Urokinase-type Plasminogen Activator (uPA) Promotes Angiogenesis by Attenuating Proline-rich Homeodomain Protein (PRH) Transcription Factor Activity and De-repressing Vascular Endothelial Growth Factor (VEGF) Receptor Expression*

Urokinase-type plasminogen activator (uPA) regulates angiogenesis and vascular permeability through proteolytic degradation of extracellular matrix and intracellular signaling initiated upon its binding to uPAR/CD87 and other cell surface receptors. Here, we describe an additional mechanism by which uPA regulates angiogenesis. Ex vivo VEGF-induced vascular sprouting from Matrigel-embedded aortic rings isolated from uPA knock-out (uPA−/−) mice was impaired compared with vessels emanating from wild-type mice. Endothelial cells isolated from uPA−/− mice show less proliferation and migration in response to VEGF than their wild type counterparts or uPA−/− endothelial cells in which expression of wild type uPA had been restored. We reported previously that uPA is transported from cell surface receptors to nuclei through a mechanism that requires its kringle domain. Intranuclear uPA modulates gene transcription by binding to a subset of transcription factors. Here we report that wild type single-chain uPA, but not uPA variants incapable of nuclear transport, increases the expression of cell surface VEGF receptor 1 (VEGFR1) and VEGF receptor 2 (VEGFR2) by translocating to the nuclei of ECs. Intranuclear single-chain uPA binds directly to and interferes with the function of the transcription factor hematopoietically expressed homeodomain protein or proline-rich homeodomain protein (HHEX/PRH), which thereby lose their physiologic capacity to repress the activity of vehgr1 and vegfr2 gene promoters. These studies identify uPA-dependent de-repression of vegfr1 and vegfr2 gene transcription through binding to HHEX/PRH as a novel mechanism by which uPA mediates the pro-angiogenic effects of VEGF and identifies a potential new target for control of pathologic angiogenesis.

