Ero1-Lα plays a key role in a HIF-1-mediated pathway to improve disulfide bond formation and VEGF secretion under hypoxia: implication for cancer


Oxygen is the ultimate source of oxidizing power for disulfide bond formation, suggesting that under limiting oxygen proper protein folding might be compromised. We show that secretion of vascular endothelial growth factor (VEGF), a protein with multiple disulfide bonds, was indeed impeded under hypoxia and was partially restored by artificial increase of oxidizing equivalents with diamide. Physiologically, the oxireductase endoplasmic reticulum oxidoreductin-1 (Ero1)-Lα, but not other proteins in the relay of disulfide formation, was strongly upregulated by hypoxia and independently by hypoglycemia, two known accompaniments of tumors. Further, we provide genetic evidence that induction of Ero1-Lα by hypoxia and hypoglycemia is mediated by the transcription factor hypoxia-inducible factor 1 (HIF-1) but is independent of p53. In natural human tumors, Ero1-Lα mRNA was specifically induced in hypoxic microenvironments coinciding with that of upregulated VEGF expression. To establish a physiological relevance to modulations in Ero1-Lα levels, we showed that even a modest, two- to three-fold reduction in Ero1-Lα production via siRNA leads to significant inhibition of VEGF secretion, a compromised proliferation capacity and enhanced apoptosis. Together, these findings demonstrate that hypoxic induction of Ero1-Lα is the key adaptive response in a previously unrecognized HIF-1-mediated pathway that operates to improve protein secretion under hypoxia and might be harnessed for inhibiting tumor growth via inhibiting VEGF-driven angiogenesis.


Reduced oxygen availability (hypoxia) negatively impacts on vital cellular processes and, accordingly, induces host adaptive and corrective responses. Upregulated transcription of most genes participating in adaptive responses to hypoxia is controlled by the transcription factor hypoxia-inducible factor 1 (HIF-1) (Czyzyk-Krzeska, 1997). HIF-1 target genes play key roles in multiple pathways, including angiogenesis, vascular reactivity and remodeling, glucose and energy metabolism, cell proliferation and survival, erythropoiesis, iron homeostasis and others (Carmeliet et al., 1998) (for a review see Semenza, 2000).

Another process likely to be impaired under hypoxia, and hence benefit from a HIF-1-mediated adaptive response, is disulfide bond formation. Disulfide formation is essential for the correct folding of proteins in the endoplasmic reticulum, primarily those destined for secretion. Protein disulfide isomerase (PDI) is the direct donor of disulfides to newly synthesized proteins (Ferrari and Soling, 1999). PDI is kept, in turn, in the oxidized state by the action of endoplasmic reticulum oxidoreductin-1 (Ero1), a FAD-containing, endoplasmic reticulum membrane-associated protein (Tu et al., 2000). The ultimate source of oxidizing power for PDI oxidation and hence to disulfide formation is molecular oxygen that directly oxidizes Ero1 (Tu and Weissman, 2002) (for a recent review, see Tu and Weissman, 2004).

Thus, when oxygen availability becomes limited, the whole cascade might be halted. The resultant accumulation of misfolded proteins in the ER triggers a process known as the unfolded protein response (UPR) marked by increased transcription of chaperons, and folding catalysts and proteins that still remain misfolded are relocated to the cytosol for proteasomal degradation (for a review see Rutkowski and Kaufman, 2004). Certain ER proteins are known to be regulated by hypoxia, including Grp78, Grp94 (Sciandra et al., 1984) and ORP150 (Kuwabara et al., 1996).

Vascular endothelial growth factor (VEGF), examined here as an example for a secreted protein, is a homodimer whose proper folding through formation of three intramolecular disulfide bonds and two intersubunit disulfide bonds is a prerequisite for its function (Potgens et al., 1994; Claffey et al., 1995; Muller et al., 1997). VEGF is regulated under hypoxia at multiple levels: transcription of VEGF is induced via direct binding of HIF-1 to a hypoxia response element (HRE) in its promoter (Forsythe et al., 1996). Hypoxia also leads to stabilization of the otherwise short-lived VEGF mRNA (Shweiki et al., 1992; Ikeda et al., 1995; Shima et al., 1995) and an internal ribosome entry site (IRES) refractory to hypoxia is used to secure efficient translation (as cap-dependent mRNA translation is generally impaired under hypoxia) (Stein et al., 1998). Yet, it is possible that a failure to properly fold and secrete the protein under hypoxia, circumstances where it is most needed, might jeopardize its bioavailability. Neutralizing antibodies to VEGF are already in clinical use, having proven effective in prolonging life of cancer patients. Thus, additional new modalities to reduce the amount of VEGF secreted by hypoxic tumors might also be useful for inhibition of tumor growth. Accordingly, we searched for a HIF-1 transcriptional target that normally operate to increase the capacity of disulfide formation under conditions of diminishing oxygen and, therefore, might serve as a target for reducing VEGF secretion. Here we show that Ero1-Lα, but not other components in the protein relay participating in disulfide formation, is a HIF-1 target gene and a key player in the hypoxic response to disulfide bond formation. Moreover, we demonstrate that VEGF secretion is reduced in the tumor cells following RNAi-mediated Ero1-Lα suppression and that even moderate downregulation of Ero1-Lα has a cytotoxic effect.


