Targeting Energetic Metabolism

Abstract

This perspective highlights advances in the understanding of the role of cellular metabolism in the pathogenesis of pulmonary hypertension. Insights gained in the past 20 years have revealed several similarities between the cellular processes underlying the pulmonary vascular remodeling in pulmonary hypertension and those seen in cancer processes. In line with these insights, there is increasing recognition that abnormal cellular metabolism, notably of aerobic glycolysis (the “Warburg effect”), the potential involvement of hypoxia-inducible factor in this process, and alterations in mitochondrial function, are key elements in the pathogenesis of this disease. The glycolytic shift may underlie the resistance to apoptosis and increased vascular cell proliferation, which are hallmarks of pulmonary hypertension. These investigations have led to novel approaches in the diagnosis and therapy of pulmonary hypertension.

Understanding of the pathogenesis of pulmonary hypertension (PH) has undergone transformative advances since its recognition more than a century ago (1). Key in the pathophysiological characterization of the disease was the introduction of pulmonary artery catheterization more than 50 years, shaping the field of PH research into the twenty-first century (2). Elevated pulmonary artery pressures documented by pulmonary catheterization have remained the gold standard for diagnosis and have provided key prognostic indications of outcome (3). Further, it prompted studies aimed at elucidating the role of pulmonary artery vasoconstriction in the pathogenesis of PH, induced by alveolar hypoxia or vasospastic agents. These were followed by experimental modeling of pulmonary vascular remodeling, notably the molecular signaling underlying these processes. These approaches are used to the present day to dissect the pathogenesis of PH, including the most serious form of PH known as idiopathic pulmonary arterial hypertension (IPAH), which is representative of clinical entities listed under pulmonary arterial hypertension (PAH). These insights resulted in key therapeutic advances that have become the mainstay of treatment, including prostacyclin, endothelin receptor antagonists, and enhancers of nitric oxide signaling (4).

We may be on the verge of a second transformative conceptualization of pulmonary hypertension, perhaps of the same impact as that revealed by pulmonary hemodynamics. The past two decades have witnessed paradigm-shifting discoveries, the most groundbreaking involving the realization that most cases of the familial form of IPAH present with heterozygous loss-of-function mutations of bone morphogenetic protein receptor (BMPR)-2 (5, 6). Mutations in BMPR-2 are also found in a minority of cases of sporadic IPAH (7). This discovery, along with the concept advanced in the past two decades that PH pathology involves an imbalance of cell proliferation versus cell death, led to the hypothesis that the cellular and molecular features seen in PAH resemble hallmark characteristics described for cancer (8), including dysregulated angiogenesis (i.e., proliferation of abnormal capillaries and overexpression of vascular growth factors [9–11]), unchecked cellular growth (with expression of proliferation markers [12]), and resistance to apoptosis (12, 13), among others. These data were preceded by findings of dysregulated vascular cell growth within the pulmonary vasculature (10) and clonal expansion of endothelial cells in IPAH plexiform lesions (14), which altogether suggested the “quasi-neoplastic” nature of PAH (15) and indicated that the disease might involve loss of tumor suppression genes. These include members of the transforming growth factor (TGF)-β family (16), leading to enhanced TGF-β signaling in vascular cells (17, 18). An important supportive insight of the “neoplastic-like” hypothesis was subsequently provided by the identification of increased survivin expression in PAH smooth muscle cells (SMCs) (19), a key antiapoptotic mediator involved in the pathogenesis of apoptosis resistance seen in cancer cells (20).

In the update of their landmark review on “hallmarks of cancer,” Hanahan and Weinberg have included the properties of deregulated cellular energetics, genomic instability, tumor-promoting inflammation, and avoidance of immune destruction as emerging cancer features (21). Of interest, data have documented genetic instability in IPAH (22), in line with our prior observation of microsatellite instability in IPAH plexiform lesions (16), and the potential role of inflammation in the pathogenesis of PAH (23, 24). Furthermore, altered cell metabolism is essential in allowing cancer cells to proliferate (21, 24). This growth advantage relies on energy metabolic adaptations, which are controlled differently than in resting cells. Furthermore, several growth factors converge in affecting how cells use key carbon sources in energy generation and dispose of them as macromolecules. The mechanisms underlying the metabolism in proliferative processes, such as cancer, have become emerging targets for novel diagnostic and therapeutic approaches that may have relevance to pulmonary hypertension (25). It is the goal of this perspective to highlight the present experimental evidence regarding the role of cellular metabolic alterations in the pathogenesis of pulmonary hypertension by interfacing the investigations of cancer cell metabolism and cellular metabolic alterations in PH.