identify uPA-dependent de-repression of vegfr1 and vegfr2 gene transcription through binding to HHEX/PRH as a novel mechanism by which uPA mediates the pro-angiogenic effects of VEGF and identifies a potential new target for control of pathologic angiogenesis.
In the healthy adult blood vessels show little or no growth or extension except during the ovarian cycle and during placental development, and there is minimal physiologic turnover of vascular endothelial cells (1). Angiogenesis, the process by which new blood vessels develop from pre-existing vasculature, is activated by diverse pathophysiological stimuli, such as hypoxia, inflammation, or wounding, and the vessels return to the quiescent state once these stimuli are removed or wound closure has been attained (1,2). These complex processes of physiological and adaptive angiogenesis require a finely tuned balance between integrins, angiopoietins, chemokines, junctional molecules, oxygen sensors, matrix components, endogenous inhibitors, and many other factors (3).
However, there are prevalent and important settings (malignancy, inflammation, diabetic retinopathies, and development of atherosclerotic plaques, among others) in which pro-angiogenic stimuli predominate, resulting in what has been referred to as the "angiogenic switch" (4,5). Persistent excessive neoangiogenesis may be deleterious to the host, e.g. enhancing tumor growth or proliferation of "leaky" retinal vessels subject to rupture. A more thorough understanding of the process underlying the angiogenic switch that are not shared by normal vessels might identify steps in the process that could be subject to therapeutic intervention aimed at suppressing excessive neoangiogenesis or safely inducing therapeutic angiogenesis.
uPA has also been implicated in the process through which VEGF stimulates endothelial cell proliferation and forms new blood vessels. For example, exogenous VEGF does not induce angiogenesis when injected into infarcted myocardium in uPA knock-out mice (uPA Ϫ/Ϫ mice) (33). VEGF-induced endothelial permeability also depends on uPA and uPAR (34). Endothelial cells derived from uPA Ϫ/Ϫ mice do not overexpress the X-linked inhibitor of apoptosis (XIAP), which maintains endothelial survival in response to VEGF unless uPA is restored (35). We have also reported that uPA enhances endothelial permeability through intracellular signaling pathways shared with VEGF (36). However, the possibility that uPA contributes to VEGF-induced signaling through pathways unrelated to proteolysis and receptor-mediated intracellular signaling has not been explored.
We recently reported that single-chain uPA (scuPA) translocates to the nuclei of proliferating cells (37) where it regulates transcription factor HOXA5 (38), which is involved in endothelial cell proliferation and repair (39,40). In this manuscript we provide insight into a novel mechanism through which uPA mediates the pro-angiogenic effects of VEGF. We show that scuPA translocates to the nuclei of endothelial cells where it binds to the homeobox transcription factor HHEX, a repressor of vegfr1 and vegfr2 gene promoters, and in doing so interferes with their function and thereby induces VEGF receptor expression. These findings delineate a novel mechanism that contributes to the regulation of endothelial proliferation and a potential new approach toward control of aberrant angiogenesis.
Construction and Expression of Mouse ⌬K-uPA Domain Deletion Mutant-The muPA/pMT/BiP plasmid encoding mouse uPA inserted between the Bgl2 (5Ј) and Xba1 (3Ј) sites used to express the recombinant WT mouse uPA in Drosophila S2 cell-based expression system has been described previously (41). To obtain a vector that encodes kringle-deficient muPA (⌬K-muPA), the muPA/pMT/Bip plasmid was used as a template, and an overlap PCR strategy was applied. In the first step, two fragments were generated using the primer sets 1) P1 forward (5Ј-CGCTCGGGAGATCTGGCAGTGTACTTGGAGC-3Ј) and P2 reverse (5Ј-CTACAGACGAAGAAGGCTTTGC-ATCTATCTCACAGTGCTkCCCCCTGGAATTTCC-3Ј) to generate the fragment 1.1, with the Bgl2 restriction site at the 5Ј end, and 2) P3 forward (5Ј-GGAAATTCCAGGGGGAGCAC-TGTGAGATAGATGCAAAGCCTTCTTCGTCTGTAG-3Ј) and P4 reverse (5Ј-CGAAGGGCCCTCTAGACTATTAGAA-GGCCAGACC TTTCTCTTC-3Ј) to generate the fragment 1.2 having the Xba1 restriction site at the 3Ј end. In the second step, the 1.1 and 1.2 fragments were mixed at an equimolar ratio, and the mixture was used as the template for the overlap PCR using the P1 forward and P4 reverse primers to generate the ⌬K-muPA fragment having the Bgl2 and Xba1 restriction sites at the 5Ј and 3Ј ends, respectively (⌬K-muPA(Bgl2/Xba1) fragment). The muPA/pMT/BiP plasmid was cut with the Bgl2 and Xba1 restriction enzymes to remove the muPA insert, and the restricted plasmid was used to ligate the ⌬K-muPA(Bgl2/Xba1) fragment to obtain the ⌬K-muPA/pMT/BiP vector. This vector was then used to express the recombinant ⌬K-muPA lacking amino acids 48 -144 in the Drosophila S2 cell-based expression system as described (42).
end of the sequence encoding muPA (muPA/pMT/sign1 intermediate vector). Second, the primer pair (P7 forward (5Ј-GTT-GGCCTCTCGCTCGGGAGATCTATGAAAGTCTGGCT-GGCG AGCCTGTTCCTCTGCGCCTTGGTGGTGAAA-AACTCTG-3Ј)) and P8 reverse (5Ј-CAGAGTTTTTCAC-CACCAAGGCGCAGAGGAACAGGCTCGCCAGCCAGAC-TTTCATAGATCTCCCGAGCGAGAGGCCAAC-3Ј)) was used to introduce the sequence encoding the MKVWLASLFL peptide into the muPA/pMT/sign2 intermediate vector, which encodes full-length mouse uPA with native signal peptide allowing the protein to be secreted in mammalian cells. The following primer pair (P9 forward (5Ј-GCCTAAGCTTACGC-GTATGAAAGTCTGGCTGGCG-3Ј) and P10 reverse (5Ј-GTAATCCAGAGGTTGATTATCATATGACTAGTCTAT-TAGAAGGCCAGACCTTTCTC-3Ј)) was used to amplify the fragment muPA with the natural signal peptide from the muPA/pMT/sign2 intermediate vector template to introduce an MluI site in the 5Ј-untranslated region and an SpeI site in the 3Ј-untranslated region of the fragment, respectively. The fragment was cut with the Mlu1 and Spe1 restriction enzymes and then cloned into the pWPXL vector (Addgene and D. Trono laboratory, EPFL-SV-GHI-LVG, Station 19, CH-1015, Lausanne, Switzerland) and digested with MluI and SpeI to obtain the muPA/pWPXL lentiviral transfer vector, which encodes full-length mouse uPA.
To obtain the pWPXL-based lentiviral transfer vector, which encodes ⌬K-muPA, the muPA/pMT/sign2 intermediate vector was used as a template, and an overlap PCR strategy was applied as described above. The P9 forward primer (5Ј-GCCTAAGC-TTACGCGTATGAAAGTCTGGCTGGCG-3Ј) and the P2 reverse primer (5Ј-CTACAGACGAAGAAGGCTTTGCATC-TATCTCACAGTGCTCCCCCTGGAATTTCC-3Ј) were used to generate the fragment 2.1, having the Mlu1 restriction site at the 5Ј end. The P3 forward primer 5Ј-GGA AAT TCC AGG GGG AGC ACT GTG AGA TAG ATG CAA AGC CTT CTT CGT CTG TAG-3 and the P10 reverse primer 5Ј-GTAATCC-AGAGGTTGATTATCATATGACTAGTCTATTAGAAGG-CCAGACCTTTCTC-3Ј were used to generate the fragment 2.2 having the Spe1 restriction site at the 3Ј end. As above, the mixture of 2.1 and 2.2 fragments was used as the template for the overlap PCR using the P9 forward and P10 reverse primers to generate the ⌬K-muPA fragment having the Mlu1 and Spe1 restriction sites at the 5Ј and 3Ј ends, respectively. This vector was treated with the above restriction enzymes and cloned in pWPXL vector as above to generate ⌬K-muPA/pWPXL. Production of lentivirus using empty pWPXL and muPA/pWPXL or ⌬K-muPA/pWPXL as transfer vectors was performed as described previously (41) Animals uPA Ϫ/Ϫ and WT mice were obtained under a Material Transfer Agreement (MTA) between the Russian Cardiology Research and Production Center (Moscow, Russia) and the FIRC Institute for Molecular Oncology (Milan, Italy). The colony was maintained at the Pushchino nursery (Pushchino, Russia). All experimental procedures were performed according to the "Rules for carrying out experiments using laboratory ani-mals" of the Russian Cardiology Research and Production Center.

