A redox mechanism controls differential DNA binding activities of hypoxia-inducible factor (HIF) 1alpha and the HIF-like factor.

Hypoxia-inducible factor 1alpha (HIF-1alpha) and the HIF-like factor (HLF) are two highly related basic Helix-Loop-Helix/Per-Arnt-Sim (bHLH/PAS) homology transcription factors that undergo dramatically increased function at low oxygen levels. Despite strong similarities in their activation mechanisms (e.g. they both undergo rapid hypoxia-induced protein stabilization, bind identical target DNA sequences, and induce synthetic reporter genes to similar degrees), they are both essential for embryo survival via distinct functions during vascularization (HIF-1alpha) or catecholamine production (HLF). It is currently unknown how such specificity of action is achieved. We report here that DNA binding by HLF, but not by HIF-1alpha, is dependent upon reducing redox conditions. In vitro DNA binding and mammalian two-hybrid assays showed that a unique cysteine in the DNA-binding basic region of HLF is a target for the reducing activity of redox factor Ref-1. Although the N-terminal DNA-binding domain of HIF-1alpha can function in the absence of Ref-1, we found that the C-terminal region containing the transactivation domain requires Ref-1 for full activity. Our data reveal that the hypoxia-inducible factors are subject to complex redox control mechanisms that can target discrete regions of the proteins and are the first to establish a discriminating control mechanism for differential regulation of HIF-1alpha and HLF activity.

Hypoxia-inducible factor 1␣ (HIF-1␣) and the HIF-like factor (HLF) are two highly related basic Helix-Loop-Helix/Per-Arnt-Sim (bHLH/PAS) homology transcription factors that undergo dramatically increased function at low oxygen levels. Despite strong similarities in their activation mechanisms (e.g. they both undergo rapid hypoxia-induced protein stabilization, bind identical target DNA sequences, and induce synthetic reporter genes to similar degrees), they are both essential for embryo survival via distinct functions during vascularization (HIF-1␣) or catecholamine production (HLF). It is currently unknown how such specificity of action is achieved. We report here that DNA binding by HLF, but not by HIF-1␣, is dependent upon reducing redox conditions. In vitro DNA binding and mammalian two-hybrid assays showed that a unique cysteine in the DNA-binding basic region of HLF is a target for the reducing activity of redox factor Ref-1. Although the N-terminal DNA-binding domain of HIF-1␣ can function in the absence of Ref-1, we found that the C-terminal region containing the transactivation domain requires Ref-1 for full activity. Our data reveal that the hypoxia-inducible factors are subject to complex redox control mechanisms that can target discrete regions of the proteins and are the first to establish a discriminating control mechanism for differential regulation of HIF-1␣ and HLF activity.
Cells of aerobic organisms depend upon a plentiful supply of oxygen for energy production, and consequently, a number of mechanisms have evolved to sense and respond to disruptions of oxygen homeostasis. Although several signal transduction and gene regulatory mechanisms have been established to operate in response to oxidative stresses induced by intracellular oxygen radical production or cell penetration by oxidants, mammalian gene regulatory pathways that respond to hypoxia (low oxygen) have only recently begun to be understood. Analysis of hypoxia-induced activation of the erythropoietin gene led to the cloning of HIF-1␣ 1 (1), a transcription factor that exhibits marked increases in stability and transcription po-tency at low intracellular oxygen levels (for a recent review, see Ref. 2). HIF-1␣ is a member of a bHLH protein subfamily that contains PAS dimerization domains juxtaposed with the bHLH. A number of bHLH/PAS factors have now been cloned and found to have crucial roles in physiological functions as diverse as xenobiotic