The Redox Protein Thioredoxin-1 (Trx-1) Increases Hypoxia-inducible Factor 1α Protein Expression: Trx-1 Overexpression Results in Increased Vascular Endothelial Growth Factor Production and Enhanced Tumor Angiogenesis¹

Disruption of oxygen homeostasis is a major aspect of disease pathophysiology, particularly in heart disease, stroke, cerebrovascular disease, tumor angiogenesis, and metastases (reviewed in Ref. 1). The transcription factor HIF-1³ is one of the key regulators of oxygen homeostasis (2). HIF-1 is a heterodimer of an α and β subunit, both of which belong to the basic-helix-loop-helix PER-ARNT-SIM family of transcription factors (3). HIF-1β, also known as ARNT-1, is a constitutive nuclear protein involved in a range of transcription systems, whereas the major pathway regulating gene induction in response to hypoxia is under the control of the HIF-1α and HIF-2α subunits (reviewed in Refs. 4, 5). HIF-1α expression is increased in a number of human cancers (6, 7, 8, 9, 10, 11, 12, 13).

In cells replete with oxygen, HIF-1α and HIF-2α subunits are rapidly degraded by the ubiquitin-proteasome pathway (3, 14). Destruction is mediated by a ubiquitin E3 ligase complex in which the pVHL, in association with elongin B and elongin C, binds oxygen-dependent destruction domain(s) in the HIF-1α subunits (15). The interaction between human pVHL and HIF-1α is regulated through hydroxylation of HIF-α oxygen-dependent destruction domain proline residues (P402 and P564) by one or more members of the prolyl-4-hydroxylase family (reviewed in Ref. 16). Because such enzymes require molecular oxygen as a cosubstrate, it has been suggested that these enzymes provide a link between HIF regulation and the availability of molecular oxygen. However, pVHL is unlikely to be the only ubiquitin protein ligase that regulates the half-life of HIF-1α. The MDM2 ubiquitin protein ligase is recruited to HIF-1α by the binding of the tumor suppressor p53, which may also result in a decrease in HIF-1α levels (17, 18). In addition, activation of the phosphatidylinositol 3′-kinase pathway has been shown to increase HIF-1α protein levels under nonhypoxic conditions (19, 20).

During hypoxia, HIF degradation is suppressed and HIF-1 and HIF-2α protein levels increase and dimerize with HIF-1β subunits whose expression is not changed by hypoxia (21). The HIF complex interacts with coactivators such as p300/CBP (mediated by hydroxylation of asparagine 851 of HIF-1α; Ref. 22), SRC-1, and TIF2 (23, 24, 25, 26, 27) and binds to specific enhancer elements called HREs to activate transcription of a range of genes (28). The number of known HIF-1 target genes continues to increase and includes genes such as the VEGFs, iNOS, endothelin-1, and others that attract new vasculature and increase oxygenation (29). This may be the reason that hypoxic tumors are the most proangiogenic (30) and exhibit an aggressive tumor phenotype (31, 32).

Trx-1 is a small (104 amino acids) redox protein that undergoes reversible NADPH-dependent reduction by selenocysteine containing flavoprotein Trx-1 reductases (for reviews see Refs. 33, 34). Trx-1, through its redox activity, regulates the activity of enzymes such as apoptosis signal-regulating kinase 1 (35) and protein kinases C α, δ, ε, and ζ (36) and increases the DNA binding and transactivating activity of transcription factors, including nuclear factor κB (37), the glucocorticoid receptor (38), and p53 (39). Trx-1 is a potent cell growth and survival factor. Mouse WEHI7.2 lymphoma cells transfected with human Trx-1 form tumors in immunodeficient scid mice that grow more rapidly and show less spontaneous and drug-induced apoptosis than vector-alone-transfected cells. A redox inactive mutant Trx-1 acts as a dominant negative to inhibit MCF-7 human breast cancer xenograft tumor growth in scid mice (40, 41). Trx-1 expression is increased in several human primary cancers, including lung, colon, cervix, liver, pancreatic, colorectal, and squamous cell cancer (42, 43, 44, 45, 46), and has been linked to aggressive tumor growth and inhibited apoptosis (45, 47). More recently, increased Trx-1 levels have been correlated with decreased patient survival in non-small cell lung cancer (48).

