Proteomics-based discovery of a novel, structurally unique, and developmentally regulated plasminogen receptor, Plg-R_KT, a major regulator of cell surface plasminogen activation

Localization of plasminogen on cell surfaces is a crucial control point for positive regulation of cell surface plasmin proteolytic activity that facilitates both physiologic and pathologic processes,^1,2  including macrophage recruitment during the inflammatory response,^3-6  tissue remodeling,⁷  wound healing,^8,9  tumor cell invasion and metastasis,^10-12  skeletal myogenesis,¹³  neuroendocrine prohormone processing,^14,15  and neurite outgrowth.^16,17  Cell surface plasminogen binding sites promote plasminogen activation by reducing the Km (11- to 60-fold) for plasminogen activation.^18-24  Active plasmin also associates with the cell surface, where its activity is protected from inhibitors.^25,26 

Plasminogen binding sites are very broadly distributed on both eukaryotic and prokaryotic cells.²⁷  Of the many eukaryotic cells examined to date, only erythrocytes do not bind plasminogen.²⁸  The interactions of plasminogen with eukaryotic cells are mediated by lysine binding sites within the disulfide-bonded kringle domains of plasminogen.^18,29  Therefore, plasminogen binding to eukaryotic cells is blocked in the presence of lysine and lysine analogs, including ϵ-aminocaproic acid (EACA).²⁷  Because most cell types have a very high capacity for plasminogen, no single molecule can account for the entire plasminogen binding capacity of a given cell type.²⁷  However, a subset of plasminogen binding proteins exposing C-terminal basic residues on cell surfaces are predominantly responsible for the ability of eukaryotic cells to enhance plasminogen activation because carboxypeptidase B (CpB) treatment abrogates cell surface–dependent plasminogen activation.²⁴  Correspondingly, plasminogen-dependent macrophage recruitment in vivo is mediated by CpB-sensitive plasminogen receptors, and plasminogen binding to recruited macrophages is increased, compared with peripheral blood monocytes.^6,30  Therefore, we probed the monocyte proteome as a source of an inducible integral membrane plasminogen receptor(s) exposing a C-terminal basic residue on the cell surface.

Several plasminogen binding proteins with established intracellular functions that are synthesized with C-terminal lysines associate with the monocytoid cell surface (eg, α-enolase,^29,31  TIP49a,³²  histone H2B, and p11³³ ). Other functional plasminogen binding proteins that are not synthesized with C-terminal basic residues are present on monocytoid cells, including annexin II,³⁴  amphoterin,³⁵  tissue factor,³⁶  and α_Mβ₂.³⁷  However, no integral membrane plasminogen binding proteins that are synthesized with C-terminal basic residues have been identified to date. The existence of a receptor with such a structure would constitute a novel mechanism for stimulating plasminogen activation because its induction would endow cells with the ability to bind plasminogen and promote plasminogen activation, without requiring release and rebinding of intracellular proteins or proteolytic cleavage of a membrane protein to reveal C-terminal basic residues. Therefore, we used the exquisite sensitivity of multidimensional protein identification technology (MudPIT) to search for an integral membrane plasminogen receptor(s), exposing a C-terminal basic residue on the cell surface and up-regulated during differentiation.

Glu-plasminogen was purified from fresh human blood as described.^38,39  Lys-plasminogen was from Enzyme Research Laboratories. Single-chain tissue plasminogen activator (t-PA) was from Calbiochem. Polyclonal antibodies were raised in rabbits and monoclonal antibodies were raised in rats and mice against the synthetic peptide, CEQSKLFSDK, thiol coupled to keyhole limpet hemocyanin. Antibodies were selected for direct binding to immobilized CEQSKLFSDK coupled to bovine serum albumin (BSA) and for the ability to inhibit specific plasminogen binding to CEQSKLFSDK. Anti–α-enolase monoclonal antibody (mAb) 9-C12³¹  and polyclonal antiplasminogen⁴⁰  were prepared in our laboratory. Anti–urokinase plasminogen activator receptor (uPAR; product no. 3936) was from American Diagnostica. Fluorescein isothiocyanate (FITC)–conjugated CD marker antibodies and the relevant FITC-conjugated isotype control antibodies were from Pharmingen. The mouse macrophage differentiation antibody rat anti–mouse F4/80 FITC conjugate was from Serotec. Goat anti–mouse FITC-conjugated polyclonal antibody and goat anti–rabbit FITC-conjugated polyclonal antibody were from Calbiochem. Alexa 488-F(ab′)₂ of goat anti–rabbit immunoglobulin G (IgG) and Alexa 568-F(ab′)₂ fragment of goat anti–mouse IgG were from Invitrogen.

For differentiation experiments, monocytes were further isolated from the mononuclear cell population by plating onto tissue culture dishes (Corning), before the addition of human recombinant macrophage colony stimulating factor (M-CSF; Calbiochem).