Glycolytic shift in PAH: the role of hypoxia-inducible factor-1α

Oxygen levels are a central metabolic regulator in the balance between cellular glycolysis versus oxidative phosphorylation. Normal cells shift their ATP generation preferentially toward glycolysis when exposed to low oxygen levels and reverse this shift toward oxidative phosphorylation in the presence of sufficient oxygen. Cancer cells, however, consume 10 times as much glucose as normal tissues, even with adequate oxygen, and exhibit decreased oxidation of pyruvate into acetyl-CoA for the citric acid (Krebs) cycle, leading to excessive intracellular and extracellular lactate levels. This is the so-called Warburg effect (24, 26), based on the observation by Otto Warburg that cancer cells use predominantly glycolysis under normal oxygen supply rather than mitochondrial oxidation for their ATP generation (see the online supplement). This metabolic reprogramming in cancer cells arises from their need to use varied metabolic pathways that support uncontrolled neoplastic growth. This adaption occurs despite the significantly lower yield of ATP provided by glycolysis (only 2 ATP molecules) when compared with the 15- to 18-fold higher yield of ATP generated when a molecule of glucose is fully oxidized via the citric acid cycle and oxidative phosphorylation. This adaption also allows for the maintenance of increased ADP/ATP ratios (essential for the continuous generation of ATP), NADPH synthesis, and the generation of building blocks for macromolecular synthesis (27). Both acetyl-CoA and pentose monophosphate shunt intermediates are used to generate macromolecules, such as carbohydrates, nucleotides, and fatty acids, which are necessary to sustain rapidly dividing cells. As becomes apparent further on in this article, pulmonary vascular cells share a similar metabolic shift and altered molecular controls with those exhibited by cancer cells and proliferating cells during embryonic development.

Both the physiological adaptation of normal cells to marked hypoxia and the metabolic dysregulation in cancer cells are largely credited to up-regulation of hypoxia-inducible factor (HIF)-1α. HIF-1α, when combined with HIF-1β (or aryl-receptor nuclear translocator), activates more than 100 genes involved in energy metabolism, extracellular matrix remodeling, apoptosis, cell migration, cell cycle control, pH regulation, and angiogenesis (28, 29) (Figure 1). HIF-2α, with more cell-restricted expression than HIF-1α, acts in these cells to regulate similar pathways (collectively, we refer to HIF-1α, -2α, and -1β as hypoxia-inducible factors [HIFs]).

Figure 1.

Hypoxia-inducible factor (HIF) activation in pulmonary hypertension (PH). In certain conditions, such as hypoxia, reduced nitric oxide (NO) and manganese superoxide dismutase (MnSOD2), or inhibited prolyl hydroxylases, HIF-1α becomes stabilized and forms a heterodimer with HIF-1β to activate the transcription of over 100 genes, some of which are shown above. Up-regulation of glycolytic enzymes such as glucose transporters (GLUT), hexokinases (HK), phosphofructokinase (PFK), pyruvate kinases (PK), and pyruvate dehydrogenase kinase (PDK) results in a metabolic shift toward glycolysis and away from fatty acid β-oxidation. The up-regulation of certain growth factors including platelet-derived growth factor (PDGF), fibroblast growth factor (FGF), and transforming growth factor (TGF)-β leads to hyperplastic and hypertrophic effects in the pulmonary vasculature. Increased expression of vascular endothelial growth factor (VEGF) contributes to misguided angiogenesis. The up-regulation of stromal-derived growth factor (SDF)-1 with its receptors (CXCR4 and CXCR7) aids in the recruitment of bone marrow precursors which likely contribute to the signaling driving vascular remodeling.

