Responding to Hypoxia: Lessons From a Model Cell Line

See allHide authors and affiliations

Science's STKE  20 Aug 2002:
Vol. 2002, Issue 146, pp. re11
DOI: 10.1126/stke.2002.146.re11


Mammalian cells require a constant supply of oxygen to maintain adequate energy production, which is essential for maintaining normal function and for ensuring cell survival. Sustained hypoxia can result in cell death. It is, therefore, not surprising that sophisticated mechanisms have evolved that allow cells to adapt to hypoxia. "Oxygen-sensing" is a special phenotype that functions to detect changes in oxygen tension and to transduce this signal into organ system functions that enhance the delivery of oxygen to tissue in various organisms. Oxygen-sensing cells can be segregated into two distinct cell types: those that functionally depolarize (excitable) and those that do not functionally depolarize (nonexcitable) in response to reduced oxygen. Theoretically, excitable cells have all the same signaling capabilities as the nonexcitable cells, but the nonexcitable cells cannot have all the signaling capabilities as excitable cells. A number of signaling pathways have been identified that regulate gene expression during hypoxia. These include the Ca2+-calmodulin pathway, the 3′-5′ adenosine monophosphate (cAMP)-protein kinase A (PKA) pathway, the p42 and p44 mitogen-activated protein kinase [(MAPK); also known as the extracellular signal-related kinase (ERK) for ERK1 and ERK2] pathway, the stress-activated protein kinase (SAPK; also known as p38 kinase) pathway, and the phosphatidylinositol 3-kinase (PI3K)-Akt pathway. In this review, we describe hypoxia-induced signaling in the model O2-sensing rat pheochromocytoma (PC12) cell line, the current level of understanding of the major signaling events that are activated by reduced O2, and how these signaling events lead to altered gene expression in both excitable and nonexcitable oxygen-sensing cells.


Mammalian cells require a constant supply of oxygen to maintain adequate energy production, which is essential for maintaining normal function and for ensuring cell survival. Even brief episodes of hypoxia can result in cessation of oxidative phosphorylation and depletion of cellular adenosine triphosphate (ATP), which results in profound deficiencies in cellular function. Sustained hypoxia can result in cell death. It is, therefore, not surprising that sophisticated mechanisms have evolved that allow cells to adapt to hypoxia before ATP depletion occurs. During the last ten or so years, there has been a growing number of reports on hypoxia-induced transcription of specific genes that mediate such cellular functions as erythropoiesis, pulmonary ventilation and blood flow, angiogenesis, and energy metabolism. Progress has also been made in understanding how cells detect changes in oxygen tension and how specific signal transduction pathways and transcription factors are activated by hypoxia. Here, we present a view of the current level of understanding of the major signaling events that are activated by reduced O2, and how these signaling events lead to altered gene expression. We have focused our discussion specifically on hypoxia, rather than on the more complex phenomenon of ischemia, which involves deprivation of nutrients in addition to reduced oxygen levels, although undoubtedly the two will share some common signaling mechanisms. In addition, we have included, wherever possible, discussion on what is known about the involvement of the various signaling pathways in cell death, survival, and hypoxic and ischemic tolerance.

Oxygen-Sensing Cells

All cells are sensitive to alterations in oxygen level, but not all cells are oxygen-sensing. "Oxygen-sensing" is a special phenotype that functions to detect changes in oxygen tension and transduce this signal into changes in organ system functions that enhance the delivery of oxygen to tissues. For example, oxygen-sensing cells in the kidney release erythropoietin in response to hypoxia. This stimulates the production of red blood cells and, therefore, the oxygen-carrying capacity of the blood. Another example is the vascular smooth muscle cells in the lung, which constrict in response to reduced oxygen tension. This diverts blood from poorly oxygenated regions of the lung to better-oxygenated regions, thus ensuring optimal loading of oxygen into red blood cells and subsequent delivery of oxygen to tissues. A third example is the type I cells in the carotid body, which signal the central nervous system to stimulate ventilation of the lung. This hyperventilation response serves to enhance the oxygen levels in the lung. The sum total of these functions is to enhance the delivery of oxygen to tissues. In addition to the ability to detect changes in oxygen levels, oxygen-sensing cells in these different tissues possess a remarkable tolerance to hypoxia. This is an important property, because it allows these unique cells to continue to perform the important function of oxygen sensing even during hypoxia.

Oxygen-sensing cells can be segregated into two distinct cell types: excitable and nonexcitable. The oxygen-sensing cells in the pulmonary circulation and carotid body are examples of excitable oxygen-sensing cells, whereas the oxygen-sensing cells in the kidney are nonexcitable. Excitable oxygen-sensing cells are defined as such because they undergo depolarization due to changes in ionic conductances through voltage-sensitive channels in response to decreased oxygen tension. This depolarization is coupled to distinct signaling events, such as changes in calcium flux across the plasma membrane that in turn can lead to changes in the intracellular concentration of calcium ([Ca2+]i) and activation of Ca2+-dependent signaling pathways. The distinction between excitable and nonexcitable cells is important because, theoretically, excitable cells can possess all the signaling pathways that nonexcitable cells have, but some of the calcium-mediated pathways will most likely be absent in nonexcitable cells. This functional depolarization is in contrast to the pathological depolarization that all cells will eventually undergo in response to sustained oxygen deprivation as ATP levels are depleted and the ion pumps that maintain membrane potential cease to function.

The key event in most signaling pathways is the interaction between a chemical factor and a receptor associated with the plasma membrane or located in the cytoplasm or nucleus. Regardless of receptor location, its activation by the chemical factor sets in motion a signaling cascade that ultimately regulates cell functions, including gene expression. The idea that O2 can act as a chemical factor to specifically regulate signal transduction events is somewhat controversial. Nevertheless, there is growing evidence that this is indeed the case.

A major obstacle in gaining insights into how hypoxia activates signal transduction and gene expression mechanisms is the scarcity of oxygen-sensing cells in intact tissues. A popular approach for overcoming this problem is to use clonal cell lines whose response to reduced O2 closely resembles the in vivo function of O2-sensing cells. These model systems provide valuable insights into the basic mechanisms of O2-sensing, including signal transduction and gene regulatory processes. We established the dopaminergic pheochromocytoma (PC12) cell line as a reliable and valuable model for this purpose (1-3).

PC12 cells serve as a model for carotid body type 1 cells [reviewed in (4)]. Carotid body type 1 cells sense O2 levels in the bloodstream and elicit a complex set of events that help to maintain O2 homeostasis (5, 6). These cells have a hypoxia-tolerant phenotype that allows them to survive and function in a low-oxygen environment. PC12 cells closely resemble type I cells morphologically and phenotypically (7). Both cell types depolarize rapidly (within seconds) during hypoxia through inhibition of an O2-sensitive outward K+ current (3, 8-10); this depolarization is followed by an increase in [Ca2+]i (6, 11, 12). Longer-term responses (minutes to hours) include stimulation of tyrosine hydroxylase, c-fos, and junB gene expression (1, 13). Both also synthesize and release dopamine in response to acute (1 to 3 hours) hypoxia (14-17). In many of the studies based on PC12 cell responses to hypoxia, the cells remain viable and continue to proliferate during and following periods of prolonged hypoxia (1% O2 for 24 to 48 hours). Thus, the PC12 cell line is a useful system in which to study the molecular and cellular basis of O2 chemosensitivity and the mechanisms by which O2-responsive genes are regulated by hypoxia.

We have used the excitable PC12 cells to identify a number of signaling pathways that regulate gene expression during hypoxia. These include the 3′-5′ adenosine monophosphate (cAMP)-protein kinase A pathway (18), the Ca2+-calmodulin pathway (18), the ERK1 and ERK2 mitogen-activated protein kinase (MAPK) pathway (19), the stress-activated protein kinase (SAPK, also known as p38 kinase) pathway (20), and the phosphatidylinositol 3-kinase (PI3K)-Akt (21) pathway as regulators of gene expression. Although the responses of different cell types will probably vary (especially between excitable and nonexcitable cells), it is likely that there will be a core set of responses to hypoxia that is common to all cells.