Hoxa9-ER4 cells were cultured as described⁴¹  and were differentiated either with murine M-CSF (Calbiochem) or with conditioned media produced by LADMAC cells (ATCC), as a source of M-CSF.

Quantitative flow cytometric equilibrium binding of fluorescein isothiocyanate (FITC)–plasminogen to monocytoid cells was analyzed using beads impregnated with FITC as described.⁴²  Briefly, the output from the flow cytometer was standardized into mean equivalent standard fluorescence units using beads impregnated with different mean equivalent standard fluorescence units of FITC. The fluorescence intensity change of the EACA-inhibitable FITC-plasminogen conformational change (Q = ΔI_f/ΔI_i; where I_i and I_f are the initial and final fluorescence intensities of FITC-plasminogen, respectively) was approximated using a F500 fluorometric plate reader (Bio-Rad Model 680; Bio-Rad) equipped with a FITC filter set (excitation, 490 nm; emission, 520 nm). Binding parameters were determined using the program, NLREG, Version 4.1. (Phillip H. Sherrod; www.nlreg.com)

Confocal images were captured using a Zeiss 710 laser confocal scanning microscope (LCSM), with a 63 × objective (1.4 na), running the latest Zen 2009 Zeiss software suite (Carl Zeiss Inc). All images were then imported and further analyzed for quantitative colocalization using 2 independent software packages: LSM examiner (Zeiss Inc) and ImageJ (National Institutes of Health [NIH] imaging; http://rsb.info.nih.gov/ij). Colocalization between fluorescently labeled aggregates of Plg-R_KT with either uPAR or plasminogen was quantified by obtaining the threshold range of real over background signal and then using the average real threshold range to calculate the correlation coefficients (M values) of at least 40 cells in 2 separate experiments. To define the number and size of each labeled aggregate, images were imported into Image Pro Plus (Media Cybernetics Inc) where each cell was outlined and a similar threshold range was used to define a real signal within each cell. Once this range was defined the software then automatically extracts parameters such as area, perimeter total number, and average fluorescence intensity of the fluorescently labeled aggregates per cell.

Progenitor and M-CSF–differentiated Hoxa9-ER4 cells (5 × 10⁸) were separately biotinylated, using EZ-Link Amine-PEO₃-Biotin (Pierce). The cells were then subjected to dead cell removal on annexin V–coated magnetic microspheres (Miltenyi) that resulted in a 99% enrichment of viable cells (as determined in fluorescence-activated cell sorting [FACS] analysis with propidium iodide and annexin V). Membrane fractions were prepared from the viable cells by Dounce homogenization in the presence of Complete Protease Inhibitor Cocktail (Roche) in Invitrosol (Invitrogen), followed by centrifugation steps as used in our laboratory^32,43  and 3 mg was applied to a 1-mL plasminogen-Sepharose affinity column as described.²⁹  The column was washed in phosphate-buffered saline (PBS) containing 1 × Invitrosol until no protein was detected at 280 nm followed by elution with the washing buffer containing 0.2M of EACA. The eluant from the plasminogen-Sepharose column was incubated with 50 μL of immobilized avidin for 30 minutes at 4°C. Proteins bound to the immobilized avidin were resuspended in 5 μL of Invitrosol and heated at 60°C for 5 minutes. Then, 45 μL of 80% acetonitrile was added and the samples were digested with trypsin at 37°C for 18 hours. After 24 hours, the solvent was evaporated in a speedvac (ThermoFisher Scientific Model), and peptides were dissolved in 50 μL of buffer A (95% H₂O, 5% acetonitrile, and 0.1% formic acid).

The protein digest was subjected to MudPIT (reviewed in Eng et al⁴⁴ ). Peptide mixtures were resolved by strong cation exchange liquid chromatography followed by reversed-phase liquid chromatography.^45-48  Eluting peptides were electrosprayed into an LTQ ion trap mass spectrometer equipped with a nano–liquid chromatography electrospray ionization source (ThermoFinnigan). Full mass spectrometry (MS) spectra were recorded over a 400- to 1600-m/z range, followed by 3 tandem mass spectrometry (MS/MS) events sequentially generated in a data-dependent manner on the first, second, and third most intense ions selected from the full MS spectrum (at 35% collision energy). Mass spectrometer scan functions and high-performance liquid chromatography solvent gradients were controlled by the Xcalibur data system (ThermoFinnigan).

Tandem mass spectra were extracted from raw files, and a binary classifier,⁴⁹  previously trained on a manually validated dataset, was used to remove low-quality MS/MS spectra. Remaining spectra were searched against a Mus musculus protein database containing 50 370 protein sequences downloaded as FASTA-formatted sequences from European Bioinformatics Institute's International Protein Index (EBI-IPI; database Version 3.23, released on November 2, 2006), and 124 common contaminant proteins, for a total of 66 743 target database sequences.^50,51  To calculate confidence levels and false-positive rates, we used a decoy database containing the reverse sequences of the 66 743 proteins appended to the target database, and the SEQUEST algorithm⁵²  to find the best matching sequences from the combined database.