Migration Assay
mLMVECs isolated from WT or uPA Ϫ/Ϫ mice or uPA Ϫ/Ϫ mLMVECs infected with control or mouse WT uPA-or mouse ⌬K-uPA-encoding LVs were deprived for 24 h in EBM-2 basal medium supplemented with 0.5% FBS (EBM-2/0.5% FBS medium). Starved cells were detached by trypsin, washed, and resuspended in EBM-2/0.5% FBS. Wells in 24-well plates were filled with either EBM-2/0.5% FBS or EBM-2/0.5% FBS medium supplemented with 25 ng/ml mouse VEGF A (R&D Diagnostics). FluoroBlok TM transwell inserts (BD Biosciences) were inserted into the wells, and cell suspensions, prepared as above (5 ϫ 10 4 cells/ml), were added to the transwells and allowed to migrate for 18 h per the manufacturer's instructions. No increase in total cell number was observed during this time (not shown). Migrating cells were loaded with Calcein AM dye and Hoechst33342 dye for visualization and fixed in 4% paraformaldehyde. Cells were photographed using either a Leica DM4000 inverted fluorescence microscope equipped with a 10ϫ objective or with an EVOS FL Auto Cell Imaging System (Thermo-Fisher Scientific) using a 4ϫ objective. Cell numbers in each microscopic field were quantified using EVOS software.

In Vitro Endothelial Cell Tube Formation Assay
mLMVECs isolated from WT or uPA Ϫ/Ϫ mice or uPA Ϫ/Ϫ mLMVECs infected with control or mouse WT uPA-or mouse ⌬K-uPA-encoding LVs were studied. In vitro endothelial cell tube formation assay was performed and quantified as described by us previously (44 -46). Movies showing time courses of endothelial tube network formation by WT and uPA Ϫ/Ϫ mouse ECs were taken using the EVOS FL Auto Cell Imaging System equipped with the EVOS Onstage Incubator.

Ex Vivo Aortic Sprouting
Thoracic aortae were isolated from 8 -12-week-old mice as described by us previously (47,48). Aortic segments were embedded into Matrigel (Corning) containing EGM with VEGF (5 nM). Aortic ring sprouts on days 7 or 10 were fixed in 4% paraformaldehyde and photographed as below. In some experiments the samples were stained with DAPI and photographed to include all sprouts along the perimeter of the projection of the vessel ring and the entire length of the sprouts. Sprout lengths were determined by morphometric analysis using ImageJ software. To determine the entire area occupied by sprouts, several fluorescent images of the same vessel ring taken at different positions were superimposed and stitched to reconstruct the entirety of the vessel including the outgrown sprouts. Sprout areas were determined by morphometric analysis using ImageJ software and dividing the area of the sprouts by the aortic perimeter. Images of sprouting were obtained using an EVOS FL Auto Cell Imaging System with a 4ϫ objective or a Leica DMI6000 microscope using a 5ϫ objective.

Transcription Factor Protein Binding Array Analysis
Binding of uPA to transcription factors was analyzed using the TranSignal TM TF Protein Array kit (Version I, catalog #MA3501, Panomics) per the manufacturer's instructions as described by us previously (38). Imaging of the membranes was performed using the ChemiDoc-It Imaging System (UVP, LLC, Upland, CA) with the LabWorks Image Acquisition and Analysis Software (UVP Inc. Bioimaging Systems).

Indirect Immunofluorescence Staining
To analyze the intracellular distribution of exogenously added uPAs in primary human pulmonary microvascular endothelial cells (hLMVECs), the cells were grown in 8-well chamber slides (LabTek, Campbell, CA) and incubated with 20 nM WT-scuPA or ⌬GFD-scuPA or ⌬K-scuPA for 30 min. The cells were stained as described (37) using the anti-uPA Abs (American Diagnostica, catalog #389, Lot 198), and nuclei were counterstained with DAPI (0.5 g/ml). Stained cells were mounted as described (37) and examined with a confocal laser-scanning microscope (Zeiss LSM 510; Carl Zeiss, Heidelberg, Germany). Images taken in DAPI channel were pseudocolored in red.

Immunohistochemistry Analysis
Paraffin-embedded sections of non-small lung carcinoma tissue were deparaffinized in xylene and rehydrated in graded ethanol solutions. Blocking of endogenous peroxidase and permeabilization was performed in 2.2% H 2 O 2 solution in methanol. Permeabilized sections were blocked for 20 min using 10% horse serum diluted in 1% BSA/1ϫ automation buffer solution (blocking buffer) at room temperature. Sections were incubated overnight at 4°C with primary mouse monoclonal antihuman uPA Ab (American Diagnostica, catalog #3689, Lot 060406) diluted in the blocking buffer, washed, and incubated with biotinylated horse anti-mouse secondary Ab diluted in the blocking buffer for 30 min at 37°C followed by incubation with streptavidin-HRP (Vectastain kit; Vector). The tyramide signal amplification (TSA) reaction was performed using Alexa 488 Fluor TSA kit (Invitrogen) as per the manufacturer's instructions. Tissue sections were then incubated with Cy3-conjugated mouse monoclonal anti-SM ␣-actin Ab (Sigma, catalog #C6198, Lot 024M4838V) in blocking buffer for 1 h. The slides were incubated with SYTO-6 fluorescent dye (Invitrogen) to visualize the nuclei, mounted in Gel-Mount medium (Electron Microscopy Sciences, Hatfield, PA), and analyzed using confocal microscopy, as above. Images taken in far-red channel for nuclear staining (SYTO-6 stain) were pseudo-colored in red. Images taken in red channel (Cy3-conjugated anti-SM ␣-actin) were pseudocolored in blue.
Immunohistochemical staining of sections was performed for the endothelial cell marker von Willebrand factor. Fresh specimen sections and de-mounted sections previously stained for uPA and SM ␣-actin were subjected to antigen retrieval using Dako Target Retrieval Solution after blocking of endogenous peroxidase as above. Sections were blocked using 10% goat serum diluted in 1% BSA/1ϫ automation buffer and incu-bated overnight at 4°C with primary rabbit polyclonal antivon Willebrand factor Ab (Dako, catalog #A008229 -5, Lot 20020455) or whole rabbit Ig as a negative control. Subsequent staining steps were as above except that 3,3Ј-diaminobenzidine (Thermo Fisher) was used as an HRP substrate, and nuclei were counterstained with hematoxylin solution (Thermo Fisher). Images were taken using EVOS FL Auto Imaging System.