To further understand the role of Trx-1 in promoting tumor growth, we have studied the effects of increased Trx-1 expression and of a redox inactive mutant Trx-1 on HIF-1α levels and the activation of HIF downstream genes. We have shown that increased Trx-1 expression is associated with increased HIF-1α levels and HIF transactivation in cancer cells together with an increase in VEGF production and enhanced tumor angiogenesis.

MCF-7 human breast cancer, HT-29 colon cancer, and WEHI7.2 mouse lymphoma cells were obtained from the American Tissue Type Collection (Manassas, VA). The stably transfected cell lines used were MCF-7 human breast cancer cells transfected with Trx-1 (MCF-7/Trx; clone 9), with the redox inactive Cys³²→Ser/Cys³⁵→Ser mutant Trx-1 (MCF-7/C32S/C35S; clone 4) or with empty vector (MCF-7/neo; Ref. 40). The corresponding HT-29 human colon cancer cells were HT-29/Trx (clone 7), HT-29/C32S/C35S (clone 7), and HT-29/neo cells and WEHI7.2 mouse lymphoma cells WEHI7.2/Trx (clone 5), WEHI7.2/C32S/C35S (clone 5), and WEHI7.2/neo cells (47). Cells were grown under humidified 95% air, 5% CO₂ in an incubator at 37°C in DMEM supplemented with 10% fetal bovine serum and 200 μg/ml G418 where appropriate. For exposure to hypoxia, the culture flasks were incubated at 37°C for various times with a humidified gas mixture containing 5% CO₂/95% N₂ and air. A Pro:Ox 110 oxygen sensor (Reming Bioinstruments Co., Redfield, NY) was used to regulate the oxygen level in the gas phase to 1% oxygen. After treatment, cells were washed twice with PBS (pH 7.5) at 4°C and stored as cell pellets at −80°C for further analysis. One ml of media from each flask was removed after treatment and stored at −80°C for measurement of VEGF levels. Recombinant human Trx-1 (40) was stored at −80°C in 10 mm DTT, which was removed immediately before use as described previously (40).

Approximately 10⁷ cells were lysed at 4°C for 1 h in 200 μl of lysis buffer [150 mm NaCl, 50 mm Tris buffer (pH 7.5), 1 mm phenylmethylsulfonyl fluoride, 2 μg/ml aprotinin, 2 μg/ml leupeptin, 0.1 mm sodium orthovanadate, 1% NP40, and 0.2% SDS]. The lysate was centrifuged (15 min, 4°C, 12,000 × g) and the supernatant was collected. A 20-μl aliquot was removed and stored at −80°C before the protein content was determined using the Bio-Rad assay (Bio-Rad, Hercules, CA). The amount of human VEGF in cell lysates and VEGF secreted into the medium was determined using an ELISA kit that measures VEGF₁₆₅ and VEGF₁₂₁ isoforms (Human VEGF-ELISA; R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. Mouse VEGFs in the cell lysate and medium of mouse WEHI7.2 cells were determined using a mouse VEGF-ELISA kit (R&D Systems) that also measures VEGF₁₆₅ and VEGF₁₂₁ isoforms. Cell lysate VEGF was expressed as pg VEGF protein/mg of total cell protein and VEGF in the medium corrected to pg VEGF protein/mg of total cell protein from the same flask.

Total RNA was isolated from cells using Trizol (Invitrogen Life Technologies, Inc., Carlsbad, CA) following standard protocols, and murine leukemia virus reverse transcriptase (Roche Molecular Biochemicals, Indianapolis, IN) was used to generate cDNA. The cDNA was then amplified by the PCR using Platinum Taq DNA polymerase (Invitrogen Life Technologies, Inc.). The PCR primers for the VEGF isoforms have been previously published (49) and detect all of the alternatively spliced VEGF RNA transcripts. PCR-amplified products were electrophoresed in 1% agarose gel and stained with ethidium bromide.