metabolism (the dioxin or Ah receptor), maintenance of circadian rhythms (Clock and Period), and neurogenesis (Single-minded) (for a recent review, see Ref. 3). Like the dioxin receptor, HIF-1␣ responds to environmental cues to form a functional heterodimer with a general bHLH/ PAS partner protein, Arnt (Ah receptor nuclear translocator). A closely related factor, the HIF-like factor (HLF (4); also known as EPAS1 (endothelial PAS1) (5), HIF-related factor (6), and MOP2 (member of PAS family 2) (7)), shares 48% amino acid identity with HIF-1␣ and is correspondingly induced by hypoxia to form a transcriptionally active HLF/Arnt heterodimer. In vitro mechanistic analysis of HIF-1␣ and HLF revealed that they dimerize with Arnt with similar affinities (4); bind to the same consensus hypoxia response elements found in genes such as erythropoietin (EPO) and vascular endothelial growth factor; and in transient transfection assays, activate reporter genes with similar intensities when under hypoxic stress (4,5,8). RNA and protein blot analyses showed both HLF and HIF-1␣ to be widely expressed, with higher expression of HLF in the lung and endothelial cells (4,5,8). Despite strong sequence and mechanistic similarities and prevalent coexpression of HIF-1␣ and HLF, targeted disruption of their genes in mice results in embryonic lethalities with startling differences of phenotype. HIF-1␣ null mice (9, 10) die at embryonic day 10 with defects in vascularization, whereas HLF null mice (11) die mid-gestation (days 12.5-16), suffering bradycardia and reduced catecholamine levels. These results establish that, at the very least, HIF-1␣ and HLF perform critically distinct functions during embryogenesis and indicate that they stimulate different target genes and/or undergo undiscovered differences in their in vivo mechanisms of activation.
To address the issue of how specificity of function between HIF-1␣ and HLF may be achieved, we have analyzed their DNA binding capacities under varying experimental conditions. The N-terminal DNA-binding basic regions of HIF-1␣ and HLF differ by a single amino acid, serine 28 in HIF-1␣ being replaced by cysteine 25 in HLF (see Fig. 3). Notably, single cysteines in the basic regions of Fos and Jun subunits of the AP-1 transcription factor complex render DNA binding subject to redox control. A protein with reducing activity, Ref-1 (redox factor 1), has been characterized as a factor that imparts strong DNA binding activity to the AP-1 complex by maintaining the cysteines as reduced sulfhydryl groups (12). Subsequently, a number of proteins containing cysteines in their DNA-binding domains have been found to have their DNA binding activities controlled by reducing factors such as Ref-1 (e.g. CREB, NF-B, and Myb (12); p53 (13); and Pax5 and Pax8 (14,15)) or thioredoxin (e.g. the glucocorticoid receptor (16) and NF-B (17)). We report here that Ref-1 imparts DNA binding to the cysteine-containing HLF bHLH/PAS A region, but does not influence the DNA binding activity of the serine-containing HIF-1␣ bHLH/PAS A region. Mammalian two-hybrid assays indicated that Ref-1 can interact with the HLF N-terminal sequence, but not the HIF-1␣ N terminus. As expected, expression of antisense Ref-1 RNA reduces the ability of HLF to function as a transcription factor, but surprisingly also compromises the transcription potency of HIF-1␣. Through analysis of chimeric proteins, we have obtained evidence that Ref-1 is important for HIF-1␣ activity in a region C-terminal to the bHLH domain. Thus, the activities of both HIF-1␣ and HLF are under complex modes of redox control, with some distinction between domains subject to redox regulation, providing the first instance of differential control for activation of these two proteins. These results suggest a novel mechanism whereby specificity of hypoxia-induced gene regulation may be achieved.
Purification of His-tagged Ref-  lyzed by SDS-PAGE, and protein concentration was determined by the Bradford assay (Bio-Rad).
