Hypoxia Stimulates Urokinase Receptor Expression Through a Heme Protein-Dependent Pathway

HYPOXIC STRESS UNDERLIES a number of biologically-important processes in which cellular migration and invasion occur. For example, hypoxia within an expanding tumor leads to the release of vascular endothelial growth factor (VEGF) and stimulation of angiogenesis,1 the success of which depends on endothelial cell migration and invasion. Hypoxia may also play an important role in promoting tumor metastasis and invasion of the extracellular matrix.2-5 It is likely that specific responses to hypoxia also promote cellular migration during processes such as atherosclerotic plaque development, as suggested by the expression of hypoxia-inducible proteins by plaque macrophages,6 as well as during early human pregnancy, where blastocyst implantation and uterine invasion by trophoblast cells occur under conditions of relative hypoxia.7 8 

The expression of pericellular plasminogen activator activity may facilitate the cellular migration and invasion that occur in the above settings.9 Generation of such activity is dependent on the binding of urokinase-type plasminogen activator (uPA) to the urokinase receptor (uPAR), a 55-kD glycoprotein anchored to the plasma membrane by a glycosyl-phosphatidylinositol moiety.10 The urokinase receptor promotes cellular migration through several mechanisms, one of which involves its ability to amplify and focus the expression of plasminogen activator activity to discrete sites on the cell surface.11,12 In addition, the uPAR mediates cellular adhesion to vitronectin,13promotes integrin-dependent migration,14 and initiates intracellular signaling events.15 

In light of the above considerations, we have investigated whether hypoxia plays a role in regulating the expression of uPAR by invasive first trimester human trophoblasts and human umbilical vein endothelial cells (HUVEC). Our studies suggest that hypoxia stimulates uPAR expression and cellular invasion through a reconstituted extracellular matrix and that these effects are mediated through an oxygen-binding heme protein.

Tissue culture medium was obtained from GIBCO-BRL (Grand Island, NY). Fetal bovine serum (FBS) was from Hyclone (Logan, UT) or from GIBCO-BRL. Cobalt chloride and 4,5-dihydroxybenzene-1,3-disulfonate (Tiron) were purchased from Sigma Chemical Co (St Louis, MO). Control mouse IgG_2a and fluorescein isothiocyanate (FITC)-conjugated secondary antibodies were obtained from Dako Corp (Carpinteria, CA). The fluorogenic plasmin peptide substrate, HDVALL-AMC, was purchased from Bachem (King of Prussia, PA). Urokinase enzyme-linked immunosorbent assay (ELISA) kits, plasminogen, and the monoclonal anti-uPAR antibody used to measure uPAR expression by flow cytometry (monoclonal antibody [MoAb] 3937), were obtained from American Diagnostica, Inc (Greenwich, CT). Prourokinase was a generous gift from Drs Jack Henkin and Andrew Mazar (Abbott Laboratories, Abbott Park, IL). Matrigel was obtained from Collaborative Biomedical Products (Bedford, MA), and invasion chambers (8.0 μmol/L pore size) from Costar Corp (Toronto, Canada). Endothelial cell growth supplement was purified as described by Maciag et al.16 Recombinant VEGF was purchased from R & D Systems (Minneapolis, MN). Radiochemicals and Reflection NEF autoradiographic film were from Dupont/New England Nuclear (Mississauga, Canada), and Pharmacia Oligolabelling kits were obtained from Pharmacia Biotech (Piscataway, NJ). Iodobeads were purchased from Pierce (Rockford, IL). Nylon membranes were from Micron Separations, Inc (Westboro, MA). Airtight chambers used for culture of cells under hypoxic conditions were obtained from BellCo Biotechnology (Vineland, NJ).

HTR-8/SVneo cells were obtained from explant cultures of human first trimester placenta and immortalized by transfection with a cDNA construct containing the SV40 large T antigen.17 These cells have been previously characterized17 and have been maintained in culture for more than 120 passages in RPMI 1640 supplemented with 5% FBS. They exhibit a high proliferation index and share various phenotypic similarities with the nontransfected parent HTR-8 cells such as in vitro invasive ability and lack of tumorigenicity in nude mice.17 HUVEC were isolated from pooled umbilical cords as previously described18 and cultured in Medium 199 containing 75 μg/mL endothelial cell growth supplement, 2 mmol/L glutamine, and 5% FBS.

For culture under hypoxic conditions, cells were placed in airtight chambers, which were flushed with a gas mixture containing 5% CO₂ and 95% N₂ until the oxygen concentration within the chamber, measured using a Miniox 1 oxygen analyzer (Catalyst Research Corp, Owings Mills, MD), was less than 0.5%. Cells were then incubated within the sealed chambers for up to 24 hours, at 37°C. Under these conditions, the O₂ concentration equilibrates within 1 to 2 hours and remains at or below 1% throughout the incubation period.