HIF-1α transcriptionally up-regulates glucose uptake via increased expression of glucose transporters GLUT1 and GLUT3, and glycolytic flux via activation of the hexokinase (HK)-1 and -2 isoforms, phosphofructokinase (PFK)-2, glyceraldehyde dehydrogenase, enolase-1, phosphoglycerate kinase isoform 1, and pyruvate kinase isoform M2 (PK M2). In malignancy, activated oncogenes can also contribute to the glycolytic shift: Ras increases glucose uptake, Src phosphorylates several glycolytic enzymes, Myc can activate several glycolytic enzymes including PK M2, and Akt activates HK-2 and PFK-1 and -2 (30). Moreover, pyruvate dehydrogenase kinase (PDK)-1 is regulated by HIF-1α, and the subsequent inhibition of the pyruvate dehydrogenase complex (PDC) during hypoxia facilitates increased ATP generation by anaerobic glycolysis while inhibiting mitochondrial respiration and reactive oxygen generation (31, 32) (see the online supplement). Last, HIF-1α increases the expression of cytochrome oxidase (complex 3) isoform 4.2 with concomitant down-regulation of expression of the 4.1 isoform, which has relatively higher oxidative properties under hypoxia; moreover, cytochrome oxidase 4.2 enhances ATP generation under hypoxia conditions (33). Although the aggregate properties of HIF-1α support its role as a key transcriptional regulator of glycolysis in the setting of the Warburg effect, there is evidence linking E2F transcription factor to metabolic adaptation to cell growth (34).

“Hypoxia signaling” and the activation of HIF in PH

Alterations of cellular metabolism driven largely by HIF are paradigmatic of cellular responses to hypoxia. Chronic hypoxia (i.e., less than 50% of the normal partial pressure of oxygen) leads to mild/moderate PH, largely characterized by an increase in pulmonary artery SMC numbers (hyperplasia) and cell mass (hypertrophy) (35, 36). However, the mechanisms underlying these changes remain unclear. Acute intraalveolar hypoxia triggers pulmonary artery vasoconstriction (HPV) due to the sensitization of SMCs to calcium influx. The mediators of acute hypoxic vasoconstriction include Rho kinase, endothelin, altered redox state, and transient receptor potential channel activation, among others (37). Many PH models exhibit both acute hypoxic vasoconstriction and chronic PH, which is suggestive of a causal link between the acute and chronic responses of pulmonary arteries. However, there are examples in which these processes are uncoupled. Knockout mice lacking the serotonin receptor-2b have preserved HPV but are protected from chronic PH (38), whereas the converse occurs with mice lacking endothelial nitric oxide synthase (39). There are compelling data indicating that hypoxic PH is instead caused by increased expression of growth factors, including platelet-derived growth factor (PDGF) (40), fibroblast growth factor (11), TGF-β (41), and additional hyperplastic/hypertrophic effects of HIF-1α on SMCs (42). It remains unclear whether enhanced HPV is required for the activation of these pathogenetic processes of remodeling that underlie the extreme pulmonary vascular lesions in severe forms of PAH (43, 44). A fraction of patients with severe PAH will exhibit excessive vasoconstrictive responses, possibly related to underlying pulmonary vascular remodeling.

It is conceivable HIF activation triggers cellular events culminating in severe PAH; HIF activation might occur via direct hypoxia-like stabilization of HIF (such as caused by reactive oxygen species or a decrease in nitric oxide) or cell signaling converging into enhanced HIF transcription under normoxia. Although the inciting events leading to severe human PAH are unknown, severe PH with endothelial cell proliferation is seen experimentally when rats are exposed to chronic hypoxia combined with vascular endothelial growth factor (VEGF) receptor blockade (12). This model relies on early endothelial cell apoptosis, followed by focal proliferation of apoptosis-resistant endothelial cells. The proliferative lesions have evidence of local hypoxia (with a local O₂ concentration equal to or less than 2%) and an HIF-1α inhibitor protects against the development of severe PH and pulmonary vascular lesions in the model (R. M. Tuder and colleagues, unpublished observations). The finding of hypoxia in these vascular lesions is somewhat surprising as the effective diffusion path of oxygen is approximately 145 μm (45), well within the reach of remodeled pulmonary arteries abutting oxygenated alveolar spaces.