Solid Phase Protein-Protein Binding Assay
To study HHEX-FLAG binding to uPA deletion mutants, 96-well plates were coated with 0.7 M each of recombinant uPA deletion mutant or 1% BSA in Dulbecco's modified PBS, blocked with Dulbecco's modified PBS supplemented with 1% BSA, and incubated with the nuclear extract from 293HEK cells transfected with the HHEX-FLAG-pcDNA3.1 construct for 1 h. After washing with Dulbecco's modified PBS supplemented with 1% BSA, bound HHEX-FLAG was detected using anti-HHEX rabbit polyclonal Ab (Abcam, catalog #34222 Lot 513969) followed by incubation with HRP-conjugated anti-rabbit antibody (Jackson ImmunoResearch Laboratories, catalog #111-035-144) and color development with 3,3,5,5-tetramethylbenzidine. The reaction was terminated by adding 1 mM H 2 SO 4 , and the absorbance at 450 nm was measured using a microplate reader.

Cell Proliferation Assay
Proliferation of human LMVECs (Lonza) was measured using the [ 3 H]thymidine incorporation assay as described (49). Briefly, cells were plated at 60% confluency in EGM-2 complete medium (Invitrogen), starved in EBM-2 medium containing 0.5% FBS and antibiotic/antimycotic supplement (EBM-2/0.5% FBS) (Invitrogen) for 24 h, and stimulated with human VEGF (R&D Diagnostics) in EBM-2, 0.5% FBS for an additional 24 h. [ 3 H]Thymidine was added during the last 3 h of incubation of the cells with VEGF (final concentration 1 Ci/ml), the cells were washed to remove unincorporated [ 3 H]thymidine and harvested, and the incorporated radioactivity was quantified as described (49). Proliferation of mouse LMVECs was measured using the BrdU incorporation assay as described (50) with several modifications; cells were plated in black 96-well plates with clear bottoms, and Alexa488-conjugated secondary Ab was used to detect bound anti-BrdU antibodies. Fluorescence was measured using the Synergy TM 2 Multi-Detection Microplate Reader (BioTek Instruments, Inc.) and analyzed using the Gen5 TM Microplate Data Collection & Analysis Software.

Quantitative Reverse Transcriptase-mediated PCR
K562 cells (1 ϫ 10 7 ) were transfected with 10 g of pMUG1-PRH or pSIH-uPA and empty pMUG1 and pSIH plasmids for 48 h. RNA was produced according to standard protocols. Quantitative PCR was performed in triplicate as described (51) using the primers listed below, and the data were analyzed using Rotorgene 6 software (Corbett Research; Rotorgene RG-3000). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the internal control. Results for relative expression ratios were calculated according to the efficiency calibrated mathematical model.

Statistical Analysis
Differences between groups were compared using the oneway analysis of variance statistical test. Statistical analyses were performed using the EZAnalyse add-in to Microsoft Excel software. Significance was set at a p value of Ͻ0.05.