SEQUEST searches were done on an Intel Xeon 80-processor cluster running under the Linux operating system. The peptide mass search tolerance was set to 3 Da. No differential modifications were considered. No enzymatic cleavage conditions were imposed on the database search, so the search space included all candidate peptides whose theoretic mass fell within the 3-Da mass tolerance window, despite their tryptic status.

The validity of peptide/spectrum matches was assessed in DTASelect2⁵³  (Scripps Research Institute; http:/fields.scripps.edu/?q=content/software) using SEQUEST-defined parameters, the cross-correlation score (XCorr), and normalized difference in cross-correlation scores (DeltaCN). The search results were grouped by charge state (+1, +2, and +3) and tryptic status (fully tryptic, half-tryptic, and nontryptic), resulting in 9 distinct subgroups. In each one of the subgroups, the distribution of XCorr and DeltaCN values for direct and decoy database hits was obtained, and the 2 subsets were separated by quadratic discriminant analysis. Outlier points in the 2 distributions (for example, matches with very low Xcorr but very high DeltaCN) were discarded. Full separation of the direct and decoy subsets is not generally possible; therefore, the discriminant score was set such that a false-positive rate of 5% was determined based on the number of accepted decoy database peptides. This procedure was independently performed on each data subset, resulting in a false-positive rate independent of tryptic status or charge state.

Either the C-terminal peptide of Plg-R_KT (CEQSKLFSDK coupled to BSA), the reverse peptide (KDSFLKSQEC), or BSA alone was coated onto wells of microtiter plates at 10 μg/mL. The wells were blocked with 5% BSA. Either Glu-plasminogen or Lys-plasminogen was then incubated with the wells, followed by antiplasminogen mAb51 raised in our laboratory,⁵⁴  and detection with horseradish peroxidase (HRP)–conjugated goat anti–mouse IgG (Biosource). Binding of single-chain t-PA was determined using anti–t-PA mAb (Calbiochem) followed by HRP-conjugated goat anti–mouse IgG (Biosource).

Cells (8 × 10⁴/mL) were preincubated with rat anti–Plg-R_KT mAb 35B10 or rat IgG2a for 30 minutes at 37°C. Glu-plasminogen (2.7 μM) was then added and incubated for an additional 30 minutes at 37°C. Then 20 nM of single-chain recombinant t-PA was added. Plasmin activity was measured after 3 minutes by diluting the reaction mixture 1:10 into S-2251 (DiaPharma Group) to a final concentration of 1mM and monitoring absorbance at 405 nm as described previously.^24,42 

Immuno-Fluore Mounting Medium was from ICN Biomedicals Inc.

To establish a system whereby plasminogen receptors could be induced, we examined the effect of differentiation on plasminogen binding to monocytes. Peripheral blood monocytes were obtained from mononuclear cells isolated from freshly donated human blood as described.⁵⁵  In FACS analysis, as described,¹⁰  the cells were identified by their distinct forward scatter/side scatter properties compared with the other cells within the mononuclear cell preparations. Plasminogen binding was analyzed by dual-color FACS analysis using FITC-plasminogen and propidium iodide to gate for only viable cells, as described.¹⁰  Specific binding was defined as binding that was inhibited by EACA. The viable peripheral blood monocytes bound plasminogen specifically, as defined by the shift in fluorescence in the presence of EACA (Figure 1A left-most column, [FITC-plasminogen]). Monocytes were further purified from the mononuclear cell preparation by adherence as described under “Cells” and then treated with the differentiation-inducing cytokine, M-CSF. The fluorescence signal indicating specific FITC-plasminogen binding to the viable M-CSF–differentiated cells was 11-fold greater than that observed with the unstimulated monocytes (compare Figure 1A and B left-most columns [FITC-plasminogen]). The CD antigen profiles were consistent with differentiation into macrophages (compare Figure 1A and B), and the morphology of the cells changed from a rounded shape to a spindle-type morphology (data not shown). Thus, differentiation of the monocytes toward the macrophage phenotype markedly up-regulated plasminogen binding.

Our search for an integral membrane plasminogen receptor(s), exposing a C-terminal basic residue on the cell surface and up-regulated during monocyte differentiation, required large quantities of cells. Therefore, we tested whether the Hoxa9-ER4 cell line would behave as peripheral blood monocytes with respect to plasminogen binding capacity. The Hoxa9-ER4 cell line is derived from primary murine bone marrow myeloid precursors immortalized with an estrogen-regulated conditional oncoprotein, Hoxa9-ER4.⁵⁶  The Hoxa9-ER4 line is factor dependent (granulocyte-macrophage colony-stimulating factor), and differentiates to monocytes when estrogen is removed from the medium, thereby inactivating the Hoxa9-ER protein. The mature monocytes respond to M-CSF.⁴¹  We compared plasminogen binding to undifferentiated and M-CSF–differentiated Hoxa9-ER4 progenitor cells. No specific binding of plasminogen to the Hoxa9-ER4 progenitor cells was detected (Figure 1C left-most column [FITC = Plasminogen]). However, as the cells differentiated along the monocytic pathway, specific plasminogen binding to the cells was observed (Figure 1D left-most column [FITC = Plasminogen]). The CD antigen profiles were consistent with differentiation into the macrophage phenotype (compare Figure 1 C and D).