Expression of HIF-1α and HIF-1β was documented in plexiform lesions of IPAH lungs in combination with increased expression of the HIF target VEGF (10). These descriptive findings supported the concept that plexiform lesions may represent a form of misguided angiogenesis, similar to that found in hyperplastic blood vessels supplying glioblastoma multiforme (46), which are histologically similar to IPAH plexiform lesions. The potential pathogenetic role of HIF-1α and HIF-2α in PH was highlighted by the protection against experimental hypoxic PH in mice heterozygous for either gene (47, 48). A similar finding was reported in mice carrying the mutation of Chuvash polycythemia in the von Hippel-Lindau tumor suppressor protein at codon 200 (R200W), which is associated with a disease known for the stabilization of HIF-1α and HIF-2α (49). Further evidence of multifaceted pathogenic mechanisms of HIF-1α in PH include smooth muscle cell alkalinization via up-regulation of Na⁺/H⁺ exchanger isoform 1 (50), mitochondrial dysfunction, Kv1.5 channel down-regulation, and enhanced intracellular Ca²⁺ influx (51) (Figure 1).

In cultured IPAH endothelial cells, HIF-1α is possibly stabilized via decreased NO levels due to decreased expression of manganese superoxide dismutase-2 (MnSOD2) (52), which was in line with prior observations in IPAH lungs (53). Of note, a similar link between decreased MnSOD2 expression and up-regulation of HIF-1α occurs in fawn-hooded rats with spontaneous PH, possibly linked to CpG methylation and silencing of the MnSOD promoter (51).

These observations are in line with the documentation that IPAH cells are more glycolytic, produce less ATP, and have fewer mitochondria than normal pulmonary artery endothelial cells (54). In fact, the decrease in mitochondria numbers in IPAH cells is due to abnormally activated HIF-1α, as its knockdown with small interfering RNA results in increased numbers of mitochondria in IPAH endothelial cells (52). The aggregate of these results provides compelling evidence of the glycolytic shift in PH pathogenesis, similar to observations made in cancer biology. The aggregate of these cell-based data is supported by the observation of increased uptake of ¹⁸F-labeled deoxyglucose (FDG) in the lungs of patients with IPAH when compared with control subjects (Figure 2) (54).

Figure 2.

Increased ¹⁸F-labeled deoxyglucose (FDG) uptake in pulmonary hypertension (PH). (A) Representative positron emission tomography (PET) and computed tomography (CT) images from a patient with idiopathic pulmonary arterial hypertension (IPAH) (bottom) and a healthy control subject (top). PET images are shown on the right, and CT images are shown on the left. (B) Standardized uptake values (SUV) of ¹⁸F-labeled deoxyglucose (FDG)-PET scans in the lungs of patients with IPAH (n = 4) and healthy control subjects (n = 3) at 1.5 and 3 hours after injection (^*,†P = 0.01). Reprinted by permission from Reference 52.

Increased HIF expression and preferential use of glycolysis are also relevant to more efficient mobilization and higher proangiogenic effects of bone marrow precursor cells (55). The potential contribution of circulating cells to the pathogenesis of PH has been supported by several human and experimental in vivo studies. There is evidence of increased numbers of circulating progenitor cells in patients with IPAH (56, 57) and accumulation of CD131-positive precursor cells in experimental and IPAH remodeled pulmonary arteries (58). The potential relevance of bone marrow–derived cells in PH was underscored by the improved phenotype in rats depleted of monocytic precursors and monocytes (55, 59). The local expression of HIF-1α in IPAH vessels (10, 51) may aid in the recruitment of bone marrow precursors via enhanced production of stromal-derived factor-1 and increased expression of its receptors CXCR4 and CXCR7 (60) (Figure 1). The local niche with enhanced “hypoxia-like” signaling may be favorable to proliferation of vascular progenitor cells, which have been postulated to give rise to the proliferative endothelial cells seen in plexiform lesions (15). This favorable microenvironment for endothelial cell proliferation has been postulated in nonpulmonary settings, as demonstrated by the beneficial effects of angiogenic bone marrow precursors in peripheral revascularization models (55).