Electrophoretic Mobility Shift Assay (EMSA)-Annealed doublestranded oligonucleotides were end-filled with 32 P-labeled dNTPs using Klenow enzyme and used as probes for gel shift assays. The W18 probe contains the hypoxia response motif from the erythropoietin enhancer (21). The XRE is a probe containing a dioxin receptor response motif from the cytochrome P4501A1 gene (18). Gel mobility shift assays were carried out in a 20-l volume with the bacterially expressed and purified protein HIF-1␣-(1-245), HLF-(1-265), or DR-(1-287) with the partner factor Arnt as specified in a two-step process. Briefly, protein samples (100 -400 ng/reaction) in the presence or absence of the reducing agent Ref-  or DTT were allowed to dimerize and interact at room temperature for 30 min. Binding buffer (final concentrations: 10 mM Tris (pH 7.5), 10% glycerol, 1 mM EDTA, 1 mM MgCl 2 , 50 mM KCl, 50 mM NaCl, 0.1 mM DTT, 50 ng/l poly(dI⅐dC), and 0.5 g/l bovine serum albumin) and the radiolabeled probe W18 or XRE (Ͼ10,000 cpm) were then added, and the reactions were transferred to 4°C and incubated for an additional 30 min. Protein-bound DNA complexes were resolved on a 4% nondenaturing polyacrylamide gel run in 25 mM Tris/glycine (pH 8.0) and 0.1 mM EDTA at 4°C.
Mammalian Two-hybrid Assay-pCMX/Ref-1/VP16 was generated by inserting a BamHI/NheI-digested polymerase chain reaction fragment of Ref-1 into a BamHI/NheI-digested pCMX/VP16 vector. COS-1 cells were seeded onto six-well plates (3 cm) and transfected using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol. Briefly, 500 ng of a Gal4-luciferase reporter (18) was cotransfected together with 10 ng of the corresponding Gal4 chimeric expression plasmid and 1 ng of the Ref-1/VP16 fusion construct. Cells were harvested 48 h after transfection, and luciferase activity was determined using the Gene Glow assay system (BioThema).

Differential DNA Binding Activity of HIF-1␣ and HLF-We
have previously noted that the PAS A region of the dioxin receptor is important for the DNA binding activity of the DR/ Arnt heterodimer (18). As an extension of these studies, we sought to analyze the influence of the PAS A domain on the DNA binding activity of two other bHLH/PAS proteins, the closely related hypoxia-inducible factors HIF-1␣ and HLF. Nterminal regions of HIF-1␣ and HLF containing the bHLH domains, the PAS A domains, and the intervening amino acids between PAS A and PAS B (termed HIF-1␣-(1-245) and HLF-(1-265) respectively) were expressed as GST fusion proteins in bacteria. A GST fusion protein of the full-length partner protein Arnt was also expressed in bacteria, and all proteins were purified to near homogeneity by glutathione-Sepharose affinity chromatography (Fig. 1). Electrophoretic mobility shift assays have previously shown that in vitro translated HIF-1␣ and HLF, when combined with Arnt, recognize the same DNA sequences from hypoxia-responsive regions in the EPO and vascular endothelial growth factor genes (4,5). As expected, when tested in isolation, bacterially expressed HIF-1␣, HLF, and Arnt had no affinity for the hypoxia response element from the EPO gene (Fig. 2, lanes 2-4). Surprisingly, however, we found that a mixture of purified HLF and Arnt showed barely detectable DNA binding activity on the EPO hypoxia response element (Fig. 2, lanes 5-7), whereas equivalent quantities of HIF-1␣ and Arnt exhibited clear DNA binding (lanes 8 -10). This striking difference in activity was unexpected as sequence analysis of HLF and HIF-1␣ revealed a strong homology between the two DNA-binding and dimerization regions, with 83% identity between the bHLH domains and just under 70% identity over the PAS domains. These results stimulated us to investigate whether other factors might invoke differential control over DNA binding activity between HIF-1␣ and HLF.
