Urokinase receptor mediates lung fibroblast attachment and migration toward provisional matrix proteins through interaction with multiple integrins

Abstract

Fibroblasts from patients with pulmonary fibrosis express higher levels of the receptor for urokinase, and the extent of fibrosis in some animal models exhibits a dependence on the urokinase receptor. Recent observations have identified the urokinase receptor as a trans-interacting receptor with consequences on signaling and cell responses that vary depending on its interacting partner, the relative levels of expression, and the state of cellular transformation. We undertook this study to define the urokinase-type plasminogen activator cellular receptor (u-PAR)-integrin interactions and to determine the functional consequences of such interactions on normal human lung fibroblast attachment and migration. u-PAR colocalizes in lammelipodia/filopodia with relevant integrins that mediate fibroblast attachment and spreading on the provisional matrix proteins vitronectin, fibronectin, and collagens. Inhibitory antibody studies have revealed that human lung fibroblasts utilize α_vβ₅ to attach to vitronectin, predominantly α₅β₁ (and α_vβ₃) to attach to fibronectin, and α₁β₁, α₂β₁, and α₃β₁ to attach to collagen. Blocking studies with α-integrin subunit decoy peptides and u-PAR neutralizing antibodies indicate that u-PAR modulates the integrin-mediated attachment to purified provisional matrix proteins, to anti-integrin antibodies, or to fibroproliferative lesions from fibrotic lungs. Furthermore, these decoy peptides blunt fibroblast spreading and migration. We show that u-PAR can interact with multiple α-integrins but with a preference for α₃. Taken together, these data demonstrate that u-PAR may interact with multiple integrins in normal human lung fibroblasts thereby promoting attachment, spreading, and migration. Modulation of fibroblast invasion would be expected to lead to amelioration of fibroproliferative diseases of the lung.

studies in vitro and in transgenic animals have demonstrated that the fibrinolytic system has protean effects on numerous disease processes including acute and chronic inflammatory disorders, cancer invasion and metastasis, angiogenesis, and fibroproliferative disorders of the vasculature, lung, kidney, and eye (3, 4, 11, 16, 22, 23, 27, 29, 37, 55). These disorders have the common feature of a complex, yet coordinated movement of cells and the biochemical and structural remodeling of the extracellular matrix. The molecular interactions and cell biological processes that mediate the effects of the fibrinolytic system on cell-matrix interactions are active areas of investigation.

Protease-protease inhibitor mechanisms of the urokinase/plasmin system that relate to intravascular or alveolar fibrin turnover are deranged in pulmonary fibrosis but are clearly only a part of the relevant biology (38). The receptor for urokinase is a 50-kDa, cell surface glycoprotein receptor that binds urokinase with high affinity (K_d = 2 nM) and that is linked to the plasma membrane via a glycosylphosphatidyl inositol side chain (5, 8, 9, 44). This receptor can interact in trans with other cell surface receptors including integrins, EGFR, and the G protein-coupled receptor (FPXLR1; Refs. 32, 46, 67). Interaction of urokinase-type plasminogen activator cellular receptor (u-PAR) with the extracellular domain of integrins is demonstrable by coimmunoprecipitation and colocalization (12, 65, 67). u-PAR also exhibits specific, saturable binding to immobilized integrins (12, 65, 67). Further, these interactions modulate several biological processes including adhesion, migration, phagocytosis of fibrinogen, intracellular signaling, and tumor dormancy in a cell type-specific manner (2, 10, 42, 50, 67).

Integrins are a large family of heterodimeric transmembrane receptors (composed of an α- and a β-subunit) that typically cluster in the cell membrane upon integrin receptor recognition of its ligand in the extracellular matrix (7, 18). Integrin clustering leads to the assembly of a signaling complex that is also an adhesion structure (known as a focal contact or focal adhesion) and is localized to the submembranous region at the base of the cell. These adhesion structures/signaling complexes regulate cell adhesion, spreading, migration, survival, and proliferation through modulation of the actin cytoskeleton. Other signaling molecules, such as kinases, cytoskeletal proteins, and adaptor/scaffolding proteins, are recruited to and become concentrated at these structures (7, 18). Regulation of integrin signaling occurs at multiple levels and includes the specific heterodimeric (αβ) subunit pairing, the association of the integrin with other cell membrane molecules (i.e., u-PAR or tetraspans), the available divalent cations, the cell type, and the environmental context (7, 18, 24, 52, 67). The integrin β-subunit cytoplasmic tail is necessary for most integrin signaling events due to its association with specific cytoskeletal and signaling molecules, and the integrin α-subunit can also modulate integrin signaling, for example, through its extracellular W4 domain association with u-PAR (65, 72).

