|Home | About | Journals | Submit | Contact Us | Français|
Growing evidence indicates that ubiquitin ligases play a critical role in the hypoxia response. Among them, Siah2, a RING finger ligase, is an important regulator of pathways activated under hypoxia. Siah2 regulates prolyl hydroxylases PHD3 and 1 under oxygen concentration of 2% to 5%, thereby allowing accumulation of hypoxia-inducible factor (HIF)-1α, a master regulator of the hypoxia response within the range of physiological normoxic to mild hypoxic conditions. Growing evidence also indicates an important function for Siah2 in tumor development and progression based on pancreatic cancer, mammary tumor, and melanoma mouse models. This review summarizes our current understanding of Siah2 regulation and function with emphasis on hypoxia and tumorigenesis.
Organisms respond to decreased oxygen (hypoxia) by altering fundamental physiological activities such as respiration and metabolism to maintain homeostasis. Without a proper response to hypoxia, multiple systems could become deregulated resulting in potential cell death. Organisms from fungi or yeast to mammals can mobilize a hypoxia response. However, biological responses differ from species to species. Fungi mostly control their metabolism to alter growth, whereas mammals have acquired more complex responses in accordance with their need to maintain multiple systems. In humans, the hypoxia response can include physiological responses ranging from altered metabolism, respiration, blood cell production, and neovascularization (1).
Two cellular responses to hypoxia are possible: an acute/transient hypoxia response or a chronic/sustained response. In an acute response, cells immediately begin conserving ATP by limiting the cellular processes consuming energy and decrease the O2 consumption. They also increase respiration and produce specific cytokines for the cell survival, migration, or vascularization. During chronic hypoxia, increases in red blood cell number or formation of new vasculature occur, adapting an organism to decreased oxygen conditions. Normally, cells in our body exist at around 5% oxygen concentration. Although it is lower than the oxygen concentration in the atmosphere (21%), which is generally referred to as normoxia, we will refer to the range of 5% to 2% oxygen as “physiological normoxia to mild hypoxia conditions” in this review. Cells have developed highly sensitive mechanisms to detect subtle decreases (e.g., from 5% to 2%) in oxygen levels in order to trigger the hypoxia to response. Tumorigenic environments also create a state of chronic hypoxia. As cancer cells grow, hypoxic regions with insufficient connections to the vascular network form. If this condition persists, a hypoxia response is induced, resulting in production of cytokines such as vascular endothelial growth factor (VEGF), which promote vascularization. Eventually a tumor establishes connection to the vascular network, enhancing tumor cell growth and promoting metastasis (recently reviewed in refs. 2, 3). In this review, we focus on activity of the ubiquitin ligase Siah2 in regulating the hypoxia response and discuss its role in tumor development.
Cells exposed to decreased oxygen alter fundamental cellular activities, such as translation or transcription. Hypoxia-inducible factor (HIF) transcription factors are key regulators of the latter. HIFs are basic helix-loop-helix/PAS proteins consisting of an α subunit (e.g., HIF-1α) and a β subunit (e.g., ARNT). Together they form a heterodimer that regulates transcription by binding to a hypoxia-responsive element (HRE) (refs. 3, 4). HRE has a core five-nucleotide sequence RCGTG (R: A/G), which is well conserved among numerous hypoxia-responsive genes. Those targets can be categorized into several groups on the basis of their physiological roles, such as cell cycle (cyclin, ref. 5; IGF2, ref. 6; TGF-α, ref. 7); cell death (NIP3, NIX, refs. 8, 9); metabolism (GLUT1, ref. 10; PGK-1, refs. 11, 12; G3PDH, ref. 13; PDK1, refs. 14, 15); angiogenesis (VEGF, refs. 16–18); erythropoiesis (EPO, ref. 19); and cell migration (c-met, ref. 20; CXCR4, ref. 21). As many of these genes function in the hypoxia response, HIF is considered as a central regulator of hypoxia.