Hypoxia-Responsive Signal Transduction Pathways


The immediate response of PC12 cells exposed to hypoxia is membrane depolarization and increased [Ca2+]i (2), which is similar to the response measured in carotid body type I cells (10, 22). Whole-cell voltage-clamp experiments revealed that this membrane depolarization in PC12 cells is mediated by an O2-sensitive potassium channel (23). The outward K+ current through this channel becomes progressively inhibited when PC12 cells are exposed to a graded reduction in O2 tension (2), which leads to a stepwise depolarization and an increase in [Ca2+]i that is proportional to the magnitude of depolarization. Electrophysiological and pharmacological studies revealed that the hypoxia-sensitive K+ current in PC12 cells is most likely mediated by the Kv1.2 channel (3, 23). Kv1.2 antibodies dialyzed through the patch-recording pipette completely blocked the O2-sensitive K+ current, whereas antibodies against another K+ channel (Kv2.1) had no effect. Introduction of the Kv1.2 channel gene into Xenopus oocytes leads to expression of an O2-sensitive K+ current (23). Thus, these oocytes can be transformed to possess an O2-sensing phenotype simply by introducing the Kv1.2 channel. It remains unclear how reduced O2 tension regulates the activity of this hypoxia-sensitive channel. This regulation could occur through a direct effect of hypoxia on the hypoxia-sensitive channels or by indirect effects through other O2-sensitive signaling molecules. In other types of excitable O2-sensing cells, such as pulmonary smooth muscle cells, other K+ channels (for example, Kv1.5) may mediate membrane depolarization.

The increase in [Ca2+]i that occurs in response to depolarization suggests that various Ca2+-dependent protein kinases and phosphatases may be regulated by reduced O2 (24). Activation of these Ca2+-signaling enzymes can have profound affects on cell function, including altered gene expression, and probably plays a major role in the physiological adaptation (tolerance) to hypoxia in excitable O2-sensing cells. Depolarization of PC12 cells by hypoxia activates Ca2+-calmodulin-dependent protein kinases (CaMKs), which in turn can stimulate transcription of certain genes (24, 25). For example, the gene that encodes tyrosine hydroxylase (TH), the rate-limiting enzyme in the synthesis of dopamine and other catecholamines, is stimulated by hypoxia by a mechanism that requires increased [Ca2+]i (1, 26, 27). Removal of extracellular Ca2+, or chelation of intracellular Ca2+ with BAPTA-AM, inhibits the increase in cytosolic free [Ca2+] and blocks the hypoxia-induced increased expression of tyrosine hydroxylase (TH) (28). Blockade of the L-type Ca2+ channel with nifedipine partially inhibits the increased expression of TH in PC12 cells during hypoxia. In addition, pharmacological blockade of calmodulin inhibited the hypoxia-induced stimulation of a hypoxia-response element (HRE)-luciferase reporter gene (19). Thus, in excitable O2-sensing cells depolarization and an increase in [Ca2+]i are a primary mechanism for regulating gene expression and, therefore, cell function.

Depolarization and increased [Ca2+]i are not unique to O2-sensing cells. There is a long history of evidence that depolarization can mediate signaling events that regulate gene expression in response to different stimuli in different cell types. For example, membrane depolarization and Ca2+ entry into neurons lead to transcription of genes involved in plasticity and long-term potentiation of synaptic transmission (28). This Ca2+-regulated transcription involved the phosphorylation of cAMP-response element binding protein (CREB) and binding of other protein factors that were recruited to the promoter region of regulated genes. Elevated [Ca2+]i is also involved in the regulation of genes involved in phenotype changes in arterial smooth muscle in hypertensive rats (29). This response was associated with increased [Ca2+]i and an increased activation of CREB and the immediate early gene c-fos. The importance of increased [Ca2+]i for activating cascades of enzymatic reactions and protein-protein interactions in the cytoplasm and nucleus that lead to transactivation of genes has been extensively studied (30). It is clear that Ca2+ entry into the cell can differentially affect cell functions involved in cell survival, synaptic strength, and cell death (31). These cellular functions are mediated by Ca2+ binding to calmodulin and the subsequent activation of various enzymes, including the calmodulin kinases and Ca2+-regulated adenylate cyclases, which transduce the Ca2+ signal into nuclear events associated with gene expression.

The Stress- and Mitogen-Activated Pathways

The SAPK and MAPK pathways play critical roles in responding to cellular stress and promoting cell growth and survival (32, 33). SAPKs and MAPKs are the downstream components of three-member protein kinase modules (34). Five homologous subfamilies of these kinases have been identified, and the three major families include the p38s (SAPK2), c-Jun NH2-terminal kinases (JNK), and the ERK1 and ERK2 MAPKs. In general, the SAPKs (p38 and JNK) are activated primarily by noxious environmental stimuli, such as ultraviolet light (UV), osmotic stress, inflammatory cytokines, and inhibition of protein synthesis (35-39). Increasing evidence suggests that under certain conditions the p38 and JNK pathways can also be activated by mitogenic and neurotrophic factors (40, 41). In contrast, ERK1 and ERK2 are stimulated primarily by mitogenic and differentiative factors in a Ras-dependent manner (42, 43). There is also evidence that the ERK pathway can be activated by other environmental stimuli (such as heat shock or changes in osmolarity) (32-34). Hypoxia is a prevalent physiological stressor in many disease states and stimulates the activity of the SAPK and MAPK signaling pathways.

The effects of hypoxia on the SAPK and MAPK signaling pathways were studied in PC12 cells (20). Western blot analysis revealed a time-dependent phosphorylation of p38α and p38γ by hypoxia. The level of total p38 (phosphorylated and nonphosphorylated forms) remained unchanged. Cyclin D1, a downstream target of p38 (44), has been implicated in regulating progression through the G1 phase of the cell cycle (45, 46). In PC12 cells, hypoxia causes a decrease in cyclin D1 protein levels and also causes cells to accumulate at the boundary between the G0 and G1 phases of the cell cycle (20). Both of these effects are partially reversed by pharmacological blockade of p38α with SB203580 or transfection with a kinase-inactive form of p38γ. In addition, the hypoxia-induced activity of transfected p38γ is attenuated, but not abolished, in the absence of extracellular calcium (47). These results indicate that p38α and p38γ are activated by hypoxia. Inhibition of progression through the cell cycle by activation of the p38 pathway may be an important adaptive mechanism in PC12 cells and in other cells that are undergoing cell division (such as in embryonic cells or tumors).

The effect of hypoxia on the ERK pathway in PC12 cells has also been studied. Exposure to moderate hypoxia (5% O2 for 6 hours) induced a marked increase in tyrosine phosphorylation of both ERK1 and ERK2, indicating activation of this pathway (20). The phosphorylation of ERK1 and ERK2 in response to hypoxia was somewhat less than that measured when PC12 cells were exposed to prototypical activators of ERK signaling, such as nerve growth factor and UV light.

In contrast to the effect of hypoxia on the p38 and ERK pathways, the JNK pathway is not activated by hypoxia in PC12 cells (20). However, the JNK pathway is functional, because exposure of PC12 cells to UV light increased JNK activity (20). Thus, hypoxia regulates the p38 SAPK and ERK pathways, but not the JNK pathway in PC12 cells. Activation of these pathways may play an important role in hypoxia regulation of gene expression and in cellular adaptation to hypoxia. The downstream transcription factors and protein kinases that are targeted by these pathways are beginning to be elucidated. Nevertheless, little is known about the specific genes that are regulated by these pathways in response to extracellular stress.

The ERK and SAPK pathways are also involved in the activation of hypoxia-induced gene expression in various other cell types. For example, the hypoxia-induced proliferative response in vascular fibroblasts is mediated by the ERK pathway (48). There is also evidence that the ERK pathway is involved in the stimulation of expression of vascular endothelial growth factor (VEGF) and hypoxia-induced angiogenesis in tumors (49). This is based on results showing that Ras-mediated activation of ERK1 and ERK2 stimulated transcription from the VEGF promoter, and that ERK1 and ERK2 activation led to direct phosphorylation of hypoxia inducible factor 1-α (HIF-1α), a transcription factor that binds to the HRE. Moreover, ERKs and SAPKs that are activated by various cellular stresses also contributed to the increased abundance of VEGF in hamster fibroblasts by stabilizing VEGF mRNA (49, 50). In a related study, hypoxia enhanced survival of nutrient-depleted HepG2 tumor cells, which are also O2-sensing cells, by reducing susceptibility to apoptosis by increased abundance of VEGF through a mechanism involving ERK1 and ERK2 phosphorylation (51). Additionally, ERK1 and ERK2 are activated by hypoxia in human microvascular endothelial cells-1 (52). Experiments using dominant-negative mutants in these cells revealed that ERK1 is required for hypoxia-induced transactivation by the HIF-1 complex.