Nuclear and cytoplasmic extracts were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Pierce, Rockford, IL) according to the manufacturer’s instructions. Total protein extracts were prepared as described for the VEGF ELISA. The protein concentration of each sample was measured using the Bio-Rad Protein assay (Bio-Rad), and Western blotting was performed as described previously (43). Blots were probed overnight at 4°C with a 1:250 dilution of mouse antihuman HIF-1α (Transduction Labs, Lexington, KY), a 1:200 dilution of mouse antihuman HIF-1β (Santa Cruz Biotechnology, Santa Cruz, CA), a 1:100 dilution of mouse antihuman iNOS (Transduction Labs), a 1:1000 dilution of goat antihuman actin (Santa Cruz Biotechnology), or a 1:1000 dilution of goat antihuman lamin A (Santa Cruz Biotechnology). Antimouse or antigoat horseradish peroxidase-conjugated secondary antibodies (Amersham Pharmacia, Uppsala, Sweden) were used at a dilution of 1:5000 for detection by chemiluminescence, and blots were quantified using ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

Northern blotting was carried out as described previously (42) using 15 μg of total RNA. Full-length cDNA probes for HIF-1α and 18S rRNA were used. Blots were imaged using the MD Storm 860 phosphorimager and quantified with ImageQuant software (Molecular Dynamics, Sunnyvale, CA).

The pGL3 firefly luciferase reporter plasmid containing the HRE from phosphoglycerate kinase (50) was supplied by Professor Ian Stratford (University of Manchester, United Kingdom). Plasmid DNA was prepared using a commercial kit (Qiagen, Valencia, CA). The empty pGL3 control plasmid and the pRL-CMV Renilla luciferase containing plasmid used to control for transfection efficiency were obtained from Promega (Madison, WI). Cells were transfected with 5 μg of HIF-1α reporter plasmid or pGL3 control plasmid and 0.025 μg of pRL-CMV Renilla luciferase plasmid using LipoTAXI mammalian transfection reagent (Stratagene, TX). Forty-eight h later, cells were exposed to hypoxia as described previously. Firefly and Renilla luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI) according to the manufacturer’s instructions.

WEHI7.2 cells (10⁷) or WEHI7.2/Trx cells were injected s.c. into the flanks of scid mice, and tumors were allowed to reach ∼0.5 g before being removed, formalin fixed, and embedded in paraffin. Sections were stained with antibodies to factor VIII-related antigen (Dako-Patts, Santa Barbara, CA) or VEGF (Santa Cruz Biotechnology) using an automated immunostainer (GenII; Ventana Medical Systems, Tucson, AZ). Detection of bound antibody was assessed through the use of indirect avidin-biotin-peroxidase methodology with 3,3′-diaminobenzidine as the color substrate (Dako-Patts). A Ventana Medical Systems antibody diluent was used as a negative control and human placental tissue as a positive control. Nuclei were counterstained with hematoxylin. Microvessel density was assessed by light microscopy using the criteria of Weidner et al. (51). Vessels were identified based on the combination of positive endothelial staining for factor VIII-related antigen and morphology. Branched structures were counted as one vessel. When an area was identified, individual microvessels were counted at ×200 (each field represents an area of 0.74 mm²). Three to five areas were counted/sample, and the mean number of vessels/unit area was determined. All samples were coded to eliminate potential observer bias in counting.

MCF-7/neo cells exposed to air had very low levels of HIF-1α protein measured by Western hybridization but when exposed to hypoxia (1% oxygen) for 16 h HIF-1α protein increased (Fig. 1). In MCF-7 cells transfected with Trx-1, HIF-1α protein was increased 2.3-fold in normoxia and 2.1-fold in hypoxia compared with empty vector control cells (P = <0.001 in both cases; Fig. 1). In MCF-7 cells transfected with the redox inactive C32S/C35S Trx-1, HIF-1α protein was decreased 0.5-fold in normoxia and 0.4-fold in hypoxia (P = <0.001 in both cases; Fig. 1). There was no change in HIF-1β protein levels caused by hypoxia, Trx-1, or C32S/C35S transfection (Fig. 1). Similar effects on HIF-1α protein levels by Trx-1 and C32S/C35S transfection were seen in human HT-29 colon cancer and mouse WEHI7.2 lymphoma cells (results not shown).

MCF-7/Trx cells showed a significantly increased activation of the HRE compared with MCF-7/neo cells with a 24-fold hypoxia-induced increase seen in Trx-1 transfects compared with a 7.5-fold hypoxia-induced increase in MCF-7/neo cells (P = <0.001). MCF-7/C32S/C35S cells showed a 0.7-fold decrease in the hypoxia-dependent response compared with MCF-7/neo cells, although it was still increased compared with normoxic controls (Fig. 2). Similar effects of Trx-1 and C32S/C35S on HRE activation were observed in cells grown in air. The ratios of luciferase activity under hypoxia:normoxia were not significantly different between MCF-7/neo, MCF-7/C32S/C35S, and MCF-7/Trx cells with values of 9.4 ± 3.7, 12 ± 5 and 10 ± 4.5, respectively (P = <0.05 in all cases).