Of the three HIF-α proteins (1α, 2α, and 3α), HIF-1α and HIF-2α are particularly critical for the hypoxia response; recent reports show a distinct function of HIF-1α and HIF-2α such as transcriptional regulation in cooperation with c-Myc on cancer progression (22, 23). The role of HIF-3α is less well understood, but it has been suggested that the alternative spliced form of HIF-3α binds to and inhibits transcriptional activity of HIF-1α (24). Although HIF-1α and HIF-2α share several common targets such as VEGF, it has recently been shown that Oct4 is a specific, direct target of HIF-2α but not of HIF-1α (25). Moreover, it is suggested that HIF-1α functions in early stages of neuroblastoma development, whereas HIF-2α promotes growth and metastasis in later stages (23). Expression patterns of these two genes also differ: HIF-1α is expressed ubiquitously, whereas HIF-2α expression is limited to endothelium, kidney, lung, heart, and small intestine (22). Finally, in mice, knockout of either gene results in an embryonic lethal phenotype. However, embryos die of different causes: HIF-1α embryos die around E10.5 due to cardiac and vascular defects (26–28), whereas HIF-2α null embryos die of bradycardia, vascular defects, and incomplete lung maturation (29–32).
HIF-1α and HIF-2α are similarly regulated by oxygen levels. In decreased oxygen concentrations, HIF-α levels increase mainly because of regulation at the protein level. In normoxia, HIF-α subunits are actively ubiquitinated and degraded by the proteasome. The E3 ligase responsible for this ubiquitination is a Skp, Cullin, F-box (SCF)-type ubiquitin ligase complex consisting of pVHL, elongin B, C (also called the VBC complex), and Rbx1 (33, 34). pVHL is the molecule recognizing the HIF-α subunit (35). Mutation of pVHL is seen in patients with renal clear cell carcinoma. pVHL/HIF-α interaction requires hydroxylation of HIF-α prolines 402 and 564 located in what is termed the oxygen-dependent degradation domain (ODD; refs. 36, 37). Once hydroxylated, HIF-αs are captured by the pVHL binding pocket and ubiquitinated with subsequent degradation (38, 39).
HIF-1α undergoes other posttranslational modifications. It can be phosphorylated by ERK-MAPK, resulting in increased transcriptional activity (40). HIF-1α can also be SUMOylated, which can either positively or negatively regulate its stability (41–43). Hydroxylation of HIF-1α asparagine 803, located in the C-terminal transcriptional activation domain, inhibits transcriptional activity by blocking recruitment of the coactivators p300/CBP (44).
HIF-α is hydroxylated on prolines 402 and 564 by a family of enzymes called prolyl hydroxylase domain-containing proteins or PHDs (EGLN, HPH) and on asparagine 803 by FIH (factor inhibiting HIF; refs. 37, 44). Although they hydroxylate different residues, PHD and FIH belong to the Fe(II) and 2-oxoglutarate-dependent oxygenase protein family. Enzymes in this family require oxygen, ascorbic acid, 2-oxoglutaric acid, and Fe2+ for catalytic activity, and loss of any one of these factors reduces activity (45). Thus, PHD and FIH actively hydroxylate and thereby negatively regulate HIF under normoxic conditions when oxygen is abundant and are inhibited in hypoxia, enabling HIF to become active.
Three PHD proteins have been identified: PHD1, 2, and 3 (37). In vitro all three robustly hydroxylate HIF-1α (46). However, in vivo PHD2 plays a major role in normoxic HIF-α regulation, and it and PHD3 function in either mild or prolonged hypoxic conditions (47, 48). The role of PHD1 in HIF-1α regulation in vivo is not clear, although it reportedly hydroxylates IKK proteins (49). Recent reports also indicate that FIH hydroxylates and regulates the activity of Notch and IκB, both in the ankyrin repeat domains (50–52).
Several groups have generated PHD knockout mice. PHD2−/− mice showed embryonic lethality between E12.5 and 14.5, whereas PHD1 and 3−/− mice were apparently normal (53). PHD1 nulls exhibited altered anaerobic metabolism and lower oxygen consumption in skeletal muscle. This metabolic defect causes hypoxic tolerance in mice mainly because of impaired HIF-2α regulation (54). Conditional inactivation of PHD2 in mice increases EPO production, leading to polycythemia and congestive heart failure, possibly due to continuous expression of high levels of HIF (55). PHD3−/− mice show decreased rates of neuronal apoptosis, resulting in increased size of the superior cervical ganglion, adrenal medulla, and carotid body and overall impaired sympathoadrenal development (56). These animal models indicate an important role for PHDs in physiological normoxia and mild hypoxic conditions.
