agonism of the thrombin receptor, but had no effect on hyperpermeability (45). Rather, ROCK2 activation was established as the main mediator of pulmonary edema, via regulation of basal tension and cell-cell adhesion. Interestingly, dual ROCK1/ROCK2 inhibition was additive. These results challenge the notion that formation of actin stress fibers is sufficient for reducing barrier function and suggest that F-fiber formation only increases permeability when cell junctions are weakened. The MAP kinase pathway has also been implicated in exerting control over actin and adherens junction organization (44,321,327).

Other kinase signaling cascades, including members of the Src family kinases (SFKs), have been shown to be activated as part of the global increase in kinase/phosphatase ratio that occurs following an increase in ROS. As such, various extracellular mediators known to increase ROS, including TNFα and thrombin, activate SFK members (273). SFK have been subcellularly localized near cell junctions, and interact with virtually all of the major components of the junctional/contractile apparatus (272). In particular, SFKs have been shown to directly phosphorylate adherens junction targets, including VE-cadherin, and facilitate adherens junction disassembly (207). SFKs can also interact with components of transcellular flux pathways such as caveolin-1 (272). While SFKs are known to be expressed in lung ECs (273) and phosphorylate targets critical to barrier function in various types of ECs in vitro (727), the role of specific SFK members in regulating macro- and microvascular EC function and their participation in mechanisms mediating barrier dysfunction and lung injury are largely not understood.

Finally, NO is another signaling molecule that has been studied intensely as a regulator of endothelial barrier function. The effect of NO on the microvasculature is complex and not fully understood, with lines of evidence supporting both barrier protective and barrier disruptive behavior (149,726). For instance, cGMP may offer protection from oxidative stress in microvascular ECs (598) and ROS-induced endothelial barrier dysfunction in vitro (427), yet NO-induced elevations in cGMP levels have been implicated in permeability increases following high tidal volume ventilator injury in mice (551). The reasons for these discordant results is unclear; however, emerging evidence suggests that subcompartmentalization of messengers like cGMP, cAMP, kinases and eNOS (541) may allow variable effects on barrier function based on subcellular localization. Lipid raft domains may serve as platforms for aggregation of several culprit second messenger proteins such as various protein kinases, tyrosine kinases, receptors such as VEGR and TRPs and other Ca²⁺ channels, allowing for rapid, localized signal transduction to induce rises in intracellular Ca²⁺. Interactions between signaling pathways may also complicate the issue, as the cGMP-dependent injury observed with high tidal volume was coupled with increased hydrolysis of cAMP (551), which offers barrier protective effects when localized near the plasma membrane (548,646).

For most of the pathways outlined earlier, the effect of various second messaging systems on barrier function likely varies as a function of injury type (e.g., ROS vs. shear stress-induced endothelial injury) as well as the endothelial phenotype under study (macrovascular vs. microvascular). Moreover, other experiment-dependent characteristics, such as species used, the presence of flow and in vivo versus in vitro preparations may also have an impact.

Targets of barrier disruption signaling

As touched upon in the preceding section, activation of the multiple kinase families, intracellular Ca²⁺ and other messenger systems ultimately converge on several key effector proteins that control endothelial barrier function. These include: (i) the contractile apparatus (Table 1), which is comprised of actin filaments, phosphorylated myosin light chains, and various associated proteins which control actin fiber polymerization, rearrangement, degradation and actin-myosin binding (146); (ii) members of the adherens junctions, encompassing VE-cadherin, catenins, and other adaptor molecules that connect the junctions to the contractile apparatus; (iii) components of focal adhesions such as integrins, which tether the cell to the underlying matrix; and (iv) pathways of transcellular fluid transport, controlled primarily via interactions between second messenger kinases and caveolin-1 (303).

Table 1.

Proteins Involved in Regulating Endothelial Cell Barrier Function [Adapted from Dudek and Garcia (146)]

Protein	Function
Spectrin	Cross-links F actin at periphery and stimulates myosin II ATPase
α-Actinin	Links actin cytoskeleton to focal adhesions; displacement from actin stress fibers disrupts the microfilaments
Fimbrin	Links actin cytoskeleton to vimentin network at cell adhesion sites
Cortactin	Phosphorylation by p60src reduces actin bundling activity
Cofilin	Rho pathway inhibits depolymerization activity during stress fiber formation
hsp27	Phosphorylation by p38/MAPKAP induces stress fiber formation
VASP	Stimulates actin nucleation and polymerization at focal adhesions
Arp2/3	Produces branching actin network at cell periphery through interaction with Wiskott-Aldrich syndrome protein and cortactin
Profilin	Overexpression inhibits actin stress fiber formation
Gelsolin	Inhibition of activity decreases stress fiber-dependent contraction
MLCK	Activation produces stress fibers, cellular contraction, EC permeability
Filamin	Phosphorylation by CaMKII permits reorganization of cortical actin
Caldesmon	Facilitates actomyosin interaction
Vinculin	Binds actin and catenin at junctional sites

Abbreviations: hsp27, heat shock protein 27; VASP, vasodilator-stimulated phosphoprotein; MLCK, myosin light chain kinase; CAMKII, Ca²⁺/calmodulin-dependent protein kinase II; MAPKAP, MAP kinase-activated protein kinase 2.

Of all of these systems, the cytoskeleton plays a critical role in regulating barrier strength. Complete reviews of the role of the cytoskeleton in regulating permeability have been published (146, 651). Adherens junctions, in particular, are directly linked to cytoskeletal elements, like actin, such that actin/myosin contraction “tugs” on the adherens junction and VE-cadherin pairs from adjacent cells to dissociate them from each other. Thus, the major regulation of junction integrity occurs either at the level of expression of adherens junction proteins (i.e., assembly or disassembly of VE-Cadherin on the cell membrane) or at the level of contractile apparatus attachment (146). Actin filament formation and degradation is controlled by a family of proteins (i.e., cofilin) and these proteins then serve as targets for kinase systems as a means of regulating paracellular barrier function. Other pathways that are independent of actin and myosin light chain interaction have been identified as well (646), including pathways that utilize nonactin components of the cytoskeleton, such as intermediate filaments and microtubules, although these mechanisms are less well understood (146,632). Lastly, while paracellular leak is thought to be the primary mechanism driving increases in endothelial permeability, the transcellular pathway has been shown to be important in maintaining basal barrier homeostasis as well contributing (albeit to a smaller extent) to lung injury. Transcellular permeability involves caveolae-mediated transport of molecules like albumin and, like paracellular mechanisms, is tightly regulated by multiple families of kinases, including Src kinases (272).

