Transactivation of the Urokinase-type Plasminogen Activator Receptor Gene through a Novel Promoter Motif Bound with an Activator Protein-2α-related Factor*

The urokinase receptor overexpressed in invasive cancers promotes laminin degradation. The current study was undertaken to identify cis elements and trans-acting factors activating urokinase receptor expression through a footprinted (−148/−124) region of the promoter containing putative activator protein-2- and Sp1-binding motifs. Mobility shifting experiments using nuclear extract from a high urokinase receptor-expressing cell line (RKO) indicated that Sp1, Sp3, and a factor similar to, but distinct from, activator protein-2α bound to this region. Mutations preventing the binding of the activator protein 2α-related factor diminished urokinase receptor promoter activity. In RKO cells, the expression of a negative regulator of activator protein-2 function diminished urokinase receptor promoter activity, protein, and laminin degradation. Conversely, urokinase receptor promoter activity in low urokinase receptor-expressing GEO cells was increased by activator protein-2αA expression. Although using GEO nuclear extract, little activator protein-2α-related factor bound to the footprinted region, phorbol 12-myristate 13-acetate treatment, which induces urokinase receptor expression, increased complex formation. Mutations preventing the activator protein-2α-related factor and Sp1/Sp3 binding reduced urokinase receptor promoter stimulation by this agent. Thus, the constitutive and phorbol 12-myristate 13-acetate-inducible expression of the urokinase receptor is mediated partly throughtrans-activation of the promoter via a sequence (−152/−135) bound with an activator protein-2α-related factor.

(K D ϳ0.5 nM) to a 45-60-kDa heavily glycosylated cell surface receptor (u-PAR) 1 (4,5) composed of three similar repeats approximately 90 residues each (6,7). The amino-terminal domain binds the plasminogen activator with the carboxyl terminus domain serving to anchor the binding protein to the cell surface via a glycosyl-phosphatidylinositol chain (6,7).
The u-PAR has multiple functions. First, urokinase bound to the u-PAR activates plasminogen at a much faster rate than fluid phase plasminogen activator (8,9), and this contributes to type IV collagenase activation (10). Second, the binding site clears urokinase-inhibitor complexes from the extracellular space (11) via a ␣ 2 -macroglobulin receptor-dependent mechanism (12). Third, the u-PAR interacts with the extracellular domain of integrins to connect to the cytoskeleton, thereby mediating cell adhesion and migration (13)(14)(15). Fourth, the u-PAR is chemotactic for human monocytes and mast cells, and this may require the cleavage of the binding site between domains 1 and 2 (16,17).
The u-PAR gene is 7 exons long and is located on chromosome 19q13 (18,19). Transcription of the u-PAR gene yields a 1.4-kilobase mRNA or an alternatively spliced variant lacking the membrane attachment peptide sequence (20,21). The amounts of u-PAR are controlled mainly at the transcriptional level, but altered message stability and receptor recycling may represent other means of controlling the amount of this gene product at the cell surface (22)(23)(24)(25).
The transcriptional regulation of the u-PAR gene is still poorly understood. Soravia et al. (26) reported that the basal expression of the gene was regulated via Sp1 motifs proximal and upstream of the transcriptional start site. Our laboratory showed that both the constitutive and PMA-inducible expression of the gene required a footprinted region (Ϫ190/Ϫ171) of the promoter containing an AP-1 motif (22). We also observed a second footprinted region of the promoter (Ϫ148/Ϫ124), and deletion of this region caused a dramatic reduction in the constitutive u-PAR promoter activity in a colon cancer cell line characterized by its high expression of this gene. Interestingly, this region of the promoter contained noncanonical AP-2 (Ϫ142/Ϫ134) and Sp1 (Ϫ147/Ϫ138) motifs overlapping with each other as well as nonconsensus polyomavirus activator 3 (Ϫ133/Ϫ127) motifs.
The u-PAR has been implicated in a number of physiological and pathological processes involving tissue remodeling, al-though it is not critical for mouse development (27). In cancer, several experimental and clinical findings support the view that the u-PAR plays a prominent role in tumor cell invasion and metastasis. For example, the u-PAR mRNA is expressed in the tumor cells of invasive colon cancers (28,29), and a high u-PAR protein level is predictive of short survival times for patients with this disease (30). Further, earlier studies have shown that the overexpression of a human u-PAR cDNA increased the ability of human osteosarcoma cells to invade into an extracellular matrix-coated porous filter (31). Conversely, down-regulating u-PAR levels using antisense expression constructs, oligonucleotides, or synthetic compounds reduced the ability of divergent invasive cancers to invade in vitro and in vivo (32)(33)(34)(35)(36). Since the u-PAR is a key factor in promoting tumor-associated proteolysis, down-regulation of its expression could be a promising strategy for inhibiting cancer invasion and metastasis. We therefore undertook a study with two objectives: (a) to identify cis-elements and trans-acting factors regulating constitutive and PMA-inducible u-PAR gene expression via the footprinted region spanning nucleotides Ϫ148/ Ϫ124 and (b) to determine the effect of interfering with transcription factors binding to this region on u-PAR-directed laminin degradation.

EXPERIMENTAL PROCEDURES
Vectors and Antibodies-The u-PAR CAT reporter consisted of 449 base pairs of sequence (37) stretching from Ϫ398 to ϩ51 (relative to the transcription start site) cloned into the XbaI site of the pCAT-Basic vector (Promega, Madison, WI). Reporter constructs regulated by truncated u-PAR promoter fragments were as described previously (22). The urokinase CAT reporter consisted of 2345 base pairs of 5Ј-flanking region fused directly to the reporter (38). Antibodies to Sp1, Sp2, Sp3, and AP-2 isoforms were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The Jun-D expression construct was described elsewhere (22). Oligonucleotides were purchased from Genosys Biotechnologies (The Woodlands, TX). Recombinant AP-2␣A and Sp1 (full-length human proteins) were obtained from Promega (Madison, WI). Expression vectors for AP-2␣A, AP-2␣B, and AP-2 antisense (39) consisted of the cloned sequences inserted into the EcoRI site of pSG5 (Stratagene, La Jolla, CA) and were kindly provided by Dr. Michael Tainsky. The AP-2 pBLCAT2 reporter construct contained three consensus AP-2 motifs 5Ј of the pBLCAT2 reporter (40). For the generation of the R2 CAT reporter construct, an oligonucleotide spanning nucleotides Ϫ154/ Ϫ128 was cloned into the XbaI site of pCATbasic (Promega).