Integrins have been shown to play a role in the fibroproliferative response in the lung. For example, integrins α_vβ₆ and α_vβ₈ activate TGFβ, a potent profibrotic growth factor (25, 33, 36, 45, 47). Furthermore, proliferation and migration of idiopathic pulmonary fibrosis fibroblasts, in response to integrin ligation, are aberrantly regulated through reductions in the activity of the tumor suppressor phosphatase and tensin homologue protein (70, 71). In addition, integrin-initiated signaling is vital to the trans-differentiation of fibroblasts to myofibroblasts and for the epithelial-mesenchymal transformation (15, 28, 60). The importance of u-PAR modulation of integrin function in fibroblasts to pulmonary fibrosis has yet to be explored.

In the case of mononuclear cells, it is known that their activation and transvascular emigration of mononuclear cells (neutrophils and monocytes) into the pulmonary interstitium are a urokinase receptor-dependent process in several inflammatory states, including those of infectious etiology, chemokine-driven (KC), and fibrogenic stimuli (bleomycin-induced lung injury; Refs. 1, 20–22, 53, 54, 57, 63). Recently, renal podocyte cell u-PAR has been shown to play a role in the renal dysfunction due to podocyte foot process effacement and proteinuria seen after endotoxin administration (64). In the specific case of fibroblasts, invasion, and remodeling of a fibrinogen, the vitronectin and fibronectin-rich provisional matrix within the pulmonary alveolus is a fundamental feature of the normal and pathologic repair process after lung injury (8a, 58, 59). We and others (48, 56) have shown that fibroblasts derived from patients with fibrotic lungs exhibit a promigratory phenotype in vitro and exhibit upregulation of u-PAR. However, the role and mechanisms by which the urokinase receptor mediates migration of fibroblasts have not been fully elucidated. We sought to characterize and determine the mechanism of the effect of u-PAR on human lung fibroblast attachment, spreading, and migration that are critical to the fibrotic process.

MATERIALS AND METHODS

Materials.

Eagle's modified minimum essential medium (EMEM) and DMEM were purchased from Mediatech (Herndon, VA). Gelatin Sepharose 4B, horseradish peroxidase (HRP)-conjugated secondary antibodies, and ECL Western blot reagents were from Amersham Pharmacia Biotech (Piscataway, NJ). Human vitronectin, fibronectin (from human plasma), and collagen (type I; from calf skin) were from Molecular Innovations (Southfield, MI), Boehringer Mannheim (Indianapolis, IN), and ICN Biomedicals (Irvine, CA), respectively. Fibronectin-depleted human fibrinogen was from Enzyme Research Laboratories (South Bend, IN). Protein G-agarose was from Pierce (Rockford, IL). Fluorochrome-conjugated secondary antibodies were from Molecular Probes (Eugene, OR). Mouse MoAb and rabbit polyclonal anti-human u-PAR antibodies were from American Diagnostica (Greenwich, CT). Mouse MoAb anti-human integrins β₁ (AB1965 and P4C10), α_vβ₃ (LM609), α_vβ₅ (P1F6), α₃ (P1B5), α₅ (SAM-1), and rabbit polyclonal anti-integrin β₁ (AB1952) were from Chemicon International (Temecula, CA). Mouse MoAb anti-human integrins α₂ (Clone A2-IIE10) and β₁ (Clone DE9) were from Upstate Biotechnology (Lake Placid, NY). Mouse MoAb anti-integrin α_v (clone L230) was from the University of Alabama at Birmingham Hybridoma Core Facility and was purified in Dr. Gladson's laboratory. Mouse MoAb anti-vinculin antibody (M0520) was from American Qualex Antibodies (San Clemente, CA). Peptides homologous to the β-propeller repeat regions of the extracellular domains of the α₃-integrin chain (AA 273–289) α₃₂₅ PRHRHMGAVFLLSQEAG, the scrambled control peptide S-325 HQLPGAHRGVEARFSML, the α₅-integrin chain (AA 294–310) PKGNLTYGYVTILNGSD, and the α_v-integrin chains (AA 274–290) PRAARTLGMVYLYDGKN were provided by Dr. H. Chapman, Jr. or were purchased from Anaspec (San Jose, CA; Ref. 66).

Flow cytometry analysis of cell surface expressed proteins.

CCD-19 Lu normal adult human lung fibroblasts (ATCC; passages 4–9) were maintained and propagated as described (41). Washed, resuspended fibroblasts (4 × 10⁵ cells in 300 μl of HBSS/1% BSA) were incubated with primary antibodies (10 μg/ml IgG or 1:100 dilution of ascites) at 4°C for 45 min. followed by incubation with fluorochrome (488 or 594 nm)-conjugated rabbit anti-mouse IgG (20 μg/ml) at 4°C for 30 min. in the dark. Flow cytometric analysis was performed on EPICS XL-MCL flow cytometer (Beckman-Coulter, Miami, FL).