It is clear that under pathological conditions, numerous stimuli can act at a variety of points to modify EC behavior and promote EC barrier dysregulation. In addition to investigation of signaling pathways that worsen barrier function, there has been recent interest in upstream proteins that exert barrier protective effects, such as S1P and angiopoietins. Interactions between S1P and one of its receptors, S1P1R, are thought to strengthen the endothelial barrier through a variety of mechanisms, including stabilization of the adherens junction and rearrangement of the actin cytoskeleton in a protective configuration [reviewed in (452,640,668)]. Consistent with these suppositions, S1P administration attenuated LPS-induced increases in bronchoalveolar lavage fluid protein and Evans blue dye extravasation in a murine model of endotoxin-induced lung injury (493) and attenuated high tidal volume induced increases in bronchoalveolar lavage fluid protein in a canine model of lung injury (608). Interestingly, prolonged applications of S1P promoted barrier dysfunction, suggesting that S1P-S1P1R binding can have temporally different effects, perhaps due to receptor internalization following ligation (564). Unlike S1P1R, ligation of S1P2R has opposing effects, increasing vascular permeability in response to exogenous ROS (542). The exact effects of the other members of the S1P receptor family on vascular permeability remain incompletely understood.

Ultimately, understanding the pathways involved in the pathogenesis of edema in patients with acute lung injury and ARDS is required to develop therapeutics to reduce mortality. Whether pharmacological targeting to disrupt pathways that have already been identified to be involved in reducing barrier integrity, or to enhance pathways that may improve barrier function, can be translated to decrease mortality associated with ARDS remains to be determined.

Pulmonary Hypertension

Pulmonary hypertension (PH) is a complex, progressive, and ultimately fatal condition arising from a variety of etiologies. Defined by the hemodynamic criteria of mean P_PA ≥ 25 mmHg at rest, PH has been classified into five major categories: (i) pulmonary arterial hypertension (PAH), including idiopathic, heritable and drug/toxin-induced PH; (ii) PH due to left heart disease; (iii) PH due to interstitial lung diseases and/or hypoxia; (iv) chronic thromboembolic PH; and (v) PH with unclear and/or multifactorial origin, including hematologic and systemic disorders (27,197). Due to the complicated nature of the disease, the exact pathogenesis of PH is poorly understood; however, genetic mutations have been identified as an underlying cause in some cases of idiopathic and familial PAH, while global alveolar hypoxia is thought to be a causal factor in PH due to high altitude exposure, sleep-disordered breathing, and chronic lung diseases. In all cases, however, owing to increased afterload, right ventricular hypertrophy becomes a serious complication with the potential for ensuing right ventricular failure.

While the exact causes of PH remain under investigation, and are likely to vary with the underlying pathogenic or genetic cause, it is widely recognized that pathophysiology results from sustained active vasoconstriction and pulmonary vascular remodeling. Dysfunction of the pulmonary endothelium, which is a rich source of both vasodilators and vasoconstrictors, is a likely contributing factor. In the normal lung, the balance of vasodilators to vasoconstrictors favors low PVR. However, with EC dysfunction, release of NO and PGI₂ is impaired while release of ET-1 and thromboxane may be augmented. A shift in the balance between endogenous vasodilators and vasoconstrictors combined with remodeling of pulmonary blood vessels results in a net vasoconstriction and increased vascular resistance.

The remodeling of the pulmonary vasculature in PH is characterized to varying degrees by thickening of the intimal and/or medial layer of muscular vessels and the appearance of cells expressing smooth muscle specific markers in precapillary arterioles (distal muscularization), resulting from excessive proliferation and migration of PASMCs (567, 596, 650). In some forms of severe PAH, increased medial muscularity is coupled with the development of occlusive lesions, involving PASMCs, ECs and possibly cells of nonvascular origin (597, 637). Reductions in arteriolar caliber, whether from vasoconstriction or occlusion, clearly exert the greatest influence on PVR; however, decreased compliance (i.e., increased stiffness) in the elastic proximal pulmonary arteries may also increase right ventricular afterload (199,394,596,647).

Depending on the underlying cause, the relative contributions of vasoreactivity and vascular remodeling to elevated P_PA vary. Initially, all forms of PH were thought to arise from “fixed” constriction due to inward remodeling and narrowing of the vascular lumen. Over the last 15 years, however, accumulating evidence suggests a large portion of the “fixed” component was actually due to incomplete relaxation and that remodeling resulting in luminal encroachment is less prevalent (286, 410, 595), restricted primarily to severe forms of PAH. This is particularly true in the case of hypoxia-induced PH, where medial remodeling was shown to occur in an outward manner, and likely contributes to elevated PVR via hyperreactivity to constricting agents. Several recent reviews have provided detailed descriptions of the evidence for remodeling in various forms of PH, as well as in depth coverage of the potential cellular mechanisms (508, 549, 567, 595, 597). In the following section, we will briefly review the current knowledge and highlight recent advances in identifying some of the factors involved in both the contractile and remodeling process.

Pulmonary hypertension in humans

In humans, vasoconstriction is relatively easy to identify in PH with administration of vasodilators, including Ca²⁺ channel blockers, inhaled NO, prostacyclin or ROCK inhibitors (597, 627). In contrast, most evidence for remodeling comes from postmortem or postoperative tissue specimens. For example, examination of lung tissue from PAH patients revealed early abnormalities that included medial hypertrophy, adventitial thickening, and extension of muscle down the vascular tree, with vaso-occlusive lesions developing at later stages (596, 637). Similarly, COPD patients with PH (31, 697, 703) and South Americans living at altitude (22, 253, 446, 447, 492) also exhibited arteriolar muscularization, increased intimal and/or medial thickness and reduced lumen area.

While development of PH is common in individuals that are born at sea level and subsequently move to high altitude, several populations have been identified that have evolutionarily adapted to living at altitude, including highlander populations in the South American Andes, Tibet and the East African Plateau. All reside at high altitude without developing substantial PH, yet genetic diversity appears to underlie the adaptation achieved by these populations, as can be inferred from the differences in both phenotype and genotype. For example, Tibetans and Andeans differ substantially in resting ventilation, hypoxic ventilatory response, oxygen saturation, and hemoglobin concentration [reviewed in (40,41)]. In contrast to both the Tibetans and Andeans, Ethiopians have high arterial oxygen saturations at altitude (43). While data is not available from the African population, the native high-altitude-adapted population of Tibet does not have increased pulmonary vascular smooth muscle (235,242), while healthy Andeans exhibit some remodeling (22).

Given that Tibetans have resided at high altitude longer than any other population, the difference in vascular response may be due to variations in genetic adaptation. Indeed, a loss-of-function mutation in the Hif2a gene was identified in Tibetans and found to correlate with reduced P_PA (42,58,583,644,716). Conversely, genetic mutations leading to HIF-2α gain of functionality are associated with development of PH in humans (179, 195) and in mice (619). Finally, individuals with Chuvash polycythemia, a rare congenital genetic disorder resulting in hypomorphic alleles for the gene encoding von Hippel-Lindau protein, a main component in the HIF degradation pathway, exhibit elevated HIF levels, increased pulmonary vascular sensitivity to hypoxia and susceptibility to developing PH at sea level (85,585).