Activation of MAPKs has also been associated with the cardioprotective effects that occur in heart cells following brief preconditioning exposures to hypoxia and ischemia. ERK1 and ERK2, p38 SAPK, and protein kinase C (PKC) were activated simultaneously during preconditioning by brief simulated ischemia followed by reoxygenation (53). Specifically, the inhibition of p38α led to protection against combined hypoxic and ischemic injury in neonatal rat ventricular myocytes. However, the role of p38 in mediating protection in cardiomyocytes appears to be somewhat complex, because another study showed that the duration of p38 activation might be important (54). Short durations of p38 activation during preconditioning stimulus protect cardiomyocytes from ischemic cell death, whereas longer durations can cause cell death. Thus, p38 kinase is activated in cardiomyocytes by ischemia-hypoxia preconditioning episodes, and activation of p38 may play a key role in determining the fate of cardiomyocytes during sustained ischemia-hypoxia insults.

These studies demonstrate that hypoxia can regulate the SAPK and ERK signaling pathways, and that activation of these pathways by hypoxia can mediate changes in function in various cell types and tissues. Further investigation is needed to gain insights about the specific mechanisms by which these pathways are activated by hypoxia and about the downstream targets of these pathways that lead to altered cell function.

The MAP Kinase Phosphatase-1 Pathway

MAPK phosphatase-1 (MKP-1, also known as CL100 and 3CH134) is a member of a family of dual-specificity phosphatases that oppose the effects of the MAPKs and SAPKs (55). Phosphorylation of MAPKs and SAPKs can be induced by many cellular stimuli. Upon phosphorylation of Thr-X-Tyr motifs, these signaling enzymes become activated and translocate to the nucleus, where they phosphorylate various transcription factors, thereby regulating gene expression. The MKP family of enzymes is capable of dephosphorylating both phosphothreonine and phosphotyrosine in Thr-X-Tyr motifs, such as those found in the MAPKs and SAPKs. Activation of MAPKs and SAPKs is frequently associated with activation of MKPs, suggesting that MKPs play a role in feedback control of MAPK signaling (56).

MKPs can generally be classified as either being localized primarily in the nucleus (MKP-1 and MKP-2) or in the cytosol (MKP-3, MKP-4, MKP-5 and the protein M3/6) (57). The physical interaction of MAPKs and SAPKs with MKPs can stimulate the catalytic activity of both cytosolic and nuclear MKPs (58-61). The expression of the nuclear MKPs is inducible, and these are considered to be immediate-early genes. The increase in MKP gene expression may represent another level of negative feedback regulation on MAPK signaling pathways (56, 62). Our laboratory found that MKP-1 mRNA and protein levels are both strongly increased by hypoxia in PC12 cells (63). We also found that this regulation is not unique to PC12 cells, in that it also occurred in other cell lines (HepG2, Hep3B) that have been used to study O2-responsive gene regulation (63). However, the increased abundance of MKP-1 in response to hypoxia does not occur in non-oxygen-sensing cell lines (HEK 293, COS-7) (63). In addition, the hypoxia-mimicking agents cobalt chloride and deferoxamine caused an increase in MKP-1 gene expression that was similar to that caused by hypoxia. Both agents mimic the effects of hypoxia, in part by increasing the binding activity and protein levels of the HIF family of transcription factors. This suggests that the hypoxia-induced increase in MKP-1 mRNA may depend on the activation of HIFs.

Several studies have implicated calcium as being involved in the regulation of MKP-1 gene expression (62, 64, 65), and other studies have indicated that the SAPKs are involved (66, 67). Interestingly, the increase in MKP-1 mRNA induced by hypoxia in PC12 cells was unaffected by the removal of extracellular and intracellular Ca2+, whereas the increase in MKP-1 mRNA due to KCl-induced depolarization was abolished in the absence of Ca2+ (63). It has been suggested that induction of MKP-1 gene expression occurs as a compensatory response to activation of MAPK (68, 69). Although pharmacological blockade of the MAPK kinase MEK prevented hypoxia-induced phosphorylation of ERK1 and ERK2, it did not alter the effect of hypoxia on MKP-1 mRNA levels (63). Therefore, the hypoxia-induced activation of ERK1 and ERK2 is not essential for the increase in MKP-1 gene expression. This finding is consistent with those from other studies in which inhibition of MEK was insufficient to prevent the induction of MKP-1 gene expression by other stimuli (70, 71). The PI3K-Akt pathway is also activated by hypoxia in PC12 cells, and this effect is blocked by wortmannin (21). However, wortmannin treatment had no effect on the increase in MKP-1 mRNA levels, which indicates that the PI3K-Akt pathway is not involved in hypoxic regulation of this gene (63).

The activation of MKP-1 gene expression by hypoxia was markedly attenuated, but not completely prevented, by a drug (SB203580) that inhibits the p38 family of SAPKs (63). There is increasing evidence that activation of p38 may have both apoptosis-promoting and cell-protective functions (72-76). Of the five known members of the p38 kinase family, only p38α and p38γ are activated by hypoxia in PC12 cells. The p38α and p38β subtypes are inhibited by SB203580, but the p38γ and p38δ subtypes are not. This suggests that the inhibition of the hypoxia-induced expression of MKP-1 in cells pretreated with SB203580 is probably due to blockade of p38α. The residual increase in MKP-1 (that which is not blocked by SB203580) may be mediated by p38γ or by other (non-p38) signaling pathways. The elucidation of the role of p38γ in hypoxic gene regulation awaits new tools, such as the development of a pharmacological inhibitor of this protein.

The Phosphatidylinositol 3-Kinase-Akt Pathway

The cellular processes related to survival and apoptosis are mediated in part by the PI3K-Akt pathway. Akt [also known as protein kinase B (PKB)] is a cytosolic serine-threonine kinase critical for cell survival under adverse conditions, and is also involved in different cellular processes, such as cell growth and differentiation (77, 78). Akt blocks apoptosis induced by a number of "death stimuli" through its effects on several downstream targets, including the proapoptotic Bc1-2 family member Bad, Forkhead transcription factors, and CREB. In a number of cell types, withdrawal of growth factors triggers programmed cell death, and activation of Akt can block this process (77-79). Certain stressors, including osmotic stress, H2O2, and sodium arsenite, also stimulate Akt (80, 81). However, this is a controversial area, because other studies have failed to find effects of various stress stimuli on Akt (82).

The cellular mechanisms involved in Akt regulation are beginning to be understood. Akt has a pleckstrin homology (PH) domain that preferentially binds phosphatidylinositol 3,4,5-trisphosphate [PI(3,4,5)P3] and phosphatidylinositol 3,4-bisphosphate [PI(3,4)P2] phospholipids, which are generated by PI3K. Akt translocates from the cytosol to the inner leaflet of the plasma membrane to bind PI(3,4,5)P3 and PI(3,4)P2 (83). At the lipid bilayer, Akt becomes phosphorylated by phosphatidylinositol-dependent kinase (PDK) (77, 78). Akt becomes phosphorylated on Ser473 (in the regulatory domain) and Thr308 (in the catalytic domain), resulting in the activation of the enzyme. Once activated, Akt detaches from the plasma membrane and translocates to the nucleus (83). At both the plasma membrane and in the nucleus, activated Akt phosphorylates various transcription factors and other regulatory proteins (77, 78), including glycogen synthase kinase (GSK), Bad, caspase 9, the Forkhead family of transcription factors, and inhibitor of κB (IκB) kinase (IKK). Some of these targets are involved in mediating cell death (77).

A dramatic increase in phosphoAkt (Ser473) appeared in PC12 cells after 6 hours of exposure to mild hypoxia (5% O2) and persisted through 24 hours of exposure (21). Activation of Akt, and its associated antiapoptotic effects, is typically initiated by PI3K activation (77, 84). In PC12 cells, inhibition of PI3K by wortmannin completely abolished the effect of hypoxia on phosphorylation of Akt, but did not alter the total Akt protein level (21).