Protein ubiquitination plays a critical role in the hypoxia response. For example, under hypoxia, the ubiquitination of HIF-1α and p53 is inhibited, whereas MyoD or IκB ubiquitination is promoted (57–59). Ubiquitination is achieved by sequential reaction of ubiquitin-activating enzyme E1, ubiquitin-transferring enzyme E2, and a substrate-specific ubiquitin ligase E3 (60). Substrates recognized by E3 are either mono- or poly-ubiquitinated and degraded, or in some cases, altered in terms of subcellular localization or protein-protein interaction (61). Siah2 is a RING finger type ubiquitin ligase with a catalytic RING domain on its N-terminus, followed by two zinc fingers and a C-terminal substrate binding domain (SBD; 62). Siah is a mammalian homolog of Sina, a Drosophila protein functioning in eye development (63, 64). Mammals express two isoforms, Siah1 and 2, and mice exhibit two Siah1 isotypes, 1a and 1b (65). A number of Siah substrates have been identified, including N-CoR, β-catenin, TRAF2, 2-oxoglutarate dehydrogenase-complex protein E2 (OGDC-E2), and TIEG (66–70; see Table 1 for detail). In some cases, interaction requires an adaptor protein such as phyllopod (PHYL) and Siah-interacting protein (SIP) to degrade Drosophila Tramtrack and mammalian β-catenin, respectively (see below; 67, 71). Siah2 null mice exhibit mild phenotypes, such as minor increases in the number of hematopoietic progenitor cells (72). However, Siah1a and Siah2 double nulls exhibit embryonic lethality, indicating an essential function in embryogenesis.
To search for proteins interacting with Siah2 we undertook mass spectrometric and co-immunoprecipitation analyses. These analyses identified PHDs 1, 2, and 3. Active degradation of PHD3 and a lesser degree of PHD1 were seen in 293T cells. Degradation was inhibited by treatment with the proteasome inhibitor MG132, supporting the idea that Siah2 targets PHD3 and PHD1 for degradation via the ubiquitin-proteasome pathway (73). Interestingly, although PHD2 was also found to associate with Siah2, PHD2 stability was not affected by the ligase. Among the mechanisms underlying selective targeting of PHD3 but not PHD2 by Siah2 is their distinct subcellular localization.3
Because PHD3 regulates HIF-1α stability through hydroxylation, the hypoxia response was monitored in mouse embryonic fibroblasts (MEF) established from either Siah2 knockout or wild type mice. HIF-1α levels were significantly reduced in MEFs derived from Siah2 nulls, and induction of a HIF-1α target, VEGF, was decreased in response to hypoxia. Importantly, MEFs derived from Siah1a/Siah2 double nulls exhibited more severe phenotypes and almost no HIF-1α expression or VEGF induction. These outcomes likely underlie the embryonic or neonatal lethality of Siah1a/2 double null mice (72). Because PHD3 is a primary Siah2 target and likely plays an important role in HIF-1α expression in some cell types, PHD3 expression was evaluated in Siah2 null MEFs relative to wild type controls. Importantly, inhibition of PHD3 expression by RNAi in Siah1a/2 double null MEFs rescued HIF-1α expression, suggesting that PHD3 mediates decreased HIF-1α expression in the absence of Siah2. Furthermore, a hypoxic environment enhanced the ability of Siah2 to degrade PHD3. The Siah2 mRNA levels increased in as little as 2 hours after cells were exposed to hypoxic conditions, an increase that was also seen under physiological normoxic to mild hypoxic conditions (5% O2; Fig. 1, also see below).