Binding isotherms were constructed for the specific interaction of plasminogen with the monocytoid cells using dual-color quantitative flow cytometry (Figure 2). FACS data from the gated viable cell populations that were treated with different concentrations of FITC-plasminogen, in the presence or absence of EACA, were used to derive the apparent K_d and B_max of the interactions by fitting these data to the single site binding equation (LR) = ([L]B_max)/([L]+K_d). (The FACS signal that was inhibitable by EACA was calculated as specific binding.) The plasminogen binding isotherms with M-CSF–treated monocytes and M-CSF–treated Hoxa9-ER4 cells were saturable (Figure 2A-B). The binding parameters determined for the M-CSF–differentiated Hoxa9-ER4 cells were apparent K_d = 0.95 μM and B_max = 1.9 × 10⁵ sites/cell. An apparent K_d of 0.57 μM and a B_max of 9.1 × 10⁵ molecules/cell were calculated for M-CSF–treated monocytes. In contrast, with peripheral blood monocytes, the specific plasminogen binding isotherms did not reach saturation at a plasminogen concentration of 10μM (Figure 2A), suggesting that the K_d is more than 5μM and the B_max is more than 4 × 10⁵ molecules/cell. (Based on the data in Figure 2A, at blood and extracellular concentrations of plasminogen [2 μM⁵⁷ ], ∼ 10⁵ molecules of plasminogen are predicted to be bound to each circulating peripheral blood monocyte, whereas ∼ 5 × 10⁵ molecules of plasminogen are predicted to be bound per macrophage.) Thus, differentiation increased both the affinity and capacity of monocytes and progenitor cells for plasminogen.

We examined whether the M-CSF up-regulated plasminogen binding sites were accessible to CpB. The M-CSF–differentiated Hoxa9-ER4 cells were incubated with 100 U/mL CpB and washed, before determining FITC-plasminogen binding to the viable cells in FACS analysis. CpB treatment decreased specific FITC-plasminogen binding to the viable cells by 78%, consistent with our results with other monocytoid cells.^24,29,58 

We used MudPIT to probe the membrane proteome of M-CSF–differentiated Hoxa9-ER4 cells for the presence of a regulated integral membrane plasminogen receptor(s), exposing a C-terminal basic residue on the cell surface. Intact M-CSF–differentiated Hoxa9-ER4 cells were biotinylated as described in “Plasminogen receptor isolation.” Initially, using FACS analysis we verified that (1) viable cells were effectively biotinylated using antibiotin FITC-labeled monoclonal antibodies, (2) specific plasminogen binding by the cells was not decreased when cells were treated with the biotinylating reagent, and (3) biotinylated cells exhibited the same reduction in plasminogen binding capacity in response to CpB as nonbiotinylated cells, as criteria that the biotinylated plasminogen receptors behaved as native plasminogen receptors. The cell membranes were then isolated, subjected to affinity chromatography on plasminogen-Sepharose, bound to immobilized avidin, digested with trypsin, and subjected to MudPIT analysis, as described in “Multidimensional chromatography and tandem mass spectrometry.” Using this method, only 1 protein with a predicted transmembrane sequence and a C-terminal basic residue was identified: the hypothetical protein, C9orf46 homolog (IPI00136293), homologous to the protein predicted to be encoded by human chromosome 9, open reading frame 46. Table 1 lists the peptides corresponding to C9orf46 homolog that were obtained when the membrane fraction from M-CSF–treated Hoxa9-ER4 cells was subjected to MudPIT. Our isolation of peptides corresponding to C9orf46 homolog is, to our knowledge, the first demonstration of the existence of this protein. We have designated the protein, Plg-R_KT, to indicate a plasminogen receptor with a C-terminal lysine and having a transmembrane domain.

The C9orf46 homolog/Plg-R_KT murine DNA sequence encodes a protein of 147 amino acids with a calculated molecular mass of 17 261 Da (Figure 3A top line). Notably, a C-terminal lysine is encoded. We blasted the C9orf46 homolog/Plg-R_KT sequence against all species using NCBI Blast⁵⁹  and obtained unique human, rat, dog, and cow predicted orthologs, with high homology (eg, human vs mouse = 94% similarity), high identity, and no gaps in the sequence (Figure 3A and Table 2). Of key importance, a C-terminal lysine was predicted for all of the mammalian orthologs obtained in the blast search (Figure 1A). In a query of the Ensembl Gene Report,⁶⁰  DNA sequences of all 9 other fully sequenced mammalian orthologs (Table 3) and 13 other partially sequenced mammalian orthologs encoded C-terminal lysines.