Hypoxia also induced phosphorylation of GSK-3 (21), a well-characterized downstream target of Akt (85). Pretreatment with wortmannin blocked the hypoxia-induced phosphorylation of GSK-3, suggesting that hypoxia activates this signaling cascade through the PI3K pathway. The mechanism by which hypoxia activates the PI3K-Akt pathway is unknown, but the hypoxia-induced activation of this pathway is prevented by pretreatment of PC12 cells with cycloheximide or actinomycin D, which suggests that de novo protein synthesis is required (84).

Results from several studies suggest that Akt might play an important role in mediating cell survival during hypoxia. For instance, Mazure and co-workers (86) demonstrated that hypoxia-induced stimulation of VEGF gene expression was attenuated by wortmannin. It has also been suggested that activation of Akt promotes stabilization and accumulation of the hypoxia-inducible transcription factor HIF-1α (87). It is of interest to note that the HIF-1β subunit contains an Akt consensus phosphorylation site (77). The Akt -GSK-3 pathway regulates events involved in cellular metabolism, including the metabolism of glycogen (85, 88). In addition to its metabolic functions, GSK-3 also regulates cell survival in PC12 cells (89). Transfection of a constitutively active GSK-3 drives PC12 cells into apoptosis, whereas transfection with kinase-inactive GSK-3 blocks apoptosis (89). The activation of Akt (that is, phosphorylation of Akt) and the inactivation of GSK-3 (that is, phosphorylation of GSK-3) by hypoxia are consistent with the PI3K-Akt pathway promoting cell survival and tolerance to hypoxia in PC12 cells.

There is new evidence that phosphorylation of Akt may be involved in determining cell survival or cell death after transient focal cerebral ischemia (90). Immunohistochemistry showed that the expression of phosphorylated Akt was markedly increased in the cerebral cortex by 4 hours after the onset of reperfusion, whereas by 24 hours it was decreased. An increase in phosphorylated Akt was observed in cortical cells that survived cerebral ischemia, whereas the level of phosphorylated Akt was decreased in the ischemic core, a region of cell death.

PI3K-Akt is also a potential signaling pathway for expression of hypoxia-inducible genes in tumors and cancer cells. In hypoxic regions of tumors, HIF levels are increased and trigger the increased expression of genes encoding oncogenes and angiogenic factors, such as VEGF. The signal transduction pathways that mediate the activation of HIF in tumors remain unknown, but appear to require phosphorylation. PI3K-Akt is a major signaling pathway for the regulation of oncogenes and tumor suppressor genes and is involved in regulation of VEGF and, thus, is a good candidate for mediating HIF activation in tumor cells. Pharmacological inhibition of PI3K (by pretreatment with LY294002) inhibited the hypoxic induction of HIF-1α, but not HIF-2α (also known as EPAS-1), in breast cancer cells, and reduced both the basal and hypoxia-induced expression of VEGF mRNA and protein (91). Similar results were seen in other cancer cell lines, including PC-3 prostate carcinoma cells (87, 92). In these cells, growth factors, such as insulin and epidermal growth factor, increased the protein levels of the inducible HIF-1α, but not of the constitutively expressed HIF-1β. Both accumulation of HIF-1α and HIF-dependent gene expression were blocked by pharmacological inhibition of PI3K, by transfection with dominant-negative PI3K or dominant-negative Akt, and by transfection with the tumor suppressor PTEN, which is a PI3K antagonist.

There is also evidence that the ERK and PI3K-Akt pathways may interact synergistically in the activation of HIF-1α (93). Transfection with oncogenic Ras stimulated activity of a reporter gene construct that contained the VEGF promoter. This activity was blocked by inhibition of MEK1 with PD98059 or by inhibition of PI3K with wortmannin, and the effects of PD98059 and wortmannin were additive. The synergistic effect appears to arise from ERK phosphorylation of the HIF-1α regulatory-inhibitory domain and GSK-3 phosphorylation of the HIF-1α oxygen-dependent degradation domain. These results indicate that the use of PI3K inhibitors may be an effective approach for inhibiting the effects of hypoxia and oncogenes on tumor growth.

The PI3K-Akt pathway regulates signaling events that control cell survival and death. Abundant data suggest that this pathway is a major mediator of gene expression in a number of different cell types and tissues in response to hypoxia. Manipulation of this pathway with pharmacological agents might be a useful approach for either promoting or inhibiting the effects of hypoxia on cell function. In some cases (for example, brain ischemia), the desired effect would be cell survival, whereas in others (for example, tumors) it would be cell death.

Cyclic AMP, Protein Kinase A, and CREB

CREB is a transcription factor that promotes gene expression in response to many physiological signals, including neurotransmitters, depolarization, synaptic activity, mitogenic and differentiative factors, and stressors (24, 94-98). Multiple signaling pathways converge at the level of CREB, which regulates expression of genes that contain a specific DNA sequence called the cAMP response element (CRE) (99). Upon phosphorylation on Ser133, CREB can facilitate transcriptional activation of genes containing the CRE motif (100). Several protein kinases, including protein kinase A (PKA), calcium-calmodulin-dependent protein kinases (CaMKs), protein kinase C (PKC), ribosomal S6 kinase-2 (RSK-2), and mitogen-activated protein kinase-activated protein (MAPKAP) kinase-2, can mediate phosphorylation of CREB (25, 94, 96, 100-103). Transcriptional activation by CREB may require the recruitment of cofactors, such as CREB binding protein (CBP) or p300, to the promoter (104, 105), and phosphorylation of CREB at Ser133 facilitates interactions between CREB and its cofactors. Some studies have indicated that the cofactors themselves must be modified by distinct signal transduction events before complex formation can take place (106). However, there is also evidence that Ser133 phosphorylation of CREB alone is sufficient for recruitment, and that the additional signal-dependent modifications of CBP or p300 are not required (107).

Because of the major role for CREB in regulating genes that mediate a wide diversity of cellular functions in response to different environmental stimuli, the potential role of CREB in regulating the cellular response to reduced O2 levels was investigated in PC12 cells. Although the level of CREB protein remained unchanged, hypoxia induced a robust Ser133 phosphorylation of CREB that was greater than that produced by either forskolin (through activation of PKA) or by KCl-induced depolarization, two prototypical stimuli used to activate CREB (108). Thus, CREB phosphorylation could be an important mechanism by which PC12 cells respond to hypoxia. An important finding was that the hypoxia-induced phosphorylation of CREB was not mediated by any of the previously known pathways that activate CREB, including PKA-dependent and Ca2+-dependent protein kinases. This conclusion is based on the finding that phosphorylation of CREB stimulated by hypoxia occurred in PKA-deficient PC12 cells, in the absence of both extracellular and intracellular Ca2+, or after pharmacological blockade of the Ca2+-dependent isoforms of PKC. Lack of stimulation of MAPKAP kinase activity by hypoxia, and pharmacological inhibition of ERK1 and ERK2 and of p38 SAPK, indicated that these pathways are also not required for the hypoxia-induced phosphorylation of CREB in PC12 cells (108).

Unlike the phosphorylation of CREB, hypoxia-induced activation of TH gene expression was blocked completely by either removal of Ca2+ or inhibition of PKC (27). In both type I carotid body (1) cells and PC12 cells (26), TH gene expression is increased by hypoxia. The TH promoter contains a CRE that is critical for cAMP- and Ca2+-induced activation of expression (109, 110). TH gene expression was attenuated significantly in studies using a reporter gene [TH promoter coupled to chloramphenicol acetyltransferase (TH-CAT)] in which the CRE had been mutated (19). The TH gene contains several regulatory motifs that participate in hypoxia-induced activation of gene expression, including AP1-like and HRE-like cis elements located between –284 to –190 nucleotides relative to transcription start site (1). Thus, one or more of these upstream regulatory elements presumably mediates the residual activation of the TH gene in the absence of the CRE, and CREB is likely to be insufficient to mediate the entire induction by hypoxia.

Hypoxia has been implicated in a number of pathological functions that involve the cAMP-PKA-CREB axis. For example, in T84 intestinal epithelial cells, inflammatory mediators, such as tumor necrosis factor-α (TNF-α), are stimulated by hypoxia through a mechanism that involves reduction of CREB and phosphorylated CREB levels (111). Overexpression of wild-type CREB, but not mutated CREB (S133A), by retroviral-mediated gene transfer reverses hypoxia-elicited induction of TNF-α, defining a causal relationship between hypoxia-elicited CREB reduction and TNF-α induction (111). This indicates a role for CREB in the hypoxia-elicited epithelial phenotype, and implicates changes in intracellular cAMP concentrations as an important second messenger in differential induction of proinflammatory mediators by hypoxia.

