Urokinase-type Plasminogen Activator Receptor Induces Conformational Changes in the Integrin αMβ2 Headpiece and Reorientation of Its Transmembrane Domains*

The glycosylphosphatidylinositol-linked urokinase-type plasminogen activator receptor (uPAR) interacts with the heterodimer cell adhesion molecules integrins to modulate cell adhesion and migration. Devoid of a cytoplasmic domain, uPAR triggers intracellular signaling via its associated molecules that contain cytoplasmic domains. Interestingly, uPAR changes the ectodomain conformation of one of its partner molecules, integrin α5β1, and elicits cytoplasmic signaling. The separation or reorientation of integrin transmembrane domains and cytoplasmic tails are required for integrin outside-in signaling. However, there is a lack of direct evidence showing these conformational changes of an integrin that interacts with uPAR. In this investigation we used reporter monoclonal antibodies and fluorescence resonance energy transfer analyses to show conformational changes in the αMβ2 headpiece and reorientation of its transmembrane domains when αMβ2 interacts with uPAR.

The glycosylphosphatidylinositol-linked urokinase-type plasminogen activator receptor (uPAR) interacts with the heterodimer cell adhesion molecules integrins to modulate cell adhesion and migration. Devoid of a cytoplasmic domain, uPAR triggers intracellular signaling via its associated molecules that contain cytoplasmic domains. Interestingly, uPAR changes the ectodomain conformation of one of its partner molecules, integrin ␣ 5 ␤ 1 , and elicits cytoplasmic signaling. The separation or reorientation of integrin transmembrane domains and cytoplasmic tails are required for integrin outside-in signaling. However, there is a lack of direct evidence showing these conformational changes of an integrin that interacts with uPAR. In this investigation we used reporter monoclonal antibodies and fluorescence resonance energy transfer analyses to show conformational changes in the ␣ M ␤ 2 headpiece and reorientation of its transmembrane domains when ␣ M ␤ 2 interacts with uPAR.
The integrins are heterodimeric cell adhesion molecules, each formed by an ␣ and ␤ subunit. Twenty-four specific pairs of integrins are expressed in humans. They serve as adhesion molecules and are bona fide signaling receptors. Integrin ␣ M ␤ 2 (Mac-1, CR3, CD11bCD18) is expressed primarily on cells of the myeloid lineage and natural killer cells (1,2). Structural data of ␣ M ␤ 2 are lacking. A model of ␣ M ␤ 2 in a bent conformation was generated based on the crystal structure of ␣ V ␤ 3 (3) as template is shown to illustrate its domain organization in general (see Fig. 1). ␣ M ␤ 2 binds a multiplicity of ligands that include intercellular adhesion molecule 1, complement protein iC3b, blood coagulation protein fibrinogen, saccharides on microbes, and denatured proteins (4 -7). It mediates adhesion, migration, phagocytosis, and degranulation of monocytes and neutrophils and has a role in immune tolerance (2, 8 -12). In addition, the functional couplings of ␣ M ␤ 2 and partner molecules Fc␥RIIIB, Fc␥RIIA, low density lipoprotein receptor, and urokinase-type plasminogen activator receptor (uPAR) 2 have been reported (13)(14)(15)(16).
The glycosylphosphatidylinositol-anchored uPAR (CD87), its ligand urokinase plasminogen activator (uPA), and ␣ M ␤ 2 form a trimolecular complex that promotes fibrinolysis (17,18). uPAR also interacts with the ␤ 1 and ␤ 3 integrins (19,20). The uPAR-integrin complex is important in tumor biology and metastasis because of the proteolytic activity of uPA-uPAR and signaling capacity of the uPAR-integrin complex (21). Although uPAR does not contain a cytoplasmic domain, it can trigger cytosolic signaling by changing the shape of its integrin partner. uPAR changes the conformation of integrin ␣ 5 ␤ 1 , which promotes RGD-independent adhesion to fibronectin (22). Furthermore, the uPAR-␣ 5 ␤ 1 complex induces extracellular signal-regulated kinase signaling in tumor cells (23).
Integrins undergo marked conformational changes under specific cellular or extracellular conditions. One of these changes involves a switch-blade-like conversion from a bent to an extended conformation (24,25). The integrin uPAR binding sites are located in the ␤-propeller of the integrin ␣ subunit, and the inserted (I)-like domain of the ␤ subunit (22,26). The marked difference in the heights of uPAR and an extended integrin on a cell membrane does not favor such interaction as compared with an integrin that is bent ( Fig. 1) (27,28). uPAR changes the conformation of the ␤1 ectodomain when it interacts with a bent ␣ 5 ␤ 1 (22), but the molecular details of signal transmission at the C-terminal half of the integrin remain unclear. It was reported that the separation of the integrin ␣ IIb ␤ 3 TMs is an important event for outside-in signaling (29).
Here, we showed that the association of uPAR with a bent ␣ M ␤ 2 induces movement of the hybrid domain in the ␤ 2 subunit and the reorientation of the ␣ M ␤ 2 TMs.
Expression Constructs-The numbering of the integrin and uPAR amino acids is based on the mature protein as described in Barclay et al. (37). The amino acid numbering with the initiation Met considered as the first amino acid is also included and indicated in parentheses in Figs. 3, 4, and 6. The integrin ␣ M and ␤ 2 pcDNA3 expression plasmids were described previously (35). uPAR cDNA that encodes the amino acid sequence Leu 1 -Thr 313 having a stop codon introduced after Thr 313 was amplified using the forward primer 5Ј-GAAGATCTCTGCGGT-GCATGCAGTG-3Ј and reverse primer 5Ј-TCCCCGCGGTCA-TTAGGTCCAGAGGAGAG-3Ј that contained the BglII and SacII restriction enzyme sites, respectively. The digested PCR product was ligated into the BglII and SacII sites of the pDisplay vector (Invitrogen) to allow the uPAR cDNA to be in-frame with the N terminus HA tag (in the pDisplay vector) to generate the expression plasmid HA-uPAR. The D1-deleted uPAR mutant referred to as D2D3 (Leu 93 -Thr 313 ) was reported previously (35). It was cloned into the pDisplay vector to generate HA-D2D3 using the same procedure as for the full-length uPAR. The forward and reverse primers used were 5Ј-GAAG-ATCTCTCGAATGCATTTCCTGTGGC-3Ј and 5Ј-TCCCC-GCGGTCATTAGGTCCAGAGGAGAG-3Ј, respectively.