In addition, the primary sequence of C9orf46/Plg-R_KT is apparently tightly conserved in humans, with no polymorphisms (cSNPs) within the 6 exons encoded by the gene (on chromosome 9p24.1) yet identified in the NCBI human genome sequence variation database (dbSNP, http://www.ncbi.nlm.nih.gov/SNP).⁶¹ 

The C9orf46 homolog/Plg-R_KT sequence was analyzed in the TMpred site (Swiss Institute of Bioinformatics; http://www.ch.embnet.org/software/TMPRED_form.html). The strongly preferred model predicted 2 transmembrane helices extending from F₅₃-L₇₃ (secondary helix, oriented from outside the cell to inside) and P₇₈-Y₉₉ (primary helix, oriented from inside the cell to outside; Figure 3B). The 2 transmembrane helices with the same orientation were also predicted using the HMMTOP⁶²  site (Institute of Enzymology, Biological Research Center, Hungarian Institute of Science; http://www.enzim.hu/hmmtop. This model takes into account the “positive inside” rule in which positively charged residues in membrane proteins (eg, residues K₇₄R₇₅K₇₆K₇₇) are found predominantly on the cytoplasmic side of the membrane.⁶³  Notably, a highly positively charged loop (residues 74-77) between the 2 predicted transmembrane domains is conserved in the predicted protein sequences of all mammalian orthologs for which either full or partial sequences are available (eg, Table 3). When analyzed in the Split 4.0 Server site (Membrane Protein Secondary Structure Prediction Server; http://split.pmfst.hr/split/4),⁶⁴  only the region extending from F₈₀-Q₉₄ was identified as a predicted transmembrane helix. Thus, the possibility that the N-terminus may be embedded in the cytoplasm cannot be excluded. Nonetheless, the conservation of the highly charged positive loop (residues 74-77) strongly favors the model in which F₅₃-L₇₃ also is a transmembrane domain (with the N-terminus outside) because the presence of positively charged residues at the C-terminus of apolar domains in proteins favors the insertion of the upstream apolar domains into the cell membrane.⁶⁵ 

Several key aspects of our experimental paradigm support the topologic model shown in Figure 3B. First, recovery of peptides corresponding to C9orf46 homolog relied on accessibility to the biotinylation reagent in the context of intact cells, supporting the exposure of Plg-R_KT domains on the cell surface. The biotinylation reagent we used reacts with the carboxyl moieties of the R-groups of either Asp or Glu. Thus, biotinylation of Plg-R_KT on intact cells would not occur if the highly basic loop between the transmembrane helices (K₇₄-K₇₇) was exposed on the cell surface. Hence, a model in which both the amino and C-termini are on the cytoplasmic face of the membrane can be excluded.

Second, a key prediction from the model in Figure 3B is the presence of a 48 amino acid C-terminal tail with a C-terminal lysine exposed on the cell surface. We addressed this prediction experimentally. When intact M-CSF–treated Hoxa9-ER4 cells were incubated with CpB before subcellular fractionation, purification, and MudPIT analysis, no peptides corresponding to C9orf46 homolog were detected. These data are consistent with exposure of the C-terminal lysine of Plg-R_KT on the cell surface and, hence, accessibility to CpB,³²  so that CpB treatment of intact cells resulted in loss of the ability of Plg-R_KT to bind to the plasminogen-Sepharose column.

To assess whether expression of C9orf46 homolog/Plg-R_KT was increased during cellular differentiation, biotinylated membrane fractions from undifferentiated progenitor Hoxa9-ER4 cells were subjected to the purification method described in “Plasminogen receptor isolation.” In the absence of M-CSF treatment, no peptides corresponding to C9orf46 homolog/Plg-R_KT were detected, suggesting that membrane expression of Plg-R_KT was markedly up-regulated by M-CSF treatment.

To further assess the M-CSF–responsiveness and membrane localization of Plg-R_KT we raised a monoclonal antibody in rats against the synthetic peptide, CEQSKLFSDK (corresponding to the 9 C-terminal amino acids of murine Plg-R_KT with an amino terminal cysteine added for coupling). Membrane and cytoplasmic fractions from progenitor and M-CSF–differentiated Hoxa9-ER4 cells were electrophoresed and Western blotted with either anti–Plg-R_KT mAb or isotype control. A specific immunoreactive band migrating with an Mr_app of approximately 17 000, consistent with the predicted molecular mass of C9orf46 homolog/Plg-R_KT, was detected in membranes from M-CSF–differentiated cells (Figure 4A), clearly demonstrating the existence of this novel protein. The protein was not detected in membrane fractions from progenitor Hoxa9-ER4 cells or in cytoplasmic fractions from either progenitor or differentiated cells (Figure 4A). These data further demonstrate the responsiveness of the C9orf46 homolog gene to a maturation-inducing signal, and verify the presence of C9orf46 homolog/Plg-R_KT in the cell membrane.