The expression of the antiapoptotic Bcl-2 gene is stimulated by hypoxia and produces a protective effect in neuronal cells (111). This effect depends on the CRE in the Bcl-2 promoter. Using a reporter gene system in PC12 cells, the CRE in the Bcl-2 promoter was determined not to be essential for stimulation of Bcl-2 expression in response to nerve growth factor, but mutation of this element abolished the increase in expression in response to hypoxia (112). The isolated Bcl-2 CRE can also confer hypoxia-responsiveness on a heterologous promoter. Cotransfection of a dominant-negative CREB with the reporter gene construct abolished hypoxia-responsiveness. Hypoxia also increased the transcriptional activity of CPB (112). This finding demonstrates the importance of CRE-CREB interactions in the induction of Bcl-2 gene expression by hypoxia, allowing the Bcl-2 protein to protect neuronal cells against hypoxia.

CREB is also a molecular determinant of smooth muscle cell (SMC) proliferation that is stimulated by hypoxia (113). In arterial sections from the systemic and pulmonary circulation, the concentration of CREB was high in proliferation-resistant SMCs and low in SMCs prone to proliferation. Exposure to proliferative stimuli, such as hypoxia or platelet-derived growth factor (PDGF), decreased CREB concentration. In addition, transfection of SMCs with constitutively active CREB decreased PDGF-induced proliferation, whereas transfection with dominant-negative CREB increased proliferation. Assessment of CREB gene transcription by nuclear run-on analysis and transcription from a CREB promoter-luciferase construct indicated that CREB levels in SMC were in part controlled at the level of transcription. Overexpression of wild-type or constitutively active CREB in primary cultures of SMC arrested cell cycle progression. Active CREB decreased the expression of multiple cell cycle regulatory genes, as well as expression of genes encoding growth factors, growth factor receptors, and cytokines.

The varying effects on protein levels and phosphorylation in the different cells discussed illustrates that there is not a single canonical cellular response to hypoxia involving CREB. In PC12 cells, hypoxia stimulates CREB phosphorylation but CREB activation is not the only requirement for stimulation of expression of some CRE-containing genes (TH, for example). In intestinal epithelial cells, hypoxia decreases CREB activity, allowing the expression of proinflammatory mediators; in SMCs, hypoxia decreases CREB activity, allowing proliferation to occur. Thus, the cellular response to hypoxia can be to stimulate or inhibit CREB activity, depending on the cell type.

The second messenger cAMP is a mediator of cell protection during hypoxia. The cAMP-PKA signal cascade plays a central role in the regulation of both neural activity and energy metabolism. Under conditions of increased demand and decreased availability of energy (such as hypoxia, hypoglycemia, or excessive neuronal activity), an increase in adenosine (ADO) levels provides a powerful protective feedback mechanism (114). Acute (30 min) exposure to hypoxia causes the release of adenosine from the PC12 cells (115). Investigations into the intracellular and extracellular mechanisms underpinning the secretion of ADO in PC12 cells exposed to hypoxia revealed changes in gene expression and activities of several key enzymes associated with ADO production and metabolism, as well as decreased mRNA levels for the rat equilibrative nucleoside transporter rENT1, which is one of the major ADO transporters in PC12 cells (115). Decreases in the enzymatic activities of ADO kinase and ADO deaminase, accompanied by an increase in those of cytoplasmic and ecto-5′-nucleotidases, result in an increased capacity to produce intracellular and extracellular ADO. This increased potential to generate ADO and decreased capacity to metabolize ADO indicate that PC12 cells shift toward an ADO producer phenotype during hypoxia. In addition to the effect of hypoxia on ADO production and secretion, the A2A ADO receptor mRNA level, but not that of the A1 or A3 receptor, was substantially increased after exposure to hypoxia (116, 117).

Adenosine decreases Ca2+ current (ICa) and attenuates the hypoxia-induced increase in [Ca2+]i in PC12 cells. These effects are mediated through the adenosine A2A receptor, a heterotrimeric guanine nucleotide-binding protein (G protein)-coupled receptor that stimulates adenylate cyclase and PKA through activation of Gs (118). Calcium imaging studies using fura-2 revealed that anoxia (100% N2) produced a rapid rise in [Ca2+]i and that this rise was attenuated by ADO. Pre-exposure to chronic hypoxia (10% O2 for 24 or 48 hours) enhanced the subsequent anoxia-induced rise in [Ca2+]i. In contrast, pre-exposure to chronic hypoxia blunted the inhibitory effects of ADO on both ICa and on the anoxia-induced increase in [Ca2+]i.. Northern blot analysis and radioligand binding with [3H]CGS21680, a selective A2A receptor ligand, showed that the A2A receptor levels were increased by chronic hypoxia. Chronic hypoxia did not alter the immunoreactivity levels of Gαs or activity of adenylate cyclase. However, PKA enzyme activity was significantly inhibited when PC12 cells were exposed to 10% O2 for 24 or 48 hours. This suggests that reduced activity of PKA is the primary mechanism by which the actions of adenosine on ICa and [Ca2+]i are attenuated by chronic hypoxia in PC12 cells. This mechanism may also serve to limit negative feedback on ICa and [Ca2+]i by adenosine and therefore maintain enhanced membrane excitability of PC12 cells during long-term hypoxia. Similar results were obtained in whole-cell voltage clamp and fura-2 calcium imaging studies on isolated rat carotid body type I cells (116).

Protein Kinase C

PKC is a family of serine-threonine protein kinases that are involved in signal transduction pathways that regulate various cellular functions, including growth, proliferation, and apoptosis. The growth and proliferation functions of PKC have been implicated in hypoxia-induced responses. For instance, activation of PKC enhances the angiogenic process, and PKC is involved in the signaling of VEGF in retinal cells exposed to hypoxia (119). A dramatic increase in the angiogenic response to hypoxia and ischemia was observed in the retina of transgenic mice overexpressing PKCβ2, but not by the expression of the α, δ, or ϵ isoforms of PKC. The hypoxia-induced angiogenesis was inhibited by overexpression of a dominant-negative PKCβ2. A potential mechanism for this PKCβ-mediated response may involve VEGF-induced endothelial cell proliferation (119).

Pharmacological data suggest that PKC is also involved in regulating the increased pulmonary vascular resistance that occurs during hypoxia (120). Hypoxia significantly increased pulmonary arterial resistance, pulmonary venous resistance, and pulmonary capillary pressure and decreased total vascular compliance by decreasing both microvascular and large-vessel compliances. The nonspecific PKC inhibitor staurosporine, the specific PKC blocker calphostin, and the Ca2+-dependent PKC isozyme blocker Gö-6976 inhibited the effect of hypoxia on pulmonary vascular resistance and compliance. In addition, the PKC activator thymeleatoxin increased pulmonary vascular resistance and compliance in a manner similar to hypoxia. These results show that PKC inhibition blocks the hypoxic pressor response and that the pharmacological activation of PKC mimics the hypoxic pulmonary vasoconstrictor response.

Hypoxia leads to the activation of heat shock transcription factor (HSF) and HIF-1 in RIF (for radiation-induced fibrosarcoma) tumor cells (121). Heat shock protein (HSP) and angiogenic factor genes are stimulated by HSF and HIF-1 and are thought to contribute to the malignant progression of hypoxic tumor cells. A major factor in this response to hypoxia was the translocation of PKCδ to the cell membrane. Inhibition of PKCδ activation, either pharmacologically or genetically by transient transfection of a dominant-negative PKCδ, significantly inhibited the transcriptional activation of HSF and HIF-1 by hypoxia. These results strongly substantiate a view that the PKCδ isozyme plays an important role in transmitting hypoxia signals to both HSF and HIF-1 in RIF tumor cells. In addition, the membrane translocation of PKCδ is dependent on the activation of PI3K. Treatment with a PI3K inhibitor (wortmannin or LY294002) abrogated not only PKCδ translocation, but also the subsequent activation of HSF and HIF-1 by hypoxia (121). Together, these results show that the PKCδ isozyme acts to transmit hypoxia-induced signals that stimulate the activity of both HSF and HIF-1, and that an upstream regulator of PKCδ is PI3K.