For fluorescence resonance energy transfer (FRET) experiments, the FRET pair fluorophores monomeric cyan fluores-cent protein (mCFP) and the monomeric yellow fluorescent protein (mYFP) were fused to the C termini of the integrin cytoplasmic tails. The ␣ L mCFP and ␤ 2 mYFP expression constructs were reported previously (38,39). The ␣ M mCFP, ␣ ML mCFP, ␣ ML *mCFP constructs were generated using standard molecular biology procedures. All point mutations that introduce Cys into ␣ M and ␤ 2 TMs were made using the QuikChange TM site directed mutagenesis kit (Stratagene, La Jolla, CA) with relevant primer pairs. All expression constructs were verified by sequencing (Research Biolabs sequencing service, Singapore).
Cell Culture and Transfection-293T and K562 cells (ATCC) were maintained in either DMEM medium or RPMI1640 medium, respectively, each containing 10% (v/v) heat-inactivated fetal bovine serum, and 100 IU/ml penicillin and 100 g/ml streptomycin (Hyclone, Logan, UT) at 37°C in a 5% CO 2 atmosphere. Transient transfection of 293T was performed by the calcium phosphate method (35). 5 g of plasmid was used for each indicated constructs unless otherwise stated. Transient transfection of K562 was performed by electroporation using the Amaxa Nucleofector device and reagents as recommended by the manufacturer (Amaxa Gmbh). ␣ ML CFP and variants (3.5 g each), ␣ L CFP (3.5 g), ␤ 2 YFP and variants (3.5 g each), HA-uPAR, and HA-D2D3 (2.7 g each) were used unless otherwise indicated.
Flow Cytometry and FRET Analyses-Flow cytometry analyses of transfectants were performed essentially as described (35). Briefly, cells were stained with 20 g/ml primary mAb followed by fluorescein isothiocyanate-conjugated sheep anti-mouse IgG (1:400 dilution) (Sigma). Stained cells were analyzed on a FACSCalibur using the CellQuest software (BD Biosciences).
Photobleach FRET was performed essentially as reported (39). K562 transfectants expressing the integrin ␣ and ␤ subunits with mCFP or mYFP fused to their cytoplasmic tails were spun onto poly-L-lysine-coated glass slides. Photobleach FRET was performed on a Zeiss LSM510 confocal microscope with an oil immersion 63ϫ objective (Carl Zeiss Inc., Thornwood, NY). mCFP was excited with the 458ϭnm argon laser line, and the emission signal was detected with a BP 470 -500-nm emission filter. mYFP was excited using the 514-nm argon laser line, and the emission signal was detected with a LP 530-nm emission filter. mCFP and mYFP were excited, and emission signals were detected for 10 time points. The time interval between each time point was 3 s, except that between the fifth and sixth time points, when photobleaching of mYFP was performed. For mYFP photobleach, the entire cell was scanned 20 times with the 514 argon laser line set at the maximum intensity. The duration of the photobleach was ϳ60 s depending on the size of the cell that was scanned. For analysis, only the cell membrane was selected as the region of interest. Fluorescent signals that were detected in the perinuclear region of the cell (integrins in the Golgi) were not considered. The mCFP signals in the region of interest at the fifth time point (pre-mYFP photobleach) and sixth time point (post-mYFP photobleach) were acquired. FRET efficiency (E F ) was calculated as a percentage based on the equation E F ϭ (I 6 Ϫ I 5 ) ϫ 100/I 6 , where I n is the mCFP intensity at the nth time point. Similar analyses of cells that The model of ␣ M ␤ 2 was built using Modeler8v1 and PyMOL (W. L. DeLano 2002) using these structure coordinates. The bent ␣ M ␤ 2 was generated using ␣ V ␤ 3 coordinates 1L5G as template (3). The ␣ M I domain 1BHO (67) was included. The structures of ␤ 2 plexin-semaphorin-integrin (PSI), hybrid, integrin epidermal growth factor 1 (EGF1), -2, and -3 were from 2P26 and 2P28 (68). uPAR coordinates were from 2FD6 (66). The ␣ M ␤ 2 model only serves as an illustration in the absence of a complete structure of an I domain-containing integrin. The detail position of the I domain in an intact integrin has not been determined.
were not subjected to mYFP photobleach (referred to as unbleached) were included and plotted against samples that were subjected to photobleach. The mean noise computed as N F ϭ (I 5 Ϫ I 4 ) ϫ 100/I 5 in which the mCFP signals at the fourth and fifth time points before the bleaching process was close to zero in all cases. For experiments involving uPA, transfectants were incubated in RPMI media containing 0.1% (w/v) bovine serum albumin and 100 nM uPA for 45 min at 37°C (35). Thereafter, unbound uPA was removed by centrifuging cells through a heat-inactivated fetal bovine serum cushion. Cells were later spun onto poly-L-lysine slides and subjected to FRET analyses.
Cell Surface Protein Biotinylation, Immunoprecipitation, and Detection-Labeling of cell surface proteins with biotin was performed as described previously (39). 293T transfectants were washed twice in PBS and incubated in PBS containing 0.5 mg/ml sulfo-NHS-biotin (Pierce) for 15 min at room temperature. The labeling reaction was terminated by washing the cells in PBS containing 10 mM Tris-HCl, pH 8.0, and 0.1% (w/v) bovine serum albumin. For reporter mAb immunoprecipitation analysis, the labeled cells were incubated in DMEM medium containing 5% (v/v) heat-inactivated fetal bovine serum and 10 mM HEPES (DMEM-FH) with the relevant mAb (3 g each) in the presence or absence of 2 mM MnCl 2 (referred to as Mn 2ϩ in main text) for 30 min at 37°C. The unbound mAbs were removed by washing the cells twice in the DMEM-FH and lysed in lysis buffer (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% (v/v) Nonidet-P 40) with protease inhibitors (Roche Applied Science) for 30 min at 4°C. Integrins bound to the mAbs were precipitated with protein A-Sepharose beads (GE Healthcare). Precipitated proteins were resolved on 7.5% SDS-PAGE under reducing conditions, electroblotted onto polyvinylidene difluoride membrane (Millipore, Billerica, MA), and probed with streptavidin-conjugated HRP (GE Healthcare). Protein bands were visualized by enhanced chemiluminescence (ECL) using the ECL kit (GE Healthcare) according to the manufacturer's instructions.