To test the prediction that Plg-R_KT is an integral membrane protein, M-CSF–differentiated Hoxa9-ER4 cell membranes were subjected to phase separation in Triton X-114 as described.^66,67  In this technique, integral membrane proteins form mixed micelles with the nonionic detergent and are recovered in the Triton X-114 detergent phase, whereas hydrophilic proteins remain in the aqueous phase. An immunoreactive band migrating with a Mr_app of approximately 17 000 was detected in the detergent phase in Western blotting with anti–Plg-R_KT mAb, but was not detected in the aqueous phase (Figure 4B). These data further support the prediction that Plg-R_KT is an integral membrane protein.

Confocal microscopy was performed to examine the subcellular expression of Plg-R_KT and to examine whether Plg-R_KT colocalized with uPAR, an additional key component of the cell surface plasminogen activation system. Plg-R_KT was immunodetected in small aggregates (∼ 0.2μm²) dispersed over the cell surface (Figure 5A green). uPAR was also immunodetected in small aggregates (∼ 0.2 μm²) dispersed over the cell surface (Figure 5A red), consistent with published data.⁶⁸  Plg-R_KT colocalized with uPAR as revealed by merged images (Figure 5A). The extent of colocalization of Plg-R_KT with uPAR was 73% (± 3%). These results suggest that Plg-R_KT and uPAR are present in very close proximity on the cell surface in an orientation to promote plasminogen activation.

To determine whether the epitope recognized by anti–Plg-R_KT mAb functioned as a plasminogen binding site, the cells were either preincubated with buffer or plasminogen, fixed, and then stained with polyclonal antiplasminogen or anti–Plg-R_KT mAb. As with the polyclonal anti–Plg-R_KT, Plg-R_KT was immunodetected in small aggregates (0.16 ± 0.03 to 0.26 ± 0.03 μm²) dispersed over the cell surface (Figure 5B red). However, after preincubation with plasminogen, the ability to immunodetect Plg-R_KT was reduced by half (Figure 5B-C). These results suggest that plasminogen and the anti–Plg-R_KT mAb (directed against the C-terminal peptide) share binding sites on the cell and that plasminogen binds to the C-terminal domain of Plg-R_KT on the cell surface.

To further explore the functional importance of the C-terminal lysine, we tested whether the synthetic peptide, corresponding to the C-terminus of Plg-R_KT could bind plasminogen. The peptide, CEQSKLFSDK, was coupled to BSA and then coated onto wells of microtiter plates. Glu-plasminogen was incubated with the wells, followed by our antiplasminogen mAb⁵⁴  and detection with HRP-conjugated goat anti–mouse IgG. Glu-plasminogen bound to the peptide in a concentration-dependent manner, achieving half saturation at a concentration of 7.6nM (Figure 6A). The binding was specific because it was blocked in the presence of EACA, consistent with the ability of EACA to inhibit plasminogen binding to differentiated Hoxa9-ER4 cells (Figure 1D). Lys-plasminogen also bound specifically to the peptide (Figure 6B) and the concentration for half saturation was less than 2.7nM, consistent with the higher affinity of Lys-plasminogen compared with Glu-plasminogen for cell surfaces.^69,70  We investigated the interaction of t-PA with the C-terminal peptide because t-PA and plasminogen share binding sites on monocytoid cells and t-PA binding to monocytoid cells is sensitive to CpB.⁵⁸  Concentration-dependent binding of t-PA to the peptide was observed (Figure 6C) and the concentration for 50% saturation was 3.2nM, consistent with the relative affinities obtained when comparing Glu-plasminogen and t-PA binding to cell surfaces.⁷¹  We noted that the concentration for 50% saturation of plasminogen binding to immobilized CEQSKLFSDK coupled to BSA was much less than the K_d value we had determined for plasminogen binding to cells. These differences in apparent affinity when plasminogen binds to the immobilized peptide compared with the cell surface may be due to our use of BSA-conjugated peptide to coat the plate. If multiple peptides were conjugated to the BSA, that may provide a higher affinity surface than the cell surface receptor, because plasminogen has multiple kringle domains that may interact in a cooperative lysine-dependent manner with multiple Plg-R_KT peptides on a single BSA molecule. To resolve this issue, we tested the ability of the soluble C-terminal peptide to inhibit Glu-plasminogen binding under solution phase equilibrium conditions. The soluble peptide competed for Glu-plasminogen binding in a dose-dependent manner with an IC50 of 2μM (Figure 6D), similar to the K_d values we had determined for Glu-plasminogen binding to both human M-CSF–treated monocytes and M-CSF–treated Hoxa9-ER4 cells (see “Plasminogen binding is enhanced when monocytes differentiate”). In addition, a mutated peptide with the C-terminal lysine substituted with alanine did not compete for plasminogen binding at concentrations up to 1mM (Figure 6D), further supporting the role of the C-terminal lysine in the interaction of Plg-R_KT with plasminogen.