Results
uPA Is Required to Induce Angiogenesis by VEGF in Vivo-To further assess whether uPA is involved in VEGF-induced angiogenesis (33,52), we first asked if uPA modulates VEGF-induced capillary sprouting using aortic rings isolated from WT and uPA Ϫ/Ϫ mice. Aortae from WT mice embedded in Matrigel developed sprouts in response to VEGF that were Ͼ3 times longer those emanating from the aortae of uPA Ϫ/Ϫ mice (7.01 Ϯ 0.24 versus 2.32 Ϯ 0.15, respectively) (p Ͻ 0.001) (Fig.  1A). The lengths of the sprouts developed from unstimulated WT aortae were 1.43 times longer those from uPA Ϫ/Ϫ mice (p Ͻ 0.001) (Fig. 1B), suggesting additional growth factors in Matrigel that affect sprouting also depend on uPA, although not to the same extent as VEGF.
uPA Potentiates VEGF-induced Microvascular Endothelial Cell Proliferation, VEGF-induced Signaling, and Migration-Angiogenesis requires that endothelial cells proliferate and migrate. To examine the role of uPA in VEGF-induced EC proliferation, mLMVECs from uPA Ϫ/Ϫ and WT mice (53) were incubated with VEGF. VEGF (10 ng/ml) induced a Ͼ3-fold increase in DNA synthesis in WT mLMVEC, whereas uPA Ϫ/Ϫ mLMECs were almost completely non-responsive ( Fig. 2A). uPA Ϫ/Ϫ mLMVECs were then transfected with a lentiviral construct encoding mouse uPA (KO EC uPA LV) or with empty lentiviral construct as a negative control (KO EC con LV). Transfected cells were then stimulated with 5-25 nM mouse VEGF for 18 h. KO EC uPA LV showed significantly greater proliferation in response to VEGF than did KO EC con LV cells (p Ͻ 0.001) (Fig. 2B). Erk1,2 and PI3K-Akt-S6K pathways are activated in endothelial cells stimulated by VEGF (54 -56) and have been implicated in mediating proliferative responses in endothelium (57,58) and in other cell types (for review, see Ref. 59). VEGF induced sustained phosphorylation of Erk1,2 and the S6 ribosomal subunit (the substrate of p70 S6 kinase, one of the effectors of the PI3K/Akt/ mTOR pathway; Ref. 60) in KO EC uPA LV within 2 min, whereas either delayed phosphorylation of Erk1,2 (after 10 min of stimulation) or almost no S6 ribosomal subunit phosphorylation was seen in the KO EC con LV cells (Fig. 2C). uPA also potentiated the proliferative response of human LMVECs to VEGF (Fig. 2D).
We also compared migration of mLMVEC isolated from WT and uPA Ϫ/Ϫ mice in response to VEGF. Unstimulated WT and uPA Ϫ/Ϫ cells showed little or no migration through the porous membrane (Fig. 3A). When stimulated with VEGF (25 ng/ml), the migration response of WT mLMVEC was Ͼ3 times greater than uPA Ϫ/Ϫ LMVEC.
We then utilized an in vitro endothelial tube formation assay to compare angiogenic activity of WT and uPA Ϫ/Ϫ mLMVECs. Although WT mLMVEC formed branching endothelial tubes, uPA Ϫ/Ϫ mLMVEC formed clusters with few branches (Fig. 3B). Time-lapse videomicroscopy revealed that random migration of uPA Ϫ/Ϫ mLMVEC was slower than WT cells, and although there was some evidence of elongation, the intercellular con- nections were unstable (supplemental Movie 1). In contrast, cells isolated from the WT mice readily formed tubular network structures (supplement Movie 2). uPA Translocates to the Nucleus of Endothelial Cells-uPA is secreted as a scuPA (61,62) that is able to bind uPAR (12) and other receptors (14,63) and can be converted into a two-chain active enzyme by plasmin and some other proteases (64). We previously reported that scuPA, but not two-chain uPA (tcuPA) or a kringle-deficient uPA mutant (⌬K-scuPA) (See Fig. 4A for schematic representation of uPA variants), translocates rapidly to the nuclei of diverse types of proliferating cells in a kringledependent, uPAR-independent, manner (37). To determine which scuPA variants translocated to the nuclei of human LMVECs, WT-scuPA, ⌬GFD-scuPA, and ⌬K-scuPA were radiolabeled with Na 125 I and incubated with hLMVECs for 1 h. Exogenously added recombinant 125 I-scuPA and 125 I-⌬GFD-scuPA, but not 125 I-⌬K-scuPA, translocated to the nuclei of proliferating hLMVECs (Fig. 4B). Fig. 4C shows the subcellular distribution and immunofluorescence labeling of exogenously added unlabeled scuPA, ⌬GFD-scuPA variant, that does not bind uPAR (37,42) but retains the kringle and translocates to the nuclei of endothelial cells and ⌬K-scuPA variant. Untreated hLMVECs are shown in the right panel. These data confirm that uPAR is not essential for nuclear translocation of the kringle-containing WT uPA or its variant in lung microvascular endothelial cells.