In vivo experiments with Siah2 knockout mice exposed to hypoxic conditions (7% O2) were undertaken to examine hemoglobin concentration and ventilatory responses. Wild type mice exposed to hypoxia exhibit a moderately higher, albeit statistically significant, increase in blood hemoglobin concentration within 14 days. In terms of the ventilatory response, wild type mice undergoing acute phase hypoxia (in 15 minutes) exhibited decreased metabolism and increased ventilation. By contrast, Siah2 null mice significantly decreased their metabolism but could not increase ventilation. Changes seen in the Siah2 knockout animals resemble those seen in the HIF-1α heterozygous animals (74) and are expected to be more pronounced in the absence of Siah1a and Siah1b genes.
Gel filtration analyses indicate that a proportion of PHD3 exists as a protein complex in cultured cells (75). This complex exhibits a higher molecular weight upon hypoxia treatment, which is reversible when cells become reoxygenated. PHD3 protein in the complex exhibits decreased HIF hydroxylation activity. Thus, complex formation may serve as a secondary mechanism to negatively regulate PHD3 activity such as in prolonged hypoxia. Siah2 was mostly found in a smaller molecular weight fraction, suggesting that PHD3 subject to active degradation by Siah2 does not reside in the large complex. Recently, PHD3 aggregates have been observed under normoxia condition by immunocytochemistry (76). The aggregate formation is dependent on the prolyl-hydroxylase activity of PHD3. These aggregates contain the proteasome component LMP-2 and appear to be scattered in the cytoplasm. The larger PHD3 complex and these aggregates likely represent two independent structures. However, the fact that PHD3 is found in both suggests that its activity is dynamically regulated by association and dissociation with specific complexes. The key question remains what would be the components of those structures.
The hypoxia response involves multiple signaling cascades, including Akt or MAPK kinase pathways (77). Among those kinases, p38 MAPK is activated under hypoxic conditions in a manner partly dependent on reactive oxygen species (ROS) produced in mitochondria (78). p38 activity or that of the upstream kinases MKK3 or 6 is important to stabilize HIF-1α under hypoxia conditions, as shown in MEFs derived from various kinase knockout mice (79). Siah2 is subjected to phosphorylation, which increases its activity under hypoxia conditions (Fig. 1; 80). This phosphorylation of Siah2 was inhibited when cells were treated with p38 inhibitor SB203580. Siah2 contains the SQ/TQ amino acid sequence motif, a conserved motif for p38 phosphorylation. Phosphopeptide mapping or mutational analyses revealed S29 and T24 as major sites of Siah2 phosphorylation. Phosphomimetic mutants (T24E or S29D) showed gain-of-function activity and enhanced HIF-1α expression. Overall these studies indicate that phosphorylation-dependent Siah2 activation is sufficient to activate a component of the hypoxia response.
Akt signaling functions to induce HIF-1α expression. First, growth factors, such as insulin-like growth factor (IGF-1), induce HIF-1α expression independent of hypoxia (81). Second, in some cell types, such as glioblastoma cells, hypoxia itself induces Akt activation and concomitant HIF-1α stabilization through an unknown mechanism (82). Siah2 mRNA induction by hypoxia (Fig. 1) is blocked by treatment with the PI3K inhibitor LY294002.4 LY294002 treatment also decreases Akt activity. Conversely, Siah2 is induced when an active form of Akt is introduced into cells. Therefore, hypoxic activation of Akt would be one of the critical signals to induce Siah2 expression. Ongoing studies are addressing what transcription factors mediate Akt-dependent regulation of Siah2 transcription.
Tumors generate local hypoxic regions, triggering a response necessary to recruit new vasculature required for tumor cell survival. Because of the key role played by HIF-1α in tumor development, we examined potential involvement of Siahs in tumorigenesis and metastasis using a syngeneic xenograft model of SW1 melanoma cells. SW1 melanoma cells were stably transfected with two types of Siah inhibitors (PHYL peptides or a dominant negative Siah2 RING mutant). Peptides derived from the PHYL sequence (PHYL peptides) bind to the Siah SBD with high affinity (83) and disrupt PHD3 binding, thereby elevating PHD3 levels and reducing HIF-1α levels (84). Whereas PHYL peptide expressing cells affect Siah2 targeting of substrates mediated by SBD, Siah2 RING mutant-expressing cells attenuate general function of Siah2 by serving as a dominant negative form. The two types of cells (PHYL and RING mutant expressing) were subcutaneously injected into C3H mice to monitor potential changes in tumor formation and lung metastases.