We found that plasminogen activation was stimulated 12.7-fold in the presence of M-CSF–treated Hox9-ER4 cells, compared with the reaction in the absence of cells and that cell-dependent plasminogen activation was stimulated 4.4-fold on differentiated cells, compared with undifferentiated cells (Figure 7). To verify the role of Plg-R_KT in plasminogen activation, we tested the effect of a monoclonal antibody (anti–Plg-R_KT mAb 35B10) raised in rats against the synthetic peptide, CEQSKLFSDK. The IgG fraction reacted with the C-terminal peptide of murine Plg-R_KT, blocked plasminogen binding to CEQSKLFSDK (Figure 6E), and specifically recognized a protein with a Mr_app of approximately 17 000 in Western blotting of M-CSF–treated Hoxa9-ER4 cells (Figure 4). In FACS analysis of M-CSF–treated Hox9-ER4 cells, anti–Plg-R_KT mAb 35B10 reacted specifically with the cell surface giving a mean fluorescence intensity value of 93.2 compared with the mean fluorescence intensity value for the rat IgG_2a isotype control of 32.8. Anti–Plg-R_KT mAb 35B10 substantially suppressed cell-dependent plasminogen activation by 46% and suppressed cell differentiation–dependent plasminogen activation by 58% (Figure 7). In controls, plasminogen activation in the absence of cells or on undifferentiated cells was not affected by anti–Plg-R_KT mAb.

Localization of plasminogen on cell surfaces is a crucial control point for positive regulation of cell surface plasmin proteolytic activity that facilitates both physiologic and pathologic processes.^1,2  Plasminogen binding sites, with C-terminal basic residues, are predominantly responsible for the ability of eukaryotic cells to bind plasminogen and promote plasminogen activation.²⁴  In the current study, we identified a novel protein, Plg-R_KT, the first integral membrane plasminogen receptor, synthesized with a C-terminal lysine.

Our results support a mechanism in which Plg-R_KT binds plasminogen via an interaction requiring the C-terminal lysine of Plg-R_KT that is exposed on the cell surface. When intact, differentiated cells were pretreated with CpB before membrane fractionation, Plg-R_KT was not detectable in the MudPIT analysis, indicating that the C-terminal lysine of Plg-R_KT was accessible to CpB on the cell surface and, thus, lost the ability to bind plasminogen-Sepharose after CpB treatment of intact cells. Second, Glu-plasminogen bound specifically to a synthetic peptide corresponding to the C-terminus of Plg-R_KT and the binding interaction was inhibited by the lysine analog, EACA, that inhibits the interaction of plasminogen with the cell surface. Lys-plasminogen also bound specifically to the peptide consistent with the ability of Lys-plasminogen to also bind to the cell surface.^69,70  Third, plasminogen binding to a mutated peptide with the C-terminal lysine substituted with alanine was not detected, further supporting the key role of the C-terminal lysine in the interaction with plasminogen. Of key importance, anti–Plg-R_KT antibodies directed against the C-terminal peptide of Plg-R_KT detected Plg-R_KT dispersed over the cell surface in small aggregates. A similar immunolocalization pattern of small aggregates dispersed over the cell surface has been published for other cell surface plasminogen binding proteins, α-enolase, histone 2B, p11, and annexin 2.⁶  Of key importance, after preincubation of the cells with plasminogen, the ability to immunodetect Plg-R_KT was reduced by half (Figure 5B-C). These results suggest that plasminogen and the anti–Plg-R_KT mAb share binding sites on the cell and that plasminogen binds to the C-terminal domain of Plg-R_KT on the cell surface. Notably, a C-terminal lysine is encoded for all 26 mammalian orthologs of Plg-R_KT for which the C-terminal sequence is available, consistent with a fundamental role of Plg-R_KT to bind plasminogen in mammals.

t-PA also bound specifically to the C-terminal peptide of Plg-R_KT, consistent with sharing of binding sites with plasminogen and with the CpB sensitivity of the interaction of t-PA with cells.⁵⁸  Despite potentially sharing a binding site on Plg-R_KT, the relative concentrations of t-PA and plasminogen in the circulation should permit simultaneous binding of both ligands to the cell surface, and each t-PA molecule should be bound proximally to several plasminogen molecules.⁵⁸  We also found that Plg-R_KT was spatially highly colocalized with the uPAR. These results suggest a mechanism for colocalization of plasminogen with plasminogen activators to promote cell surface plasminogen activation. Of key importance, treatment of the cells with anti–Plg-R_KT mAb markedly reduced the ability of M-CSF–treated Hoxa9-ER4 cells to stimulate plasminogen activation.

Our experimental data support the prediction that Plg-R_KT is a multipass transmembrane protein, exposing a C-terminal lysine on the cell surface. Our results demonstrating CpB-sensitivity of Plg-R_KT on the cell surface supported the cell surface exposure of the C-terminus. Regarding transmembrane topology, Plg-R_KT was immunochemically detected in the membrane fraction, but not the cytoplasmic fraction of differentiated cells. And, in phase partitioning experiments, Plg-R_KT partitioned to the detergent phase.