Site-directed Mutagenesis-This was performed according to the protocol of the Site-Directed Mutagenesis Kit (5Z701) of CLONTECH (Palo Alto, CA). For generation of the Sp1/Sp3mt u-PAR CAT, pCATbasic (Promega) regulated by 398 base pairs of the u-PAR promoter (37) served as a template. A mutation primer substituting T for G at positions Ϫ148, Ϫ147, Ϫ144, and Ϫ142 and its corresponding selection primer (5Ј-CTTATCATGTCTGGTACCCCCGGAATTC-3Ј) converting the BamHI site of pCATbasic to a KpnI site were annealed to the denatured template plasmid, and the plasmid was amplified according to the protocol. Remaining wild-type plasmid was eliminated by two BamHI digestions for 4 h, each of them followed by transformation of nondigested DNA into BMH 71-18 mutS cells (CLONTECH, Palo Alto, CA). DNA of selected clones was isolated and sequenced using the Amersham Pharmacia Biotech T7-Sequenase 2.0 Kit.
For generation of the AP-2/Sp1/Sp3mt u-PAR CAT construct, the Sp1/Sp3mt u-PAR CAT plasmid served as a template. The second mutation primer substituted A for C at positions Ϫ146, Ϫ145, Ϫ142, and Ϫ141 of the u-PAR promoter, and the selection primer (5Ј-CTTAT-CATGTCTGGATCCCCCGGAATTC-3Ј) changed the KpnI site gener-ated above back to BamHI. Selection for AP-2/Sp1/3-mutated plasmids was done by KpnI digestion. The procedure was continued as described above.
CAT Assays-Cells were transfected at 60% confluency using poly-L-ornithine as described previously (41). All transient transfections were performed in the presence of a luciferase expression vector (4 g), and transfection efficiencies were determined by assaying for luciferase activity. CAT activity was measured as described previously (22). The amount of acetylated [ 14 C]chloramphenicol was determined using a Storm 840 PhosphorImager (Molecular Dynamics, Sunnyvale, CA) using ImageQuant software. Student's t test analysis was performed two-sided using the SPSS for Windows statistical software (release 6.1.3) (SPSS Inc., Chicago, IL). Statistical significance was defined as p Յ 0.05.
Magnetic Separation of Transfected and Nontransfected Cells-Transfected cells were enriched by the MACS-Select method of Miltenyi Biotech (Auburn, CA). RKO cells were co-transfected with the AP-2␣B expression construct and a plasmid encoding a mutated CD4 molecule (pMACS 4) in a 3:1 ratio. Cells were harvested after 42 h in 320 l of PBE buffer (phosphate-buffered saline, 0.5% bovine serum albumin, 5 mM EDTA) and incubated for 15 min with 80 l of a magnetic beadconjugated antibody directed against the mutated CD4 molecule. The cell suspension was then run through VSϩ separation columns using the VarioMACS magnet according to the manufacturer's protocol.
Western Blotting and Enzyme-linked Immunosorbent Assay for u-PAR Protein-RKO cells were extracted into a buffer (10 mM Tris, pH 7.4, 0.15 M NaCl, 1% Triton X-100, 0.5% Nonidet P-40, 20 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 1 mM EGTA, 1 mM EDTA) for 10 min at 4°C. Insoluble material was removed by centrifugation, and 750 g protein of cell extract was immunoprecipitated at 4°C for 16 h with 0.25 g of a polyclonal anti-u-PAR antibody and protein A-agarose beads. The polyclonal antibody (kindly provided by Dr. Andrew Mazar, Angstrom Pharmaceuticals, San Diego, CA) was raised in rabbits against amino acids 1-281 of the human u-PAR and purified on a Sepharose-immobilized u-PAR column. The immunoprecipitated material was subjected to standard Western blotting (42), and the blot was probed with 5 g/ml of an anti-u-PAR monoclonal antibody (catalog no. 3931, American Diagnostica, Greenwich, CT) and a horseradish peroxidase-conjugated goat anti-mouse IgG. Bands were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech).
For the determination of u-PAR by enzyme-linked immunosorbent assay, resected tissue was prepared as described by the manufacturer (American Diagnostica).
Laminin Degradation Assays-These were carried out as described previously (43). RKO cells were harvested with 3 mM EDTA/phosphatebuffered saline, washed twice, and seeded (500,000 cells) on radioactive laminin-coated (2 g/dish) dishes. The cells were allowed to attach overnight. Subsequently, cell surface u-PARs were saturated by incubating the cells at 37°C for 30 min with 5 nM urokinase, and unbound plasminogen activator was removed by washing. The cells were then replenished with serum-free medium with or without 10 g/ml plasminogen (final concentration). After varying times at 37°C, aliquots of the culture medium were withdrawn and counted for radioactivity. Solubilized laminin represents the degraded glycoprotein (43).