Redox Factor Ref-1 Stimulates DNA Binding of the HLF/ Arnt Heterodimer, but Does Not Influence HIF-1␣/Arnt DNA Binding-Upon closer inspection of the amino acid sequences for HIF-1␣ and HLF, it was noticed that the only difference within a 13-amino acid stretch of the DNA-binding basic regions between the two proteins was that cysteine 25 in HLF correlated with serine 28 in HIF-1␣. This single amino acid difference is conserved for both human and mouse proteins. (Despite the original published sequence of HLF describing amino acid 25 as a serine, resequencing has confirmed a cysteine at this position, consistent with EPAS1 (5), HIF-related factor (6), and MOP2 (7) sequences.) The position of this difference was intriguing, as cysteines in the DNA-binding basic regions of the oncoproteins Fos and Jun were found to mediate susceptibility for redox control of DNA binding activity (25). Strikingly, these cysteines are in a highly similar immediate environment, flanked by basic amino acids (Fig. 3). In the case of Fos and Jun, mutation of the conserved cysteines to serines provided a higher affinity DNA-binding complex that was resistant to redox control and displayed increased cell transforming ability (26). Xanthoudakis and Curran (12) isolated a protein from cell extracts (Ref-1) that was necessary for strong DNA binding activity of Fos and Jun. Ref-1 was found to be identical to the DNA repair enzyme APE/HAP-1 and to harbor two distinct activities, an N-terminal reducing function capable of interacting with cysteines and a C-terminal DNA repair function (20).
To test whether the redox factor Ref-1 could influence DNA binding of HIF-1␣ and HLF, we expressed a histidine-tagged N-terminal redox domain of Ref-1 in bacteria and purified the protein by nickel affinity chromatography (Fig. 4A). Upon addition of this purified Ref-  polypeptide to the EMSA reaction containing HLF and Arnt with the EPO hypoxia response element, a single DNA-binding entity was clearly observed (Fig. 4B, lanes 7-9), whereas none was evident without Ref-   ) supplies a reducing function critical for HLF DNA binding, the addition of dithiothreitol to the protein/DNA incubation also allowed conversion of the HLF/Arnt heterodimer to a DNA-binding form (Fig. 4B, lane 6). In total contrast, the addition of either Ref-  or dithiothreitol to the HIF-1␣/Arnt EMSA incubation did not influence the strength of DNA binding (Fig. 4C, compare lanes 6 and 7-9 with lane 5). These results ascertain that a reducing activity needs to be present to invoke DNA binding of the HLF/Arnt heterodimer, but not the HIF-1␣/Arnt heterodimer, and imply that a cysteine in the basic region of HLF may be a target of the Ref-1 reducing protein that also acts to provide strong DNA binding activity for the Fos-Jun complex. To determine whether the effect of Ref-1 was specific for HLF and was not a general effect on bHLH/PAS factors, an N-terminal region of the dioxin receptor encompassing the bHLH/PAS A domain was expressed as a GST fusion protein in bacteria and purified (Fig. 5A). When combined with bacterially expressed Arnt, a complex was formed that bound the XRE, the known target sequence of the DR/Arnt heterodimer (Fig. 5B). The intensity of this EMSA complex was not influenced by the addition of Ref- , showing that Ref-1 does not nonspecifically enhance the DNA binding activity of bHLH/PAS proteins.