For the cross-linking study, transfectants were incubated in PBS containing 0.5 mg/ml cross-linker 3,3Ј-dithiobis(sulfosuccinimidylpropionate) (DTSSP) (Pierce) for 30 min at room temperature. The reaction was quenched by washing cells in PBS containing 1 mM Tris, pH 7.5. Cells were incubated in DMEM-FH with 3 g each of control mouse IgG or KIM185 for 30 min at 37°C. Unbound antibodies were removed by washing cells twice in DMEM-FH. Cells were lysed, and bound mAbs were precipitated as described above. Precipitated proteins were resolved on 6% SDS-PAGE under nonreducing (Ϫdithiothreitol (ϪDTT) conditions or on 10% SDS-PAGE under reducing (ϩDTT) conditions. The HA-uPAR cross-linked to ␣ M ␤ 2 was detected by immunoblotting with anti-HA antibody. Protein bands were detected using HRP-conjugated donkey anti-rabbit IgG followed by ECL.
For examining integrins with cysteine-lock TMs, transfectants were labeled with sulfo-NHS-biotin as described above. Instead of using reporter mAbs, 3 g of mAb IB4 was added to cell lysate and incubated for 1 h at 4°C to immunoprecipitate the integrins. Thereafter, proteins were precipitated with protein A-Sepharose and resolved on 6% SDS-PAGE under either nonreducing conditions or on 7.5% SDS-PAGE under reducing conditions followed by ECL detection aforementioned. For HA-uPAR coimmunoprecipitation analyses, transfectants were incubated in DMEM-FH with the relevant mAb (3 g each) for 30 min at 37°C. After washing in DMEM-FH, cells were lysed in lysis buffer. The HA-uPAR-integrin complex was precipitated as described above and resolved on 10% SDS-PAGE under reducing conditions. To detect HA-uPAR that co-precipitated with the integrins, blots were probed with anti-HA antibody followed by HRP-conjugated donkey antirabbit IgG. Protein bands were visualized by ECL.
For the detection of actin in cell lysates of transfectants, lysates were immunoblotted with mouse anti-actin antibody followed by HRP-conjugated sheep anti-mouse IgG. Protein bands were visualized by ECL.

RESULTS
Interaction with uPAR Induces the Movement of the ␣ M ␤ 2 Hybrid Domain-The association of uPAR with ␣ M ␤ 2 has been reported (14,17,26,35,40). Here, we further verify the association of uPAR with ␣ M ␤ 2 in 293T transfectants. 293T cells were transiently transfected with ␣ M ␤ 2 alone, ␣ M ␤ 2 with HA-uPAR, or ␣ M ␤ 2 with HA-D2D3 (uPAR without D1). HA-D2D3 was included as a control because we reported previously that D1 of uPAR is critical for its association with ␣ M ␤ 2 (35). Flow cytometry was performed, and the expression of ␣ M ␤ 2 was detected by mAb KIM185 that is ␤ 2 subunit-specific and an activating mAb (32), and the expressions of HA-uPAR and HA-D2D3 were detected by anti-uPAR D2-specific mAb ( Fig. 2A). Comparable levels of ␣ M ␤ 2 were detected in all transfectants, and a high level of HA-uPAR and HA-D2D3 was detected in respective transfectants.
Next we performed chemical cross-linking experiment using the water-soluble and membrane-impermeable compound DTSSP containing two NHS-ester functional groups that react with primary amines and which has a thiol-cleavable spacer with an arm length of 12 Å. The close proximity of uPAR with ␣ M ␤ 2 when they interact would allow cross-linking of both molecules by DTSSP. 293T transfectants were incubated in PBS containing DTSSP followed by immunoprecipitation as described under "Experimental Procedures" (Fig. 2B). Immunoprecipitated proteins were resolved on denaturing SDS-PAGE under nonreducing (ϪDTT) or reducing (ϩDTT) conditions. Under nonreducing conditions an apparent high molecular weight HA-uPAR was detected only in the KIM185 precipitate of ␣ M ␤ 2 /HA-uPAR transfectants. We conjectured that the high molecular weight signal detected for HA-uPAR was because of its association with ␣ M ␤ 2 covalently crosslinked by DTSSP (uPAR ϳ 50 kDa, ␣ M ϳ 170 kDa, ␤ 2 ϳ 95 kDa; uPAR-␣ M ␤ 2 ϳ 315 kDa). When DTT was included, HA-uPAR was dissociated from the KIM185-immunoprecipitated ␣ M ␤ 2 , and it migrated as an ϳ50-kDa protein band. By contrast, HA-D2D3 was not detected in the KIM185 precipitate of ␣ M ␤ 2 /HA-D2D3 transfectants, which was consistent with the requirement of D1 for the interaction of uPAR with ␣ M ␤ 2 (35). Thus, the physical association of uPAR with ␣ M ␤ 2 can be detected in the 293T transfection system.
The integrin headpiece comprises from the ␣ subunit the inserted (I) domain (for I domaincontaining integrins that include ␣ M ␤ 2 ; this domain is also referred to as the A domain), ␤-propeller, and thigh domain and from the ␤ subunit the I-like domain, hybrid domain, plexin-semaphorin-integrin, and integrin epidermal growth factor 1 ( Fig. 1) (24). Three distinct integrin conformations have been reported based on electron microscopy analyses of ␣ L ␤ 2 and ␣ X ␤ 2 . These are the bent conformer, the extended conformer with a closed headpiece, and an extended conformer with an open headpiece (41). One of the key features distinguishing a closed headpiece from an open headpiece lies in the orientation of the hybrid domain. In the closed headpiece the hybrid domain and plexin-semaphorin-integrin are in juxtaposition to the ␣ subunit, and in the open headpiece they "swingout" (42) by 62°in the crystal structure of ligand-mimetic bound platelet integrin ␣ IIb ␤ 3 (43).