Region II of the u-PAR Promoter Footprinted by Nuclear
Extract from a High u-PAR-expressing Cell Line Is Bound with Sp1, Sp3, and an AP-2␣-related Factor-We previously reported (22) that nuclear extract from a high (3 ϫ 10 5 binding sites/cell) u-PAR-expressing colon cancer cell line (RKO) footprinted a region (referred to as region II) of the u-PAR promoter (nucleotides Ϫ148/Ϫ124). As a first step to identifying transcription factor(s) bound to this region, EMSA was carried out using an oligonucleotide spanning nucleotides Ϫ154/Ϫ128 (Fig.  1A). The oligonucleotide was not extended to the 3Ј-end (Ϫ124) of the footprinted region II, since preliminary EMSA utilizing a probe that included sequences 3Ј of Ϫ128 had failed to reveal any specific binding complexes. Employing the Ϫ154/Ϫ128 oligonucleotide, three slower migrating bands (indicated by a brace, arrow, and asterisk) were apparent (Fig. 1B, lanes 2 and  8) with a 100-fold excess of the nonradioactive oligonucleotide eliminating (lane 3) all of these bands. Computer analysis of this region of the u-PAR promoter revealed the presence of putative AP-2-(Ϫ142/Ϫ134) and Sp1/Sp3-(Ϫ147/Ϫ138) bind-ing motifs all bearing one mismatch with the corresponding canonical (AP-2, GCCNNNGGC; Sp1, RYYCCGCCCM) sequences. The addition of a 100-fold excess of a consensus AP-2-containing oligonucleotide (Fig. 1B, lane 7 and 11) from the human metallothionein IIa promoter sequence (44) eliminated one of the shifted bands (*). In contrast, substitution of this oligonucleotide at the AP-2 motif (CC to TT) prevented it from competing for the shifted bands (Fig. 1B, lane 10). Increasing  (22) and the oligonucleotide used in EMSA in this figure. B, RKO nuclear extract (8 g) was incubated at 21°C for 20 min with an end-labeled oligonucleotide (Ϫ154/Ϫ128 u-PAR) in the presence or absence of a 100-fold excess of competitor sequences. After this time, 2 g of the indicated antibodies were added, and complexes were subsequently analyzed by gel electrophoresis. C, EMSA was carried out as described for B with the exception that the amount of the Sp1 consensus sequence was varied. D and E, RKO nuclear extract was incubated where indicated with an anti-AP-2␣A, an anti-AP-2␤, or an anti-AP-2␥ antibody (2 g) or an equivalent amount of an unspecific IgG for 50 min at 4°C followed by protein A-agarose beads. Beads were subsequently removed by centrifugation, and treated and untreated nuclear extract was analyzed by EMSA as described for B. The data are typical of duplicate experiments. amounts of a nonradioactive oligonucleotide bearing a consensus Sp1 motif caused a dose-dependent decrease in the intensity of two of the shifted bands (indicated with a brace and arrow) (Fig. 1C) while having little effect on the band (*) competed with the AP-2 motif-containing oligonucleotide.
The ability of the consensus Sp1 motif to compete for the binding of nuclear-extracted proteins to the u-PAR promoter footprinted region II oligonucleotide suggested that transcription factors recognized by these motifs were bound to the u-PAR promoter. To examine this possibility, "supershifting" experiments were carried out. The addition of an anti-Sp1specific antibody to the RKO nuclear extract resulted (Fig. 1B, lane 4) in a slower migrating band (indicated with a line) with a concomitant decreased intensity of the complex (indicated by a brace) competed with the Sp1 consensus sequence. On the other hand, the addition of an antibody against Sp2 had no effect (lane 5) on the migration pattern, while an antibody directed at the DNA-binding domain of Sp3 completely abolished (Fig. 1B, lane 6) the shifted band with the intermediate mobility (arrow). These data suggested that the region of the u-PAR promoter footprinted with nuclear extract from a high u-PAR-expressing cell line (22) is bound with Sp1 and Sp3.
Since the fastest migrating complex evident in the EMSA (indicated by an asterisk) was competed with an AP-2 consensus motif (Fig. 1B), we hypothesized that the bound protein was an AP-2 isoform. To test this hypothesis, two sets of experiments were carried out. First, RKO nuclear extract was mixed with 2 g of an anti-AP-2␣ antibody or an equal amount of IgG and subsequently with protein A-agarose beads. The supernatant (depleted of AP-2␣-immunoreactive proteins) was then used in band shifting experiments. The fastest migrating band (*), which was competed with an excess of the AP-2 consensus, was practically abolished (Fig. 1D) by treatment of the nuclear extract with the anti-AP-2␣ antibody, whereas the bands recognized by the Sp1 (brace) and Sp3 (arrow) antibodies were unaffected. Further, the IgG control had no effect on the intensity or mobility of any of the shifted bands and antibodies specific for AP-2␤, or AP-2␥ failed to deplete the RKO nuclear extract of the binding factor (Fig. 1E). Second, we determined if authentic AP-2␣ could bind to the u-PAR promoter region II. In EMSA using an oligonucleotide spanning Ϫ154/Ϫ128, authentic AP-2␣ (Promega) gave rise to a shifted band (line) (Fig. 2), which could be competed with an excess of either the u-PAR region II oligonucleotide (Ϫ154/Ϫ128) or a consensus AP-2 motif. The authentic AP-2␣ bound to the Ϫ154/Ϫ128 probe had a different mobility from the retarded band (which was removed by immunoprecipitation with the anti-AP-2␣ antibody; Fig. 1D) observed using RKO nuclear extract (*).
To delineate the minimal sequence of footprinted region II required for the binding of Sp1/Sp3 and the factor recognized by the anti-AP-2␣ antibody, band-shifting experiments were carried out with the u-PAR promoter oligonucleotide truncated from either the 5Ј-or 3Ј-end (Fig. 3A). The removal of two base pairs at the 5Ј terminus (generating oligonucleotide Ϫ152/ Ϫ128) had little effect on the intensities of the Sp1-(brace) and Sp3-(arrow) bound complexes when compared with probe Ϫ154/Ϫ128. However, a severe attenuation in the binding of these factors was apparent with further truncation from the 5Ј-end as evident with probe Ϫ150/Ϫ128. These data suggested that nucleotides 5Ј of the nonconsensus Sp1 motif (Ϫ147/Ϫ138) are required for the optimal binding of Sp1 and Sp3 to region II of the u-PAR promoter. On the other hand, the fastest migrating band (*), which is recognized by the anti-AP-2␣ antibody, was unaffected by the removal of 5Ј nucleotides with binding maintained with probe Ϫ143/Ϫ128. These data rule out the possibility that an AP-2-like motif located further upstream (Ϫ151/Ϫ143) is mediating the binding of this factor to region II of the u-PAR promoter. While the binding of Sp1/Sp3 and the AP-2␣ antibody-reactive factor demonstrated different 5Ј requirements, the binding of these transcription factors showed identical requirements for 3Ј sequences. Thus, the removal of up to 5 base pairs from the 3Ј terminus of oligonucleotide Ϫ154/Ϫ130 (generating oligonucleotide Ϫ154/Ϫ135) had little effect on transcription factor binding. However, the deletion of an additional 3 base pairs from the 3Ј-end (oligonucleotide Ϫ154/Ϫ138) completely abolished the binding of these three transcription factors. Thus, sequences in the u-PAR promoter extending 3Ј to Ϫ135 are required for the optimal binding of Sp1, Sp3, and the factor recognized by the anti-AP-2␣ antibody.