Genetics also play a role in some cases of PAH. Initial discovery of a gene locus at chromosome 2q31–32 (435,459) led to the subsequent identification of bone morphogenetic protein receptor II (BMPR2) as the gene of interest (138, 290). BMPR2 is a serine/threonine kinase receptor that binds members of the TGF-β superfamily, including BMPs. BMP ligands bind type-II receptors (BMPR2 and Act2A and 2B), which recruit type-I receptors (ALK2, 3, and 6) to form a heterodimeric complex whereby type-II receptors induce type-1 receptor phosphorylation, leading to initiation of intracellular signaling cascades that ultimately lead to activation of nuclear transcription factors to upregulate or suppress target genes (577).

Mutations in the gene encoding BMPR2 occur in ~70% of patients with heritable PAH (10,112,113,138,385,628) and ~25% of idiopathic PAH patients (154). Although the number of unique BMPR2 mutations identified in patients approaches 300, all result in loss of protein function, encompassing insufficient protein production, dysfunctional protein trafficking and localization, or abnormal protein signaling (126). Interestingly, the penetrance of the BMPR2 mutation is low with an estimated lifetime risk of 20% (343), suggesting a “second hit” may be required for disease manifestation. Factors influencing development of PAH in those with BMPR2 mutations include female gender (343). Additional genes that may be associated with susceptibility include ALK-1, SMAD9, ENG, BMPR1B (ALK6), CAV1, KCNK3, and EIF1AK4 [reviewed in (25,384)].

Pulmonary hypertension in animal models

To better understand functional and structural changes in the hypertensive lung and to begin to identify underlying causes, several animal models of PH have been utilized (685). Unfortunately, no single model perfectly replicates human PAH. Nevertheless, these models provide opportunities for studying the development, progression and mechanistic underpinnings of PH and evaluating potential therapeutic treatments.

Hypoxia

In 1915, altitude was identified as a primary cause of lower chest edema, cardiac dilation and enlargement, and eventual heart failure in cattle living in the Colorado mountains (>8000 ft) (221). This recognition provided the first animal model of hypoxic PH, which has been used to explore the pulmonary and cardiac changes induced by exposure to chronic hypoxia. In later studies, extensive vascular remodeling was reported, characterized by collagen deposition, adventitial and medial thickening, and arteriolar muscularization (527, 594, 596). As with humans, there is variability in the development of hypoxic PH in cattle, which may have a genetic basis (575, 681, 698). Recently, SNP analysis of four candidate genes in cattle with severe hypoxic PH failed to uncover obvious SNPs associated with high-altitude susceptibility, although SNPs in additional candidate genes were identified and will require further investigation (456). Moreover, genetic comparison between altitude-susceptible and -resistant cattle revealed alterations in several gene networks whose biological function involve signaling pathways known to contribute to human PH, including BMP, TGF-β, VEGF, ERK/MAPK, NF-kB, and HIF signaling (456). The extent to which each of the identified pathways confers susceptibility to hypoxia-induced PH in these animals remains to be determined, but it should be noted that all of the pathways identified have been implicated in humans and other species.

Rodents exposed to CH develop predictable and reproducible vasoconstriction and vascular remodeling reminiscent of humans with hypoxia-associated PH. Rats exposed to several weeks of simulated high-altitude exhibit substantial increases in P_PA and right ventricular mass, accompanied by increased vascular wall thickness, due to PASMC hypertrophy and hyperplasia, and distal muscularization (513,596). While imaging studies initially appeared to suggest that remodeling in these models reduced luminal diameter and resulted in rarefaction, or vascular pruning (263, 512), later observations showing that remodeling in this model appears to occur in an outward manner (271) and that administration of ROCK inhibitors acutely normalized P_PA (448), suggest that the imaging results may have reflected incomplete reversal of vasoconstriction or variable perfusion (50,406,595). Mice also develop PH upon exposure to CH, although PASMC proliferation is minimal (38, 470, 596) and remodeling is most often observed as muscularization of distal vessels and thickening and functional stiffening of proximal, conduit arteries (593,594,596). Across species, the phenotype observed with chronic hypoxia exposure is consistent, with lack of occlusive lesions and reversal of P_PA with return to normoxic conditions.

Other models in which hypoxia plays a contributing factor have been developed, including the Fawn-hooded rat (FHR) and the Sugen/hypoxia (Su/Hx) model. FHRs spontaneously develop PH with age, with mild CH accelerating the process (450, 544). In this model, extension of muscle into peripheral vessels and medial hypertrophy of proximal arteries was observed, as was reduced vessel filling in imaging studies (544), but whether this reflected development of vaso-occlusive lesions, which have not been reported, or severe vasoconstriction (595) is unclear. In light of findings that acute administration of ROCK inhibitors resulted in near normalization of P_PA, it would appear that the latter is more likely. In the Su/Hx model, a single dose of the VEGF receptor inhibitor, SU5416, is followed by exposure to CH (3,621). VEGF receptor inhibition initially causes EC apoptosis; subsequent hypoxic exposure may stimulate proliferation of a subset of ECs that are apoptosis-resistant, causing occlusive lesions reminiscent of PAH in humans and PASMC proliferation. PH induced in the Su/Hx model is irreversible even after return to normoxia, and ROCK inhibitors only partially reversed P_PA (633), suggesting a significant portion of the increased PVR may be due to remodeling and occlusion of flow in small vessels. Interestingly, a detailed temporal characterization of the Su/Hx model revealed that while vaso-occlusive lesions are present early (2-week posthypoxia) in the course of the Su/Hx model (621,633) concentric neointimal, or plexiform, lesions did not form until later (i.e., 8-to 13-week posthypoxia) time points (3). These results would suggest that the plexiform lesions observed in PAH patients, while perhaps contributing to progression/severity of the rise in P_PA in late disease, may in fact be a consequence of EC damage during PH rather than the inciting cause. The Su/Hx model has been adapted to mice, which exhibited increased arterial muscularization and collagen deposition in the media and adventitia, vaso-occlusive lesions and ultimately right heart dysfunction; however, in contrast to rats, mice required weekly injections of SU5416 and PH resolved upon return to normoxia (110).

Monocrotaline

Following ingestion of seeds from Crotalaria spectabilis, monocrotaline (MCT) is metabolized in the liver into MCT pyrrole, a toxic substance that causes inflammation and endothelial injury. In rats, a single injection of MCT results in the development of PH within 2 to 3 weeks (317). The MCT model is widely used due to the ease of inducing severe PH without requiring specialized equipment or housing as is needed with hypoxia models. Histological analysis has shown that while smooth muscle content increases in arteries of all sizes, occlusive lesions do not develop (416). However, when MCT-induced wall injury/EC dysfunction is combined with pneumonectomy, intimal changes occur, including development of distal vascular occlusive lesions (467, 620). Despite substantial remodeling, ROCK inhibitors acutely normalize P_PA in MCT-treated rats (448), suggesting that sustained contraction is a major component of PH in this model. Unlike human disease, the fact that >30 agents effectively reverse MCT-induced PH (596) raises questions as to the relevance of the MCT model.