We examined whether ␣ M ␤ 2 interaction with uPAR induces conformational change to the integrin headpiece. 293T cells were transfected with ␣ M ␤ 2 or ␣ M ␤ 2 /HA-uPAR. Cell surface expressions of ␣ M ␤ 2 and HA-uPAR were determined by flow cytometry analyses using the mAb KIM185 and mAb VIM5, another uPAR-specific mAb, but it recognizes D1 of uPAR (Fig.  2C). Comparable expressions of ␣ M ␤ 2 were detected in both transfectants, and high levels of HA-uPAR were detected in ␣ M ␤ 2 /HA-uPAR transfectants.
We then performed immunoprecipitation analyses with ␤ 2 integrin activation reporter mAbs. 293T cells expressing ␣ M ␤ 2 or ␣ M ␤ 2 /HA-uPAR were surface-labeled with biotin, and ␣ M ␤ 2 was immunoprecipitated with mAbs MEM148, KIM127, and To detect HA-uPAR and HA-D2D3 expressions, cells were stained with anti-uPAR mAb (D2-specific). Cells were subsequently stained with fluorescein isothiocyanateconjugated secondary antibody followed by flow cytometry analyses. Shaded histograms represent staining with KIM185 or anti-D2 (uPAR). Open histograms, irrelevant mAb. B, cross-linking of ␣ M ␤ 2 and HA-uPAR on 293T transfectants. Cells expressing ␣ M ␤ 2 , ␣ M ␤ 2 with HA-uPAR, or ␣ M ␤ 2 with HA-D2D3 were incubated in PBS containing the chemical DTSSP for 30 min at room temperature. Thereafter, cells were subjected to immunoprecipitation with KIM185 as described under "Experimental Procedures." An irrelevant mAb was included as control. Precipitated proteins were resolved on a 6% SDS-PAGE under nonreducing conditions (ϪDTT) or on a 10% SDS-PAGE under reducing conditions (DTT). Detection of HA-uPAR was performed by anti-HA immunoblotting followed by ECL. C, flow cytometry analyses of 293T transfectants expressing ␣ M ␤ 2 with or without HA-uPAR. mAb KIM185 was employed to detect ␣ M ␤ 2 expression, and mAb VIM5 (D1-specific) was used for HA-uPAR detection. Shaded histograms represent staining with KIM185 or VIM5. Open histograms, irrelevant mAb. D, the conformation of ␣ M ␤ 2 interacting with uPAR was assessed by immunoprecipitation analyses. 293T transfectants expressing ␣ M ␤ 2 or ␣ M ␤ 2 and HA-uPAR were surface-biotinylated, lysed, and immunoprecipitated with mAbs KIM127, MEM148, or KIM185. Immunoprecipitated proteins were resolved on a 7.5% SDS-PAGE gel under reducing conditions. Biotin-labeled ␣ M ␤ 2 subunits were probed with streptavidin-HRP and detected by ECL. E, conformational change in ␣ M ␤ 2 when it interacts with uPAR was assessed by a different approach. 293T transfectants bearing ␣ M ␤ 2 and HA-uPAR were immunoprecipitated with the indicated mAbs, and HA-uPAR that co-precipitated with ␣ M ␤ 2 was detected by immunoblotting with anti-HA antibody. Proteins were resolved on a 10% SDS-PAGE under reducing conditions. IP, immunoprecipitation. uPAR Induces ␣ M ␤ 2 Conformational Changes KIM185. The mAb MEM148 was shown to recognize a neoepitope in the hybrid domain of the ␤ 2 subunit that is masked in the absence of hybrid domain movement (36). Hybrid domain movement is a key feature of an integrin with an open headpiece (24). The mAb KIM127 recognizes a neo-epitope in the I-EGF2 of the ␤ 2 subunit and is masked in the bent conformation but expressed in the extended conformation (33,34). KIM185 was also included in the analysis as a control, and its epitope lies in the ␤2 I-EGF4/␤-tail domain region (44). Transfectants were incubated in media containing these mAbs with or without Mn 2ϩ , which activates ␣ M ␤ 2 (45), for 30 min at 37°C before immunoprecipitation was performed. KIM127 and MEM148 did not immunoprecipitate a significant level of ␣ M ␤ 2 heterodimer from lysate of cells transfected with ␣ M ␤ 2 (Fig.  2D). The ␤ 2 signal detected in the MEM148 sample was attributed to unassociated ␤ 2 as described previously (46). When supplemented with Mn 2ϩ , a high level of ␣ M ␤ 2 was precipitated by KIM127 and MEM148. The control mAb KIM185 precipitated ␣ M ␤ 2 without requirement of Mn 2ϩ treatment. These data showed that under resting conditions, ␣ M ␤ 2 is in a bent conformation with a closed headpiece because of the lack of reactivity of ␣ M ␤ 2 with KIM127 and MEM148. When cells coexpressing ␣ M ␤ 2 and HA-uPAR were analyzed, the profiles of KIM127 and KIM185 samples were similar to that of cells expressing ␣ M ␤ 2 only, but significant ␣ M ␤ 2 signal was detected by MEM148 even without Mn 2ϩ treatment. The addition of Mn 2ϩ further increased the amount of ␣ M ␤ 2 precipitated by MEM148. These data suggest that when uPAR interacts with ␣ M ␤ 2 , the hybrid domain of ␤ 2 was displaced as reported by MEM148, although it was also apparent that the population of ␣ M ␤ 2 with hybrid domain displacement was relatively small in the presence of uPAR interaction.