Secretion of VEGF is impeded under hypoxia due to unfavorable oxidative balance

Certain genes that help the cell to cop with hypoxic insults encode membrane bound or secreted proteins that must themselves pass through the ER secretory pathway. Since a correct oxidative balance in the ER is critical for the proper function of oxireductases implicated in disulfide formation, ER hypoxic stress might impede secretion of these proteins.

To test this contention, we focused on VEGF, a secreted protein with three intramolecular disulfide bridges and two intermolecular bonds that is considered the most important angiogenic factor and the leading target in perspective anti- and proangiogenic therapies. VEGF ELISA was used to monitor amounts of VEGF secreted by mouse embryonic fibroblasts under normoxia and hypoxia, relative to the levels of VEGF mRNA and intracellular protein produced. Comparing secreted VEGF to intracellular VEGF is, however, problematic since, in general, misfolded, nonsecreted proteins are at least partially degraded and hence intracellular VEGF is likely to be underappreciated (Claffey et al., 1995; Walter et al., 1996; Rutkowski and Kaufman, 2004). For this reason, we compared the amount secreted VEGF to the level of VEGF mRNA, based on earlier studies indicating that translation of VEGF mRNA proceed at a 100% efficiency also under hypoxia (Stein et al., 1998). As evident from Figure 1a, the 10-fold increase in VEGF mRNA induced in mouse embryonic fibroblasts under hypoxia was not matched by a similar increase in the amount of secreted protein, which was merely two-fold. In HIF-1α−/− cells subjected to hypoxia, VEGF secretion was lower than the normoxic level. A similar deficit in secreted VEGF was also observed using several tumorigenic cell lines (data not shown). Thus, the deficit in secreted VEGF is most likely due to a protein maturation/secretion problem. To further show that impaired VEGF secretion is associated with hypoxia, cells proficient and deficient for HIF-1 were compared with regard to VEGF secretion. Again, significantly less protein was secreted under hypoxia per VEGF mRNA expressed and less protein was secreted by HIF-1-null cells. Notably, this difference was observed already under ‘normoxia’, presumably reflecting incidental stress experienced by the cultured cells even without a deliberate insult (Figure 1b). It should be pointed out, however, that VEGF served in this study as a representative example for a secreted protein with multiple disulfide bonds and other proteins are presumably affected similarly. Indeed, a recombinant secreted alkaline phosphatase (SeAP) activity detected in the medium of transfected cells was also reduced under hypoxia and, in HIF-1α null cells, was only 40% of the normoxic level (data not shown).

Figure 1

Secretion of VEGF is impeded under hypoxia. (a) Cultures of transformed mouse embryonic fibroblasts HIF-1α+/+ and HIF-1α−/− were grown under normoxia or 20 h exposure to hypoxia as indicated in ‘Materials and methods’. Steady-state levels of VEGF mRNA were determined by densitometric scanning of Northern blots. VEGF protein secreted by the same cells was determined using VEGF ELISA. Results are expressed as fold induction under hypoxia relative to the normoxic levels. (b) Levels of VEGF mRNA and secreted protein were determined at the indicated times points after the onset of hypoxia in an isogenic pair of HIF-1α+/+ and HIF-1α−/− cells (Ryan et al., 2000) and compared to cells maintained under normoxia. Results are expressed as VEGF mRNA/secreted protein ratio (in arbitrary units)

To examine whether the secretion deficit under hypoxia was due to unfavorable oxidative balance in the ER, we tested the effect of diamide. Diamide causes −SH oxidation in the ER and was previously shown to functionally rescue a genetic deficiency in the yeast oxireductase Ero1p (Frand and Kaiser, 1998). When added to cultured HIF-1α−/− transformed mouse fibroblasts at a concentration of 0.2 mM for 5 h, diamide improved VEGF secretion by 1.8-fold, without changing VEGF mRNA levels (the average of five independent experiments; s.d.=0.165). These cells were chosen in order to provide a sensitized background, and since HIF-1 activity is redox sensitive (Wang et al., 1995b).

Ero1-Lα is induced by hypoxia in vitro and co-induced with VEGF in hypoxic tumor microenvironments

Results described above clearly point that there is a bottleneck in optimal protein maturation/secretion under hypoxia and suggest that a positive feedback mechanism to increase the ER oxidizing capacity might exist. In the cascade of redox reactions leading to disulfide bond formation, the oxireductase Ero1 is the most upstream (Tu and Weissman, 2002), rendering it an attractive candidate for a hypoxia-inducible gene (see scheme in Figure 2a). Expression of Ero1-Lα mRNA in hepatocellular carcinoma cells was indeed strongly upregulated shortly after exposure to hypoxia, reaching maximal steady-state levels within 12 h (Figure 2b). Elevated levels of Ero1-Lα protein accompanied increased expression of Ero1-Lα mRNA (Figure 2c). Noteworthy, upregulation of Ero1-Lα by hypoxia was demonstrated in a variety of tumor cell lines, as well as in nontransformed, primary cells, including hepatocellular carcinoma, C6 glioblastoma, transformed mouse embryonic fibrobalsts, primary MEFs and embryonic stem cells. Fold induction of Ero1-Lα under hypoxia in these cells was: 22, 7, 11, 4 and 2, respectively. These results indicate that hypoxic induction of Ero1-Lα is a general phenomenon. Hypoxic induction of Ero1-Lα in some cell lines was recently reported by Gess et al. (2003).