Genetically engineered murine models

With advances in knowledge regarding some of the underlying causes of human PH, attempts have been made to recapitulate the disease using genetically engineered mice. For example, it is now clear that 5-HT was involved in the development of PAH associated with use of the diet medications, aminorex or fenfluramines, which functioned to increase 5-HT availability. Under normal conditions, the vast majority (>99%) of 5-HT is stored in platelets and rapid metabolism of 5-HT results in near negligible plasma levels (390). With elevations in plasma 5-HT, activation of cell surface receptors can mediate pulmonary vasoconstriction (386,389,429), or 5-HT can be taken up into SMCs. Receptor engagement and/or internalization and receptor-independent signaling results in generation of ROS, activation of MAPKs and ROCK, and induction of genes involved in regulating cell growth (390). In PASMCs, 5-HT influx and efflux is controlled by membrane SERTs (390), the expression of which are upregulated in PASMCs from PAH patients (390), and increased 5-HT transport into PASMCs enhances, and SERT inhibition prevents, proliferation (150, 387). To mimic the enhanced SERT expression observed in PAH patients, mice overexpressing the gene for human SERT were generated, with SERT overexpression inducing medial thickening and distal muscularization in normoxic female, but not male, mice (390, 690), mimicking the disproportionate female predominance seen in human PAH. On the other hand, male mice deficient in tryptophan hydroxylase 1, the enzyme responsible for peripheral 5-HT synthesis, were resistant to hypoxia-induced PH (428). While increasing 5-HT internalization via SERT over-expression is sufficient to induce ROCK-dependent remodeling in mice (397), coordination between SERT and 5-HT receptor binding may also occur. For example, inhibiting 5-HT_1B receptors by silencing or pharmacological blockade reduced PASMC proliferation and CH-induced remodeling (390,430), responses mediated by SERT (150,387).

Another model of PAH generated based on data from patients is a murine model with loss-of-function mutations in BMPR2. Transgenic mice with global heterozygous BMPR2 mutations exhibit variable PH morphology, with some, but not all, developing increased wall thickness compared to wild type mice (49). Subjecting these mice to hypoxia for 3 weeks had no effect on right ventricular pressure (182,588), whereas extending hypoxic exposure to 5 weeks resulted in mild disease presentation (182), suggesting that the mutation alone may not be sufficient to generate PH. When BMPR2 loss-of-function was restricted to smooth muscle, mice developed PH with increased medial thickness and muscularization of distal pulmonary arteries, but no intimal lesions (684). Clearly, these models present complex morphologies and the results are consistent with the fact that only ~20% of individuals with heterozygous BMPR2 mutations develop PAH, suggesting a second genetic or environmental “hit” may be required (457).

Factors involved in the pathogenesis of pulmonary hypertension

The animal models described above have been utilized extensively to identify mechanisms responsible for development of PH and test potential treatments. Given the multiple classifications of human PH, none of the preclinical models perfectly captures the pulmonary-specific characteristics of severe human disease; however, these models have provided valuable insight into the cellular modifications and signaling pathways contributing to disease pathogenesis in PH, with the majority of work centered on PASMCs and ECs (Fig. 8). Detailed reviews regarding the cellular mechanisms involved in the pathogenesis of vasoconstriction and vascular remodeling in PH have been published (549,567,593,596). In this section, we will briefly review the most well-established candidates.

Figure 8.

Illustration of putative mechanisms involved in the pathogenesis of pulmonary hypertension. Abbreviations: 5-HT, 5-hydroxytryptamin; K- and Ca-channels, potassium and calcium channels; AEC, alveolar epithelial cells; BMP, bone morphogenetic protein; cGMP, cyclic guanosine monophosphate; ECM, extracellular matrix; EGF, epidermal growth factor; EPC, endothelial progenitor cells; HIF, hypoxia inducible factor; MMPs, matrix metalloproteinases; NADPH, nicotinamide adenine dinucleotide phosphate; NO, nitric oxide; PDE, phosphodiesterase; PDGF, platelet-derived growth factor; PGI₂, prostaglandin I₂; Rho-Ki, Rho kinases; ROS, reactive oxygen species; sGC, soluble guanylate cyclase; TGF, transforming growth factor-β; TK, tyrosine kinase; TKi, tyrosine kinase inhibitor; TRPC, transient receptor potential cation channels; VEGF, vascular endothelial growth factor. Reproduced, with permission, from Schermuly et al. (549).

Ion channels

In chronic hypoxia models, the earliest study examining functional changes in PASMCs revealed depolarization of the membrane potential (605), a finding confirmed in other species [reviewed in (569)]. Reduced K⁺ channel expression and activity were subsequently identified as contributing factors to depolarization in hypoxic PH (81, 419, 569, 584, 662). Experimental manipulations that augmented PASMC K⁺ channel expression/activity reduced remodeling and reduced right ventricular systolic pressure in hypoxic animals (419, 504). The initial reports of downregulated K⁺ channels and membrane depolarization were hypothesized to drive activation of voltage-gated Ca²⁺ channels and Ca²⁺ influx, leading to elevated [Ca²⁺]_i. Increased [Ca²⁺]_i results in enhanced PASMC contraction (572), growth (223,342,426) and migration (348) and has been documented in cells from hypoxic animals (572). However, inhibitors of voltage-gated Ca²⁺ channels have little effect on PASMCs from hypoxic animals (355,571,666), but can be activated by agonists (259,382) and contribute to stimulated proliferation (533). Thus, voltage-gated Ca²⁺ channels may participate in the remodeling process in PH, particularly in the presence of excessive growth factors.

It is now well established that sustained Ca²⁺ influx is mediated primarily by upregulation of NSCCs (565, 569), which are composed of TRPC proteins (355, 666). Unlike voltage-gated Ca²⁺ channels, NSCCs are not activated by depolarization, but can be modulated by phosphorylation, receptor activation or store depletion (260, 464, 665). Increased abundance of TRPC proteins was observed in PASMCs derived from animals subjected to chronic hypoxia (355,666). Decreasing the activity of NSCCs, either pharmacologically or by RNA silencing, reduced [Ca²⁺]_i (355,666).

Consistent with results obtained in the chronic hypoxia model, PASMCs from humans with PAH are also depolarized (725) and exhibit reduced K⁺ channel expression and activity (73, 730). Similar reductions in K⁺ channel activity were reported in the FHR model (72). Augmenting PASMC K⁺ channel expression/activity increased apoptosis in PASMCs from PAH patients (73). Thus, modulating K⁺ channel activity modifies PASMC growth; likely by conferring resistance to apoptosis (81,526).