Serine-to-Cysteine Point Mutation Renders DNA Binding of HIF-1␣ Susceptible to Redox Control-The similarities of redox control for DNA binding activities in the Fos/Jun heterodimer and the HLF/Arnt heterodimer suggest that the cysteine in the basic region of HLF is a likely target for redox control. However, other cysteines in the HLF region being analyzed might also be affected by the redox conditions of the experiment. We therefore sought to obtain direct evidence that the basic region Cys 25 is the target for redox control of HLF DNA binding. To this end, we created a point mutant in the HIF-1␣-(1-245) N-terminal sequence that resulted in a switch from Ser 28 to Cys 28 . This point mutant polypeptide was expressed as a GST fusion protein in bacteria and purified to provide HIF-1␣-(1-245) S28C (Fig. 6A). When incubated with the radiolabeled probe from the EPO hypoxia response element, neither HIF-1␣-(1-245) nor HIF-1␣-(1-245) S28C alone showed any DNA binding activity (Fig. 6B, lanes 3 and 4). When combined with bacterially expressed and purified Arnt, HIF-1␣-(1-245), but not HIF-1␣-(1-245) S28C, exhibited DNA binding in the absence of reducing agents (Fig. 6B, compare lanes 6 and 9). 190) did not influence the DNA binding activity of the HIF-1␣-(1-245) complex, both reducing agents were able to invoke strong DNA binding of the HIF-1␣-(1-245) S28C complex (Fig.  6B, compare lanes 6 -8 with lanes 9 -12). The ability to convert the redox-resistant DNA binding activity of HIF-1␣ into the redox-sensitive DNA binding activity seen analogous to that of HLF establishes that placing a cysteine in the DNA-binding basic region of these bHLH/PAS proteins renders them susceptible to redox control. Moreover, a dose dependence for the influence of Ref-1 on the DNA binding activity of HIF-1␣-(1-245) S28C was observed (Fig. 6C, lanes 7-12), consistent with Ref-1 reducing HIF-1␣-(1-245) S28C through protein/protein interaction.  (Fig. 7). As the N-terminal regions of HLF and HIF-1␣ lack transactivation domains (27), the increased reporter gene activity seen for these constructs may be due to the recruitment of cellular Arnt, which is known to harbor a strong transactivation domain, via the HLH regions of the chimeras. ported to be expressed in HeLa cells during normoxia (8). The decreased reporter activities seen in the presence of antisense Ref-1 were not due to nonspecific detrimental influences on transcription, as a control Rous sarcoma virus-luciferase reporter gene was unaffected by expression of antisense Ref-1 (Fig. 8B).

Ref-1 Interacts with the bHLH N Terminus of HLF, but Not with That of HIF
Ref   7-12, respectively)). DNA binding activity was measured using gel mobility shift assays with the 32 P-labeled W18 probe and identical buffer conditions and reaction volumes for each incubation. Lanes 1 in B and C contain probe alone; and lanes 2-5 in B and lanes 1-4 in C show a lack of DNA binding for the indicated proteins in isolation.
The arrows indicate the positions of HRE retarded complexes, and the asterisks denote free probe. 31). Cysteines that could potentially be targets for Ref-1 activity appear in this complex HIF-1␣ C-terminal transactivation region. To gain evidence for a function of Ref-1 outside the DNA-binding bHLH region of HIF-1␣, we constructed a chimeric protein in which the bHLH domain of HIF-1␣ was replaced by the corresponding domain of the dioxin receptor, forming DR bHLH/HIF-1␣ (Fig. 9). EMSAs of the purified dioxin receptor bHLH/PAS A region and Arnt showed that Ref-1 activity was not needed for DNA binding of the DR/Arnt heterodimer (Fig. 5B). When the DR bHLH/HIF-1␣ fusion protein was coexpressed with Arnt in the presence of an XREluciferase reporter gene, the activity in HeLa Tet-on cells was 3-fold above that of the reporter gene alone (Fig. 9). In the presence of antisense Ref-1, however, reporter gene activity was reduced to basal levels seen upon transfection of the reporter gene alone. In contrast, antisense Ref-1 had no effect on the basal levels of the XRE reporter gene (Fig. 9). These results imply that Ref-1 is important for some activity of HIF-1␣ outside the N-terminal bHLH dimerization and DNA-binding domains and are consistent with the previous experiments showing that Ref-1 is important for the transcriptional activity of the full-length HIF-1␣ protein (Fig. 8A), but not for DNA binding of the HIF-1␣ bHLH/PAS A domain (Fig. 4C).