Our data also indicate that Plg-R_KT is up-regulated during differentiation. In the MudPIT analysis, Plg-R_KT peptides were detected in M-CSF–differentiated cells, but not in progenitor cells. Similarly, in Western blotting, Plg-R_KT was detected in membranes of differentiated cells, but not in the progenitor cells.

The C9orf46 homolog/Plg-R_KT transcript is broadly expressed in normal human and mouse tissues (as determined in high-throughput gene expression profiling in which RNA samples from human and murine tissues were hybridized to high-density gene expression arrays^72,73 ), including spleen; thymus; lymph node; lung; intestine; bone marrow; adrenal, pituitary, and other endocrine tissues; vascular tissue; kidney; liver; stomach; bladder; and neuronal tissue (hippocampus, hypothalamus, cerebellum, cerebral cortex, olfactory bulb, and dorsal root ganglion). We examined Plg-R_KT protein expression in murine tissue and detected Plg-R_KT in lymph node, bone marrow, and intestine, tissues rich in leukocytes but not in adipose tissue, as a negative control (supplemental Figure 1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Differences in mRNA expression data and Western blotting results may be explained by high expression by specific cells within tissues. In Western blotting, we also have detected Plg-R_KT in rat hippocampal neurons, bovine adrenal tissue, and in rat pheochromocytoma PC12 cells (R.J.P. et al, unpublished observations, June 2007).

We searched for C9orf46 homolog/Plg-R_KT mRNA microarray expression data at http://www.ebi.ac.uk/gxa⁷⁴  (Table 4). C9orf46 homolog/Plg-R_KT mRNA is present in monocytes; leukocytes; natural killer (NK) cells; T cells; myeloid, dendritic, and plasmacytoid cells; breast cancer cells; acute lymphoblastic leukemia cells; and Molt-4 acute lymphoblastic leukemia cells. These data are consistent with previous reports documenting expression of plasminogen binding sites on peripheral blood leukocytes,⁵⁵  breast cancer cells,^10,75  and other tissues (reviewed in Miles et al²⁷ ). In addition, results obtained by searching the ArrayExpress Warehouse (http://www.ebi.ac.uk/gxa) indicated that the C9orf46 homolog gene is also regulated in other tissues by lipopolysaccharide, canrenoate, H₂O₂, and dexamethasone (Table 5). The broad distribution and regulation in tissues that express plasminogen binding sites suggest that C9orf46 homolog/Plg-R_KT provides plasminogen receptor function that may serve to modulate plasmin proteolytic functions in these tissues, as well. In genome-scale quantitative image analysis, overexpression of more than 86 cDNAs, including C9orf46 homolog, conferred dramatic increases in cell proliferation, whereas knockdown of C9orf46 homolog mRNA resulted in apoptosis.⁷⁶  In microarray studies, C9orf46 homolog mRNA expression has a high power to predict cervical lymph node metastasis in oral squamous cell carcinoma.⁷⁷  We have previously shown that cell-bound plasminogen protects monocytoid cells from tumor necrosis factor-α–induced apoptosis.⁷⁸  Future studies should elucidate the relationship of these roles of C9orf46 homolog/ Plg-R_KT to its plasminogen receptor function.

In summary, regulation of the cell surface expression of plasminogen receptors is a mechanism by which cells modulate the proteolytic activity of their environment. We have identified a novel plasminogen receptor with unique characteristics: Plg-R_KT is an integral membrane protein that exposes a C-terminal lysine and serves to markedly promote plasminogen activation on the cell surface.

We thank Dr Ray Stevens at The Scripps Research Institute and Dr Nuala Booth and Dr Ian Booth, University of Aberdeen, for helpful discussions. This is publication 19034 from The Scripps Research Institute.

This work was supported by the National Institutes of Health (grants HL 38272, HL 45934, and HL 081046 [L.A.M.], HL 50398 [R.J.P.], NIH P41 RR011823 [J.R.Y.], National Institute of Allergy and Infectious Diseases subcontract grant UCSD/MCB0237059 [E.I.C.]), and the Department of Veterans Affairs (R.J.P.).

National Institutes of Health

Contribution: N.M.A., W.B.K., J.R.Y., R.J.P., and L.A.M. designed research; N.M.A., E.I.C., N.B., H.B., and C.M.P. performed research; M.P.K. and J.R.Y. contributed new reagents and analytic tools; N.M.A., E.I.C., N.B., H.B., C.M.P., W.B.K., R.J.P., and L.A.M. analyzed data; and N.M.A., R.J.P., and L.A.M. wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

N.M.A.'s current affiliation is FD McMaster Laboratories, Armidale, NSW, Australia. E.I.C.'s current affiliation is Department of Pharmacological Sciences, Stony Brook University, Stony Brook, NY. C.M.P.'s current affiliation is Yale University, New Haven, CT.

Correspondence: Lindsey A. Miles, Department of Cell Biology, SP30-20, The Scripps Research Institute, 10550 N Torrey Pines Rd, La Jolla, CA 92037; e-mail: lmiles@scripps.edu.