Increased [Ca²⁺]_i also has been documented in PASMCs from MCT-treated rats (365) and PAH patients (223, 587). This is consistent with findings demonstrating increased abundance of TRPC proteins in PASMCs derived from PAH patients (339,723). Moreover, a gain-of-function single nucleotide polymorphism in TRPC6 was recently identified in patients with PAH (724).

Na⁺/H⁺ exchange

The Na⁺/H⁺ exchanger (NHE) is a major contributor to maintenance of PASMC pH homeostasis (391, 511). Increased NHE activity was associated with growth factor-induced proliferation (509) suggesting a potential role for NHE in PH. Indeed, exposure to CH increased expression of NHE isoform 1 (NHE1) in PASMCs and NHE activity causes an alkaline shift in pH (531,566), while pharmacological inhibition (510) or genetic deletion of NHE1 decreased hypoxia-induced PH, vascular remodeling and PASMC proliferation and migration (720, 721). These data provided firm evidence that NHE1 plays an important role in the pathogenesis of PH, although the exact mechanism by which this occurs is unclear. Recent studies have provided tantalizing clues, suggesting that loss of NHE1 increased p27, a cyclin-dependent kinase inhibitor, and decreased activation of ROCK and E2F1, a nuclear transcription factor that controls proliferation (720, 721). These results suggest NHE1 represses a growth inhibitor pathway while stimulating proliferation.

Rho kinase

Even under basal conditions, ROCK appears to be activated and provides a minimal amount of basal tone in the lung, as demonstrated by reduced P_PA when ROCK inhibitors are infused (92). With exposure to hypoxia, the ROCK system is upregulated (78, 300), and contributes to elevated PVR. Indeed, the ROCK signaling pathway plays a central role in mediating vasoconstriction in all PH models [reviewed in (466, 671)] and is implicated in the remodeling process (167, 465). ROCK activation is necessary for PASMC migration and proliferation (212,366,465), and while acute administration of ROCK inhibitors markedly reduces right ventricular systolic pressure in most preclinical models (448, 449, 647), prolonged inhibition of ROCK reduces vascular remodeling in CH mice (160, 286). At least part of the mechanism by which ROCK modulates PASMC function occurs via cytoskeletal rearrangement in PASMCs and ECs, with RhoB contributing to hypoxia-induced migration and proliferation in both cell types (702). ROCK also converges with other mechanisms involved in remodeling, as ROCK activation is a consequence of several growth factors (546), opens Ca²⁺ channels (382,665), and stimulates the activity of NHE1 (642).

BMP signaling

BMPs signal through activation of type-I and type-II receptors, activating SMAD-dependent and -independent pathways [reviewed in (577)]. Interactions with type-I receptors activate SMAD1/5/8, which subsequently bind SMAD4, leading to activation of transcriptional responses (377); whereas SMAD-independent signaling involves MAP kinases, phosphatidylinositol 3-kinase/AKT and PKC (713). In PH, the ligands BMP2, −4, and −7 and the receptors, BMPR1 and BMPR2 have been most studied. BMP2 and BMP7 appear to play protective roles in the pulmonary circulation (434); indeed, application of BMP2 exerted an antiproliferative effect on PASMCs, and loss of BMP2 resulted in greater susceptibility to hypoxia-induced PH (17). BMP4 also exerts a growth inhibiting effect in PASMCs from proximal vessels (434), but in PASMCs from distal vessels BMP4 increased [Ca²⁺]_i via TRPC upregulation (378) and induced proliferation (714). Moreover, mice with partial deficiency for BMP4 were protected against development of hypoxia-induced PH (181,182). In addition to BMP2, −4 and −7, the BMP family contains a variety of endogenous antagonists and binding proteins which may regulate BMP receptor binding and activity in vivo (377,577).

In PAH patients, BMPR2 mutations lead to loss of BMP signaling, preventing the antiproliferative effects of BMP2 (126). Expression of both BMPR1 and BMPR2 is decreased in non-PAH forms of the disease (145) and in CH models (612), suggesting abnormal BMP signaling may be common in many forms of PH. On ECs, activation of BMPR2 by BMP2 or BMP4 was linked to NO production (201), while loss of BMPR2 results in apoptosis (625). Since EC apoptosis and/or dysfunctional production of vasodilators are likely early events in the pathogenesis of PAH, these results may explain some of the protective effects of BMPs in vivo.

Endothelin-1

A role for ET-1 in mediating PH has been suspected nearly since the discovery of the protein. Indeed, early on it was recognized that circulating ET-1 levels were increased in all animal models and human forms of PH (563, 570). In addition to vasoconstriction, ET-1 exerts promigration and proliferation actions on vascular SMCs (563). Moreover, ET receptor inhibitors prevented and reversed PH and vascular remodeling in several animal models [reviewed in (125)], and reduced proliferation in PASMCs from PAH patients (339), paving the way for use of ET receptor inhibitors in clinical practice. ET receptor inhibitors improve exercise capacity and survival (125), but whether this is due to reduced vasoconstriction or remodeling, or other effects of the drugs, is unclear.

Transcription factors

Two major transcription factors have been implicated in the development of PH: HIFs and nuclear factor of activated T-cells (NFATs). Increases in [Ca²⁺]_i activate several downstream signal transduction pathways and transcription factors are activated that could be involved in PASMC proliferation [reviewed in (336)], in particular, NFAT. In turn, NFAT reduces K⁺ channel expression and increases proliferation (73), linking alterations in [Ca²⁺]_i to dysregulated K⁺ channel expression/activity and PASMC growth. Consistent with this possibility, genetic deletion or pharmacological inhibition of NFAT prevented and/or reversed hypoxia-induced remodeling and PH (57,73).

The discovery of HIF-1 (560) provided an attractive candidate for mediating hypoxia-induced PH. As noted earlier, HIF-1β is constitutively expressed whereas HIF-1α is typically not detectable under normoxic conditions. Initial studies in mice heterozygous for a null HIF-1α allele (Hif1a^+/−) revealed a role for HIF-1 in the development of hypoxia-induced PH (568, 719), with Hif1a^+/− mice exhibiting impaired development of PH and reduced vascular remodeling in response to CH (719). PASMCs from Hif1a^+/− mice also exhibited reduced hypoxia-induced proliferation (565). Consistent with the results obtained in Hif1a^+/− mice, mice with homozygous, conditional deletion of HIF-1α in smooth muscle during exposure to chronic hypoxia exhibited attenuated development of hypoxia-induced pulmonary vascular remodeling and PH (29). Similar attenuation in CH-induced PH and vascular remodeling was observed in Hif2a^+/− mice (80).

HIFs are likely to be involved in numerous mechanisms that mediate PH. In addition to regulating both Ca²⁺ and pH homeostasis (566, 666), HIFs also regulate other factors involved in the pathogenesis of PH, including ET-1 and VEGF [reviewed in (505)]. While hypoxia is an obvious stimulant for HIF activation, HIFs also can be upregulated under nonhypoxic conditions by a number of factors (505), including ET-1 (497). Moreover, HIF-1α protein is detectable in PASMCs from FHRs (72) and lungs of PAH patients (18, 165), which is thought to be linked to mitochondrial dysfunction. Upregulation of HIF under nonhypoxic conditions could also be due to ROS generated from NADPH oxidase and activation of mTOR, both of which have been observed in cells from PAH patients (225) and lead to induction of prosurvival pathways (226,231,379).