While the anti-AP-2␣ antibody "supershifted" (arrow) authentic AP-2␣ bound to the u-PAR promoter oligonucleotide Ϫ154/Ϫ128 (Fig. 2), in contrast, we were unable to detect a "supershift" of the fastest migrating band (using RKO nuclear extract) with this antibody (data not shown). However, this could be due to the presence of multiple complexes evident in the EMSA masking a "supershifted" band. Thus, we repeated the EMSA using an oligonucleotide spanning Ϫ147/Ϫ128 (Fig.  3C). This probe is unable to bind Sp1/Sp3 as demonstrated from Fig. 3A. The addition of RKO nuclear extract to this probe yielded a slower migrating band (indicated with an asterisk), which was abolished with an excess of the unlabeled AP-2 consensus oligonucleotide. However, the addition of the anti-AP-2␣ antibody, effective in supershifting authentic AP-2␣ bound to the Ϫ147/Ϫ128 probe (arrow), failed to produce a "supershifted" complex. Likewise, antibodies to other AP-2␣ isoforms did not yield a "supershift" of the RKO-derived nuclear factor. Another possible explanation for the lack of a FIG. 2. Footprinted region II of the u-PAR promoter binds authentic AP-2␣. RKO nuclear extract (8 g) or purified AP-2␣ (10 ng) was incubated at 21°C for 20 min with radioactive oligonucleotides corresponding to nucleotides Ϫ154/Ϫ128 of the u-PAR promoter in the presence or absence of a 100-fold excess of the indicated unlabeled competitor sequences. After this time, 2 g of an anti-AP-2␣ antibody or an equivalent amount of an unspecific IgG was added to the reaction mixture where indicated. Binding complexes were subsequently analyzed by EMSA. The data are typical of duplicate experiments.
"supershift" using RKO nuclear extract is that there may be an inhibitor present in the extract. However, this is unlikely, since the addition of nuclear extract from this cell line to authentic AP-2␣ did not prevent a supershift of the latter in EMSA employing the Ϫ154/Ϫ128 probe (data not shown). These data suggest that a DNA-binding factor related to AP-2␣ (hereafter referred to as AP-2␣-related factor), derived from RKO nuclear extract, is bound to region II of the u-PAR promoter sequence.
Binding of Authentic Sp1 and AP-2␣ to the u-PAR Promoter Region II Is Mutually Exclusive-There is substantial overlap between the Sp1/Sp3 (Ϫ147/Ϫ138) and AP-2 motifs (Ϫ142/ Ϫ134) in the footprinted region II of the u-PAR promoter. Thus, we considered it unlikely that Sp1 and the AP-2␣-related factor were simultaneously binding to the u-PAR oligonucleotide probe. Nevertheless, to examine this possibility, the u-PAR oligonucleotide spanning Ϫ154/Ϫ128 was incubated with authentic Sp1 and authentic AP-2␣ proteins alone or in combination. Binding of the oligonucleotide with the individual purified proteins resulted in slower migrating complexes (Fig. 4). However, the simultaneous addition of both purified proteins to oligonucleotide Ϫ154/Ϫ128 did not yield a ternary complex and in fact resulted in a decrease in binding of either protein. The topmost band represents material remaining in the well and failed to resolve as a third band even with continuous electrophoresis. These data would suggest that the binding of Sp1 and AP-2␣ to the u-PAR promoter sequence spanning Ϫ154/Ϫ128 is mutually exclusive. However, it should be emphasized that this conclusion is based on the use of authentic AP-2␣ rather than the AP-2␣-related factor present in RKO nuclear extract.
Effect of Inhibiting Sp1/Sp3 Binding to Region II on u-PAR Promoter Activation-To determine whether u-PAR promoter activation was dependent on the binding of Sp1/Sp3 to region II, G nucleotides at positions Ϫ148, Ϫ147, Ϫ144, and Ϫ142 (within the Sp1-like motif) were substituted for T (Fig. 5A). EMSA using this substituted oligonucleotide indicated that the slower migrating bands identified as Sp1 (brace) and Sp3 (arrow) by their reactivity with specific antibodies (see Fig. 1B) were markedly diminished in their intensity when compared with an oligonucleotide corresponding to the wild type u-PAR promoter sequence (Fig. 5B). In contrast, the intensity of the complex (*) recognized with the anti-AP-2␣ antibody (see Fig.  1D) was not decreased by this change. RKO cells were then transfected with a CAT reporter driven by either the wild type u-PAR promoter (u-PAR CAT) or the promoter harboring the mutations that reduced Sp1 and Sp3 binding to region II. In duplicate experiments, the activity of the mutated u-PAR pro-  (8 g) was incubated at 21°C for 20 min with end-labeled oligonucleotides with (ϩ), or without (Ϫ) a 100-fold excess of the indicated competitor or an oligonucleotide containing a consensus AP-2 motif (AP-2 consensus). Binding complexes were resolved by EMSA as described in the legend to Fig. 1. The experiments were carried out three times. B, the nucleotide sequence of the u-PAR promoter, including the footprinted region II is shown. Underlined and overlined nucleotides indicate putative motifs identified by the Genetics Computer Group program (Madison, WI). C, RKO nuclear extract (8 g) or AP-2␣ (10 -20 ng) was incubated at 21°C for 20 min with a radioactive oligonucleotide corresponding to nucleotides Ϫ147/Ϫ128 of the u-PAR promoter in the presence or absence of a 100-fold excess of the unlabeled competitor sequences. After this time, 2 g of the indicated antibody was added to the reaction mixture where indicated. Binding complexes were subsequently analyzed by EMSA. The data are typical of duplicate experiments.
moter was similar to that achieved with the wild type u-PAR promoter (Fig. 5C).