Transfection of MCF-7 cells by Trx-1 or redox inactive Trx-1 did not significantly affect levels of HIF-1α mRNA measured by Northern blotting (Fig. 3).

Exposure of MCF-7/neo cells to hypoxia gave a time-dependent increase in the formation of VEGF protein levels, achieving a plateau value after 16 h of exposure (Fig. 4). Approximately half of the VEGF was in the cell lysate and half in the culture medium. Similar effects of hypoxia were observed on VEGF formation by HT-29 and WEHI7.2 cells (results not shown).

Trx-1 transfection significantly increased levels of VEGF in cell lysates under hypoxic conditions and VEGF secreted into the medium under both normoxic and hypoxic conditions in MCF-7, HT-29 and WEHI7.2 cells (Fig. 5). Trx-1 transfection increased total VEGF formation (VEGF in cell lysate plus that secreted into the medium) by MCF-7 breast cancer cells in air by 77% and in hypoxia by 54%; in HT-29 cells in air by 60% and in hypoxia by 46%; and in WEHI7.2 cells in air by 41% and in hypoxia by 79% (Fig. 5). Transfection with the redox inactive C32S/C35S decreased total VEGF formation under normoxic conditions by 7% in MCF-7 cells, 34% in HT-29 cells, and 27% in WEHI7.2 cells and under hypoxic conditions by 42% in MCF-7 cells, 36% in HT-29 cells, and 28% in WEHI7.2 cells compared with neo controls (Fig. 5).

Recombinant human Trx-1 added to the medium of MCF-7/neo cells gave small but significant increases in cell lysate and medium VEGF in response to hypoxia. The effect was maximal at 0.5 μm Trx-1 (Fig. 6). No effect on VEGF formation was seen in the absence of hypoxia.

VEGF isoform mRNA expression was studied in MCF-7 breast cancer cells by semiquantitative RT-PCR (Table 1). In all MCF-7 cell lines, VEGF₁₂₁ and VEGF₁₆₅ were the predominant isoforms under both normoxic and hypoxic conditions with VEGF₁₈₉ only weakly expressed. There was no change in the pattern of VEGF isoform expression in MCF-7/neo, MCF-7/Trx, or MCF-7/C32S/C35S cells under normoxic or hypoxic conditions.

Trx-1 transfection increased levels of iNOS protein after 16 h of hypoxia compared with MCF-7/neo cells (Fig. 7). Levels of iNOS were barely detectable under normoxia.

WEHI7.2/neo and WEHI7.2/Trx cells were grown as tumors in scid mice. VEGF levels were almost undetectable in the WEHI7.2/neo tumors but were markedly increased in theWEHI7.2/Trx tumors (Fig. 8). There was a 2.5-fold increase in tumor angiogenesis measured by microvessel vascular density in the WEHI7.2/Trx compared with the WEHI7.2/neo tumors.

The transcriptional activator complex HIF-1 is a key regulator of oxygen homeostasis and mediates many cellular and physiological responses necessary to adapt to changes in oxygen tension (reviewed in Ref. 2). HIF-1 has been implicated in many important aspects of tumor behavior, including increased angiogenesis, metastasis, and inhibited apoptosis (reviewed in Refs. 1, 3).

When cultured in air, cells have low levels of HIF-1α protein that markedly increase under hypoxic conditions. We observed that human MCF-7 breast cancer, human HT-29 colon cancer, and mouse WEHI7.2 lymphoma cell lines stably transfected with Trx-1 have increased levels of HIF-1α protein compared with wild-type or empty vector-transfected cells under both normoxic and hypoxic conditions. Stable transfection of the cells with a redox inactive mutant C32S/C35S Trx-1 markedly decreased HIF-1α protein levels under both normoxic and hypoxic conditions. C32S/C35S Trx-1 has previously been shown to inhibit the effects of wild-type Trx-1 by acting in a dominant negative manner (40) through competitive inhibition of Trx reductase (52). There was no change in HIF-1β/ARNT protein by hypoxia, as has been previously reported (21), or by Trx-1 or C32S/C35S transfection. The changes in HIF-1α protein were mirrored by changes in HIF transactivating activity measured using luciferase under the control of a HRE. Thus, Trx-1 enhances HIF-1α protein levels and HIF-1 activity in several cancer cell lines. Interestingly, we have previously shown that Trx-1 mRNA levels are increased 14-fold by exposure to hypoxia for 16 h (43). The mechanism by which Trx-1 increases HIF-1α protein levels remains to be established.