Cell surface labeling of integrins.

Cell surface integrins were characterized using biotinylated surface molecules followed by immunoprecipitation as previously published (19). In brief, surface labeling of the cells was accomplished by treatment of 0.5 mg/ml of sulfosuccinimidyl-6-biotinamido hexanoate (sulfo-NHS-LC-biotin) for 30 min at 22°C. Cells were lysed (1% DOC/0.5% SDS/1% Triton X-100 in TBS), and equivalent amounts of protein lysates were immunoprecipitated with Sepharose bead-bound-integrin-specific antibodies and then subjected to SDS-PAGE under nonreducing conditions followed by transfer to Immobilon membranes. Surface-labeled immunoprecipitates were detected by reaction with streptavidin-HRP (10 pg/ml) and ECL kit. The pairings of the integrin subunits are based on the known association of the α- and β-subunits, the known relative migration of the subunits, and the fluorescence-activated cell sorting (FACS) analysis results.

Western blot analysis of u-PAR and integrins.

Solubilized proteins were separated by SDS-PAGE and then immunoblotted on Immobilon-P membranes as described previously (40). The membranes were blocked (TBS-Tween-20/5% BSA; room temperature for 1 h) and incubated with either rabbit polyclonal anti-integrin β₁ or u-PAR antibodies (4°C overnight) followed by HRP-conjugated donkey anti-rabbit IgG (1:7, 500 dilution; room temperature 1 h). Specific protein bands were detected using ECL Western blot reagents on Kodak BioMax films.

Colocalization of u-PAR and integrins by immunofluorescence.

Fibroblasts were plated on human vitronectin (5 μg/ml)-, fibronectin (10 μg/ml)-, or collagen (10 μg/ml)-coated glass coverslips or chamber wells slides and were allowed to attach and spread (0.5–6 h) in attachment assay buffer. The attached cells were fixed with 3% paraformaldehyde in some cases, permeabilized (0.15% Triton X-100/PBS, room temperature for 1 min), and blocked (10% horse serum, 10 min at room temperature). For double antigen colocalization, cells were first incubated with the first mouse primary antibody (1 h at room temperature) and secondary Alexa Fluor 488-conjugated goat anti-mouse antibody (20 μg/ml; 1 h at room temperature) pair followed by the second primary/secondary pair. The cell nuclei were stained with Hoechst (33258; 20 μg/ml for 2 min at room temperature) and mounted (90% glycerol/PBS/0.01% p-phenylenediamine). Immunofluorescence was viewed under Leica inverted microscope, and digital image processing was performed using a Retiga EX camera, Compix software, and Adobe Photoshop.

Cell attachment and spreading assay.

Human lung fibroblasts were suspended in attachment assay buffer [10 mM HEPES, 140 mM NaCl, 5.4 mM KCl, 5.6 mM d-glucose, and 1% BSA, pH 7.4 (100 μM MnCl₂, 1 mM MgCl₂, ± 1 mM CaCl₂)] and seeded (in quadruplicate) on either matrix protein-coated [human vitronectin (5 μg/ml), human fibronectin (10 μg/ml), collagen (10 μg/ml), or fibrinogen (100 μg/ml)], or anti-integrin antibody-coated (10 μg/ml)/1% BSA blocked 96-well plates as described previously (43). Preliminary experiments indicated that the inclusion of 1 mM CaCl₂ in the attachment assay buffer had no effect on fibroblast attachment to vitronectin. Cells were allowed to attach for 15–30 min at 37°C, and nonadherent cells were washed off in 37°C PBS. Ovalbumin (5 μg/ml)-coated wells served as the negative control. Attached cells were counted. Wells were coated by either or the indicated concentrations of these proteins. Attachment inhibition studies were performed by incubation of the fibroblasts (either human lung fibroblasts or primary isolates of murine lung fibroblasts from u-PAR null mice) with antibodies and/or the indicated peptides (for 20 min at room temperature at the indicated concentrations) before their seeding on matrix protein-coated wells. Attachment to integrin antibodies was performed in wells precoated with Goat-anti-mouse IgG and the indicated anti-integrin mouse monoclonal function-blocking IgG or irrelevant mouse monoclonal IgG, as described previously (43). Attachment was calculated as the ratio of the cell number noted in the presence of equimolar α₃₂₅/serum-containing medium (SCM), after subtracting the number of cells present in the IgG control well. Fibroblast spreading was assessed by allowing the cells to attach (20 min at 37°C) and either scrambled (S-325; 75 μM), active peptide (α₃₂₅; 75 μM), or vehicle (DMSO) was added. Photomicrographs were taken as described above, and the individual cell area was measured for at least 200 cells/condition using Compix software.