Findings indicate that the glycolytic shift in PH may be alternatively due to alterations of the mitochondria and/or redox state. Localized hypoxia leads to increased oxidative stress via generation of superoxide radical in mitochondria complex 3, resulting in stabilization of HIF-1α by inhibition of prolyl hydroxylases (61). In adequate oxygen tension, prolyl hydroxylase uses α-ketoglutarate as a substrate to hydroxylate HIF-1α. In cultured pulmonary artery smooth muscle cells, HIF-1α activation from hypoxia is also dependent on Nogo, a key regulator of endoplasmic reticulum (ER)–mitochondrial function as part of the cell response to ER stress, suggesting that prolyl hydroxylase inhibition alone may not be adequate (62). In fact, mutations of succinate dehydrogenase (complex 2 of the electron transfer chain) and isocitrate dehydrogenase have been linked to stabilization of HIF-1α in pheochromocytomas (63) and malignant astrocytomas (64), respectively.

Metabolic transformation of the right ventricle

In parallel with the altered cellular metabolism occurring in the PH pulmonary vasculature, right ventricular myocytes also develop an altered metabolic phenotype. Under normal physiological conditions cardiomyocytes sustain a continuous workload, relying primarily on aerobic respiration for their energy supply while using a variety of substrates including glucose, fatty acids, and ketone bodies. Normally, 95% of cardiac ATP production comes from oxidative phosphorylation (65), with 50–70% derived from fatty acid oxidation alone (66). Substrate metabolism in the heart is a dynamic interaction between several aerobic and anaerobic metabolic pathways, as cardiomyocytes are able to maintain an internal flux between various metabolic substrates according to the energy demand, workload, oxidative conditions, and substrate availability (65). This flux is controlled by transcriptional regulation of protein expression, allosteric enzyme regulation, and levels of specific substrate–product pairs.

Adaptation of the diseased (left) heart involves several features that indicated a potential reversion to a fetal heart phenotype (67, 68). This phenotype is characterized by increased glycolysis, in conjunction with evidence of decreased fatty acid oxidation (see the online supplement). This metabolism is driven by increased levels of acetyl-CoA carboxylase and its product malonyl-CoA (69, 70) as well as decreased adenosine monophosphate kinase (71) and malonyl-CoA decarboxylase, which altogether favors fatty acid synthesis (see the online supplement) (69, 70).

Right ventricular (RV) dysfunction in PH arises as the increasing pulmonary arterial resistance imposes a significant strain on the right ventricle to maintain adequate cardiac output. The adaptation of the heart to PAH involves cellular and molecular events that ultimately affect glucose and fatty acid substrate metabolism. Although the mechanisms underlying right ventricular failure in PH remain poorly understood (72), there is evidence of a glycolytic shift in the right ventricle in human (73) and experimental models of PH (74).The shift in the monocrotaline (MCT)-induced PH in the rat is evidenced by increased glycolysis with up-regulation of GLUT1, and (inhibitory) phosphorylation of pyruvate dehydrogenase, associated with increased expression of PDK-4 (31). Hypoxia also produces several of these adaptive changes, including increased glucose uptake (75) with increases in the glucose transporters GLUT-1 and GLUT-4 (73, 76), an increase in rate-limiting glycolytic enzymes such as HK-2 (77), increased PDK activity (73, 74, 78), and activation of transcriptional regulators of glycolysis including HIF (31, 79, 80). However, it remains unclear whether the shifted energy generation from fatty acid oxidation to glycolysis is a homoeostatic adaptation or pathological; it is conceivable that it may initially be beneficial and later detrimental as the disease progresses (65).

Targeting the glycolytic shift in pulmonary hypertension

The aggregate of data implicating a role for a glycolytic shift in the pathogenesis of human and experimental PH (10, 51, 52) provides a compelling rationale for novel therapeutic and diagnostic modalities, including positron emission tomography scanning of the hypertensive pulmonary vasculature for increased metabolite uptake (Figure 2). Furthermore, the right ventricular glycolytic shift can be concurrently targeted. HIF-1α and -2α could be logically targeted in PAH, given their centrality in regulating glycolysis and the expression of several key genes involved in vascular cell remodeling (47, 48). 2-Metoxyestradiol, a relatively nonspecific inhibitor of HIF-1α, has shown beneficial effects in the MCT–PH model (81).