Binding of molecular oxygen to the ferrous iron atom within the porphyrin ring of a heme protein induces a change in its conformation from the deoxy, or tense, to the oxy, or relaxed state, as described for hemoglobin.19 The conformation of a putative oxygen-sensing heme protein may be regulated in a similar manner, and such a protein may acquire signaling activity after assuming the deoxy state, as described for the FixL heme protein of Rhizobium meliloti.20 To determine whether the pathway underlying the regulation of uPAR by hypoxia involves such a protein, we cultured cells for 24 hours in the presence or absence of either 100 μmol/L cobalt chloride or 30 mmol/L Tiron,21 an iron-chelating agent. Cobalt displaces iron from the porphyrin ring, but binds oxygen with lower affinity than iron,22 while Tiron chelates intracellular iron in a manner similar to that described for desferrioxamine.23 Hence, culture of cells in the presence of either of these agents induces the conformation of heme proteins into the deoxy state, thereby initiating cellular hypoxic responses. Furthermore, like oxygen, carbon monoxide also binds to heme proteins, maintaining them in the oxy or relaxed state, even under hypoxic conditions. Therefore, another strategy used to assess the involvement of such a protein in regulation of uPAR expression by hypoxia was to culture HTR-8/SVneo cells for 24 hours under hypoxic conditions in the presence of 20% carbon monoxide, with the expectation that inclusion of the latter would block heme protein-mediated hypoxic responses.

For flow cytometry, cells were released from flasks by incubation in cold phosphate-buffered saline (PBS) containing 5 mmol/L EDTA. One million cells were then incubated for 1 hour at 4°C, with either 10 μg/mL of MoAb 3937, or control mouse IgG2a, used at the same concentration. Bound antibody was detected using FITC-conjugated goat anti-mouse immunoglobulin. Cells were then fixed with 2% paraformaldehyde in PBS before analysis using a Coulter Elite Flow Cytometer (Beckman Coulter Instruments, (Miami, FL).

Measurement of uPAR expression by ligand binding analysis was performed using [¹²⁵I]-prourokinase as the ligand, as we have described previously.24 Prourokinase was radiolabeled with¹²⁵I, using Iodobeads, to a specific radioactivity of at least 10⁶ cpm/μg. Cells were then chilled to 4°C, washed with cold PBS containing 1% bovine serum albumin (BSA), and incubated for 2 hours with increasing concentrations of [¹²⁵I]-prourokinase in the absence (to determine total binding) or presence (to determine nonspecific binding) of a 100-fold molar excess of unlabeled prourokinase. Specific binding was defined as the difference between total and nonspecific binding and analyzed by nonlinear curve fitting methods (least squares method) using the Kaliedograph software program (Synergy Software, Reading, PA).

Total cellular RNA was isolated, separated by electrophoresis, and transferred to charged nylon membranes. After prehybridization at 42°C for 2 hours in 50% formamide, 5× Denhardt's solution, 0.5% sodium dodecyl sulfate (SDS), 6× SSC (1× SSC = 0.15 mol/L NaCl, 15 mmol/L sodium citrate, pH 7.0) and 100 μg/mL denatured salmon sperm DNA, membranes were hybridized overnight at 42°C with a full-length uPAR cDNA probe12 cloned in a Bluescript vector and labeled with [α-³²P] deoxycytidine 5′-triphosphate (dCTP) using a Pharmacia Oligolabelling kit. The hybridization solution consisted of 6× SSC, 0.5% SDS, 100 μg/mL denatured salmon sperm DNA and 50% formamide. After serial washes, the membrane was developed by autoradiography using Dupont Reflection NEF film. To determine whether VEGF plays a role in uPAR upregulation in HTR-8/SVneo cells, as shown previously for endothelial cells,25 relative levels of uPAR mRNA in cells cultured for 24 hours under either 20% O₂ alone, 1% O₂, or 20% O₂ + 10 ng/mL VEGF were also assessed by Northern blot analysis.