Dysregulated cell metabolism

In recent years, the role of metabolic reprogramming has emerged as a potential mechanism underlying cellular responses during PH. PASMCs from Fawn-hooded (18), chronically hypoxic and Su/Hx rats (105) and PAH patients (704) exhibit a phenotypic switch such that instead of mitochondrial metabolism, these cells preferentially use aerobic glycolysis. This type of bioenergetic shift confers a survival/growth advantage since glucose/glycolytic intermediates are substrates utilized by the pentose phosphate shunt to generate reduced nicotinamide adenine dinucleotide phosphate and nucleotides. The mechanisms underlying the switch to aerobic glycolysis may be related to overactivation of glucose-6-phosphate dehydrogenase (105) and HIF-dependent upregulation of glycolytic (710) and pro-proliferative (105,106) genes. Additionally, hypoxia-induced upregulation of glucose-6-phosphate dehydrogenase activity was correlated with reduced expression of contractile proteins in PASMCs (106), suggesting that metabolic derangements may underlie the phenotypic switch from a “contractile” to “synthetic” SMC phenotype. Activation of HIF-1 was also associated with reductions in mitochondrial number in PAH ECs (165), while mitochondrial fragmentation was observed in PAH PASMCs (72). In PASMCs from PAH patients and hypoxia- and MCT-treated rats, HIF activation was associated with mitochondrial fission mediated by dynamin related protein-1 (400), which was reduced when dynamin-related protein-1 was inhibited with the peptide Mdiv-1. Mdiv-1 also attenuated PH and proliferation (400). Consistent with the metabolic derangements observed in cultured cells and animal models, in vivo imaging studies using 18-fluorodeoxyglucose in PAH patients verified increased glucose uptake in the lungs (383,704).

The immune system

It is now recognized that in addition to, or because of, genetic influences and shifts in cell metabolism, inflammation is a contributing factor in the development of PH. While certain autoimmune diseases (i.e., systemic sclerosis and systemic lupus erythematous) and infections (i.e., HIV and schistosomiasis) are known causes of PAH, examination of patient specimens has revealed perivascular inflammation in almost all forms of PH [(638) and reviewed in (151,507,514)]. Infiltrating immune cells, including T-cells, B-cells, macrophages, dendritic cells, and mast cells have been observed in the vascular lesions of PAH patients and substantial evidence has now accumulated demonstrating a critical role for macrophages (184, 630, 648) and mast cells (30, 121, 266) in promoting vascular dysfunction in animal models of PH. High levels of procontractile and pro-proliferative cytokines are observed in both patients and animal models of PH [reviewed in (151, 507, 514)], providing a link between inflammation and the vascular hallmarks of PH. In addition, circulating anti-endothelial (141, 615), anti-fibroblast (616) and anti-nuclear (528) autoantibodies have also been reported in PAH patients, although the extent to which these autoantibodies initiate or facilitate progression of disease is still under investigation.

Surprisingly, athymic rats given an injection of SU5416, which were hypothesized to exhibit less inflammation and thus diminished PH, instead developed severe PH in the absence of hypoxic stimulus and worse inflammation (622). Subsequent studies demonstrated that loss of regulatory T-cells, which limit perivascular inflammation and vascular injury, accounted for the exaggerated phenotype in the athymic rats (617). Similarly, hypoxia was not required to induce severe PH in rats given SU5416 following sensitization with ovalbumin (425), suggesting that an allergic inflammatory phenotype can facilitate development of PH.

Therapeutics for pulmonary hypertension

Work from preclinical models allowed the identification of individual modifiers of PH and pulmonary vascular remodeling and has provided most of the rationale for the development of therapies aimed at human disease, such as PGI₂ and endothelin receptor blockade. While these studies lead to the introduction of several clinically useful therapies for PH, all current therapies are aimed at reducing vasoconstriction or slowing the progression of vascular remodeling. There are currently no therapeutic options available as a means to deremodel the pulmonary vasculature after the disease is already established. While PGI₂ and ET-1 receptor blockers have been used clinically for over a decade (180), recent drugs added to the armamentarium include the phosphodiesterase type-V inhibitor, sildenafil, and the HMG-CoA reductase inhibitor, simvastatin (314). Both drugs were originally marketed for treatment of other disorders (erectile dysfunction and dyslipidemia, respectively), but based on their pharmacological actions were hypothesized to provide benefit by targeting abnormal signaling pathways in PAH. In the case of sildenafil, increasing cGMP could provide vasodilatory and anti-proliferative actions, while statins have multiple effects on endothelial function and may inhibit ROCK. In vitro experiments revealed that sildenafil restored BMP signaling in PAH PASMCs (711), whereas simvastatin increased BMPR2 expression in ECs (274), suggesting the possibility that these drugs might rescue BMP function in vivo. Sildenafil also reduced hypoxia-induced increases in [Ca²⁺]_i and TRPC expression in PASMCs (667), attenuated PASMC proliferation (689,711) and attenuated PH and vascular remodeling in preclinical models (239,711). Sildenafil is now routinely used clinically as an add-on therapy in the PH population. Promising trials using the next generation long-acting PDE5-specific inhibitor, tadalafil (196, 469), and an activator of sGC, riociguat (69,213,214,537,582), and subsequent recent approval of these drugs for clinical use demonstrate that the cGMP pathway remains a prime target in PAH.

Simvastatin also prevented PH in preclinical models (215, 314, 733), but had variable success in reversal protocols (216, 411, 623). Unfortunately, clinical trials in the PH population have failed to find sustained improvement in PH (316, 696). Among the potential therapies suggested by data from animal models, restoration of K⁺ channel activity/expression provides an attractive target for therapeutics aimed at reducing remodeling in PAH and may be the closest to actual clinical use. Animal studies using dehydroepiandrosterone, a steroid hormone that, among other actions, opens K⁺ channels (71) or dichloroacetate, a metabolic modulator that also increases the expression of K⁺ channels (412, 419), showed promising outcomes in CH and MCT-treated animals. In the Su/Hx model, dehydroepiandrosterone attenuated PH and preserved RV contractile function via reduced RV oxidative stress, capillary rarefaction, apoptosis, autophagy and fibrosis (14, 105, 521). While it is unclear whether the beneficial effects of either drug were specifically due to actions on K⁺ channels, the use of dichloroacetate to treat PAH is currently in Phase 1 clinical trials in Canada and a small pilot study found dehydroepiandrosterone reduced mean P_PA and PVR and improved 6-min walk distances in COPD patients with PH (148).