Elimination of Sp1/Sp3 and AP-2␣-related Factor Binding to u-PAR Region II Abolishes Constitutive Promoter Activation-We were unable to interfere with the binding of the AP-2␣-related factor to region II of the u-PAR promoter without having a deleterious effect on Sp1/Sp3 binding, and this was consistent with the large overlap of these motifs and band shifting experiments (see Fig. 3A), which revealed identical 3Ј nucleotide requirements. We therefore determined the effect of eliminating the binding of these three transcription factors on constitutive u-PAR promoter activity. Nucleotide substitutions (Fig. 6A) of the Sp1-like and AP-2-like motifs, which prevented ( Fig. 6B) these transcription factors binding to oligonucleotide Ϫ154/Ϫ128, were engineered into the u-PAR promoter CAT reporter construct. The reporter construct was then compared with the wild type promoter for activation in the u-PAR-overexpressing RKO cells. We found in duplicate experiments that the activity of the mutated u-PAR promoter was reduced by over 90% in comparison with the wild type promoter construct (Fig. 6C). These data, combined with the observation that the mutation of region II, which prevented the binding of Sp1/Sp3 (but not the AP-2␣-related factor), had minimal effect on the activity of the u-PAR promoter (see Fig. 5C), implies that binding of the AP-2␣-related factor to this region is critical for the constitutive activity of this promoter in RKO cells.
Down-regulation of u-PAR Promoter Activity by the Co-expression of either a Dominant Negative or an Antisense AP-2-As indicated above, any conclusions drawn as to the role of the AP-2␣-related factor in the regulation of u-PAR promoter activity were confounded by the fact that nucleotide substitutions of region II affecting the binding of this factor also disrupted the binding of Sp1 and Sp3. To circumvent this problem, we determined the effect of interfering with AP-2␣ on u-PAR promoter activity. RKO cells were transiently transfected with an AP-2␣B expression construct. AP-2␣B is an alternatively spliced product of AP-2␣ and acts as a negative regulator of AP-2 transcriptional activity (39). In three separate experiments, increasing amounts of the AP-2␣B expression construct caused a dose-dependent repression of u-PAR promoter activity (Fig. 7). An input of 0.5 g of the dominant negative expression construct brought about an 81 Ϯ 4% inhibition of u-PAR promoter activity compared with the empty expression vector (pSG5). Increasing the amount of AP-2␣B DNA to 5 g was marginally more effective, resulting in a 93 Ϯ 5% inhibition of u-PAR promoter activity.
As an alternative to using the dominant negative expression construct, parallel experiments were also carried out using an antisense AP-2␣ expression vector. Similar to the results achieved with the AP-2␣B, the antisense expression construct caused an inhibition of u-PAR promoter activity (data not shown), with 10 g of DNA yielding a 90 Ϯ 14% repression of u-PAR promoter activity when compared with the vector backbone (pSG5). To rule out the possibility that interfering with AP-2␣ activity was having a general suppressive effect on transcription, RKO cells were transiently co-transfected with the AP-2␣B expression construct and a CAT reporter driven by the urokinase promoter. The AP-2␣B expression construct failed to repress the urokinase reporter construct using 1 g of the effector an amount that inhibited u-PAR promoter activity by over 90 Ϯ 7% (data not shown). Thus, it is unlikely that the effect of AP-2␣B on u-PAR promoter activity is due to a general shut down of transcription.
To confirm that the repressive effect of AP-2␣B on u-PAR promoter activity was mediated via the footprinted region II, two experiments were carried out. First, RKO cells were transiently co-transfected with an AP-2␣B expression construct and a CAT reporter regulated by either 148 or 105 base pairs of 5Ј-flanking sequence of the u-PAR gene. Based on band shifting experiments (see Fig. 3A) the Ϫ148 u-PAR promoter fragment can, via region II, bind the AP-2␣-related factor, whereas the u-PAR promoter regulated by only 105 base pairs of upstream FIG. 4. The binding of authentic AP-2␣ and Sp1 to region II of the u-PAR promoter is mutually exclusive. Authentic AP-2␣A (10 ng) and Sp1 (10 ng) were incubated alone or in combination at 21°C for 20 min with an end-labeled oligonucleotide (Ϫ154/Ϫ128) with or without a 100-fold excess of the unlabeled oligonucleotide. After this time, binding complexes were resolved by EMSA as described in the legend to sequence cannot. It should be noted that Sp1/Sp3 binding is preserved in the Ϫ148 u-PAR construct due to the contribution of plasmid sequences. The AP-2␣B expression construct caused a marked reduction (80%) in promoter activity using the reporter construct flanked by 148 base pairs of upstream sequence while bringing about only a modest repression (20%) of the reporter construct regulated by 105 base pairs of the u-PAR regulatory sequence (data not shown). Second, RKO cells were co-transfected with the AP-2␣B expression construct and a CAT reporter regulated by either an oligonucleotide (Ϫ154/ Ϫ128) spanning region II (R2 CAT) of the u-PAR promoter or 398 base pairs of 5Ј-flanking sequence (u-PAR CAT). The CAT reporter regulated by region II (R2 CAT) was activated (Fig. 8) in RKO cells, albeit to a lesser extent than that achieved with 398 base pairs of flanking sequence (u-PAR CAT). In two separate experiments, the activity of the R2 CAT reporter construct was inhibited by over 75% by the co-expression of the AP-2␣B when compared with an equivalent amount of the empty expression construct (pSG5). This inhibition was quantitatively similar to that achieved using the u-PAR CAT reporter. Taken together, these data suggest that the u-PAR promoter activity is indeed regulated by an AP-2␣-related factor and possibly Sp1/Sp3 in RKO cells and that this occurs via region II of the promoter.
Nuclear Extract from u-PAR-rich Cells Demonstrate Increased Binding of the AP-2␣-related Factor to the u-PAR Promoter Region II Compared with Nuclear Extract from Low u-PAR-expressing Cells-Considering the evidence implicating the AP-2␣-related factor in the regulation of u-PAR expression in RKO cells, we speculated that nuclear extract derived from cells rich in u-PAR protein would contain more of this transcription factor bound to region II of the u-PAR promoter compared with nuclear extract derived from cells characterized by their low u-PAR protein. Toward this end, we made use of another colon cancer cell line (GEO), which displays 10-fold fewer u-PARs compared with RKO as a consequence of reduced transcription of the gene (22). Nuclear extracts were generated from each cell line, and equal protein amounts were incubated with the radioactive oligonucleotide Ϫ154/Ϫ128 in the presence or absence of excess oligonucleotide competitors (Fig. 9). Binding complexes were then analyzed by electrophoresis. Nuclear extract from GEO cells gave rise to a retarded band with a mobility similar to that identified as the AP-2␣-related factor (*) using RKO nuclear extract. An excess of the AP-2 consensus sequence reduced the intensity of this band, consistent with the notion that it represents an AP-2␣-related factor-bound complex. Interestingly, the intensity of the complex (*) was reduced in nuclear extract from the low u-PAR-expressing GEO cells when compared with RKO cells, the latter of which display over 10-fold more u-PAR. Equally important, treatment of GEO cells with PMA, which increases u-PAR gene transcription in this cell line (22,45), brought about a dramatic increase in the amount of this complex (*). In contrast, the intensity of the retarded band, which was indistinguishable from Sp3 (arrow),

FIG. 5. Effect of interfering with the binding of Sp1/Sp3 to footprinted region II on the constitutive activity of the u-PAR promoter.