A previously-described invasion assay was used to determine the effect of hypoxia on the invasion of HTR-8/SVneo cells through a reconstituted basement membrane.17 Briefly, Transwell invasion chambers were coated with 100μL of a 1.3 mg/mL solution of Matrigel diluted in cold RPMI 1640 medium. The Matrigel was then air-dried for 12 hours in a laminar flow cabinet. HTR-8/SVneo cells were labeled by incubation for 24 hours in the presence of 10 μCi/mL [³H]-thymidine. The cells were then harvested, adjusted to a concentration of 5.0 × 10⁵/mL, and 100-μL aliquots added in triplicate to the upper wells of the invasion chambers. After a 24-hour incubation, cells in the upper and lower compartments of the chambers were harvested.17 The invasion index, reflecting the percentage of added cells that had invaded the Matrigel was determined by measurement of the radioactivity in the upper and lower compartments, as well as in the membrane.17 

Concentrations of gelatinase and plasminogen activators in the conditioned medium of HTR-8/SVneo cells cultured under standard or hypoxic conditions were compared using gel zymography, as previously described.17 The concentrations of uPA antigen in these samples were more accurately measured using a specific uPA ELISA.

Cell surface PA activity was determined using a modification of the fluorometric assay of Ellis et al.26 Briefly, HTR-8/SVneo cells were plated in quadruplicate wells of a 96-well tissue culture plate and allowed to grow to 95% confluency. After a 24-hour incubation under an atmosphere of either 20% O₂ or 1% O₂, the cells were washed twice and further incubated with fresh medium containing plasminogen and the plasmin peptide substrate, HDVLL-AMC, used at concentrations of 0.2 μmol/L and 0.5 mmol/L, respectively. Plasmin generation was assessed by determining the fluorescence within individual microplate wells 30 minutes later, using a Perkin-Elmer LS50B Luminescence Spectrophotometer (Perkin-Elmer Corp, Norwalk, CT) (excitation wavelength 360 nm, emission wavelength 460 nm).

We first determined the effect of hypoxia on the expression of uPAR by HTR-8/SVneo cells (Fig 1A). In comparison to cells cultured under standard conditions (20% O₂), the expression of uPAR by cells cultured in 1% O₂ for 24 hours was 68% higher (n = 7, P = .007), as determined by flow cytometry. Equal amounts of nonimmune mouse IgG_2a bound to cells cultured under both standard and hypoxic conditions, demonstrating that increased binding of anti-uPAR MoAb 3937 was not due to nonspecific interactions or increased Fc receptor expression. Similarly, assessment of uPAR expression by HTR-8/SVneo cells through measurement of [¹²⁵I]-prourokinase binding showed that hypoxia stimulated the expression of uPAR by 88% (n = 9, P = .003), an increment similar to that detected using flow cytometry. This was accompanied by a parallel increase in the kd (Table 1), as well as by an increase in the cellular content of uPAR mRNA (Fig 1B). Increased uPAR mRNA expression was first apparent after 4 hours of exposure to hypoxia and reached a maximum level after 6 hours.

It has been previously reported that VEGF increases the expression of uPAR by vascular endothelial cells.25 Furthermore, recent studies have shown that hypoxia stimulates VEGF release by increasing transcriptional activation of the VEGF gene via hypoxia-inducible factor-1 (HIF-1), as well as through mRNA stabilization.27-30 In the present study, however, culture of HTR-8/SVneo cells in the presence of 10 ng/mL of VEGF for 24 hours under normoxic conditions did not result in increased levels of uPAR transcript (Fig 2), suggesting that the hypoxia-mediated upregulation of uPAR occurs through a VEGF-independent mechanism.

We were also interested in determining whether the effects of hypoxia on uPAR expression extended to other cell types such as endothelial cells. Figure 3 shows results in which similar effects of hypoxia on upregulation of uPAR expression were detected by radioligand binding studies using HUVEC. These results were also confirmed by flow cytometry (not shown).

To examine the potential involvement of a heme protein in the hypoxia-mediated stimulation of uPAR expression, we first examined the expression of uPAR by HTR-8/SVneo cells in response to incubation in the presence of cobalt chloride or Tiron. Both flow cytometric and radioligand binding studies showed a threefold increase in the expression of uPAR in response to Tiron (n = 6, P = .004), with a more modest 28% increase observed in response to cobalt chloride (n = 8, P = .014; Table 1). In each case, the kd increased as well, as observed following exposure of cells to hypoxia. Stimulation of uPAR expression by either cobalt or Tiron was associated with increased levels of uPAR mRNA, comparable to those observed in response to hypoxia (Fig 4).

To further assess the potential role of a heme protein in the increased expression of uPAR in response to hypoxia, we examined the effects of carbon monoxide on this response. As expected, increased uPAR expression after culture of HTR-8/SVneo cells under hypoxic (1% oxygen) conditions was inhibited (56%) by inclusion of 20% carbon monoxide in the gas mixture (Fig 5); a parallel decrease was observed in the content of cellular uPAR mRNA (Fig 4). Furthermore, in two independent experiments, the inhibitory effects of carbon monoxide on the hypoxia-induced expression of uPAR were completely prevented by inclusion of cobalt chloride or Tiron in the medium (not shown), indicating that the inhibition of uPAR expression by carbon monoxide under hypoxic conditions was not due to nonspecific toxicity.