In contrast to K⁺ channels, little has been pursued with respect to other ion channels, exchangers, and PH therapeutics. Selective inhibitors for NSCCs have not been tested in PH, and given the widespread distribution of these channels; side-effects are likely to be significant. A similar problem arises with therapies aimed at targeting NHE1. Selective inhibitors of NHE1 have been developed, originally for treatment of myocardial infarction, but are associated with major side effects, including cerebrovascular events (414, 441). Given the wide distribution of TRPCs and NHE1 throughout the body, it is likely that inhibitors of these targets may only be useful in the treatment of PAH if cell-specific targeting could be achieved.

There was considerable excitement for the potential for administration of ROCK inhibitors, given that these agents reduced P_PA and PVR in PH models and PAH patients (187, 291). Unfortunately, substantial systemic hypotension accompanies intravenous delivery, limiting their use. Inhaled formulations of ROCK inhibitors can circumvent the systemic side effects (448), and were effective in acutely reducing PVR in PAH patients (186). Chronic ROCK inhibition reduced PH and remodeling in animal models (2, 160, 286) suggesting the possibility that in addition to acute vasodilator effects, prolonged use could slow and/or reverse remodeling in the patient population. Interestingly, sildenafil and statins can inhibit ROCK (216, 239), although whether the clinical dosing of these drugs inhibits ROCK activation in patients is unclear.

Studies targeting HIF activity using the pharmacologic inhibitors, digoxin or acriflavine, demonstrated reduced hypoxia-induced PH and vascular remodeling in rodents (6). Moreover, when treatment was initiated after PH was established, digoxin could slow the progression of PH (6), providing evidence supporting a therapeutic potential for drugs that block or reduce HIF in the treatment of PH. Finally, given the known role of BMPR2 dysfunction in many PAH cases, identifying means of increasing BMPR2 expression/activity in these patients remains an active area of investigation. A recent screen of clinical compounds identified low concentrations of FK506 as activators of BMPR2 signaling (591). FK506, a calcineurin inhibitor, displaces FK-binding protein-12 from type-I BMP receptors, enhancing BMP signaling. Currently in phase II trials, preliminary reports of compassionate use of FK506 in 3 advanced PAH patients revealed improvements in symptoms and clinical outcomes (590). While these initial reports are highly encouraging, whether this approach will be successful in a large population or in patients with loss-of-function BMPR2 mutations remains to be seen. Other studies in preclinical models targeting the BMPR signaling axis demonstrated that administration of low concentrations of BMP9 increased BMPR2 expression and selectively enhanced BMPR signaling in the endothelium (369). These early studies suggest that therapeutics aimed at modulating BMP signaling may be a promising strategy for treating PAH.

Pediatric pulmonary hypertension

As discussed in earlier sections, soon after birth, P_PA normally falls to levels comparable to adult values. Similar to adults, in children (term babies at sea level >3 months old), PH is diagnosed when the mean P_PA exceeds 25 mmHg in the presence of an equal distribution of blood flow to all segments of both lungs (292). While PH in children is relatively rare, persistent pulmonary hypertension of the newborn (PPHN) is more common, with an incidence of approximately 2 per 1000 live births (660). PPHN results from failure of the perinatal circulatory to transition after birth, and is characterized primarily by sustained elevation of PVR; resulting in decreased perfusion of the pulmonary circulation and continued right-to-left shunting of blood through the foramen ovale and ductus arteriosus. Common causes of PPHN include: (i) abnormal constriction of the pulmonary vasculature secondary to meconium aspiration, respiratory distress syndrome, and sepsis; (ii) structural abnormality of the vasculature due to failed remodeling after birth or congenital malformations; and (iii) bronchopulmonary dysplasia (501).

Hypoxia is a common consequence of PPHN. The majority of work addressing the effects of chronic hypoxia on the pulmonary circulation has been performed in adult animals, even though increased P_PA, remodeling and contraction also occur in the pediatric population. It is widely recognized that regulation of P_PA is influenced by postnatal age and it is unclear which mechanisms identified in the adult translate to the neonate. A recent review provided an in depth dissection of the role of hypoxia in PPHN, and the mediators and cellular mechanisms involved (203).

Summary

Regardless of inciting cause, PH is clearly a complex disorder in humans. Preclinical models have aided in understanding the remodeling that occurs, identifying cellular mechanisms involved, and investigating potential therapies. While progress has been made, much is yet to be learned regarding the mechanisms underlying the phenotypic changes in PASMCs and ECs. Unfortunately, none of the currently available treatments for PAH are curative and are often accompanied by negative side-effects or inconvenient routes of administration. Recent data from preclinical models has pointed to several promising potential therapies (224,508), including dichloroacetate, ROCK, NFAT, and HIF inhibitors. Clinical trials are needed to assess whether targeting any of these pathways will be beneficial in the patient population, and further investigation to identify additional pathways involved in the pathogenesis of PH will be required to develop new pharmacological agents aimed at targeting and reversing PH.

Bronchial Circulation

Although most agree that by the 1600s it was recognized that the lung had a second circulation, attributing the discovery of the bronchial circulation to a particular individual has been controversial, ranging from Galen to Domenici de Marchettis, to Leonardo da Vinci (99, 120, 424). In 1721, Frederich Ruysch was the first to definitively describe the bronchial circulation in detail (Fig. 9) and identify the bronchial artery originating from the aorta [reviewed in (424)]. Regardless of who initially described this circulation, it is now understood that the bronchial vasculature plays an important role in health and disease. The bronchial arteries supply oxygenated blood from the systemic circulation to the conducting airways, regional lymph nodes, nerves, walls of the pulmonary arteries and veins (vasa vasorum), and the visceral pleura, although additional functions of the bronchial circulation and its regulation in humans continue to be explored.

Figure 9.

Drawing by Fredrick Ruysch detailing the bronchial circulation. Reproduced, with permission, from Mitzner (424).

Development

The anatomy of the bronchial circulation varies substantially across species (172), with mice lacking a bronchial circulation altogether (423). Development of the human bronchial circulation occurs in two stages, with initial formation of arteries from the dorsal aorta in the fourth week of gestation. In the second trimester, these “primitive” arteries are subsequently replaced by bronchial arteries primarily arising from the superior thoracic aorta, (120, 353), although bronchial arteries either originating from, or communicating with, coronary arteries have been reported (322,374). Examination of human autopsy specimens revealed that in many, but not all, cases two bronchial arteries supply each lung (313, 353). Branching of the intrapulmonary bronchial arteries along the length of the bronchial tree down to the terminal bronchioles results in a vast network of capillaries, which form extensive anastomoses with other bronchial arteries as well as the pulmonary vasculature, creating a peribronchial plexus (120, 554). Arterioles arising from this plexus penetrate the muscular layer, forming a submucosal microvascular network (653). Venous drainage of the proximal conducting airway (first two divisions) follows the same pattern as the arterial supply, with blood being returned to the right heart via the azygos and hemiazygos veins.