Figure 2

Ero1-Lα is induced by hypoxia in vitro. (a) Schematic model of oxidative protein folding in the ER (adapted from Tu and Weissman, 2004). (b) Mouse hepa1c1c7, hepatocellular carcinoma cell line, was exposed to hypoxia for the indicated times and Ero1-Lα mRNA levels were determined using Northern blotting. Two Ero1-Lα mRNA species (4 and 1.8 kb) are indicated by arrows. The smaller transcript contains a shorter 3′-untranslated region, as determined through sequencing of Ero1-Lα cDNA clones (HK and EF, unpublished data). (c) A Western blot of protein extracts prepared from fibroblasts grown under normoxia or 20 h of hypoxia, showing increased level of Ero1-Lα protein under hypoxia. (d) Actinomycin D (7.5 μg/ml) was added to hepatocellular carcinoma cell cultures that has been exposed to hypoxia (or kept at room air) for 18 h. The samples were then withdrawn at the indicated times and levels of residual Ero1-Lα mRNA were visualized using a Northern blotting (the auroradiographic stain on the second-from-left is likely an experimental artifact)

As certain hypoxia-inducible genes are also regulated at the level of mRNA turnover (Stein et al., 1995), we also examined whether the stability of the Ero1-Lα mRNA is increased by hypoxia, using actinomycin D-chase experiments. As shown in Figure 2d, the turnover rate of Ero1-Lα mRNA is very low (determined as >6 h) and was not further increased by hypoxia, indicating that elevated steady-state levels of Ero1-Lα mRNA are predominantly due to transcription activation. Similar results were also obtained in Actinomycin D-chase experiments performed with another cell type (see Figure 4c below). In this regard, Ero1-Lα differs from VEGF mRNA that is a short-lived mRNA stabilized under hypoxia (Stein et al., 1995).

Figure 4

A functional HIF-1 is essential for induction of Ero1-Lα by hypoxia and hypoglycemia. (a) Cells nullified for HIF-1α (transformed mouse embryonic fibroblasts) and (b) cells nullified for HIF-1β (hepatocellular carcinoma line) were compared with the respective wild-type counterpart with regard to hypoxia-induced expression of Ero1-Lα. (c) Cells nullified for P53, showing an efficient hypoxic induction, were also examined for the stability of the mRNA using Actinomycin D blockade. (d) Embryonic stem cells nullified for HIF-1α were examined with respect to induction of Ero1-Lα by hypoglycemia (growth for the indicated time in glucose-free medium)

To show that Ero1-Lα expression is upregulated by hypoxia also in vivo, we performed in situ hybridization analysis of tumor sections. During expansion of the tumor mass, and due to insufficient neovascularization, certain microenvironments become hypoxic. When exceedingly hypoxic, these cells may even die and, thus, tumor cells experiencing severe hypoxia are readily distinguishable by their proximity to dying cells (Shweiki et al., 1992). As shown in Figure 3, Ero1-Lα mRNA was specifically induced in tumor cell adjacent to cells undergoing cell death, that is, in cells experiencing most severe hypoxia. This was shown both for a murine teratocarcinoma (Figure 3a–d), as well as for a natural human breast carcinoma (Figure 3e–g). Notably, Ero1-Lα mRNA was induced in the same locales as VEGF, suggesting that upregulated Ero1-Lα expression might indeed function to improve VEGF secretion.

Figure 3

Ero1-Lα is induced in hypoxic tumors in vivo. (a–d) Teratomas grown in nude mice were sectioned and hybridized in situ with an Ero1-Lα-specific probe as indicated in ‘Materials and methods’. Images shown are of bright field (a) and dark filed (b) illuminations of the same thin H&E-stained section. (c and d) Adjacent serial sections hybridized with Ero1-Lα or stained for TUNEL, respectively. Note that Ero1-Lα mRNA is induced in both viable hypoxic cells as well as cells undergoing apoptosis. (e–g) Adjacent sections of a human breast carcinoma specimen hybridized in situ with an Ero1-Lα-specific probe (e and f) or a VEGF-specific probe (g). Note upregulated expression of Ero1-Lα mRNA in cells residing at the interphase between viable (V) and necrotic (N) areas

Hypoxic and hypoglycemic induction of Ero1-Lα is mediated by the transcription factor HIF-1 and is independent of p53

Several transcription factors were shown to play a role in hypoxia-induced gene expression. Among those, HIF-1 is the major player coordinating different adaptive responses to hypoxia (Wang and Semenza, 1993; Semenza, 1999, 2000). We wished to determine, therefore, whether Ero1-Lα is also a HIF-1 target gene. The involvement of HIF-1 was initially suggested by the observation that Ero1-Lα expression is also induced by the iron chelator desferrioxamine (DFO), which is known to inhibit the prolyl-hydroxylase responsible for HIF-1 degradation (Ivan et al., 2001; Jaakkola et al., 2001) and, hence, frequently used as surrogate hypoxia marker in HIF-1-mediated responses (data not shown). To provide unequivocal evidence that HIF-1 is upstream to Ero1-Lα in the hypoxia response, we took a genetic approach. Since HIF-1 functions as an obligatory HIF-1α/HIF-1β heterodimer (Wang et al., 1995a), we examined whether upregulation of Ero1-Lα by hypoxia is abrogated in HIF-1α-null cells or, alternatively, in HIF-1β-null cells. As shown in Figure 4a and b, HIF-1α-null cells totally failed to upregulate Ero1-Lα under hypoxia. Likewise, HIF-1β-null cells were incapable of inducing Ero1-Lα under hypoxia.