There is evidence that PKC and Ca2+ act coordinately to regulate the cellular response to hypoxia. Pretreatment of PC12 cells with the L-type calcium channel blockers nifedipine or verapamil, or chelation of extracellular Ca2+, inhibited the increase in TH mRNA stimulated by hypoxia (27). Addition of chelerythrine chloride (CHL), a PKC inhibitor, to the medium before exposure to hypoxia also resulted in an inhibition of TH induction by hypoxia. These results suggest that hypoxia regulates TH gene expression by a mechanism that depends on influx of calcium partially but not exclusively through the L-type calcium channels, and that a member of the PKC family is essential for this regulation.

In addition, PKC is involved stimulating gene expression related to hypoxia in a number of different tumors. For instance, expression of the glucose-regulated protein (GRP78) can be stimulated by environmental stresses, such as glucose deprivation and hypoxia. The functional significance of GRP78 is not clear, but there is growing evidence that it is involved in promoting malignant progression and resistance to irradiation and chemotherapy in hypoxic tumor cells. The expression of GRP78 appears to be regulated by a complex signaling system involving ERK1 and ERK2 and PKC (122). A selective PKC inhibitor, GF109203X, inhibited the induction of GRP78 gene expression, as well as the activities of both ERK and Raf-1. Among six PKC isoforms expressed in MKN28 cells, PKCϵ protein level and kinase activity were increased by hypoxia. Transfection of MKN28 cells with a dominant-negative PKCϵ blocked the stimulation of GRP78 expression and the activation of ERK1 and ERK2 by hypoxia, indicating that PKCϵ directly participated in GRP78 induction under hypoxia. These results suggest that PKC activates the Raf-1-MEK-ERK signaling cascade to regulate hypoxia-induced GRP78 expression in human gastric cancer cells.

PKC also activates the ERK pathway to regulate expression of early growth response-1 (Egr-1) gene in endothelial cells during hypoxia (123). Treatment of endothelial cells with PD98059, a specific inhibitor of MEK1, inhibited hypoxia-induced Egr-1 expression. The involvement of the ERK pathway was further substantiated by the inhibition of Egr-1 promoter activity (assayed with a reporter gene) when endothelial cells were cotransfected with a dominant-negative mutant of Ras (RasN17), Raf-1 (Raf 301), or a catalytically inactive mutant of ERK2 (mERK). Addition of the PKC inhibitor calphostin C completely blocked the hypoxia-stimulated increase in Egr-1 gene expression. Hypoxia also increased ERK phosphorylation, which was abolished by administration of PD98059, calphostin C, or BAPTA/AM, a calcium chelator. Involvement of PKCα in mediating ERK activation was confirmed by the inhibition of ERK and the subsequent Egr-1 gene induction with antisense oligonucleotides to PKCα. These results indicate that the stimulation of Egr-1 expression by hypoxia requires calcium and a PKC-mediated Ras-Raf-1-ERK signaling pathway.

Finally, PKC is involved in regulating hypoxia preconditioning, which improves cellular survival during sustained hypoxia. In rat hepatocytes, evidence was presented suggesting that the A2A adenosine receptor, Gi proteins, phospholipase C, PKCδ and PKCϵ, and p38 SAPK are responsible for the development of liver ischemic and hypoxic preconditioning (124). PKC has also been implicated in regulating the signaling events in central nervous system cells involved in the regulation of respiration (125) and in oxygen-sensing cells in the carotid body (126). Thus, there is an abundance of evidence that PKC is a key mediator in the cellular the response to hypoxia.

The Pyk2 Pathway

Pyk2 (also known as CADTK, CAKp, and RAFTK) is a proline-rich nonreceptor tyrosine kinase that is activated by an increase in intracellular calcium concentration and is highly expressed in neural cell types and in PC12 cells (110, 127-130). Pyk2 can be activated by many signals, including activation of muscarinic acetylcholine (m1) receptors, PKC, growth factors, fibronectin, reactive oxygen species, and various stress-related signals (110, 127-132). Pyk2 is structurally related to the focal adhesion kinase (110, 127-129, 133). The proline-rich regions of Pyk2 provide binding sites for SH3 domain-containing proteins, such as p130cas and the small guanosine triphosphatase (GTPase)-activating protein Graf (134, 135). Activation of Pyk2 has been associated with an activation of Src, JNK, and ERKs (110, 127-130, 136).

Excitable O2-sensing cells such as PC12 cells respond very quickly to hypoxia with an increase in [Ca2+]i (2, 3). Furthermore, withdrawal of extracellular calcium blocks the hypoxia-induced increase in expression of certain hypoxia-regulated genes, including TH and junB (27). Hypoxia and depolarization with KCl cause a rapid (within 5 min of the onset of hypoxia exposure) increase in phosphotyrosine content of Pyk2, which persists throughout at least 6 hours of hypoxia exposure and which is lost in the absence of extracellular calcium (137). It is possible that the activation of Pyk2 by hypoxia specifically targets substrates that are involved in the cellular and molecular response to hypoxia. One such target is the voltage-dependent K+ channel, Kv1.2 (127, 131). Pyk2 phosphorylates Kv1.2 on one or more tyrosine residues within the cytosolic COOH-terminal portion of the channel (131). In addition to being voltage-dependent, Kv1.2 is also an O2-sensitive K+ channel (3, 23). An early event in the response to hypoxia is inhibition of conductance through Kv1.2 channels (3, 23). Furthermore, when oocytes expressing wild-type Pyk2 and Kv1.2 are treated with phorbol myristol acetate (PMA) to activate PKC, Kv1.2 currents are markedly inhibited (110). However, this effect is absent when a mutant (kinase-inactive) form of Pyk2 is coexpressed with Kv1.2 in oocytes. It appears that the primary role of Pyk2 may be to coordinate the activity of O2-sensitive ion channels, such as Kv1.2, with the intracellular Ca2+ concentration. Although the Kv1.2 channel has been identified as a downstream target of Pyk2, other hypoxia-responsive targets remain to be identified. Other downstream signaling pathways that are regulated by Pyk2 include MAPK, JNK, and c-Src (127, 129, 130, 138).

Summary: Hypoxia-Activated Signal Transduction

A myriad of signaling pathways are activated by reduced O2 in excitable O2-sensing cells (Fig. 1). These pathways are involved in regulating various cellular responses, including membrane polarization, secretion, cell fate, and gene expression. It is important to note that one of the earliest events observed in excitable cells is inhibition of O2-sensitive potassium channels (Kv1.2 in PC12 cells), which mediates depolarization and activation of voltage-sensitive Ca2+ channels. The increase in [Ca2+]i becomes a central player by regulating the cellular response to hypoxia. In addition, MAPK and SAPK signal transduction pathways and pathways involved in cell survival and death are activated by hypoxia. There is growing evidence that these different signal transduction pathways are involved in the modulation and, in some cases, the activation of hypoxia-induced transcription factors. The roles that each of these pathways contributes to survival or cell death or to the development of tolerance to hypoxic conditions remain to be determined. It is also important to note that the signaling events associated with reduced O2 are different in excitable and nonexcitable cells. Nonexcitable cells do not functionally depolarize during hypoxia and therefore do not employ an influx of Ca2+ as a signal transduction mechanism to regulate the cellular response.

Fig. 1.

Overview of the signal transduction pathways activated by hypoxia. The left part of the figure highlights the signaling pathways activated by depolarization in response to decreased O2. The middle shows how decreased O2 signals impact cell survival and growth factor signaling pathways. The right side highlights the stress-activated pathways that are involved in cellular responses to decreased O2. PDK1, phosphoinositide dependent kinase 1; SRF, serum response factor; SOS, son of sevenless; Pyk2, a proline-rich nonreceptor tyrosine kinase; MEKK, mitogen-activated protein kinase kinase kinase; MKK, mitogen-activated protein kinase kinase; FKHR, a forkhead transcription factor; p130cas, an adaptor protein; RLPK, RSK-like protein kinase; LRG, any of the many late-response genes activated subsequent to activation of early-response genes, such as c-fos and c-jun.