To determine the functional correlates of hypoxia-induced uPAR expression, we compared the invasion of HTR-8/SVneo cells through a reconstituted basement membrane (Matrigel) under both standard (20% oxygen) and hypoxic (1% oxygen) conditions, observing an increase of 41.4% ± 7.4% (P = .003) under the latter (Table 2). Similar increases were observed when invasion assays were performed under 20% oxygen in the presence of either 100 μmol/L cobalt chloride (24.6 ± 8.4%;P = .028) or 30 mmol/L Tiron (29.3 ± 12.7%;P = .035) (Table 2). The role of a heme protein in the regulation of cellular invasiveness was confirmed by experiments in which hypoxia-stimulated invasion was inhibited by 87% in the presence of 20% carbon monoxide and was not significantly different from that which occurred in the presence of 20% oxygen (Table 2).

Gel zymography showed a time-dependent reduction in the levels of PAs in medium conditioned by HTR-8/SVneo cells cultured for up to 24 hours under 1% O₂ (Fig 6). These observations were supported by additional studies in which direct measurement of urokinase antigen levels in the conditioned medium of HTR-8/SVneo cells cultured under hypoxic conditions for 24 hours were reduced by a mean of 52%. In contrast, the expression of cell surface PA activity by HTR-8/SVneo cells cultured for 24 hours under hypoxic conditions was 20% higher than that expressed by cells cultured under standard conditions (P = .000007). These findings are consistent with binding of secreted urokinase to increased numbers of cellular uPAR.

Additional studies in which the levels of gelatinases in the conditioned medium of HTR-8/SVneo cells cultured under standard or hypoxic conditions were assessed by gelatin zymography showed that hypoxia did not affect the amounts of these proteins released in response to hypoxia (not shown). These observations suggest that the increased invasiveness of HTR-8/SVneo cells observed under hypoxic conditions is not attributable to the increased production or secretion of these proteinases, and instead results from another mechanism, potentially involving increased expression of uPAR.

Our results show that uPAR expression by human trophoblasts and endothelial cells is regulated by hypoxia. Furthermore, they suggest that these responses occur via a heme protein-dependent pathway. This conclusion is based on the results of experiments in which the ability of such a protein to regulate uPAR expression was assessed through the use of cobalt chloride and Tiron, each of which causes the functional displacement of ferrous iron from the heme moiety of the putative oxygen sensing protein. We also showed that the effects of hypoxia on uPAR expression were blocked by carbon monoxide. Although these results are consistent with the hypothesis that the hypoxia-induced stimulation of uPAR expression occurs through a heme protein, their absolute dependence on such a protein remains unproven until it is identified and cloned.

A similar pathway regulating the production of erythropoietin (EPO) has been extensively characterized31 and shown to involve the interactions of specific transcription factors, such as HIF-1,32 with discrete enhancer regions in the 3′-flanking sequences of the EPO gene.33Despite restricted cellular production of EPO, hypoxic induction of reporter genes containing the EPO enhancer has been observed in a wide variety of cell types, suggesting that HIF-1 may also mediate other adaptive responses to hypoxia.34 Inspection of the 5′-flanking region of the uPAR gene35 shows three potential HIF-1 binding sequences, each of which differs by only one nucleotide from the consensus sequence 5′-XACGTGCX-3′, originally identified in the genes encoding EPO and several glycolytic enzymes.36 These sequences, 5′-TTCGTGCT-3′, 5′-TAGGTGCT-3′ and 5′-TACGGGCC-3′, are located in the same orientation as the uPAR transcriptional unit, between nucleotides -501 and -494, -578 and -573, and -1204 and -1197, respectively. We did not identify any potential HIF-1 binding sites in the 3′-flanking regions of the uPAR gene.37 

In addition to HIF-1,33 hypoxia may induce the activation of NF-κB, c-fos, c-jun, c-jun B, andjun D.38,39 While one group has also reported that hypoxia induces the activation of c-Src,40 a recent study failed to confirm these observations, despite culture of cells under hypoxic conditions that led to the induction of HIF-1, EPO, VEGF, and the glucose transporter-1, Glut-1.41 Furthermore, perturbation of c-Src activity in Hep3B cells did not affect the normoxic or hypoxic expression of the latter proteins.41 Whether other factors, in addition to HIF-1, contribute to the hypoxia-mediated stimulation of uPAR expression is unknown.