Expression of PHYL peptide in SW1 cells (SW1PHYL) used in xenograft models did not alter primary tumor size but rather inhibited lung metastasis. SW1 cells expressing IPAS, a potent inhibitor of HIF transcriptional activity (24), show similar phenotypes. Furthermore, ectopic re-expression of HIF-1α in SW1PHYL cells rescues the deficiency in lung metastasis with no change in primary tumor size, indicating overall that the Siah effect on metastasis is dependent on HIF-1α (Fig. 2). Although reduced HIF-1α staining compared to the parental SW1 cells is observed in both primary tumor cells and metastases formed by SW1PHYL cells, metastatic loci derived from SW1PHYL cells showed decreased vasculogenesis and proliferation concomitant with increased cell death. Therefore, Siah seems to promote vasculogenesis, cell growth, and survival in lung metastatic lesions, in part via its effect on HIF-1α levels. These observations indicate that the Siah2-PHD3 pathway plays an important role in tumor development, although this regulatory axis does not affect PHD2, which is considered to be the prominent PHD that regulates HIF-1α under normoxia condition. In addition, our results suggest that organ microenvironment plays a critical role in determining Siah and consequently HIF-1α effect on tumor development and progression. Specifically, conditional knockout of HIF-1α in the mammary epithelium of MMTV-PyMT mice significantly inhibits formation of lung metastases but has less effect on growth of the primary mammary tumor (85).
Unexpectedly, we found that expression of the dominant negative Siah2 RING mutant (Siah2-Rm) in SW1 melanoma cells (SW1S2RM) resulted in reduced growth of primary tumors and decreased the number of lung metastases in the xenograft model (Fig. 2). Although Siah2-Rm alone has little effect on levels of PHD3 and HIF-1α in SW1 cells, it robustly upregulates levels of Sprouty2 (SPRY2), a specific inhibitor of Ras/ERK signaling. SPRY2 suppresses lung tumorigenesis in a K-ras transgenic mouse model (86) and has been identified as a Siah2 substrate (87). ShRNA knockdown of SPRY2 in SW1Siah2-Rm cells rescues ERK signaling deficiencies in vitro and tumorigenesis in vivo, indicating that Siah2 regulates tumorigenesis via SPRY2/ERK but independent of PHD3/HIF-1α (Fig. 2). Of note, although the possibility that Siah2-Rm affects tumorigenesis through its effect on other Siah2 substrates should be considered, the primary role of SPRY2 in mediating the change in tumorigenicity is confirmed by a shRNA experiment in which inhibition of SPRY2 abolishes the effect of Siah2-Rm.
Importantly, initial analysis of a panel of human melanoma samples reveals an inverse correlation between levels of PHD3 and SPRY2 and the progression of malignancy, pointing to an oncogenic role of Siah2 in human melanomas.
Consistent with these findings, several groups report a role for Siah2 in tumorigenesis and metastasis of diverse tumor types. Siah2 activity is correlated with pancreatic tumor development likely through its positive effect on Ras signaling (88). This observation supports our studies of Siah2-SPRY2 in melanoma tumor development (84). Further, inhibiting Siah2 activity with peptides designed to interfere with Siah2 association with substrates efficiently inhibited mouse mammary tumor development and metastasis (89). Consistent with these data, crossing Siah2 null mice with the TRAMP prostate cancer mouse model resulted in marked inhibition of primary and metastatic tumors.5 Furthermore increased expression of SIP is seen in pancreatic and nasopharyngeal tumors. Increased SIP levels likely enhance targeting of substrates requiring this adaptor protein (90). However, in renal cancer cells, SIP is downregulated and overexpression leads to growth inhibition (91). Clearly, the functional significance of adaptor proteins on tumorigenesis is complex and requires further investigation.