To assess the relevance of this observation, we next asked whether endogenous uPA is present within the nuclei of endothelial cells in situ. Using confocal microscopy, we found that uPA is present within the nuclei of endothelial cells (marked by closed arrowheads) lining angiogenic vessels as well as within the nuclei of tumor cells (not shown) in a specimen of nonsmall lung carcinoma (Fig. 5, A and B). Not all nuclei stain for uPA, even in cells that express cytoplasmic or cell surface uPA. This suggests that the appearance of uPA within the nucleus might be under spatial or cell cycle control. uPAs That Are Capable of Translocation to Nuclei Up-regulate Expression of VEGFR1 and VEGFR2-Binding of VEGF-A to VEGFR1 (Flt1) and VEGFR2 (KDR) has been implicated in pro-mitogenic signaling in endothelial cells (for review, see Ref. 65). Having observed that VEGF induces a greater proliferative response in uPA-expressing ECs, we next asked whether these cells express higher levels of VEGFR1 and VEGFR2 and whether intranuclear uPA is responsible for this phenotype. To do so, we first compared the expression levels of VEGF receptors in mouse LMVECs isolated from WT and uPA Ϫ/Ϫ mice and in uPA Ϫ/Ϫ LMVECs transfected with LV vectors encoding mouse WT or ⌬K-uPA (KO EC uPA LV and KO EC ⌬K-uPA LV, respectively) or empty vector as the negative control (KO EC con LV). WT LMVEC and KO EC uPA LV cells showed higher expression levels of VEGFR1 and VEGFR2 than their uPA Ϫ/Ϫ counterparts or KO EC ⌬K-uPA LV cells (Fig. 6A), which expressed a variant that is unable to translocate to the nucleus ( Fig. 4B and Ref. 37). The addition of recombinant WT-scuPA and ⌬GFD-scuPA, but not ⌬K-scuPA (10 nM), to hLM-VECs for 24 h up-regulated VEGFR1 and VEGFR2 expression (Fig. 6B). Together, these data suggest that only uPA variants capable of translocation to nuclei up-regulate expression of VEGFR1 and VEGFR2.
Nuclear uPA in EC Enhances VEGF-induced Angiogenic Responsiveness-To relate increased expression of VEGF receptors in response to intranuclear uPA to the angiogenic potential of ECs, we re-expressed either WT-uPA or ⌬K-uPA variants in uPA Ϫ/Ϫ mLMVECs and studied their migration and . EC migration and in vitro tube formation in response to VEGF requires uPA. A, migration of lung microvascular ECs isolated from WT or uPA Ϫ/Ϫ mice in response to VEGF. LMVECs were starved in EBM-2/0.5% FBS for 24 h, detached with trypsin/EDTA, washed in starvation medium, and resuspended in the same medium. Cell suspensions were added inside Fluoroblok TM transwells, which were placed in 24-well plates containing either starvation medium or the same medium supplemented with 25 ng/ml VEGF. Transwells were incubated with calcein AM and Hoechst to visualize migrating cells and then fixed. Cells were allowed to migrate for 18 h and were then photographed with a 4ϫ objective using the EVOS FL Auto Imaging System microscope. Cells within the imaged field were counted. Each condition was set up in three wells, and three images were taken at different sites within each transwell. Typical images for each condition are shown. The bar graph on the right shows the mean Ϯ S.E. cell numbers per microscopic field of WT or uPA Ϫ/Ϫ cells that migrated in response to VEGF. Few cells migrated in starvation medium (two-three per microscopic field) and, therefore, do not appear on the graph. *, p Ͻ 0.001. B, tubular network formation by WT and uPA Ϫ/Ϫ ECs in response to VEGF. WT and uPA Ϫ/Ϫ ECs were allowed to adhere and migrate within a 24-well plate coated with Matrigel. The endothelial cell network was visualized by loading the cells with Calcein AM dye (4 g/ml), and photographs were taken using an EVOSா FL Auto Imaging System microscope. Typical images for each condition are shown on the left. Total length of the tubes was enumerated using the ImageJ software. *, p Ͻ 0.001 (right).
in vitro tube formation in response to VEGF. Fig. 6C shows that re-expression of WT-uPA caused a 2.48-fold increase in migration of uPA Ϫ/Ϫ mLMVECs (p Ͻ 0.01), whereas expression of ⌬K-uPA caused only a 1.37-fold (although significant, p Ͻ 0.01) increase in migration compared with uPA Ϫ/Ϫ mLMVEC transfected with empty lentivirus. ⌬K-uPA-transfected uPA Ϫ/Ϫ mLMVECs, like cells transfected with empty vector, formed cell clusters in the presence of VEGF in lieu of a tubular network (Fig. 6, D and E), whereas WT-uPA-expressing cells formed tubular network structures similar to WT mouse mLMVECs (Fig. 6, D and E, with reference to Fig. 3B).
We then asked if the capacity of WT-uPA to translocate to the nucleus is involved in aortic sprouting. Aortic rings isolated from uPA Ϫ/Ϫ mice were transfected with WT-uPA, ⌬K-uPA, and empty control pWPXL-based LVs immediately after isolation and embedded in Matrigel 24 h later. The parental pWPXL vector-based LV, which encodes GFP, was used to elucidate transfection efficiency of vessel ring tissues by LV. Fig. 7A shows that GFP was effectively delivered by LV. RT PCR analysis of the vessel rings transduced with the empty and WT-uPA-or ⌬K-uPA-encoding LVs confirmed transfection (Fig. 7B). VEGF induced intense sprouting from the  JULY 15, 2016 • VOLUME 291 • NUMBER 29