Previous studies have shown that the transcription factor p53 may affect cellular responses to hypoxia, possibly also through making complexes with HIF-1 (An et al., 1998; Halterman et al., 1999). We wished to determine, therefore, whether the hypoxic regulation of Ero1-Lα is modulated by the p53 status. To this end, we compared wild type and p53−/− primary embryonic fibroblasts with regard to their ability to upregulate Ero1-Lα under hypoxia. As shown in Figure 4c, the ability to upregulate Ero1-Lα under hypoxia is fully maintained in p53−/− cells. We conclude that Ero1-Lα regulation is mediated by HIF-1 in a p53-independent manner.

Hypoglycemia is another cellular insult converging on increased HIF-1 activity, evidenced by findings that certain HIF-1 target genes (e.g. VEGF) are also induced by hypoglycemia (Stein et al., 1995; Maltepe et al., 1997). As shown in Figure 4d, Ero1-Lα was also strongly induced by hypoglycemia under normoxic conditions. Furthermore, this response was abrogated in the absence of a functional HIF-1. We argue that increasing the ER oxidizing power when glucose becomes scarce (the expected consequence of hypoglycemic Ero1-Lα induction) might be of a considerable physiological significance.

The physiological relevance of modulations in Ero1-Lα levels and potential use for reducing VEGF secretion

A rate-limiting step in the folding of many newly synthesized proteins is the formation of disulfides bonds. To determine whether Ero1-Lα is unique with regard to hypoxia regulation among oxireductases participating in disulfide formation, we extended a similar analysis to other candidate genes in this protein relay. Specifically, we examined hypoxia responsiveness of PDI (the archetypal member of the PDI family), its family members Erp57 (Ellgaard and Frickel, 2003) and Erp72 (Mazzarella et al., 1990), and the Ero1-Lα-related gene ALR (Sevier et al., 2001). As shown in Figure 5, Ero1-Lα was the only gene induced by hypoxia. We also examined the Ero1-Lβ (Pagani et al., 2000) but in the cells tested could not detect a significant level of transcripts even following exposure to hypoxia (data not shown).

Figure 5

Lack of regulation by oxygen of other enzymes participating in disulfide formation. Specific probes for the indicated enzymes were prepared as described under ‘Material and methods’ and tested against RNA blots from a transformed mouse embryonic fibroblasts cell line that was either subjected (+) or not (−) to 16 h of hypoxia

To assign any physiological significance to modulations in Ero1-Lα levels, it remained to be shown that Ero1-Lα is indeed the rate-limiting step in disulfide bond formation. Specifically, we wished to determine if either up- or downmodulation of Ero1-Lα protein within the range of its hypoxic modulation will have a phenotypic manifestation. To this end, we took a siRNA approach to partially downregulate the level of Ero1-Lα protein. Ero1-Lα siRNA-expressing clones were screened and clones displaying a modest reduction in Ero1-Lα were selected for further analysis. Representative examples are shown in Figure 6. In the experiments shown, the siRNA was introduced onto both HIF-α+/+ as well as onto HIF-1α−/− cells. The constitutive level of Ero1-Lα protein in these particular clones was reduced to about half and one-third and of its normal level, respectively (Figure 6a). This reduction in Ero1-Lα led to a significant reduction in the constitutive amount of VEGF secreted to the medium (Figure 6b). Reasoning that these results reflect a general diminution in protein secretion that collectively might affect the cell physiology, we examined the effect of reduced Ero1-Lα on cell proliferation and survival. To this end, subconfluent unsynchronized and nonstressed cultures were sorted by FACS and analyzed for their position in the cell cycle (Figure 6c). The reduction in Ero1-Lα caused a significant inhibition of cell proliferation, evidenced by a two-fold decrease in cells in the S phase and in the G2 phase. Cells withdrawn from the cell cycle were found to mostly accumulate in the sub-G1 fraction rather than in G1, indicating that cells arrested in G1 underwent apoptotic death. These results demonstrate that even a modest two- to three-fold reduction in Ero1-Lα negatively impact on cell proliferation and survival, attesting that it is indeed the rate-limiting factor in disulfide formation. We extrapolate from these findings that upregulated level of Ero1-Lα under hypoxia will proportionally increase the capability for proper protein folding in face of diminution in the ER oxidizing power due to hypoxia.

Figure 6

Modest reduction in Ero1-Lα protein by siRNA results in reduced VEGF secretion, cell cycle arrest and apoptosis. (a) Ero1-Lα protein levels in a selected clones of HIF-1α+/+ and HIF-1α−/− fibroblasts transfected with Ero1-Lα siRNA, compared to cells transfected with empty vector. Production of Ero1-Lα protein was reduced to 55 and 32%, respectively, in these particular clones. (b) ELISA measurement of VEGF secreted to the culture medium. Results shown are the average of three experiments. (c) Cell cycle analysis by FACS of the same clones of mock-transfected and siRNA-transfected cells (empty bars – control, solid bars – siRNA)

While this study focuses on regulation of Ero1-Lα by oxygen, it is conceivable that, as the rate-limiting factor in disulfide formation, it is regulated by additional cues. To address this possibility we performed an extensive in situ hybridization analysis using both embryonic and adult tissues. HIF-1α−/− embryonic stem cells undergoing organogenesis in vivo were initially examined to rule out regulation of Ero1-Lα by oxygen. Ero1-Lα was dramatically upregulated during two developmental processes: cavitation in the course of forming glandular structures (Figure 7a) and placentation (Figure 7b). Interestingly, both morphogenic processes generate structures associated with massive protein secretion. Another process in which Ero1-Lα mRNA expression is strongly upregulated is formation of the corpus luteum (Figure 7c), a process involving neoangiogenesis and forming an organ specialized in hormone secretion. These findings attest for complex modes of Ero1-Lα regulation other than hypoxia that might reflect an increased requirement for disulfide bond formation in secretory structures.