“Bronchial” circulation may in fact be a bit of a misnomer, as in addition to providing systemic blood supply to the airways, these vessels also deliver systemic arterial blood to the walls of the pulmonary vessels, regional lymph nodes, nerves, and the visceral pleura (172, 405). Moreover, the distal bronchial microvasculature forms anastomoses with the pulmonary circulation. Due to these multiple connections with the pulmonary circulation, the deoxygenated bronchial venous blood drains via pulmonary veins to the left heart, producing an anatomic right to left shunt often referred to as the “bronchopulmonary anastomotic” flow or the “pulmonary collateral” circulation. While bronchopulmonary anastomoses can occur throughout the vasculature (101, 554, 656), it is now thought that drainage of bronchial venous blood is primarily a postpulmonary capillary event (482).

Determination of bronchial blood flow

The bronchial arteries are small in diameter (<1.5 mm at origin); consequently, the bronchial circulation is a high-resistance, low-capacitance system (95). The vast majority of the lung blood flow passes through the pulmonary circulation, with only a small percent carried by the bronchial circulation. Until recent years, measuring the exact blood flow carried by the bronchial circulation was a difficult task, given its low flow, small size vessels, and variable anatomic nature. Researchers relied on indicator-dilution techniques or radiolabeled microsphere tracers, both of which carried significant drawbacks (172), including the fact that these techniques measure total airway perfusion, not just that supplied by the bronchial arteries. With the use of invasive techniques, such as mounting a doppler ultrasound probe on the bronchial artery, approximations of bronchial artery blood flow were possible, with bronchial blood flow reported to be 13 to 50 mL/min in sheep (63, 100, 370). Despite the difficulties inherent in the microsphere technique, values obtained in piglets (42.1 ± 10.4 mL/min) were similar (550) to those measured using flow probes. At these levels of flow, under normal conditions bronchial flow accounts for approximately 1% to 5% of CO (370,393,550).

A number of factors can influence bronchial blood flow. Several studies have shown that flow in the bronchial arteries is increased during alveolar hypoxia (133, 673), consistent with the response observed in most systemic vascular beds. The mechanical effects of respiration also modulate flow in the bronchial circulation, with reductions in flow and increased draining via the pulmonary veins observed during lung inflation (28, 131). It is likely that these inspiration-induced alterations in bronchial flow are a consequence of changing lung volume and transpulmonary pressure compressing and/or stretching the systemic-to-pulmonary anastomotic bronchial blood vessels.

Various vasoactive factors have also been shown to alter the pattern of blood flow in the bronchial arteries. As might be expected from results in other systemic circulations, NO (100, 420) and PGI₂ (132) have each been shown to increase the bronchial blood flow. Intravenous administration of acetylcholine also vasodilates the bronchial circulation (543, 558), as does histamine (334, 358, 370). Interestingly, ET-1 also dilated bronchial arteries when administered intravenously (402, 607), perhaps due to activation of endothelial ET_B receptors.

Function of the bronchial circulation in the normal lung

A primary function of the bronchial circulation is the delivery of oxygen and nutrients to the airway wall. This becomes obvious when bronchial flow is disrupted, as can occur during lung transplant or when interventions are required for pulmonary atresia, ventricular septal defect, and major aortopulmonary collaterals. In such cases, necrosis of the airway mucosa was a frequent complication (451, 484, 555, 556). The bronchial circulation may also play an important role in acute airway responses and airway defense. Recruitment of inflammatory cells that participate in immune function to the airway mucosa also occurs via the bronchial circulation (354). Another key function of the bronchial circulation is airway clearance. Bronchial blood flow contributes to washout of bronchoconstrictors, proinflammatory mediators, and particulate matter from the airways (318, 654, 655). In instances where particulate matter causes pulmonary infarcts, the bronchial circulation also provides collateral blood flow to the lungs.

The bronchial circulation in disease

The bronchial circulation plays a substantial role in a variety of pathologic conditions, including hemoptysis, lung transplant and supplemental lung perfusion. The bronchial arteries possess a remarkable capacity for remodeling, exhibiting both enlargement and angiogenesis depending on the underlying cause. For example, infection is a common cause of bronchial arterial enlargement and increased blood flow, contributing to hemoptysis (468). Indeed, the bronchial circulation is the source of bronchial bleeding in up to 90% of patients (718). With recent development of sophisticated imaging techniques the anatomy of the bronchial arteries can be visualized (91) to accurately identify the enlargement of the bronchial arteries and the potential source of the bleed.

Remodeling of the bronchial circulation also occurs in conditions where pulmonary gas exchange is compromised, as in the case of pulmonary atresias or ventral septal defects (460). While the vast majority of gas exchange occurs in the pulmonary vasculature, the bronchial circulation can contribute to gas exchange through its anastomotic connections. In the absence of a normal pulmonary circulation, the bronchial arteries increase in number and caliber to provide the blood supply to the entire lungs (159, 333, 460, 489, 609, 649). Upon restoration of the pulmonary circulation, the bronchial blood flow returns to normal and the expanded vasculature partially regresses (159). While the exact cellular mechanisms responsible for this proliferation of vessels is still under investigation, it is clear that certain growth factors enhance the angiogenic capacity of the vessels, including VEGF (547) and stromal cell-derived factor-1 (301).

During transplantation, the pulmonary arterial, but not bronchial, circulation is surgically restored. Given that the bronchial circulation supplies approximately 50% of the airways wall perfusion (33) with the remainder coming from the pulmonary arteries, it is increasingly recognized that the bronchial circulation plays an important role in lung transplant and that loss of flow to the distal airways may result in further ischemia and hypoxic injury. In animal models, transplant with bronchial artery revascularization improved airway oxygenation compared to traditional transplant (311). It has been postulated that the loss of the bronchial circulation in lung transplant patients causes airway hypoxia, despite adequate pulmonary artery blood flow (140). In animal models, the bronchial artery circulation can be slowly restored following transplantation through the process of angiogenesis (488, 578); however, both imaging and autopsy specimens exhibit a lack of identifiable bronchial arteries and/or significant microvascular loss around airways within human transplanted lungs (140, 380). These results suggest that airway ischemia could contribute to airway fibrosis and poor outcomes with transplantation.

Conclusion

Over the past century, incredible progress has been made in understanding the structure and function of the lung circulations, spurred on by the development of advanced imaging techniques, the advent of the molecular biology revolution and availability of genetically altered animals. The adaptation of these techniques, and others, to the study of the pulmonary and bronchial vasculatures has provided a wealth of information, and while much has been learned, it is clear that there is still a great deal more to be understood. As investigators delve deeper into the cellular mechanisms that provide the basis of genetic and phenotypic changes in the lung circulations, and develop better models of pulmonary vascular disease, including lung injury and PH, undoubtedly new insights into both the physiology and factors contributing to pathology of the pulmonary circulation will be uncovered. Translation of these findings will potentially lead to better outcomes in disease states.