JOURNAL OF BIOLOGICAL CHEMISTRY 15037
aortae expressing WT-uPA, whereas significantly less sprouting was observed from the rings expressing ⌬K-uPA or transfected with empty LV (Fig. 7, C and D). These data indicate that ⌬K-uPA, which does not translocate to the nucleus, are unable to fully restore angiogenic capacity to uPA Ϫ/Ϫ ECs in response to VEGF in contrast to WT-uPA, which invests uPA Ϫ/Ϫ ECs with a responsiveness comparable to ECs isolated from WT mice (Fig. 1).

Intranuclear uPA Binds to HHEX/PRH Transcription Factor and Abolishes Repression of vegfr1 and vegfr2 Promoters-We
recently reported that uPA binds several homeobox transcription factors, including HOXA5 and Hey (38), that might be involved in regulating VEGFR1 and VEGFR2 expression (39,40). Using a transcription factor protein-protein microarray, we found that uPA also binds to HHEX/PRH transcription factor (Fig. 8A). We previously reported that HHEX/PRH binds to The slide stained for uPA was dismounted and used together with intact parallel sections for immunostaining using rabbit polyclonal anti-von Willebrand factor antibody to assess whether uPA was detected in the nuclei of endothelial cells of the vessels within tumor tissue as described under "Experimental Procedures." Total rabbit Ig served as a negative control. The brown color (top and middle panels) denotes vascular endothelial cells; nuclei are depicted in blue. Images of the same vessel as in A, marked by black arrowhead in the section previously stained for uPA and in the fresh serial section, are shown in the top and middle panels, respectively. The bottom panel represents a negative control using a fresh serial section. Images were taken using the EVOSா FL Auto Imaging system using a 40ϫ objective. Scale bar ϭ 100 m.
the promoter regions of the VEGFR1 and VEGFR-2 genes and represses their transcription (51). Therefore, we next tested the hypothesis that intranuclear uPA de-represses vegfr1 and vegfr2 promoters by binding to HHEX. In support of this hypothesis, uPA co-immunoprecipitated with HHEX/PRH when both proteins were overexpressed in 293HEK cells (Fig. 8B). As quies-cent ECs typically express low levels of uPA, to determine if uPA co-immunoprecipitates with endogenous HHEX, we preincubated hMVEC with recombinant scuPA. HHEX co-immunoprecipitated with uPA from scuPA-treated hMVECs (Fig.  8C). We then used a series of uPA domain deletion mutants listed in Fig. 4A to determine which domain(s) is required for binding to HHEX. uPA variants were immobilized onto the wells of a 96-well plate. Immobilized uPAs were then incubated with nuclear extracts from 293 HEK cells transfected with HHEX-FLAG/pcDNA3 plasmid, which encodes full-length HHEX possessing a FLAG tag at the C terminus. WT-scuPA and ⌬GFD-scuPA bound HHEX, in contrast to ATF, which lacks the proteolytic domain, and low molecular weight-uPA or ⌬K-scuPA, which lacks the kringle (Fig. 8D). This result indicates that both the kringle and the C-terminal proteolytic domain are necessary for binding to HHEX.
To determine whether uPA induces expression of VEGFR1 and VEGFR2 by interfering with the repressor function of HHEX, we used the K562 cell line. K562 cells express low levels of endogenous uPA, and they express both VEGFR1 and VEGFR2, which can be suppressed by ectopic expression of HHEX (51). K562 cells were co-transfected with the HHEXand/or uPA-encoding vectors alone or in combination, and VEGFR1 and VEGFR2 mRNA levels were measured. The results shown in Table 1 demonstrate that overexpression of HHEX suppresses both VEGFR1 and VEGFR2 mRNA levels. Ectopic expression of uPA alone did not significantly change VEGFR1 or VEGFR2 expression, providing further evidence that uPA does not directly affect transcription of the VEGFR1 and VEGFR2 promoters. However, co-expression of uPA together with HHEX reversed HHEX-mediated repression of VEGFR1 and VEGFR2. These data suggest that binding of uPA to HHEX inhibits binding of the transcription factor to its target DNA sequence. To examine this possibility in more detail, we performed an EMSA using vegfr2 promoter-derived double-stranded oligonucleotide that contained a HHEX consensus region (see spotted in duplicate with proteins expressed from full-length transcription factor cDNAs were incubated with scuPA (10 nM) for 2 h. Bound scuPA was detected with rabbit anti-uPA polyclonal antibodies, HRP-conjugated secondary goat anti-rabbit antibodies, and chemiluminescence substrate. B, co-immunoprecipitation (co-IP) of uPA and HHEX from 293HEK cells ectopically expressing HHEX-FLAG and uPA. HEK 293 cells were transfected with HHEX-FLAG in pcDNA3.1 and uPA/pcDNA3.1 vectors. Two days later, cells were harvested, and nuclear extracts were prepared using the NucBuster protein extraction kit (Novagen). uPA or HHEX was immunoprecipitated using mouse monoclonal anti-uPA or anti-FLAG antibodies immobilized on agarose beads. Normal mouse Ig immobilized on agarose beads was used as the negative control. Immune complexes were analyzed by Western blot (WB) using anti-uPA rabbit polyclonal antibodies and HRP-conjugated anti-FLAG mouse monoclonal antibodies. C, co-immunoprecipitation of uPA and HHEX from the nuclear extracts of human scuPA-treated human LMVECs. uPA and HHEX were co-immunoprecipitated from the nuclear extracts using mouse monoclonal anti-uPA and anti-HHEX antibodies, respectively. Normal mouse Ig immobilized on agarose beads was used as the negative control. Immune complexes were analyzed by Western blot using anti-uPA rabbit polyclonal antibodies and anti-HHEX rabbit polyclonal antibody. D, binding of HHEX to recombinant uPA variants. Recombinant uPA variants (33 nM in PBS) or BSA (1%) as the negative control were immobilized in 96-well plates in triplicate and incubated with nuclear extract from HEK293 cells transfected with pcDNA 3.1/HHEX-FLAG. Bound HHEX-FLAG was detected using an anti-HHEX polyclonal antibody/HRP-anti-rabbit antibody sandwich ELISA. Optical density was read at 450 nm (OD 450

Nuclear uPA Regulates Expression of VEGF Receptors
"Experimental Procedures"). The data show that WT scuPA did not bind directly to the HHEX DNA consensus sequence, but it did inhibit binding of HHEX to its DNA target site (Fig. 8E). These hypotheses were further supported by the results of the VEGFR1 and VEGFR2 promoter luciferase reporter assay. Expression of HHEX resulted in a 3.49-and 2.68-fold repression of VEGFR1 and VEGFR2 promoter activities, respectively, whereas co-expression of HHEX with uPA partially but significantly reversed inhibition of VEGFR1 and VEGFR2 promoter activity by HHEX (Fig. 8F). These results provide additional support for the hypothesis that binding uPA to HHEX leads to de-repression of the VEGFR1 and VEGFR2 promoters.