Figure 7

In situ hybridization of Ero1-Lα during selected developmental processes. (a) HIF-1α−/− teratomas grown in nude mice. Note upregulated expression of Ero1-Lα mRNA in a structure undergoing cavitation (marked by a white arrow and by a green arrow in the structure where it has been only partially completed) but not in the structure where this process has already been completed (marked by a red arrow). Left and right figures are bright and dark field images, respectively, of a thin H&E-stained section. (b) Implantation site of an early postimplantation embryo showing strong Ero1-Lα expression in trophoblast cells and ectoplacental cone. (c) Corpus luteum


Hypoxia is a major environmental stress that might negatively impact cellular homeostasis through compromising the activity of many enzymatic pathways. Realizing that molecular oxygen serves as the terminal electron acceptor for disulfide formation (Tu and Weissman, 2002), it is anticipated that proper folding of proteins destined for secretion, a process necessitating oxidizing equivalents, will also be compromised under hypoxia. Yet, the likelihood that protein maturation and secretion might be severely inhibited under hypoxia has not been appropriately appreciated. Here we used VEGF as an example to demonstrate this phenomenon. VEGF was chosen because its activity to promote formation of more blood vessels is mostly needed in circumstances of insufficient tissue oxygenation and since its production is accordingly increased under hypoxia (Shweiki et al., 1992). We showed that secretion of VEGF under hypoxia is indeed suboptimal and could be improved by artificial increase of oxidizing equivalents in the ER. Notably, there is an efficient mechanism to guarantee maximal translation of VEGF mRNA under hypoxia, thus rendering protein folding and/or secretion the ‘bottleneck’ in its bioavailability (Stein et al., 1998). Obviously, the case of VEGF is an example for a more general problem manifested by impaired secretion under hypoxia, a problem requiring some corrective mechanism. Importantly, tumor growth relies on proper secretion of many proteins, including growth factors, proteases and ECM proteins. Since hypoxia is prevalent in growing tumors, secretion of these proteins might be impeded.

Adaptive and corrective responses to hypoxia involving increased transcription are orchestrated by dedicated transcription factors, primarily by HIF-1 (Wang and Semenza, 1993; Semenza, 1999, 2000). We, therefore, examined individual components in the protein relay involved in disulfide formation with regard to hypoxia-inducibility, in general, and for being a HIF-1 target gene, specifically. Multistep pathways regulated by HIF-1 can be regulated at each step of the enzymatic cascade or, alternatively, by hypoxic induction of a critical element within the pathway (e.g. the most upstream or the rate-limiting factor). Glycolysis exemplifies the former situation, as for each step of the pathway a particular hypoxia-inducible isoform has been identified (Firth et al., 1994; Semenza et al., 1994). Findings reported here clearly show that among the oxireductases in the relay of disulfide formation that were examined (albeit not examining all PDI-like proteins) Ero1-Lα is is the only hypoxia-induced gene. Notably, Ero1-Lβ (Pagani et al., 2000) and ALR (also known as Erv2p), playing a role parallel to Ero1-Lα in ER oxidative folding in yeast (Sevier et al., 2001) were not induced by hypoxia. This is similar to the situation in glycolysis where only a single isoform of each glycolytic enzyme is hypoxia-inducible (Firth et al., 1994; Semenza et al., 1994). Ero1-Lα is indeed the natural candidate in the cascade of disulfide formation for regulation by oxygen, considering that it is in a substoichiometric concentration relative to PDI (the later constituting 2% of the protein in the ER), that each FAD-bound Ero1-Lα molecule can support multiple rounds of PDI oxidation and that free excess of free FAD cannot drive Ero1-catalyzed disulfide formation under anaerobic conditions (for a review see Tu and Weissman, 2004). The facts that PDI is a multigene family, that its different members display different tissue distributions and that Ero1 is unable to interact with some of the PDI family members, suggest that tissue-specific differences might exist with regard to hypoxic-regulation of disulfide formation, as indeed was recently demonstrated for endothelial cells (Sullivan et al., 2003).

Here we have provided clear genetic evidence that Ero1-Lα is downstream of HIF-1 in the regulatory cascade. Direct targets of HIF-1 are distinguished by possessing hypoxia regulatory elements (HREs) within their promoter or elsewhere in the gene that serve as HIF-1 binding sites (Semenza, 1999). We have identified potential HRE sites within the Ero1-Lα promoter, including a IndexTermGCCGCACGTC sequence at position –933 (P-value <10−4) but have not yet determined whether these sequence is indeed essential for the hypoxic response. Thus, we cannot distinguish between a direct and indirect mode of HIF-1 regulation. Interestingly, hypoxic induction of Ero1-Lα was p53-independent. Since hypoxia is known to be a major selection force in p53-deficient (Graeber et al., 1996), ability to upregulate Ero1-Lα in hypoxic microenvironments of these tumors is physiologically relevant.