Hypoxia-induced Gene Expression Is Redox-sensitive-A
number of recent reports have established that the hypoxic induction of HIF-1␣ activity is sensitive to the redox state of the cell. For example, pretreatment of cells with H 2 O 2 inhibits the ability of hypoxia to stabilize the HIF-1␣ protein, thereby abrogating the hypoxic induction of EPO gene expression in Hep3B cells (32). In vitro H 2 O 2 treatment of extracts from hypoxic cells, however, did not decrease the ability of HIF-1␣/ Arnt heterodimers to bind the hypoxia response element from the EPO gene (32), in keeping with our present data describing a lack of redox influence on recombinant HIF-1␣-(1-245) DNA binding activity. Treatment of cells with CO or NO also inhibits hypoxia-induced EPO expression and the activity of hypoxiainducible reporter genes, although reports conflict as to the stability of HIF-1␣ under these conditions (33)(34)(35). The inhibitors CO and NO block hypoxic activation of the HIF-1␣ Cterminal transactivation region (35), possibly by generation of reactive oxygen species, which would be consistent with our data and the recent report by Ema et al. (60) indicating that the reducing factor Ref-1 is needed for the activity of the HIF-1␣ C terminus. In further agreement with our data, hypoxic activation of reporter genes was enhanced by cotransfection of expression vectors for Ref-1 or thioredoxin, indicating that these reducing factors may be limiting in the cell (32,60). Finally, treatment of normoxic cells with reducing agents has been shown to induce DNA binding on hypoxia response elements and activation of hypoxia-responsive reporter genes, emphasizing the importance of a reducing environment for the activity of HIF-1␣ and/or HLF (36).
The previous reports detailing the influence of oxidants on hypoxia-induced activities have interpreted their data only in terms of the effects on HIF-1␣. Whereas it is apparent that strong oxidizing conditions will prevent HIF-1␣ activity, our results suggest that subtle changes in redox conditions may play a distinct role in determining the relative activities of HIF-1␣ and HLF. This possibility is due to the cysteine in the DNA-binding domain of HLF being surrounded by basic residues, an environment that has been found to lower the pK a of cysteine thiols and to render them extremely susceptible to oxidation (37,38). During weak or transient increases in redox potential, the HLF protein may therefore be prevented from forming a DNA-binding entity, whereas HIF-1␣ would not suffer any down-regulatory effects. Notably, transient low level release of reactive oxygen species such as H 2 O 2 , O 2 Ϫ , and ⅐OH (39,40) or milder signaling oxidants such as NO (41) have been shown to oxidize cysteine sulfhydryl groups to disulfides and sulfenic acids, reversible modifications that have been proposed to be employed during redox-initiated signal transduction processes (39).
Redox Control of Eukaryotic Transcription Factors-Although both cytoplasmic and nuclear compartments of the cell are generally considered reducing environments, there is mounting evidence to suggest that oxidized forms of some proteins can exist in vivo and are subject to activation by redox signaling pathways. For example, E2A transcription factors have been found to form an intermolecular disulfide bond in B-lymphocyte cells, stabilizing homodimers to affect DNA binding. In other cell types such as muscle, reducing activities prevent E2A homodimerization and promote heterodimeric complexes with MyoD or Id (42). The Pax8 paired domain protein, a transcription factor important for thyroid development that also contains critical cysteines that require Ref-1 for full DNA binding activity, has been reported to exist predominantly in an oxidized intracellular form when Ref-1 levels are low, but in reduced form when Ref-1 levels are high (14). The potential in vivo significance of Ref-1 control of Pax8 DNA binding activity is strengthened by the observation that thyrotropin stimulates mRNA and protein expression for both Pax8 and Ref-1 (14,43). It is particularly notable that hypoxia has been reported to increase Ref-1 mRNA and protein levels (44), consistent with a role for Ref-1 in the hypoxic response. Taken together, these data suggest that during normal cellular metabolism, some eukaryotic transcription factors harbor cysteines that can readily exchange between oxidized and reduced states, a reversible switching mechanism demonstrated to be highly successful in controlling the activity of the prokaryotic transcription factor OxyR (45) and, more recently, the stressactivated chaperone hsp33 (46).