To further verify that the interaction with uPAR induces hybrid domain movement in ␣ M ␤ 2 , co-immunoprecipitation analysis was performed (Fig. 2E). The rationale of this experiment was that if the interaction with uPAR induces shape changes in ␣ M ␤ 2, uPAR would be co-precipitated with the ␣ M ␤ 2 that has undergone conformational change using the relevant reporter mAb. 293T cells transfected with ␣ M ␤ 2 and HA-uPAR were lysed and subjected to immunoprecipitation with the indicated mAbs. HA-uPAR that co-precipitated was probed and detected with anti-HA immunoblotting. The mAb LPM19c that recognizes an epitope in the ␣ M I domain was also included as a control (47). Expression of HA-uPAR was detected in total cell lysate of transfectants but not untransfected cells. HA-uPAR was not detected in the control Ig lane and was poorly detected in lanes of LPM19c and KIM127. The lack of interacting HA-uPAR with ␣ M ␤ 2 in samples containing LPM19c was consistent with a previous report and possibly due to the masking of ␣ M I domain by the partner uPAR (35). The lack of HA-uPAR co-precipitating with ␣ M ␤ 2 in the sample of KIM127 and the presence of significant HA-uPAR in samples of KIM185 and MEM148 were consistent with that aforementioned. Thus, these data lend support to the concept that uPAR interaction with ␣ M ␤ 2 induces an open headpiece with hybrid domain displacement, but it retains an overall bent conformation.
Interaction with uPAR Induces the Separation of the ␣ M ␤ 2 TMs-We next examined whether interaction of uPAR with ␣ M ␤ 2 could induce the separation or reorientation of the ␣ M ␤ 2 TMs, which is an important event in integrins ␣ L ␤ 2 and ␣ IIb ␤ 3 outside-in signaling (29,38), and this is relevant to the possible mechanism by which uPAR signals through its partner integrins. To test this hypothesis we made use of photobleach FRET detection method by fusing mCFP and mYFP to the C termini of integrin cytoplasmic tails (Fig. 3A) (38). mCFP and mYFP will be referred to as CFP and YFP henceforth. In the integrin ␣ subunit expression construct, a five-amino acid linker was engineered between the mCFP and the last amino acid of the cytoplasmic tail. In the integrin ␤ subunit, a six-amino acid linker was introduced. It was conceived that the close proximity of the cytoplasmic tails, which suggest proximity of the TMs, would allow effective FRET. A decrease in FRET would ensue if the TMs are separated, which leads to the separation of cytoplasmic tails.
When first tested, we observed that the integrin FRET pair ␣ M CFP and ␤ 2 YFP in a K562 transfectant system described previously (38,39) failed to show a significant increase in the fluorescence intensity of CFP after YFP photobleached when compared with cells that were not subjected to photobleach (unbleach) (Fig. 3B). This could be attributed to the marked length difference of the ␣ M and ␤ 2 cytoplasmic tails. We noted that FRET was used to demonstrate constitutive heterodimerization of ␣ M ␤ 2 in adherent CHO and 293T cells and was proposed to be cell type-independent (48). However, in our hands, we were unable to obtain interpretable plasma membrane FRET data because of the uneven morphology of 293T cells. Furthermore, the authors also reported possible conformational differences between ␣ M ␤ 2 expressed in Chinese hamster ovary and 293T cells (48). Thus, we retained the use of nonadherent K562 cells but with re-engineered integrin cytoplasmic tails. Previously, we analyzed FRET of ␣ L ␤ 2 cytoplasmic tails in transfected K562 (39). Hence, we re-engineered ␣ M with its cytoplasmic tail substituted with that of ␣ L having a C terminus CFP, referred to as ␣ ML (Fig. 3A). Cells transfected with ␣ ML CFP␤ 2 YFP showed an increase in CFP signal when YFP was photobleached (Fig. 3B), suggesting proximity of the cytoplasmic tails and, reasonably, the association of the ␣ M ␤ 2 TMs. To provide ease in comparison and with more cells analyzed, we plotted % FRET efficiency as described under "Experimental Procedures," and these plots were used in all subsequent figures. When compared with cells expressing ␣ M CFP␤ 2 YFP, cells expressing ␣ ML CFP␤ 2 YFP showed higher FRETs (Fig. 3C). Unbleached sample was also plotted for comparison. To exclude the possibility that FRET was detected due to microclustering of the integrins in K562 or due to sample preparation, we transfected K562 cells with ␣ ML CFP, ␣ ML YFP, and ␤ 2 . If micro-clustering occurs, it is likely that the close proximity of ␣ ML CFP␤ 2 and ␣ ML YFP␤ 2 will lead to detectable FRET. Included for comparison were cells transfected with ␣ L CFP, ␣ L YFP, and ␤ 2 . Whereas significant FRET was detected for cells expressing ␣ ML CFP␤ 2 YFP, markedly less FRET was detected in cells expressing ␣ ML CFP, ␣ ML YFP, and ␤ 2 and expressing ␣ L CFP, ␣ L YFP, and ␤ 2 . These data suggest that the FRET detected in ␣ ML CFP␤ 2 YFP was due primarily to intramolecular uPAR Induces ␣ M ␤ 2 Conformational Changes cytoplasmic tails interaction rather than microclustering. In addition, we could infer that the TMs of the integrin are in close juxtaposition.
We then analyzed the effect of uPAR on ␣ M ␤ 2 TMs. uPAR contains three domains, D1, D2, and D3 (Fig. 1). We showed previously that D1 is required for uPAR interaction with ␣ M ␤ 2 because the deletion of D1 abrogated association between uPAR and ␣ M ␤ 2 (35). The expression and identity of full-length HA-uPAR and HA-D2D3 were re-assessed in K562 transfectants with a panel of uPAR domain-specific mAbs followed by flow cytometry analyses (Fig. 4A). Although cells expressing full-length HA-uPAR stained positive with all three mAbs, cells bearing HA-D2D3 only showed positive staining with anti-D2 and anti-D3 mAbs. Next, FRET analyses were performed on K562 expressing fulllength HA-uPAR or HA-D2D3 with ␣ ML CFP␤ 2 YFP (Fig. 4B). Cells expressing only the integrins and cells expressing integrins with HA-D2D3 showed comparable levels of FRET. By contrast, a significant diminution of FRET was observed for cells expressing integrins with full-length HA-uPAR. These data suggest the separation of the integrin cytoplasmic tails in uPAR co-expressing cells, and they suggest reorientation or separation of the ␣ M ␤ 2 TMs in these cells.