Several HIF-1 target genes are also regulated by hypoglycemia (Stein et al., 1995; Maltepe et al., 1997). Here we showed that Ero1-Lα is also induced by hypoglycemia in a HIF-1-dependent manner and independently of its regulation by hypoxia. We believe that this finding is of physiological significance considering that ischemia is usually accompanied by both hypoxia and hypoglycemia that, depending on the circumstances, their relative severity may vary. Hence, the ability to upregulate Ero1-Lα expression independently by both stresses provides a more robust feedback response to ischemia.

To attribute any physiological advantage to the modulations in Ero1-Lα levels reported in this study, it was essential to demonstrate that it is indeed the limiting factor in protein folding and secretion. More specifically, it had to be shown that even a modest reduction in Ero1-Lα is sufficient to compromise protein secretion and, consequently, to impair vital cellular functions. RNAi approach was taken to simulate a modest shortage in protein, within the range of shortage in FAD-bound Ero1-Lα expected under hypoxia. Results clearly showed that a 45 and 68% reduction in Ero1-Lα were sufficient to impair VEGF secretion, to cause cell cycle arrest and to promote apoptosis. Thus, we conclude that that upregulation of Ero1-Lα by hypoxia and hypoglycemia (and possibly by additional factors) provides a clear benefit in the cellular ability to cop with these insults.

Co-induction of Ero1-Lα and VEGF in hypoxic tumor cells, together with findings reported here that downregulation of Ero1-Lα impedes VEGF secretion, suggest that Ero1-Lα might be harnessed for reducing the levels of this key proangiogenic factor. The fact that the levels of Ero1-Lα mRNA in hypoxic tumor cells are much higher than its levels in nontumor cells provides a rationale for selectivity. Interestingly, a recent study has identified Ero1-Lα as included in the small group of eight genes predicting poor survival of patients with pulmonary adenocarcinoma (Endoh et al., 2004).

In conclusion, this study has identified a previously unrecognized HIF-1 mediated pathway that operate to augment the oxidizing power of the ER in face of diminishing oxygen and thereby to enable proper protein folding and secretion. Considering that hypoxia-induced secretion of VEGF plays a key role in tumor angiogenesis, findings reported here also suggest Ero1-Lα as a potential antiangiogenic target.

Materials and methods


Diamide and actinomycin D were purchased from Sigma Inc. VEGF ELISA kit was from Oncogene Inc. and was used according to the manufacturer's instructions.

Tissue culture and transfections

Cell types: Wild type and HIF-1α-null transformed MEFs were kindly provided by Randall S Johnson (Department of Biology, University of California, San Diego, CA, USA) (Ryan et al., 2000). C6 cells, a clonal glial cell line derived from a rat glial tumor (Benda et al., 1968), p53−/− MEFs, the hepatocellular carcinoma line hepa-1c1c7 and its HIF-1β−/− counterpart (Carmeliet et al., 1998) were all grown in Dulbecco Modified Eagle Medium (DMEM) containing 10% fetal calf serum and antibiotics. ES cells were grown on gelatin coated plates with 20% FCS, βME, nonessential amino acids, sodium pyruvate, LIF (Esgro, Life Sciences) and antibiotics.

Transformed HIF-1α+/+ and HIF-1α−/− MEFs were co-transfected with 1 μg pSUPER plasmid alone or pSUPER-Ero1-Lα RNAi and 0.1 μg pMSCVneo (Clontech) as a selection marker. Selection continued for 10–14 days, and individual G418-resistant clones were tested for Ero1-Lα expression.


In total, 106–107 ES cells were implanted subcutaneously in CD1 nude mice.

Hypoxia, hypoglycemia

Hypoxia was achieved using anaerobic bacteriological jars (Beckton Dickinson, BBL) and AnaeroGen bags (Mitsubishi Gas Chemical Company Inc.) utilizing hydrogen and a palladium catalyst to remove all traces of oxygen. Hypoglycemia was achieved by using glucose free medium with normal serum.


The mammalian expression vector, pSUPER was used for expression of siRNA in transformed MEFs of both HIF-1α+/+ and HIF-1α−/− cells. Ero1-Lα RNAi was designed using xeragon site and constructed by cloning the sequence – Forward oligo: 5′-IndexTermGATCCCC-CCCTGCCATTCTGATGAAG-TTCAAGAGA-CTTCATCAGAATGGCAGGG-TTTTGGAAA-3; reverse oligo: 5′-IndexTermAGCTTTTCCAAAAA-CCCTGCCATTCTGATGAAG-TCTCTTGAA-CTTCATCAGAATGGCAGGG-GGG-3′ into the BglII HindIII cloning site of the pSUPER vector.

RNA blotting

Total RNA was extracted using TRI (trizol) reagent (Sigma) according to the manufacturer's directions. Total RNA (5–20 μg) was resolved by formaldehyde–agarose (1%) denaturing gels and blotted to positively charged nylon membrane by capillary elution. The RNA was UV crosslinked (1200 j/m2), and the membrane stained with 0.1% methylene blue to ensure equal loading and transfer. Blots were hybridized overnight with an α-32P-labeled probe with rediprime kit (Amersham). The blots were subjected to washes (2 × 30-min washes at 60°C with 2 × SSC, 1% SDS), after which they were exposed to MS sensitive film (Kodak) or a phosphoimage plate (Fuji).