In summary, Siah ligases regulate tumorigenesis and metastasis by targeting parallel pathways mediated by SPRY2/Ras and PHD3/HIF, respectively (Fig. 2). The differential effect of PHYL peptide and Siah2-RM on PHD3 and SPRY2 seems to be due to substrate specificity of the Siah isoform (Siah1 versus Siah2) and to mechanisms used by Siah to bind substrates [via the SBD (adaptor is required for some substrates) versus SBD-independent]. Our data suggest that both Siah1 and Siah2 are required to target PHD3 via the Siah SBD and both may utilize an unknown adaptor, whereas Siah2 alone is sufficient to target SPRY2, possibly via direct binding through a region outside the SBD (Fig. 3).
HIF-α proteins have been extensively studied because their expression is regulated in an oxygen-dependent manner and they regulate expression of numbers of genes enabling cellular adaptation to a hypoxia environment. However, it has also been suggested that (1) there are hypoxia responses independent of HIF, such as the mTOR pathway, and (2) that HIF may function in nonhypoxic situations following growth factor stimulation (81, 92). In addition to HIF, additional factors may be up- or downregulated following exposure to hypoxia. One potential means of regulating these factors could be through ubiquitin-dependent degradation, and Siah substrates are candidates for this process. As Siah2 activity is increased by phosphorylation and induced under hypoxia, other factors may also be regulated similarly to PHD3. It is reported that OGDC-E2 is a Siah2 target (68). Although factors signaling OGDC-E2 degradation are not well understood, chronic hypoxia in rat brains decreases OGDC activity, and Siah2 may function in this process (93). Thus far, however, we have not detected changes in levels of two Siah substrates, TRAF2 and β-catenin, under hypoxic conditions,4 raising the question of how specificity is achieved when several substrates are degraded by one E3 ligase. This could be explained by spatial expression patterns or localization of ligases and substrates. It is noteworthy that Siah substrate specificity seems to be dependent on the types of stresses cells encounter. For example, Siah1 promotes β-catenin degradation following UV-induced DNA damage, whereas Siah2 causes TRAF2 degradation upon TNF-α-induced apoptosis (67, 69). Several mechanisms likely underlie stress-dependent substrate specificity. First, different stimuli affect ligase expression. For example, Siah1 expression is induced by p53, whereas Siah2 expression is induced by hypoxia (73, 94). Second, different stimuli might alter subcellular localization of ligases or substrates. For example, acidosis induces nucleolar sequestration of pVHL to allow HIF-1α stabilization under normoxia (95). Third, different stimuli may alter posttranslational modification of ligases or substrates, which could in turn change the localization of proteins or the affinity of ligase/substrate interactions. For example, p38-induced phosphorylation of Siah2 can enhance its affinity to PHD3 (80).
Importantly, Siah proteins are found from human to fly and worm. Major functional domains of Siah2, N-terminus RING domain, Zinc-finger domain, and C-terminus SBD, are all highly conserved among the species (Fig. 3B). Thus, it is expected that Siah2 exerts its control of HIF-dependent and -independent pathways throughout the evolution of these species.
A key note in the recent studies is that Siah can promote tumorigenesis and metastasis by targeting SPRY2/Ras/ERK and PHD3/HIF-1α pathways. Because these pathways are critical in human cancers, the study highlights the potential to develop Siah inhibitors. Further investigations will clarify the role of Siah in cancer, such as whether there are any other substrates involved in the process, or a possibility of different mechanisms in different types of cancers.
We thank members of our laboratories for helpful discussions. We apologize to colleagues whose studies were not cited because of space limitations.
Grant support: KN is supported by the Program for Improvement of Research Environment for Young Researchers (MEXT, Japan) and the Takeda Science Foundation. JQ is supported by Canadian Institutes of Health Research (CIHR) postdoctoral fellowship. ZR is supported by National Cancer Institute (NCI) grant (CA111515).
3K. Nakayama and Z. Ronai, unpublished data.
4K. Nakayama, J. Qi, and Z. Ronai, unpublished data.
5J. Qi, K. Nakayama, and Z. Ronai, unpublished data.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.