A, substituted nucleotides in the u-PAR promoter region II are indicated with asterisks. Underlined and overlined nucleotides indicate putative motifs identified by the Genetics Computer Group program. B, RKO nuclear extract (8 g) was incubated with an oligonucleotide spanning nucleotides Ϫ154/Ϫ128 of the u-PAR promoter (Ϫ154/Ϫ128) or the corresponding sequence that had been substituted (G to T) at the indicated nucleotides. EMSA was performed as described in the legend to Fig. 1. C, RKO cells were transiently transfected with the wild type (u-PAR CAT) or mutated (Sp1/Sp3mt u-PAR CAT) u-PAR promoter (see A for nucleotide substitutions). After 48 h, the cells were harvested, lysed, and assayed for CAT activity. Parallel dishes were transfected with pRSV CAT and pSV0CAT as positive and negative controls, respectively. The data are representative of duplicate experiments. was decreased by PMA treatment of GEO cells.
To determine if the altered amounts of the region II-bound factors were required for the stimulation of u-PAR promoter activity by PMA, GEO cells were co-transfected with a CAT reporter flanked by the wild type or mutated u-PAR promoter and a vector bearing AP-2␣B and subsequently treated with the phorbol ester. Treatment of the cells with PMA caused a strong (10 -20-fold) induction of the wild type (u-PAR CAT) u-PAR promoter, which was prevented in cells made to coexpress AP-2␣B (Fig. 10). Mutation of the u-PAR promoter to prevent Sp1/Sp3 binding (Sp1/Sp3mt u-PAR CAT) reduced the stimulation by PMA. This stimulation was further reduced when nucleotide substitutions of the u-PAR promoter were undertaken to abolish the binding of the AP-2␣-related factor as well as Sp1/Sp3 (AP-2/Sp1/Sp3mt u-PAR CAT). Interestingly, the ability of the AP-2␣B expression construct to repress u-PAR promoter activity was diminished by mutations that prevented the binding of Sp1/Sp3 (Sp1/Sp3mt u-PAR CAT). Taken together, these data would suggest that PMA stimula-tion of u-PAR gene expression in GEO cells requires, at least in part, the binding of the AP-2␣-related factor as well as Sp1/Sp3 to region II of the promoter.
Stimulation of u-PAR Promoter Activity in the Low u-PARexpressing GEO Cell Line by the Expression of Exogenous AP-2␣A-While the above experiments suggested a requirement for the AP-2␣-related factor in the stimulation of u-PAR gene expression by phorbol ester, it was not clear as to whether this transcription factor alone was sufficient to augment u-PAR promoter activity. To address this issue, GEO cells, which have low u-PAR protein, were co-transfected on three separate occasions with a u-PAR promoter-regulated CAT reporter and an expression vector bearing the full-length form of AP-2␣ (AP-2␣A). The activity of the promoter alone was below the detection limit of the assay (Fig. 11), consistent with the low expression of the u-PAR gene in this cell line. However, the cotransfection of AP-2␣A into these cells caused a dosedependent increase in u-PAR promoter activity with up to 0.05 g of the effector plasmid. This amount of the expression con-FIG. 6. Interfering with the binding of the AP-2-related factor and Sp1/Sp3 to region II of the u-PAR promoter abolishes its constitutive activity in RKO cells. A, substituted nucleotides in region II of the u-PAR promoter sequence are indicated with asterisks. Underlined and overlined nucleotides indicate putative motifs identified by the Genetics Computer Group program. B, EMSA was carried out using oligonucleotides corresponding to the wild type u-PAR sequence (nucleotides Ϫ154/Ϫ128) or the mutated sequence (AP-2/Sp1/Sp3 mutant) and RKO nuclear extract. C, RKO cells were, at 60% confluence, transfected with a RSV-luciferase expression construct and a CAT reporter regulated by the wild type or mutated u-PAR (as indicated in A) promoter sequence. Both promoter sequences were in the context of 398 base pairs of 5Ј-flanking sequence. After 48 h, the cell were harvested, lysed, and assayed for CAT activity after normalization for transfection efficiency. The amount of acetylated chloramphenicol was determined using a Storm 840 PhosphorImager. The data are typical of duplicate experiments. struct brought about over a 13-fold stimulation of promoter activity. Higher amounts of the AP-2␣A diminished u-PAR promoter stimulation, presumably as a consequence of squelching (46). The AP-2␣A-dependent stimulation of the u-PAR promoter was greater than that achieved with 2 g of a JunD expression vector, which we had shown previously (22) to stimulate u-PAR expression via a separate footprinted region (Ϫ190/Ϫ171) of the promoter containing a classical AP-1 motif. The transfection of GEO cells with JunD and AP-2␣A together stimulated u-PAR promoter activity to an extent greater than the sum of the individual expression constructs, indicative of synergism. Thus, it is likely that that these transcription factors cooperate with each other to regulate u-PAR expression. Notwithstanding these observations, our data indicate that the expression of AP-2␣A is sufficient to up-regulate u-PAR gene expression in GEO cells.