Hybridization probes

Ero1-Lα was cloned from a λZAP cDNA library of hypoxia-treated C6 cells and subcloned into the pCDNA3.1 vector. The other probes used (all cloned into TOPO-PCR II cloning vector (Invitrogen)) were generated by PCR amplification using the following primers:


Sense: 5′-IndexTermcgcgaagcttccaccatggactttctgctctcttgggt-3′

Antisense: 5′-IndexTermcgcggatatcaccgccttggcttgtcaca-3′


Sense: 5′-IndexTermttgtcaactggctgaagaaacgca-3′

Antisense: 5′-IndexTermataaataggatcttgcccttgaa-3′


Sense: 5′-IndexTermgattgttgactacatgattgag-3′

Antisense: 5′-IndexTermtctcgtctaggatggctgcattg-3′


Sense: 5′- IndexTermcagcaacttgagagataactacc-3′

Antisense: 5′- IndexTermttatcttcttcattcactatgt-3′


Sense: 5′- at IndexTermgcggacccag cagaagcggga-3′

Antisense: 5′-IndexTermgtgttggctcagatgcactttaa-3′


Western blot analysis: cells grown in 5 cm plates were lysed in a lysis buffer containing 0.1% Triton X-100. Protein lysates were boiled prior to electrophoresis on a 10% SDS–PAGE and transfer onto a nitrocellulose membrane. The membranes were blocked with TBS-T and 5% low fat milk for 1 h at room temperature, incubated at 4°C overnight with a rabbit antiEro1-Lα antibody (1 : 6400) and subsequently with HRP-conjugated secondary antibody for 90 min at RT Immunoreactive bands were visualized using Western Blotting Luminol Reagent (Santa Cruz biotechnology, Inc.).

Production of anti-Ero1-Lα antibodies

A complete Ero1-Lα open reading frame with a 9 X His tag was cloned into bacterial vector expressed in Escherichia coli. Ero1-Lα was purified using Co-NTA in the presence of 14 mM β-mercaptoethanol in loading buffer. The purified Ero1-Lα was >95% by coomassie staining. In total, 200 μg of purified Ero1-Lα protein was injected to rabbits using a standard procedure. Anti-Ero1-Lα antibodies were purified using Ero1-Lα protein that was immobilized on Affigel-10 (BioRad).

In situ hybridization on paraffin sections was performed as described previously (Motro et al., 1990) using a Ero1-Lα-specific riboprobe.


Tunel was performed as described previously (Gavrieli et al., 1992).

FACS analysis

Cells were trypsinized, washed twice in PBS and fixed in methanol. Cells were incubated in RNaseIIa 50 μg/ml (Sigma) for 30 min. Propidium iodide was added to the cells at a concentration of 5 μg/ml. After 5 min, the cells were analyzed by fluorescence-activated cell sorting.

Bioinformatic methods

An updated Position Weight Matrix (PWM) for the HIF-1 recognition site (HRE) was constructed and used with a CIS program Barash et al., 2004 to search the human Ero1-Lα promoter.


  1. An WG, Kanekal M, Simon MC, Maltepe E, Blagosklonny MV and Neckers LM . (1998). Nature, 392, 405–408.

  2. Barash Y, Elidan G, Kaplan T and Friedman N . (2004). Bioinformatics, Sep 28 [Epub ahead of print].

  3. Benda P, Lightbody J, Sato G, Levine L and Sweet W . (1968). Science, 161, 370–371.

  4. Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans K, Dewerchin M, Neeman M, Bono F, Abramovitch R, Maxwell P, Koch CJ, Ratcliffe P, Moons L, Jain RK, Collen D, Keshert E and Keshet E . (1998). Nature, 394, 485–490.

  5. Claffey KP, Senger DR and Spiegelman BM . (1995). Biochim. Biophys. Acta., 1246, 1–9.

  6. Czyzyk-Krzeska MF . (1997). Respir. Physiol., 110, 99–111.

  7. Ellgaard L and Frickel EM . (2003). Cell Biochem. Biophys., 39, 223–247.

  8. Endoh H, Tomida S, Yatabe Y, Konishi H, Osada H, Tajima K, Kuwano H, Takahashi T and Mitsudomi T . (2004). J. Clin. Oncol., 22, 811–819.

  9. Ferrari DM and Soling HD . (1999). Biochem. J., 339 (Part 1), 1–10.

  10. Firth JD, Ebert BL, Pugh CW and Ratcliffe PJ . (1994). Proc. Natl. Acad. Sci. USA, 91, 6496–6500.

  11. Forsythe JA, Jiang BH, Iyer NV, Agani F, Leung SW, Koos RD and Semenza GL . (1996). Mol. Cell Biol., 16, 4604–4613.

  12. Frand AR and Kaiser CA . (1998). Mol. Cell, 1, 161–170.

  13. Gavrieli Y, Sherman Y and Ben-Sasson SA . (1992). J. Cell Biol., 119, 493–501.

  14. Gess B, Hofbauer KH, Wenger RH, Lohaus C, Meyer HE and Kurtz A . (2003). Eur. J. Biochem., 270, 2228–2235.

  15. Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, Lowe SW and Giaccia AJ . (1996). Nature, 379, 88–91.