Oxygen-Regulated Transcription Factors: The Hypoxia-Induced Factors (HIFs)


The mechanisms involved in regulation of gene expression in response to hypoxia are beginning to be understood. HIF-1α is critical for hypoxia-induced regulation of a number of genes in several different cell types and tissues [for review, see (139, 140)]. HIF-1α is basic helix-loop-helix (bHLH) PAS domain protein that forms heterodimers with HIF-1β, which has been identified as the aryl hydrocarbon nuclear translocator (ARNT). The helix-loop-helix structure and PAS domains are involved in heterodimerization and DNA binding. HIF-like proteins (that is, PAS domain proteins) exist in both prokaryotes and eukaryotes, which suggests that they are universal mediators of gene expression throughout nature. It has been argued that the HIF proteins, which are activated instantaneously by hypoxia, are capable of sensing reduced O2 (141). Support for this possibility comes from studies showing that HIF can be activated by iron, cobalt, and other metal ions that are known to interact directly with heme, the prototypical sensor of alterations in O2 in hemoglobin (142-146). Moreover, iron chelation increases HIF-1 levels and stimulates HIF-1-mediated gene expression (142, 143). Nevertheless, there is no evidence that HIF-1α is a heme-containing protein.

The abundance of the HIF-1 heterodimer is increased during hypoxia by stabilization (that is, reduced degradation) rather than by an increase in its mRNA (147-149). HIF-1α is rapidly degraded during normoxia and becomes stable when the O2 level is decreased. There is compelling evidence that the stabilization of HIF-1α is due to inhibition of ubiquitin-mediated degradation by proteasomes (149). It is now generally believed that the degradation of HIF-1α is mediated by the von Hippel-Lindau protein (VHL), which is an active E3 ubiquitin ligase complex (144, 145, 150, 151). Evidence for the involvement of VHL in promoting the degradation of HIF-1α comes from the finding that HIF-1α is stable in VHL-defective cells; re-expression of VHL restores O2-dependent instability of HIF-1α. In addition, VHL coimmunoprecipitates with HIF-1α (144). Renal carcinoma cell lines, which lack expression of the von Hipple-Lindau (VHL) tumor suppressor protein, maximally express HIF-1α under normoxic conditions, such that the abundance of HIF-1α is not increased by hypoxia (144). VHL mediates degradation of HIF-1α during normoxia by binding to a region of HIF that is termed the oxygen-dependent degradation domain (ODD) (149). The ODD domain consists of about 200 amino acid residues, located in the central region of HIF-1α (149). Deletion of this region leads to stabilization of HIF-1α under normoxic conditions. Moreover, the ODD region is able to confer O2-dependent instability when fused to Gal4, a yeast transcription factor.

The HIF family of transcription factors binds to the HRE, an enhancer element located within the regulatory region of many hypoxia-responsive genes (152). Once activated by reduced O2, HIF-1α dimerizes with HIF-1β in the nucleus and binds to DNA at sites represented by the consensus sequence 5′-RCGTG-3′ (153). The binding of the HIF-1 complex during hypoxia regulates genes involved in diverse physiological process, including angiogenesis, proliferation, cell survival or death, erythropoiesis, energy metabolism, and oxygen chemoreception. The HIF complex remains elevated during sustained hypoxia and is destabilized and reduced when the O2 level returns to normal physiological levels. It appears that regulation of the HIF-1 complex is associated with its stabilization and destabilization by VHL-induced proteolytic degradation. Thus, signaling events that regulate VHL interactions with the HIF-1 complex and specifically with HIF-1α are of considerable interest. In this regard, there is evidence that VHL, through its β-domain, binds directly to HIF-1α, and targets the HIF-1 complex for ubiquitination (154). It was proposed that the VHL protein has a function analogous to that of an F-box protein, which would be to recruit substrates necessary for ubiquitination to the HIF-1α.

Analysis of the O2-mediated degradation of HIF-1 complex revealed that the recognition of HIF-1α by VHL is regulated by the enzymatic hydroxylation of specific prolyl residues within the ODD of HIF-1α (155-157). Ratcliffe and colleagues (158) provided the first insights into the mechanism that regulates proline hydroxylation and degradation of HIF-1α. These workers defined a conserved HIF-prolyl hydroxylase in Caenorhabditis elegans that was identified as EGL-9. Recombinant EGL-9 activity is modulated by graded hypoxia, 2-oxoglutarate, iron chelation, and cobalt, and EGL-9 may function as an O2 sensor. Furthermore, HIF-1 was not regulated by hypoxia in C. elegans in which EGL-9 was mutated.

Several mammalian orthologs of EGL-9 have been identified. Ratcliffe and colleagues (158) identified three human HIF prolyl hydroxylases (HIF-PH) named PHD1, PHD2, and PHD3 (for prolyl hydroxylase domain containing). Like EGL-9, the activity HIF-PHs is modulated by 2-oxoglutarate, iron chelation, and cobalt. PHD2 and PHD3, but not PHD1, expression levels are regulated by hypoxia in HeLa cells. Other investigators have also identified several human and mouse homologs (159-161). The various nomenclatures and GenBank accession numbers are summarized in Table 1. To date, only one rat homolog, termed SM-20, has been identified (162, 163).

Table 1.

Mammalian HIF prolyl hydroxylases.

The mechanism responsible for activating EGL-9 and the mammalian HIF-PHs by reduced O2 is not fully understood. Of the mammalian HIF-PHs, SM-20 has been the most extensively studied. SM-20 was first identified in rat vascular SMCs, where it is stimulated by growth factors (162). SM-20 protein has been reported to be localized in the cytoplasm (164) and in the mitochondria (165). In addition to its potential function in O2-sensing and regulation of VHL-mediated HIF-1 degradation, SM-20 has been described as being the product of an immediate early gene that is involved in differentiation (166, 167) and caspase-dependent cell death (165). In rat embryo fibroblasts, SM-20 is regulated by the p53 tumor suppressor protein (168). Its stimulation by growth factors suggests that growth factor-responsive signal transduction pathways, such as the ERK pathway, might also regulate SM-20.


Another HIF-like protein called endothelial PAS domain protein 1 (EPAS-1, also known as HIF-2α) was identified in the type I O2-sensing cells of the carotid body. HIF-2α is a bHLH transcription factor that shares 48% sequence identity or similarity with HIF-1α (169-172). HIF-2α protein levels, like HIF-1α levels, are relatively low under basal (normoxic) conditions and accumulate upon exposure to reduced O2 (19). Like HIF-1α, HIF-2α translocates into the nucleus, forms a heterodimer with HIF-1β, and then binds to the HRE on hypoxia responsive genes. Multiple signals are involved in regulation of HIF-2α and, like HIF-1α, two domains of HIF-2α are required for its activation by hypoxia (170, 173). One of the critical HIF-2α domains is an internal domain that extends from amino acids 450 to 571 and shares homology to the ODD of HIF-1α, which is critical for activation by reduced O2 (149). The second important HIF-2α regulatory domain is a COOH-terminal activation domain (amino acids 824 to 876), which is the site of posttranslational modification that occurs during hypoxia (170). HIF-2α binds to cobalt, and the binding site is within the ODD (146). The ODD is necessary and sufficient for regulation of protein stability in response to hypoxia. There is a 17-amino-acid sequence within the ODD, conserved among all HIFs, that mediates the interaction between HIF and VHL (144). Recent findings suggest that hydroxylation of proline residues within the ODD regulates VHL binding to and eventual degradation of HIF (155-158). As stated above, VHL mediates the degradation of HIF-1α through a ubiquitin-proteasome pathway. The interaction between VHL and HIF is disrupted by cobalt. In HIF-2α, the cobalt and VHL binding sites overlap (146). Mutation of this overlapping site prevents cobalt binding and leads to stabilization of HIF-2α. Thus, a major component of the signaling mechanism that leads to accumulation of HIF-1α or HIF-2α might be the activation of proteins that bind to this region of the ODD and prevent degradation by VHL. In addition, it appears that activation of HIFs by cobalt and other metal ions occurs through a mechanism that prevents HIF degradation.