Most experimental evidence regarding the impact of blocking the glycolytic shift has focused on the inhibition of PDK with diisopropylammonium dichloroacetate (dichloroacetate, or DCA), an analog of acetic acid. DCA was initially described as an effective therapy for mitochondrial genetic mutations (82). Its use was then extended to probing the impact of enhanced ATP generation by glucose oxidation (82). The similarities between the vascular proliferative processes in PAH and those in neoplastic cells (83), along with the evidence of mitochondrial hyperpolarization in models of PH, led to the use of DCA in experimental PH caused by MCT and hypoxia (84) and, more recently, in mice overexpressing the serotonin transporter in pulmonary vascular SMCs (85). As with PH, cancers treated with DCA undergo increased cell death, potentially triggered by an increase in mitochondria-generated oxidants, particularly from complex 1 (85).

The mutually inhibitory balance between fatty acid oxidation and glucose oxidation, also known as the Randle cycle (see the online supplement), offers an alternative means to stimulate oxidative metabolism. This approach was introduced in the 1990s to prevent the detrimental effects of enhanced myocardial fatty acid oxidation during oxygen deprivation due to ischemia–reperfusion, based on the hypothesis that the preferential use of glucose oxidation as a source of energy (versus fatty acid oxidation) may limit cellular damage (86). Enhancement of glucose oxidation via inhibition of fatty acid oxidation has been similarly postulated to block pulmonary vascular remodeling and PH. Mice lacking malonyl-CoA decarboxylase have increased malonyl-CoA, which inhibits carnitine acyltransferase-1 activity and fatty acid oxidation (87) (see the online supplement). This inhibition enhances oxidative glycolysis, similar to the effects accomplished with DCA. These mice do not manifest hypoxic vasoconstriction and develop less hypoxic PH than wild-type mice. A complementary approach achieved by the pharmacological inhibition of the last enzymatic step (catalyzed by 3-ketoacyl-CoA thiolase [3-KAT]) of β-oxidation with trimetazidine (TMZ) also reduces fatty acid oxidation in MCT-treated rats (87). This effect is associated with an increase in Kv1.5 expression, depolarization of mitochondria, increased vascular cell death, and improved vascular remodeling and pulmonary artery pressures. The same study showed that both DCA and TMZ also decreased right ventricular hypertrophy, although this may be due to metabolic effects on the pulmonary vasculature rather than the RV myocytes. Last, interventions to activate peroxisome proliferated-activated receptor-γ, a key regulator of fatty acid metabolism, with thiazolidinediones may overcome growth-promoting signaling in pulmonary vascular smooth muscle cells in PAH (88).

In related studies focused on the RV, DCA used as treatment for MCT-induced PH or as prevention in pulmonary-artery banded (PAB) rats partially reversed right ventricular hypertrophy and increased cardiac output (74). DCA increased the rate of glucose oxidation and oxygen consumption in the MCT model, and partially prevented the increase in GLUT and the (inhibitory) phosphorylation of PDH. However, despite increased cardiac output, the right ventricular systolic pressure concurrently decreased in both models, suggesting that pulmonary vascular remodeling (which can occur even in PAB) might have also improved (74). Similarly, TMZ as a treatment for experimental MCT-induced PH increases mitochondrial oxidative phosphorylation back to baseline levels, although it may not prevent right ventricular hypertrophy (89). These data underscore that the systemic effects of metabolic modulators can make it difficult to determine the relative contributions to the pulmonary vascular or RV components in improving outcome in PAH.

Conclusions

Similarities between neoplasia and PH have moved investigators in the field to look into novel cellular mechanisms underlying the pathogenesis and the resulting clinical characteristics of PH. Awareness of the metabolic shift to aerobic glycolysis, in association with stabilization of HIF, in both the pulmonary vasculature and the right ventricle, has revealed new avenues for diagnostic and therapeutic approaches. Beyond targeting PDK and 3-KAT, many of the glycolytic enzymes represent additional targets for pharmacological intervention in PH, as has been postulated for cancer therapy. For example, imatinib, a PDGF, HK, and c-Kit inhibitor, has been shown to have beneficial effects in severe PH (25, 90). Further research elucidating the details of how these mechanisms contribute to the overall clinical picture of PH will bring us closer to understanding the initial inciting events in this disease, along with the means for effective clinical treatment. Given the fundamental nature of cellular metabolism in cell fates, these insights may have relevance for other pulmonary diseases also characterized by abnormal proliferation and dysregulated cell death.