  16. Halterman MW, Miller CC and Federoff HJ . (1999). J. Neurosci., 19, 6818–6824.

  17. Ikeda E, Achen MG, Breier G and Risau W . (1995). J. Biol. Chem., 270, 19761–19766.

  18. Ivan M, Kondo K, Yang H, Kim W, Valiando J, Ohh M, Salic A, Asara JM, Lane WS and Kaelin Jr WG . (2001). Science, 292, 464–468.

  19. Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, Kriegsheim A, Hebestreit HF, Mukherji M, Schofield CJ, Maxwell PH, Pugh CW and Ratcliffe PJ . (2001). Science, 292, 468–472.

  20. Kuwabara K, Matsumoto M, Ikeda J, Hori O, Ogawa S, Maeda Y, Kitagawa K, Imuta N, Kinoshita T, Stern DM, Yanagi H and Kamada T . (1996). J. Biol. Chem., 271, 5025–5032.

  21. Maltepe E, Schmidt JV, Baunoch D, Bradfield CA and Simon MC . (1997). Nature, 386, 403–407.

  22. Mazzarella RA, Srinivasan M, Haugejorden SM and Green M . (1990). J. Biol. Chem., 265, 1094–1101.

  23. Motro B, Itin A, Sachs L and Keshet E . (1990). Proc. Natl. Acad. Sci. USA, 87, 3092–3096.

  24. Muller YA, Li B, Christinger HW, Wells JA, Cunningham BC and de Vos AM . (1997). Proc. Natl. Acad. Sci. USA, 94, 7192–7197.

  25. Pagani M, Fabbri M, Benedetti C, Fassio A, Pilati S, Bulleid NJ, Cabibbo A and Sitia R . (2000). J. Biol. Chem., 275, 23685–23692.

  26. Potgens AJ, Lubsen NH, van Altena MC, Vermeulen R, Bakker A, Schoenmakers JG, Ruiter DJ and de Waal RM . (1994). J. Biol. Chem., 269, 32879–32885.

  27. Rutkowski DT and Kaufman RJ . (2004). Trends. Cell Biol., 14, 20–28.

  28. Ryan HE, Poloni M, McNulty W, Elson D, Gassmann M, Arbeit JM and Johnson RS . (2000). Cancer Res., 60, 4010–4015.

  29. Sciandra JJ, Subjeck JR and Hughes CS . (1984). Proc. Natl. Acad. Sci. USA, 81, 4843–4847.

  30. Semenza GL . (1999). Annu. Rev. Cell. Dev. Biol., 15, 551–578.

  31. Semenza GL . (2000). J. Appl. Physiol., 88, 1474–1480.

  32. Semenza GL, Roth PH, Fang HM and Wang GL . (1994). J. Biol. Chem., 269, 23757–23763.

  33. Sevier CS, Cuozzo JW, Vala A, Aslund F and Kaiser CA . (2001). Nat. Cell Biol., 3, 874–882.

  34. Shima DT, Deutsch U and D'Amore PA . (1995). FEBS Lett., 370, 203–208.

  35. Shweiki D, Itin A, Soffer D and Keshet E . (1992). Nature, 359, 843–845.

  36. Stein I, Itin A, Einat P, Skaliter R, Grossman Z and Keshet E . (1998). Mol. Cell Biol., 18, 3112–3119.

  37. Stein I, Neeman M, Shweiki D, Itin A and Keshet E . (1995). Mol. Cell Biol., 15, 5363–5368.

  38. Sullivan DC, Huminiecki L, Moore JW, Boyle JJ, Poulsom R, Creamer D, Barker J and Bicknell R . (2003). J. Biol. Chem., 278, 47079–47088.

  39. Tu BP and Weissman JS . (2002). Mol. Cell, 10, 983–994.

  40. Tu BP and Weissman JS . (2004). J. Cell Biol., 164, 341–346.

  41. Tu BP, Ho-Schleyer SC, Travers KJ and Weissman JS . (2000). Science, 290, 1571–1574.

  42. Walter DH, Hink U, Asahara T, Van Belle E, Horowitz J, Tsurumi Y, Vandlen R, Heinsohn H, Keyt B, Ferrara N, Symes JF and Isner JM . (1996). Lab. Invest., 74, 546–556.

  43. Wang GL and Semenza GL . (1993). Proc. Natl. Acad. Sci. USA, 90, 4304–4308.

  44. Wang GL, Jiang BH and Semenza GL . (1995b). Biochem. Biophys. Res. Commun., 212, 550–556.

  45. Wang GL, Jiang BH, Rue EA and Semenza GL . (1995a). Proc. Natl. Acad. Sci. USA, 92, 5510–5514.

Download references


This study was supported by a grant from the Belfer Foundation and from Quark Biotech Inc. We thank Yevgeni Kazanov and David Belenki for preparation of anti-Ero-1La antibodies.

Author information



Corresponding author

Correspondence to Eli Keshet.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

May, D., Itin, A., Gal, O. et al. Ero1-Lα plays a key role in a HIF-1-mediated pathway to improve disulfide bond formation and VEGF secretion under hypoxia: implication for cancer. Oncogene 24, 1011–1020 (2005).

Download citation


  • Ero1-Lα
  • VEGF
  • HIF1
  • hypoxia
  • disulfide bond formation

Further reading