A third member of the HIF family, HIF-3α, was recently discovered (174). Like HIF-1α and HIF-2α, HIF-3α contains bHLH and PAS domains. However, HIF-3α lacks one of the two transactivation domains found in the COOH-terminus of the other HIF proteins. HIF-3α is able to bind HIF-1β and can activate transcription of an HRE-containing reporter gene in a transient transfection system. The regulation of HIF-3α by hypoxia is controversial. Some investigators have reported that HIF-3α activity was stimulated by hypoxia and cobalt (174, 175). However, others failed to see hypoxic activation of HIF-3α in transiently transfected COS-7 cells (176). This latter group also reported that HIF-3α abolishes the hypoxia-induced transcriptional activity of cotransfected HIF-1α or HIF-2α, possibly by competing for limited quantities of HIF-1β. Thus, HIF-3α might be a negative regulator of hypoxia-inducible gene expression.

Signal Transduction Pathways That Regulate HIF Proteins

The specific signaling pathways involved in HIF-1α and HIF-2α activation are not well understood. The HIF-2α protein level is robustly elevated during hypoxia in PC12 cells, and an HRE-luciferase reporter gene containing several repeats of the HRE is stimulated by reduced O2 (19). Reduced O2 stimulates the ERK pathway, which activates several transcription factors, including c-Fos, JunB, CREB, and Elk-1 (169, 177, 178). We found that increased expression of recombinant HIF-2α stimulated HRE-luciferase reporter gene activity under both normoxic and hypoxic conditions (19). Pharmacological inhibition of MEK1, the enzyme immediately upstream from ERK1 and ERK2, completely blocked the effect of hypoxia on both the basal and HIF-2α-stimulated HRE-luciferase activity. In addition, transfection of a constitutively active MEK1 (pFC-MEK1) enhanced HRE-luciferase activity during both normoxia and hypoxia. When coexpressed with HIF-2α, pFC-MEK1 caused a much larger increase in the transactivation of the HRE-luciferase reporter gene by hypoxia than did transfection with HIF-2α alone. These results strongly indicate that MEK1 and ERK are important for mediating HIF-2α activation of the HRE-dependent gene expression.

Ras is the initial step in the ERK signaling pathway. Neutralization of Ras with increasing amounts of a dominant-negative Ras construct, RasN17, had no effect on the HIF-2α transactivation of the HRE-luciferase reporter (19). However, coexpression of the same amounts of RasN17 did block activation of a c-fos-luciferase reporter gene by nerve growth factor (NGF), the prototypical activator of the Ras-Raf-MEK-MAPK pathway. These findings indicate that hypoxia activates ERK signaling and HIF-2α by a Ras-independent mechanism.

As described previously, hypoxia results in depolarization and Ca2+ influx into PC12 cells and carotid body type I cells during hypoxia. Depolarization of PC12 cells results in ERK activation through a calmodulin-dependent mechanism (179, 180). We found that pretreatment of PC12 cells with calmodulin antagonists (W13 or calmidazolium chloride) caused a pronounced reduction in hypoxia-induced ERK phosphorylation and in both endogenous and exogenous HIF-2α transactivation (19). Thus, ERK activation of HIF-2α occurs through a calmodulin-dependent pathway, rather than through the prototypical mediator, Ras.

Although ERK is critical for hypoxia regulation of HIF-2α function, it is not the kinase that phosphorylates HIF-2α (19). Neither hypoxia-induced phosphorylation nor accumulation of HIF-2α protein was inhibited by pharmacological inhibition of MEK1 with PD90859 (19). This is based on the finding that blockade of MEK1 failed to prevent incorporation of [32P]-orthophosphate into HIF-2α. It is possible that multiple ERK-dependent and ERK-independent signals are required for HIF-2α activation. We propose that one ERK-independent signal leads to accumulation of HIF-2α protein, presumably by inhibition of ubiquitin-proteasome degradation (181). A second ERK-independent signal leads to the phosphorylation of HIF-2α. The mechanism of ERK-dependent activation of HIF-2α is unknown. The fact that HIF-2α phosphorylation persists in the presence of MEK1 blockade suggests that the ERK pathway does not directly target HIF-2α, but instead targets other protein(s) that are critical for the formation of the HIF-2α DNA binding complex. In contrast, ERK activation is necessary for the phosphorylation of HIF-1α, but not for HIF-1α stabilization (170). CBP interacts with HIF-1α and HIF-2α and potentiates the activation of these hypoxia transcription factors (170). The COOH-terminal regions of CBP can be phosphorylated by ERK (182). Thus, CBP might be a target of hypoxia-activated ERK, which could then recruit HIF-αs to the DNA binding complex. How or whether ERK contributes to the interaction between VHL and HIF-1α or HIF-2α protein is unknown. Finally, it has been proposed that several "general transcription factors" are present in the HIF-2α DNA binding complex (170). These proteins are potential targets of ERK regulation. Thus, it is likely that the ERK-dependent activation of HIF-2α transactivation involves the recruitment of proteins other than HIF-2α to the DNA binding complex.

Additional research is needed to elucidate the exact signal transduction mechanisms by which reduced O2 leads to the accumulation of HIF proteins and the transactivation of hypoxia responsive genes. The various HIF proteins may be regulated by different mechanisms. In addition, it is important to realize that these mechanisms will probably differ greatly between excitable and nonexcitable O2-sensing cells. In this regard, it is important to recognize that not all hypoxia-responsive genes contain an HRE and, therefore, are not targets for the HIFs. For example, the immediate early gene junB is robustly activated by hypoxia, yet does not appear to have an HRE in the 5′ regulatory promoter region. A primary mechanism for regulation of junB and other AP1 immediate early genes is Ca2+ and cAMP signaling pathways (183). Thus, a complete understanding of the gene expression pattern that is activated by hypoxia requires a comprehensive approach for identifying hypoxia-responsive genes in both excitable and nonexcitable cells, and genes whose expression is regulated by HIFs and non-HIF mechanisms.

The Future: Identification of the Hypoxia Genome

The ongoing quest to understand the role of gene expression in mediating alterations in cell phenotype (for example, a hypoxia-tolerant phenotype) has historically been restricted to the study of single genes and proteins. Yet, cellular adaptation to environmental stimuli, such as hypoxia, is a very complex physiological function that requires the simultaneous expression of hundreds of different genes and proteins. The Human Genome Project (and various other sequencing projects) is not only leading to the identification of sequences for the entire genome, but has also led to the development of new high-throughput approaches that allow identification of the genes involved in complex cellular functions. Gene expression (cDNA) microarray technology and the expanding gene sequence databases have made it possible to identify the complement of transcribed genes (the "transcriptome") within cells and tissues that are stimulated in response to an environmental stimulus or stress, such as hypoxia. The gene expression profile that results from extracellular stimuli is analyzed by various computational approaches and then classified according to gene structure and function. Thus, functional genomics offers the potential to markedly expand our understanding of biological systems and how these systems adapt to environmental stimuli.

We have used a high-throughput functional genomics approach to identify the genes that regulated by hypoxia in PC12 cells. Our approach begins with the construction of a subtracted cDNA library, which isolates the known and unknown (novel) genes that are stimulated or repressed by reduced O2. The technique for making the subtracted cDNA library is called subtractive suppression hybridization (SSH) (63, 184, 185). SSH increases the probability of obtaining low-abundance differentially expressed cDNAs. Each clone is sequenced and identified in either public or private databases. The clone inserts are then amplified and used to print microarray slides.

We used this approach to identify about 200 unique known genes and 40 unknown sequences (potential novel genes) that are stimulated by hypoxia in PC12 cells (63). We call these stimulated genes the PC12 cell "hypoxia genome." The major challenge is to group these genes into clusters based on function and the mechanism of activation. Combining cDNA microarray analysis with pharmacological approaches described above will enable mapping of the various signal transduction pathways involved in the hypoxic response and assignment of the genes to the signal transduction pathway(s) that regulate them. This is likely to be a very complex process, because there is cross-talk among the different pathways, and many genes will be regulated by more than one pathway. Complications can also arise from lack of specificity of pharmacological agents; use of techniques such as traditional gene knockouts and RNA interference (RNAi) (186) should alleviate some of these problems. This type of information will provide a more comprehensive picture of how O2-sensing cells function and adapt to a chronic hypoxic environment. In addition, information from such studies will provide insights into the development of diagnostic and therapeutic products for ischemic and hypoxic disease. This type of analysis will also enable us to determine if genes that regulate the cellular response to hypoxia stress also regulate the response to other cellular stresses.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
  187. 187.
View Abstract

Navigate This Article