Expression of AP-2␣B Decreases Endogenous u-PAR Protein Amount in RKO Cells-We then determined if the expression of AP-2␣B, which acts as a negative regulator of AP-2 function, reduces the expression of the endogenous u-PAR gene in RKO cells. Cells were co-transfected with an expression vector encoding a mutated CD4 and varying amounts of an expression construct bearing AP-2␣B. Cells were harvested 48 h later, and transfected cells were enriched with magnetic beads coated with an anti-mutated CD4 antibody (which is non-cross-reac-tive with wild type CD4) and assayed for u-PAR protein by Western blotting (Fig. 12A). A band whose molecular mass was indistinguishable from that of u-PAR (55 kDa) (47) was detected in the immunoblot. Increasing amounts of the AP-2␣B expression vector caused a dose-dependent decrease in the amount of this protein. The reduced amount of u-PAR protein was associated with the attenuated activity of a CAT reporter regulated by three tandem AP-2 motifs upstream of a thymidine kinase minimal promoter (Fig. 12B). Thus, the expression of a negative regulator of AP-2 reduces the expression of the endogenous u-PAR gene in RKO cells.
Inhibition of u-PAR-directed Laminin Degradation in RKO Cells Made to Express AP-2␣B-One of the functions of the u-PAR is to accelerate plasminogen-dependent proteolysis (8), and this is a requirement for the invasive potential of a divergent set of cancers (31,32,48). We were therefore interested in determining whether interfering with the transcriptional activation of the gene leading to reduced u-PAR synthesis would diminish extracellular matrix degradation. RKO cells in serum-free medium (RKO-SF) demonstrated minimal solubilization of laminin (Fig. 12C). However, the addition of plasminogen to the RKO cells transfected with the vector backbone (RKO-pSG5-Pl-SF) resulted in a strong time-dependent increase in laminin degradation, indicating plasmin-dependent proteolysis. After a 2-h incubation in the presence of zymogen, nearly 500,000 cpm of solubilized laminin was evident in the culture supernatant. In contrast, RKO cells transfected with the AP-2␣B expression construct (RKO-AP-2B-Pl-SF) showed markedly reduced plasminogen-dependent degradation (80% reduction) of this glycoprotein after a 2-h incubation. These data suggest that reduced u-PAR synthesis brought about by interfering with the AP-2␣-dependent transcriptional activation of the u-PAR gene in RKO cells attenuates plasminogendependent proteolysis.

DISCUSSION
The u-PAR plays a critical role in extracellular matrix degradation and tumor invasion (34,49). We have identified an important regulatory region in the u-PAR promoter (Ϫ152/ Ϫ135) that is trans-activated by an AP-2␣-related factor. Further, our finding that the expression of AP-2␣B, which interferes with the transcriptional activation of the u-PAR gene by the AP-2␣-related factor and Sp1/Sp3, diminishes u-PAR protein amounts and laminin degradation serves to illustrate that transcriptional studies of a target gene can be utilized to bring about a more indolent phenotype of cancer at least in an experimental setting.
Interestingly, in a previous report, Soravia et al. (26), using HeLa nuclear extracts, failed to detect, by EMSA, the binding of AP-2 to the urokinase receptor promoter. The reason for this difference is unclear at the present time. One possibility is that urokinase receptor expression is regulated differently in the separate cell lines used in the two studies.
The identity of the AP-2␣-related factor bound to the u-PAR promoter region II is unknown at the present time. Certainly, it is strongly related to AP-2␣, since depletion of nuclear extract with an anti-AP-2␣ antibody and protein A-agarose beads severely diminished the intensity of the retarded band in EMSA and since the u-PAR promoter region II can clearly bind authentic AP-2␣. On the other hand, in EMSA it had a distinct mobility from authentic AP-2␣, and the anti-AP-2␣ antibody was unable to supershift the RKO nuclear factor bound to a region II oligonucleotide while effecting a supershift with authentic AP-2␣ bound to this oligonucleotide. We were also unable to supershift or immunodeplete the AP-2␣-related factor from RKO nuclear extract using recently available antibodies (Santa Cruz Biotechnology) to AP-2␤ or AP-2␥, making it unlikely that the RKO-derived nuclear factor was one of these isoforms. Lee et al. (50) also noted an AP-2␣-related factor in THP-1 monocytes that regulated the expression of the B2 subunit of the V-ATPase. Similar to our study, the THP-1-derived AP-2␣-related factor was not supershifted in EMSA and had a FIG. 9. Nuclear extract from cells expressing high u-PAR levels intrinsically (RKO) or as a consequence of PMA treatment (GEO) contains more AP-2␣-related factor bound to the u-PAR footprinted region II compared with nuclear extract from the untreated GEO cells, which have few u-PAR. Nuclear extract (8 g) from 90% confluent cells treated with, or without, PMA (100 nM) for 60 min was incubated with a radioactive oligonucleotide spanning nucleotides Ϫ154 to Ϫ128 of the u-PAR promoter. Where indicated, a 100-fold excess of nonradioactive oligonucleotide competitors was included. Binding complexes were then analyzed by EMSA. The data are representative of duplicate experiments. different size from authentic AP-2␣ based on its mobility in gel retardation studies. It remains to be determined whether the AP-2␣-related factor found in the current study and in the study by Lee et al. (50) are the same.
In addition to the AP-2␣-related factor recognized by footprinted region II of the u-PAR promoter, both Sp1 and Sp3 were also bound, as evident in EMSA using nuclear extract from the high u-PAR-expressing cell line RKO. However, we found that mutations of the u-PAR promoter that prevented the binding of either of these transcription factors had a negligible effect on promoter activation, arguing against a role for these factors in regulating at least the constitutive u-PAR expression in RKO cells.