The reduction in FRET signal of ␣ ML CFP␤ 2 YFP in the presence of HA-uPAR was not an aberrant effect of HA-uPAR on the ␣ L cytoplasmic tail that was engineered into ␣ M because HA-uPAR had minimal effect on the FRET signal of ␣ L CFP␤ 2 YFP on K562 transfectants (Fig. 4C). It was also noted that in ␣ ML CFP␤ 2 YFP, the cytoplasmic tail substitution was made from ␣ M Lys 1114 onward (Figs. 3A and 4D). Recently, NMR studies reveal that the first charged residue Lys on the intracellular side of ␣ IIb and ␤ 3 does not demarcate the C-terminal end of the transmembrane domains (49,50). In ␣ IIb , two residues that are located after Lys 989 , namely Phe 992 and Phe 993 , loop back into the intracellular side of the plasma membrane (49). Comparison of ␣ M and ␣ L membrane proximal sequences reveals an amino acid difference (shaded in black) between ␣ M Lys 1114 -Arg 1120 and ␣ L Lys 1088 -Arg 1094 (Fig. 4D). The ␣ ML CFP may have a disrupted membrane insertion due to the substitution of ␣ M Leu 1115 with Val. This may complicate the analyses of ␣ ML CFP␤ 2 YFP inter-  uPAR Induces ␣ M ␤ 2 Conformational Changes acting with uPAR. Thus, we have generated an additional chimera ␣ ML *CFP in which the cytoplasmic tail exchange between ␣ M and ␣ L was made after the GFFKR sequence. FRET analyses of ␣ ML *CFP␤ 2 YFP in the presence of HA-uPAR or HA-D2D3 were performed. The FRET profile of ␣ ML *CFP␤ 2 YFP (Fig. 4D) was similar to that of ␣ ML CFP␤ 2 YFP (Fig. 4B). Thus, we reasoned that ␣ M Leu 1115 replaced by Val in ␣ ML CFP had minimal precocious effect at least on the ␣ ML CFP␤ 2 YFP FRET studies performed herein.
uPA forms a trimolecular complex with uPAR and ␣ M ␤ 2 , and it initiates Ca 2ϩ signaling in neutrophils (51). The addition of uPA to cells expressing ␣ ML CFP␤ 2 YFP and HA-uPAR did not show any significant FRET difference from cells without uPA, albeit avid uPA binding on these cells (Fig. 4E and data not shown). We went further to examine the effect of uPA on ␣ M ␤ 2 that is associated with HA-uPAR by the method of immunoprecipitation with the reporter mAbs MEM148 and KIM127 (Fig. 4F). The profile of the immunoprecipitation data was similar to that of ␣ M ␤ 2 /HA-uPAR transfectants in the absence of uPA (Fig. 2E). However, we do not preclude uPA inducing further structural changes in the ␣ M ␤ 2 TMs and ectodomain that may not be detected using our present method of FRET analyses and immunoprecipitation assays with the reporter mAbs available to us.
Next, the expression of uPAR is low on ␣ M ␤ 2 -expressing monocytes, but its expression can be upregulated markedly when these cells are treated with Escherichia coli lipopolysaccharide or Mycobacterium tuberculosis lipoarabinomannan (52). In this study the expression of uPAR is comparatively high. We asked whether reduction of uPAR expression would also exert a similar effect on the conformation of ␣ M ␤ 2 , as shown in previous sections but at a reduced level. To this end we performed FRET analyses of K562 expressing ␣ ML CFP␤ 2 YFP and different levels of HA-uPAR (by transfecting the same amount of ␣ ML CFP␤ 2 YFP but different amounts of HA-uPAR). A reduction of HA-uPAR expressed was detected in cell lysates of transfectants with decreasing amounts of plasmid HA-uPAR transfected (Fig. 5A). An anti-actin blot was included as a control. The expressions of ␣ ML CFP␤ 2 YFP in these transfectants were comparable as determined by confocal microscopy analyses (data not shown). When these cells were subjected to FRET analyses, a decrease in HA-uPAR expression had a lesser effect on ␣ ML CFP␤ 2 YFP FRET signal (Fig. 5B). We extended the analyses by performing reporter mAb immunoprecipitation of ␣ M ␤ 2 when co-expressed with HA-uPAR using 293T transfectants. A reduction in HA-uPAR plasmid transfected led to a decrease in HA-uPAR expression in 293T transfectants as determined by flow cytometry analyses using mAb VIM5 and immunoblotting with anti-HA antibody (Fig. 5C). The anti-actin control blot was included. The expressions of ␣ M ␤ 2 in these transfectants were comparable as determined by flow cytometry using KIM185. When immunoprecipitations using mAbs MEM148 and KIM185 were performed with ␣ M ␤ 2 transfectants bearing different levels of HA-uPAR, accordingly a reduction in HA-uPAR was co-precipitated from transfectants expressing reduced amounts of HA-uPAR (Fig. 5D). The reduced level of ␣ M ␤ 2 conformational changes detected with a reduction in uPAR expression may be explained by a lesser number of uPARassociated ␣ M ␤ 2 . The expression index (EI) was calculated by % gated positive ϫ geo-mean fluorescence intensity (46). M1 denotes region gated to be positive staining. The expressions of HA-uPAR in these transfectants were also detected by anti-HA immunoblot and ECL. The anti-actin immunoblot was included as protein loading control. D, 293T cells expressing ␣ M ␤ 2 with varied levels of HA-uPAR were subjected to co-immunoprecipitation analyses with the indicated mAbs. HA-uPAR was detected by anti-HA immunoblotting and ECL. In all immunoblotting experiments above, proteins were resolved on 10% SDS-PAGE under reducing conditions. Disulfide Clasp in the ␣ M ␤ 2 TMs Prevent Their Separation Induced by uPAR-To further validate the separation of the ␣ M ␤ 2 TMs induced by uPAR, we sought to generate disulfide clasp ␣ M ␤ 2 mutants. In these mutants, the extracellular membrane proximal residues of ␣ M ␤ 2 TMs were mutated singly to Cys to allow disulfide bond formation between permissible pairs of introduced Cys as indicated (Fig. 6A). Clasping of TMs by engineered disulfide bonds has been shown to attenuate ␣ IIb ␤ 3 outside-in signaling (29,53). Transfectants bearing the ␣ ML CFP␤ 2 YFP Cys mutants were surface-labeled with biotin, lysed, and immunoprecipitated with the mAb IB4 that is specific to heterodimeric ␤ 2 integrins (35). Integrins precipitated were resolved on SDS-PAGE under nonreducing (ϪDTT) or reducing (ϩDTT) conditions (Fig. 6B). Under nonreducing conditions, high molecular weight protein bands were detected that corresponded to integrin heterodimers that were covalently linked. Except for the cysteine mutant pair that contained ␣ ML L1091C, the ␣ ML P1092C and ␣ ML L1093C showed a propensity to form disulfide-bonded heterodimer with ␤ 2 I679C, ␤ 2 A680C, ␤ 2 A681C, and ␤ 2 I682C. Under reducing conditions, only three of these heterodimers were separated into their individual ␣ ML CFP and ␤ 2 YFP subunits. It was not clear why the others showed a much lower level of individual subunits. However, one of these pairs, ␣ ML L1093C ␤ 2 A680C, was selected for subsequent analyses. This cysteine clasp will be referred to as c-c in subsequent discussion.