Discussion
VEGF and uPA play prominent roles in angiogenesis (59,66). VEGF 165 induces endothelial cell motility, invasion, and proliferation by binding to specific receptors, VEGFR1 (Flt-1) and VEGFR2 (KDR) (59). Binding of VEGF to these receptors upregulates the expression of uPA and generates plasmin that converts catalytically inactive scuPA into enzymatically active tcuPA (67). tcuPA proteolyzes extracellular matrices and induces intracellular signal transduction through autocrine and paracrine pathways that promote endothelial cell proliferation and migration (for review, see Ref. 66). In this manuscript we describe an additional mechanism by which uPA promotes angiogenesis, i.e. via transcriptional up-regulation of VEGFR1 and VEGFR2 expression.
Our findings show that proliferation of human and mouse microvascular endothelial cells in response to VEGF-A requires the presence of uPA in the nucleus where it binds to the transcription factor HHEX. Binding of uPA of HHEX interferes with transcriptional repression of the vegfr1 and vegfr2 promoters leading to up-regulation of receptor expression.
Nuclear translocation of uPA in endothelial cells does not require uPAR, as both full-length uPA and a variant lacking the uPAR binding GFD are both able to translocate to cell nuclei (37) (Fig. 4C) and up-regulate both VEGF receptors (Fig. 6B). Rather, translocation of uPA to the nucleus and up-regulation of expression of VEGF receptors depends on its kringle domain (37) (Figs. 4B and 6A). We previously reported that nuclear translocation of uPA is mediated by nucleolin (37), a nucleocytoplasmic shuttle protein that transports diverse proteins from the cytoplasm to the nucleus (68). Nucleolin is located on the external plasma membrane of angiogenic endothelial lining tumor vessels but not on neighboring endothelium lining non-proliferating host vessels (69), but how its appearance on cell surfaces is regulated is unknown. Blockade of cell surface nucleolin with a specific pseudopeptide reduces tumor progression in mouse models of cancer (70,71). Therefore, the preferential translocation of nucleolin onto the surface of angiogenic endothelial cells may promote nuclear translocation of uPA and lead to de-repression of VEGFRs expression.
To find the intranuclear target for uPA responsible for its pro-angiogenic activity, we profiled uPA-binding transcription factors using a transcription factor protein-protein microarray. We found that uPA binds transcription factor HHEX, also referred to as proline-rich homeodomain protein (PRH) transcription factor, which has been identified by us previously as a transcriptional repressor of the vegfr1 and vegfr2 promoters (51).
We chose one of the potential HHEX DNA sequences derived from the vegfr2 promoter to help elucidate the mechanism by which uPA modulates the DNA binding capacity of HHEX. Our data show that uPA does not bind directly to this dsDNA oligonucleotide. Rather, uPA binds to HHEX (Fig. 8, A-C), inhibiting its capacity to dock onto its target DNA sequence (Fig. 8E). Our data also demonstrate that overexpression of HHEX/PRH reduces VEGFR1 and VEGFR2 mRNA expression and that co-expression of HHEX and uPA prevents repression of VEGFR1 and VEGFR2 by HHEX/PRH. Therefore, our data suggest that in quiescent endothelial cells HHEX represses expression of VEGF receptors. In contrast, overexpression of uPA in response to VEGF (67) or other mediators released by tumors or surrounding stroma (31,72) as well as release of uPA by tumor or stromal cells (73,74) may "sensitize" endothelial cells to pro-angiogenic stimuli through uPA-mediated de-repression of VEGFRs promoters, thereby promoting angiogenesis even in the absence of a significant change in VEGF expression per se.
uPA might also regulate angiogenesis by binding to other transcription factors that activate or repress additional genes. For example, we recently reported that uPA promotes survival of pancreatic cancer cells in part by binding to the homeobox transcription factor HOXA5 leading to down-regulation of p53 expression (38). Sustained expression of HOXA5 leads to down-regulation of many pro-angiogenic genes including vegfr2, ephrin A1, Hif1␣, and cox-2 (39). Additional studies are needed to determine whether binding of uPA to HOXA5 and   TABLE 1 Vegfr-1, and Vegfr-2 mRNA levels in K562 cells 48 hr after co-transfection with pMUG1 (empty vector), pMUG1-Myc-PRH, pSIH (empty vector), or pSIH-uPA Levels of mRNA were determined by quantitative reverse transcriptase-PCR using specific primers and compared to GAPDH mRNA. Values are shown as -fold change over control values obtained (means and S.E.) in a sample co-transfected with pMUG1 (empty vector) and pSIH (empty vector) (n ϭ 3). other homeobox transcription factors in endothelial cells promotes angiogenesis during physiological and pathological angiogenesis by inducing an angiogenic switch through coordinate up-regulation and/or down-regulation of genes involved in endothelial cell proliferation, adhesion, and migration. These data also suggest that targeting nuclear transport of uPA through its kringle or the motif that mediates its binding to transcription factors might provide novel means to control aberrant angiogenesis with minimal impact on healthy vasculature.

vegfr-1 vegfr-2 uPA PRH
Author Contributions-V. S. contributed to conception of the research, designed and performed the research, analyzed the data, and wrote the paper.