The requirement of the u-PAR promoter region II for the PMA-dependent elevation of u-PAR gene expression (53) merits discussion. We had shown previously (22) that the induction by this phorbol ester required an intact AP-1 motif (at Ϫ184) in a separately footprinted region (Ϫ190/Ϫ171) of the u-PAR promoter. However, several observations in the current study would indicate that the induction of the u-PAR promoter by PMA requires, in addition, other transcription factor binding sites including the motif bound with the AP-2␣-related factor. Thus, nucleotide substitutions of the u-PAR promoter that prevented the binding of the AP-2␣-related factor substantially reduced the stimulation of the promoter by the phorbol ester. Further, the co-expression of AP-2␣B, which is a negative regulator of AP-2 function, completely ablated the stimulation of the u-PAR promoter by PMA. Finally, an increased amount of the AP-2␣-related factor bound to the u-PAR promoter region II was evident using nuclear extract from PMA-treated GEO cells (compared with nuclear extract from untreated GEO cells). Our results are reminiscent of other studies in which the stimulation of the PAC-1 (phosphatase of activated cells) phosphatase and neuropeptide tyrosine genes by PMA was shown to be mediated partly through an AP-2-related site and coincided with induced DNA binding of AP-2 (54, 55). It was also clear that interfering with the binding of Sp1/Sp3 (but not affecting the binding of the AP-2␣-related factor) to region II of the FIG. 10. The binding of the AP-2␣related factor and Sp1/Sp3 to region II are required for the stimulation of u-PAR promoter activity by PMA in GEO cells. GEO cells were co-transfected with a CAT reporter driven by the wild type u-PAR promoter (u-PAR CAT) or the promoter that had nucleotide substitutions in region II to prevent the binding of Sp1/Sp3 (Sp1/Sp3mt u-PAR CAT) or AP-2/Sp1/Sp3 (AP-2/Sp1/Sp3mt u-PAR CAT) and an expression vector (pSG5) encoding AP-2␣B (AP-2␣B pSG5). Following transfection, cells were replenished with fresh medium supplemented with or without PMA and cultured for an additional 2 days. Cells were then harvested, lysed, and assayed for CAT activity as described in the legend to Fig. 5. The data are typical of two separate experiments.
FIG. 11. Stimulation of u-PAR promoter activity in GEO cells made to co-express AP-2␣A. GEO cells were transiently co-transfected with a u-PAR promoter-regulated CAT and an expression vector (pSG5) bearing either JunD or AP-2␣A. Parallel cultures were transfected with pSV0CAT or pRSV CAT as negative and positive controls, respectively. After 48 h, the cells were lysed and assayed for CAT activity as described in the legend to Fig. 5. The data are representative of triplicate experiments. Chloramphenicol conversions are shown as average values Ϯ S.D. for the three experiments. An asterisk indicates that differences in chloramphenicol conversions achieved by the expression of the JunD and/or the AP-2␣A when compared with the empty expression construct (pSG5) were statistically significant (p Ͻ 0.05). u-PAR promoter had a deleterious effect on the activation of the u-PAR promoter by PMA. A similar requirement for Sp1 has been reported for the stimulation of thromboxane receptor gene expression by PMA (56), although, in contrast to that study, we did not detect increased binding of this transcription factor. Taken together, it is likely that the stimulation of the u-PAR gene expression by phorbol ester is complex, requiring the interactions of multiple transcription factors with their cognate binding sites in the upstream sequence of this gene.
The mechanism by which the AP-2␣B alters u-PAR promoter activity deserves comment. Initially, we assumed that this was through a direct antagonism of the AP-2␣-like factor bound to the u-PAR promoter region II, presumably as a consequence of heterodimerization of the two proteins (39). However, one experiment suggested that this proposal might not be entirely accurate. Thus, the ability of the AP-2␣B to counter the stimulation of u-PAR promoter activity by PMA was diminished by mutations in the u-PAR promoter that prevented the binding of Sp1/Sp3. These data suggested that the expressed AP-2␣B was mediating its effect, at least in part, through the Sp1/Sp3binding motif.
Our data based on mobility shift assays using purified proteins strongly suggested that the binding of at least AP-2␣A and Sp1 to the u-PAR promoter region II is mutually exclusive, FIG. 12. Reduced amount of u-PAR protein and laminin degradation by RKO cells made to express AP-2␣B. A, RKO cells were co-transfected with the indicated amount of the expression vector encoding AP-2␣B or the vector backbone (pSG5) alone along with an expression construct bearing a mutated CD4. After 48 h, the cells were harvested and assayed for u-PAR protein by Western blotting. Ϫve Control, no cell extract. B, RKO cells were co-transfected with an expression vector (pSG5) bearing AP-2␣B (AP-2␣B) along with a CAT reporter regulated either by three AP-2 tandem repeats upstream of a thymidine kinase minimal promoter (AP-2 pBLCAT2) or the minimal promoter alone (pBLCAT2). After 48 h, the cells were lysed and assayed for CAT activity. C, RKO cells transfected and harvested as described for A were plated in serum-free medium on radioactive laminin-coated dishes. After cell attachment, surface u-PAR were saturated with exogenous urokinase, and the cultures were washed extensively and then replenished with serum-free medium supplemented with or without plasminogen. Aliquots of the culture supernatant were withdrawn at the indicated times thereafter and counted for radioactivity. After 2 h, cells were enumerated. The data are typical of duplicate experiments. ࡗ, serum-free medium only; -, serum-free medium plus plasminogen; OE, RKO-serum-free medium; f, RKO-pSG5plasminogen-serum-free medium; , RKO-AP-2B-serum-free medium; q, RKO-AP-2B-plasminogen-serum-free medium. and this is consistent with the considerable (about 50%) overlap of the Sp1/Sp3 and AP-2 motifs. Thus, it is unlikely that the AP-2␣-related factor and Sp1 are physically interacting on the u-PAR promoter region II sequence. The biological significance of having two overlapping transcription factor binding motifs in the u-PAR promoter region II, which can be bound with only one transcription factor at any one time, can only be speculated on. Certainly, for the regulation of keratin 3 gene expression in corneal epithelial cells, overlapping AP-2 and Sp1 motifs act as a switch to up-regulate expression of this gene in response to the latter transcription factor (57). For the u-PAR promoter, since this motif can be bound with Sp3, which for some genes is a trans-repressor (58), it may be that this represents a mechanism for either up-or down-regulating u-PAR expression.
In conclusion, we have shown that the constitutive and PMAinducible expression of the u-PAR gene requires, at least in part, the trans-activation of a region of the promoter spanning Ϫ152/Ϫ135 by an AP-2␣-related factor. A high u-PAR protein level in colon cancer, which promotes the invasive phenotype (31,32) and portends a poor patient outcome (30), is probably partly a consequence of trans-activation of this gene by a greater amount and/or binding activity of the AP-2␣-related factor. Further, interfering with the trans-activation of the promoter by the AP-2␣-related factor reduced endogenous u-PAR expression and diminished laminin degradation. These findings raise the exciting possibility that interfering with the AP-2␣-related factor-dependent trans-activation of the u-PAR gene may represent a novel means of diminishing extracellular matrix degradation and consequently reducing colon cancer invasion and metastasis.