Before proceeding to test the effect of c-c clasp on the ␣ M ␤ 2 TMs induced by uPAR, we verified that the clasp did not alter the ligand binding function and conformation of the ␣ M ␤ 2 . K562 cells transfected with wild-type ␣ M ␤ 2 or ␣ M ␤ 2 c-c (Fig.  6C) were allowed to adhere to the ␣ M ␤ 2 ligand fibrinogen (Fig.  6D). Cells expressing ␣ M ␤ 2 c-c showed a similar adhesion profile to cells expressing wild-type ␣ M ␤ 2 with minimal adhesion to fibrinogen in the absence of activation, and a high level of adhesion was detected when the ␤ 2 activating mAb KIM185 was included. Adhesion specificity was demonstrated with the heterodimeric-specific and function-blocking mAb IB4. Thus, the c-c clasp did not affect the cell surface expression and the ligand binding function of the ␣ M ␤ 2 ectodomain.
Next, we analyzed the ectodomain conformation of the ␣ M ␤ 2 c-c co-expressed with HA-uPAR (Fig. 7A). The profile of HA-uPAR co-precipitating with ␣ M ␤ 2 c-c was similar to that with wild-type ␣ M ␤ 2 (Fig. 2E). ␣ M ␤ 2 c-c interacting with HA-uPAR most possibly retained a bent conformation because  and in parentheses). B, assessing the formation of disulfide bond in integrin TM mutants. Cell surface-biotinylated ␣ ML CFP ␤ 2 YFP TM mutants were immunoprecipitated with the ␤ 2 integrins heterodimer-specific mAb IB4. Proteins were resolved on a 6% SDS-PAGE gel under nonreducing conditions (ϪDTT) and on a 7.5% SDS-PAGE gel under reducing conditions (ϩDTT). Proteins were probed with streptavidin-HRP and detected by ECL. The asterisk denotes the pair of integrin TM mutants that were used in subsequent analyses. This disulfide clasp is referred to as c-c henceforth. C, flow cytometry analyses of K562 transfectants expressing ␣ M ␤ 2 or ␣ M ␤ 2 with the disulfide clasp c-c. The mAb used was IB4 (shaded histogram). Open histogram, irrelevant mAb. D, K562 transfectants were examined for their adhesive properties to fibrinogen. For activation of ␣ M ␤ 2 , the activating mAb KIM185 was used. Cell adhesion specificity was demonstrated using mAb IB4, which is also a function-blocking mAb.
KIM127 failed to co-precipitate HA-uPAR, and it adopted an open headpiece conformation because it showed reactivity with MEM148. Thus, the introduction of the c-c clasp had also no apparent effect on the ectodomain conformation of ␣ M ␤ 2 interacting with HA-uPAR.
We then examined whether c-c clasp prevents reorientation of the ␣ M ␤ 2 TMs induced by HA-uPAR (Fig. 7B). K562 transfectants expressing the indicated constructs were subjected to FRET analyses. Cells expressing ␣ ML CFP␤ 2 YFP showed a significant reduction in FRET efficiency in the presence of HA-uPAR co-expression. Cells expressing ␣ ML CFP␤ 2 YFP with c-c clasp showed comparable FRETs to those without the clasp. However, co-expressing HA-uPAR with ␣ ML CFP␤ 2 YFP c-c clasp did not show significant reduction in FRET signal. This was not attributed to poor interaction of HA-uPAR with the c-c clasp integrin because HA-uPAR was co-precipitated effectively with the integrins by KIM185 and MEM148 aforementioned (Fig. 7A). Taken together, these data suggest that the TM c-c clasp prevents the separation or reorientation of the ␣ M ␤ 2 TMs induced by HA-uPAR, lending further support that uPAR interacting with ␣ M ␤ 2 not only induces the opening of the ␣ M ␤ 2 headpiece but the "activation" signal is propagated to the C-terminal halve of the ␣ M ␤ 2 that leads to the reorientation of the TMs.

DISCUSSION
The integrin ␣ M ␤ 2 is a primary receptor that promotes the adhesion and migration of polymorphonuclear leukocytes and mononuclear phagocytes (2). The activity of ␣ M ␤ 2 that is conformation-dependent is regulated at several levels. Extracellular divalent cation Mn 2ϩ stimulates adhesion of monocytes to endothelial cells mediated by ␣ M ␤ 2 and enhances ␣ M ␤ 2 ligand binding (45). From the intracellular compartment, the large cytoskeletal network protein talin interacts with the ␤ 2 cytoplasmic tail to regulate ␣ M ␤ 2 -mediated phagocytosis in the mouse macrophage line (54). On the plasma membrane ␣ M ␤ 2 interacts with Fc␥RIIIB, Fc␥RIIA, low density lipoprotein receptor, and uPAR (13)(14)(15)(16). uPAR is well reported to form a functional complex with ␣ M ␤ 2 in fibrinolysis, and it modulates the activity of ␣ M ␤ 2 (17,18,26,35,55,56). uPAR-deficient mice also showed impaired ␣ M ␤ 2 -mediated recruitment of neutrophils in response to infections (57,58). However, it is not well characterized how interaction with uPAR affects the conformation of the ␣ M ␤ 2 .