To date, eukaryotic transcription factors have been predominantly found to require reduced forms for full activity. These  (48), and the solution structure of the interaction has been solved (49). As thioredoxin is cytoplasmic and Ref-1 is nuclear in most cell types, a redox cascade is now thought to exist where Trx moves to the nucleus in response to cellular stress to activate Ref-1 and to enhance the activity of substrate transcription factors (48). Indeed, hypoxia treatment has recently been shown to translocate Trx from the cytoplasm to the nucleus (60).
We have presented data here to show that Ref-1 is important for the activities of the HIF-1␣ C terminus (Fig. 9) and the HLF DNA-binding domain. The close similarity between the HLF and HIF-1␣ C-terminal transactivation domains implies that Ref-1 might also affect the activity of the HLF C-terminal region, which in fact has very recently been observed during analysis of HLF chimeric proteins (60). A potential Ref-1 target cysteine has been suggested to exist in the C termini of both HIF-1␣ and HLF (60). Thus, a complex and multifaceted mechanism of redox control over hypoxia-inducible gene expression exists, with reducing activities being important for both DNA binding and transactivation of HLF and for transactivation, but not DNA binding, of HIF-1␣ (Fig. 10). Multiple mechanisms of redox control over function of transcription factors are not without precedent, as recent analyses of the glucocorticoid receptor have shown that activities such as ligand binding (51), DNA binding (16), and nuclear translocation (52) are all subject to redox control. Redox control of the nuclear localization of the GR involves a cysteine in the nuclear localization signal that is preceded by two basic residues, again consistent with the notion that such a cysteine would be particularly susceptible to oxidation. Indeed, mild oxidizing conditions were enough to inhibit nuclear translocation of the GR, whereas mutation of the critical cysteine to a serine abrogated this inhibition (52). Moreover, nuclear translocation of the GR showed varied sensitivities to H 2 O 2 in different cell types, with GR activity on reporter genes being affected by physiological concentrations of H 2 O 2 in mammary tumor cells (53), indicating that GR function can be influenced by endogenous redox factors and varying natural redox states according to cell type. Whereas low concentrations of H 2 O 2 (0.5-2 mM) can disrupt nuclear localization of the GR, considerably higher concentrations (20 -100 mM) are required to disrupt ligand binding in vitro (54), implying that separate receptor activities are subject to differential redox control. As HIF-1␣ has recently been shown to undergo hypoxia-induced nuclear translocation (55), it will be interesting to investigate whether this process is influenced by redox conditions.
Our data showing redox dependence of DNA binding for HLF, but not for HIF-1␣, represent the first differential control mechanism for these factors. It is proposed that HLF and HIF-1␣ have distinct target genes such as the HLF-specific Tie2 receptor tyrosine kinase (5) and the vascular endothelial growth factor receptor Flk-1 (61) as well as differential activity on common target genes, e.g. vascular endothelial growth factor (8). To understand the disparate and essential functions of HLF and HIF-1␣ in the developing embryo, it will be of major importance to fully decipher differential activation and gene regulatory mechanisms between these two factors. The realization that a redox mechanism can invoke specificity for DNA binding activity between HLF and HIF-1␣ is an initial step toward this goal. A reversible redox-dependent switch for DNA binding may also operate in paired domain proteins, which contain two DNA-binding regions (the PAI and RED domains) that recognize different DNA sequences and that can function either independently or in concert. DNA binding of the PAI domain is abrogated by oxidation, whereas the RED domain remains unaffected, creating a situation in which a change in redox conditions may switch promoter binding affinity or specificity (15). Although post-translational control of protein activity is commonly understood in terms of phosphorylation status, it is becoming increasingly apparent that other less characterized post-translational modifications such as those affected by redox factors (56,57), acetylation (58), or immunophilins with prolyl isomerase activity (59) can play important regulatory roles in the critical processes of cell signaling and transcriptional control.