The increase in HIF-1 transactivation caused by Trx-1 (3-fold) exceeds the increase in HIF-1α protein (2-fold), suggesting that mechanisms in addition to an increase in HIF-1α protein contribute to the effects of Trx-1. HIF-1 transactivating activity has been reported to be increased by Ref-1, a dual function DNA repair endonuclease and redox regulatory protein (53). This occurs through a redox-dependent increase by Ref-1 of the interaction between the transcription factor coactivators SRC-1/p160 and CREB-binding protein CBP/p300 and the COOH-terminal transactivation domain of HIF-1α (25, 26, 27). Trx-1 can directly reduce Ref-1 (54) and promotes the binding of CBP/p300 and HIF-1α (27). This could contribute toward increased HIF-1 transactivating activity by Trx-1. The transactivating activity of p53 has also been reported to be regulated by Trx-1 (39). Whether the binding of p53 to HIF-1α and subsequent proteasomal degradation through MDM2 could also be regulated by Trx-1 is not known. However, because HT-29 cells have a mutant form of p53, it seems unlikely that this could explain the results in this study. It is unlikely that the Trx-1 level is changing the overall redox state of the cell because there is no change in the ratio of reduced glutathione:oxidized glutathione or NADPH:NADP in Trx-1-transfected cells (unpublished data).

The HIF-1 complex has effects on the expression of many genes (reviewed in Ref. 1). One of the genes regulated by HIF, which is thought to be of critical importance to the progression of cancer, is VEGF (55). VEGF is the major angiogenic factor leading to the development of new blood vessels from preexisting capillary beds in solid tumors and their metastases and is an autocrine factor in hematological tumors (56, 57). Inhibition of VEGF activity by functional-blocking antibodies, expression of antisense VEGF mRNA, or disruption of VEGF receptor signaling (58, 59) reduces neoangiogenesis and tumor growth. We observed that transfection of cells with human Trx-1 increased the overall production of VEGF in MCF-7 breast cancer, HT-29 colon cancer, and WEHI7.2 lymphoma cells. The effect of exogenous recombinant Trx-1 was less than that of transfected Trx-1 and is presumably mediated by recombinant Trx-1 entering the cells because there is no known receptor for Trx-1 on the surface of cells (40). We found that both cellular and secreted levels of VEGF were increased by Trx-1. Transfection of cells with redox inactive mutant Trx-1 (C32S/C35S) inhibited the increase of VEGF. Activation of VEGF is mediated by a 47-bp HRE located 985–939 bp 5′ to the VEGF transcription initiation site (60). A HIF-1 binding site has also been demonstrated in this region (61). Therefore, the changes in VEGF protein levels in the presence of the redox-active or -inactive Trx-1 presumably can be accounted for by the changes in the transactivating ability of HIF-1. However, HIF-independent mechanisms may also play a role in controlling VEGF production and could account for the changes observed under normoxic conditions (62).

We observed that Trx-1 transfected WEHI7.2 cells grown as tumors in mice have increased VEGF levels and angiogenesis measured by microvessel density. Increased Trx-1 expression can transform cells under certain conditions (63) and stimulates the growth of tumor cells in culture (40, 64) and as xenografts in animals (47). Clinically increased Trx-1 expression is correlated with aggressive tumor growth of primary gastric cancer (45) and decreased survival of patients with non-small cell lung cancers (48). An increase in VEGF production by Trx-1 could therefore contribute to the aggressive phenotype of Trx-1 overexpressing tumors.

In summary, we have shown that Trx-1 increases levels of HIF-1α protein under both normoxic and hypoxic conditions. This is associated with an increase in hypoxia-induced HIF-1 transactivation and VEGF formation with increased tumor angiogenesis in vivo. This may provide a mechanism to explain the aggressive tumor growth observed in tumors overexpressing Trx-1.