Our data suggest that interaction of uPAR with ␣ M ␤ 2 changes the conformation of the integrin headpiece from a closed to an open conformation. A displaced ␤ 2 hybrid domain was detected using the reporter mAb MEM148. Indeed, our observation is in line with a study made on uPAR and integrin ␣ 5 ␤ 1 that showed conformational changes in the integrin using two reporter mAbs HUTS-21 and 9EG7 (22). It was observed that suppressing expression of uPAR by siRNA in human fibrosarcoma and breast cancer cell lines enhanced the epitope expressions of HUTS-21 and 9EG7, which recognize ligandinduced epitopes in ␤1 subunit (22). Thus, it was inferred that uPAR directly affects the conformation of its partner ␣ 5 ␤ 1 . Interestingly, the epitope of HUTS-21 lies in the region spanning residues 355-425 of the ␤1 subunit that is in the hybrid domain (59). The epitope of 9EG7 lies in region 495-602 of the ␤1 integrin epidermal growth factor 2, -3, and -4 (60). Together, these data suggest that conformational changes are transmitted from the headpiece of an integrin involving hybrid domain movement to the C-terminal halve of the integrin ectodomain when it interacts with uPAR.
A key feature of uPAR interaction with an integrin is the overall bent conformation of the integrin because both molecules are juxtaposed on the same membrane, and the interaction sites on both molecules disfavor association of uPAR with an extended integrin as discussed in the Introduction. Indeed, we were unable to detect extended ␣ M ␤ 2 interacting with uPAR using the reporter mAb KIM127, although association of uPAR with ␣ M ␤ 2 was verified using KIM185. However, uPAR induces displacement of the ␣ M ␤ 2 hybrid domain, which may be remi- FIGURE 7. The disulfide clasp prevented reorientation or separation of the ␣ M ␤ 2 TMs but did not attenuate the conformational changes in the ␣ M ␤ 2 ectodomain induced by uPAR. A, the ectodomain conformation of the ␣ M ␤ 2 c-c interacting with HA-uPAR was analyzed by co-immunoprecipitation analyses. Cell lysate of transfectants were immunoprecipitated with the mAbs indicated, and HA-uPAR was detected by immunoblotting with anti-HA antibody. Proteins were resolved on a 10% SDS-PAGE gel under reducing conditions. IP, immunoprecipitation. B, FRET analyses of transfectants co-expressing the indicated integrins with or without the disulfide clasp (c-c) and with or without HA-uPAR. Data are representative of three independent experiments. Data show the mean Ϯ S.E. for 20 cells. Student's t test, assuming unequal variance, was used for statistical analyses. *, p Ͻ 0.001. uPAR Induces ␣ M ␤ 2 Conformational Changes SEPTEMBER 12, 2008 • VOLUME 283 • NUMBER 37 JOURNAL OF BIOLOGICAL CHEMISTRY 25401 niscent to that of a ligand-bound ␣ 5 ␤ 1 (61) and ␣ IIb ␤ 3 (43). Thus, uPAR may be considered as an in cis ligand of ␣ M ␤ 2 .
An interesting observation made in this study is the movement of the ␣ M ␤ 2 TMs when it is associated with uPAR. This is inferred from the FRET analyses of ␣ M ␤ 2 with FRET pair fluorophore fused to the cytoplasmic tails, which report movement of the TMs. We were unable to detect significant FRET in ␣ M ␤ 2 with its native cytoplasmic tails; hence, we have made substitution of ␣ M cytoplasmic tail with that of ␣L, which was successful in the FRET system we employed previously (39). We acknowledge that there can be a limitation in the use of ␣ L cytoplasmic tail in an ␣ M construct to report ␣ M ␤ 2 conformational changes; however, we reasoned that this approach is still relevant and useful in demonstrating TM separation of ␣ M ␤ 2 . The reorientation of TMs in ␣ M ␤ 2 interacting with uPAR is further verified by clasping the TMs with a disulfide bond. This also eliminates the possibility of uPAR directly inducing spatial changes in the integrin cytoplasmic tails via other factors, which would complicate the readout and the interpretation of the data with regards to TM movement. The separation of the integrin TMs is important for outside-in signaling as shown directly in ␣ IIb ␤ 3 (29). Thus, our data suggest that uPAR changes the ectodomain conformation of its partner integrin, which is sufficient to trigger movement of the integrin TMs as illustrated in the model (Fig. 8). However, we should be cautious with the interpretation because the reorientation of the TMs or their complete separation when an integrin is activated remains to be fully characterized by structural studies. Nonetheless, the movement of the integrin TMs induced by uPAR can be one of the mechanisms by which uPAR trigger integrinmediated cytosolic signaling. Other mechanisms may involve dimerization of uPAR and its potential to partition into membrane rafts (62).
Our data and that of others (22) suggest that uPAR induces shape changes in its integrin partner that retain an overall bent conformation. Although integrin extension is one of the hallmarks of integrin activation, there are also reports of activated bent integrins (24,25,41,(63)(64)(65). A model was proposed for non-I domain-containing integrins in which engagement of a bent integrin with its ligand induces a certain degree of unbending, and in I domain containing integrins the swing out of the hybrid domain is an event of outside-in signaling in the deadbolt model of activation (25). Thus, it will be interesting to obtain structural data of an uPAR-integrin complex in future work to obtain direct measurements of these conformational changes. These will add to the ensemble of conformations adopted by the integrins under different physiological conditions that are required for fine-regulating integrin-mediated biological processes.