Although we propose that the increased expression of uPAR in response to hypoxia occurs primarily through HIF-1–mediated stimulation of uPAR gene expression, we also considered the possibility that these observations may reflect an autocrine effect of endogenously-released growth factors in response to hypoxia. For example, VEGF, the expression of which is markedly increased in response to hypoxia,1 stimulates the expression of uPAR by bovine and human endothelial cells.25 In our studies, however, culture of HTR-8/SVneo cells with exogenous VEGF did not result in increased levels of uPAR mRNA. These results suggest that, at least under the conditions used in these studies, the effect of hypoxia on uPAR expression is not mediated through VEGF. This conclusion is supported by the results of studies in which we observed that blocking anti-VEGF antibodies did not inhibit the increased expression of uPAR protein or mRNA induced by hypoxia (T. Fitzpatrick, K.R. McCrae, and C.H. Graham, unpublished observations, April 1997). However, it is possible that in other cell types, which produce greater amounts of VEGF, increased expression of uPAR in response to hypoxia may occur through additional pathways.

We observed that the levels of urokinase were progressively reduced in the conditioned medium of HTR-8/SVneo cells cultured for increasing intervals under hypoxic conditions. Although we have not determined precisely the mechanisms accountable for this observation, we believe it is likely that these decreased levels reflect increased binding of secreted uPA to cellular uPAR. This hypothesis is supported by the observation that cells cultured under hypoxic conditions expressed increased cell surface plasminogen activator activity. Nevertheless, we believe that the increased expression of uPAR induced by hypoxia is likely to be of importance in cellular migration and invasion in vivo, as it has been shown, for example, that invasive neoplastic cells may use uPA produced by neighboring stromal cells to facilitate invasion through extracellular barriers.42 Similar considerations apply to the increased kd for binding of uPA to cellular uPAR after exposure of cells to hypoxia. This modest increase, which is similar to that which accompanies the increased expression of uPAR in response to phorbol 12-myristate, 13-acetate (PMA),43 appears to result from additional glycosylation or other posttranslational modifications of the receptor. However, an increase in the kd of the magnitude observed here is not likely to significantly affect the degree of uPAR saturation with uPA in the presence of either increased endogenous uPA production or the production of significant amounts of uPA by neighboring cells. Finally, even in our system, the net effect of exposure of HTR-8/SVneo cells to hypoxia was an increase in the expression of cell surface PA activity, which we believe underlies the increased cellular invasiveness that occurred in parallel. Whether the increased in vitro invasiveness and in vivo metastatic capability of murine sarcoma cells under hypoxic conditions involve similar mechanisms has not yet been determined.5 

Our findings may be relevant to several biological processes in which cell migration and invasion occur. For example, increased expression of uPAR by extravillous trophoblast cells may facilitate their invasion through hypoxic regions of the superficial endometrial stroma.44 In support of this hypothesis are studies in which trophoblast invasion of the uterine wall has been shown to be increased in pregnant rhesus monkeys in which uteroplacental blood flow was impaired following constriction of the abdominal aorta.45 It should also be noted that the human placenta is relatively hypoxic during the first 10 weeks of gestation,8a period when trophoblast cells are maximally invasive. In addition, increased expression of uPAR is enhanced in migrating endothelial cells in vitro,46 and the fact that neovascularization of transplanted tumors in vivo is inhibited by uPAR antagonists47,48 suggests a role for hypoxia-stimulated uPAR expression in tumor angiogenesis. Additional studies have shown that uPAR antagonists inhibit tumor metastasis,49,50 as well as local invasion,50,51 perhaps explaining, in part, the association of elevated uPAR levels in extracts of breast and other types of carcinomas with a poor clinical outcome.51Finally, the expression of hypoxia-inducible proteins by macrophages within atherosclerotic plaques shows that hypoxia occurs within this setting as well,6 and suggests that the migration of these and other cell types within the plaque may be facilitated by hypoxia-stimulated increases in uPAR.52 

In conclusion, we have shown that the expression of uPAR is stimulated by hypoxia and that this response may involve an oxygen-sensing heme protein. Hence, we believe that the uPAR should be added to the expanding list of hypoxia-inducible proteins induced through such a pathway. Although it is likely that HIF-1 is involved in the hypoxia-mediated increase in uPAR expression, additional studies will be required to address this issue in different cell types and determine the physiologic importance of these observations.

We are grateful for the technical assistance of Shannyn Macdonald-Goodfellow and for the photographic work of Bob Temkin. We also thank Dr Jim Johnson for providing assistance with the flow cytometric analysis.