Cell attachment to murine lung tissue.

All animal protocols were approved by the University of Alabama at Birmingham animal resources program (Institutional Animal Care and Use Committee). C57Bl/6 (8- to 12-wk-old females) mice were instilled intratracheally with bleomycin sulfate (4 U/kg). Mice were euthanized 21 days later, and the lungs were perfused with 5 ml cold PBS through the right ventricle and then instilled with optimum cutting temperature (OCT) compound via the airway, embedded in OCT, and frozen. Normal human lung fibroblasts (19 LU) were labeled with the fluorescent dye PKH-36 (Sigma) according to the manufacturer's instructions with minor modifications. Frozen lung tissue sections (5 μm) were thawed and blocked (PBS/1% BSA, 2 h), and the labeled fibroblasts (19 LU) were allowed to adhere for 45 min (37°C, 5% CO₂). The tissue sections with cells were exhaustively washed to remove nonadherent cells. Cell number and tissue area were determined in photomicrographs (phase and fluorescent) from randomly selected areas of lung exhibiting either normal histology or fibroproliferative lesions using Compix software. In all cases, a minimum of 200 cells was counted per determination.

In vitro migration assay.

Fibroblasts were seeded at confluent density (2 × 10⁵ cells/2 cm² growth area) and allowed to attach overnight in 1–10% SCM. The confluent monolayers were then washed with serum-free medium and made quiescent (0.4% SCM for 48 h). A denuded area (wound) was created using a 200-μl pipette tip. Wounded cell monolayers were carefully washed, and the media were replaced with EMEM with or without the indicated agonists or inhibitor reagents. Digital pictures of the wounds were taken at 0.5 h (time 0) and indicated intervals later, and areas devoid of cells, “wounds,” were measured using Compix software. The effects of peptides on the wound healing process were assessed by adding the active (α₃₂₅) or scrambled (S-325) peptides, or vehicle control (DMSO) to the medium, after the wound. Fresh medium/peptides were replenished every 96 h when necessary.

Statistical analysis.

All data are means ± SD unless otherwise indicated. Continuous variables from more than two groups were compared by means of an ANOVA with a post-hoc analysis (Dunnett's test or Student-Newman-Keuls). Significance was accepted at the P ≤ 0.05 level.

RESULTS

Human lung fibroblasts express the urokinase receptor and multiple integrin receptors.

To investigate the possible involvement of u-PAR and integrin receptors in fibroblast attachment and migration during the repair of injured lung, we first characterized the cell surface expression of these integrin proteins in fibroblasts derived from normal human lung. Surface expression of u-PAR was detected, albeit at a low level, by flow cytometric analysis (FACS), by immunobloting of fibroblast lysates, and by ELISA (Fig. 1). Fibroblast cell lysates contained 2,433 pg u-PAR/10⁶ cells. By FACS analysis, the cells demonstrated a significantly weaker signal for u-PAR than for integrin β₁-chain. Cell surface labeling followed by immunoprecipitation with anti-integrin specific antibodies demonstrated the expression of α_vβ₃-, α_vβ₅-, and multiple β₁-containing integrins (Fig. 2A). Bands at 95 and 125 kDa and in the α_v immunoprecipitate suggest that α_v (150 kDa) is pairing with β₃/β₅ and β₁, respectively (Fig. 2A). Bands at 150 and 220 kDa in the β₁-immunoprecipitate suggest that β₁ (125 kDa) is pairing with α_v-integrin and at least α₁-integrin subunits, respectively (Fig. 2A). FACS analysis further revealed the surface expression of α₁-, α₂-, α₃-, α₅-, α_vβ₃-, α_vβ₅-, and β₁-integrins (Fig. 2B), as summarized in Table 1.

Fig. 1.

Human lung fibroblasts express urokinase-type plasminogen activator cellular receptor (u-PAR) and integrin β₁. A: flow cytometry detection of cell surface expression of u-PAR, integrin β₁, and IgG controls on human lung fibroblasts were performed using specific primary antibodies. Cells express u-PAR and integrin β₁ with fluorescence intensities of 5 and 50 times higher, respectively, compared with IgG control. B: Western blotting and immunoprecipitation analysis of cell lysates. Shown are the following: fibroblasts attached to vitronectin (20 min) lysed in either Triton X-100 lysis buffer (lane 1) or RIPA buffer (lane 2); human su-PAR (10 ng) control (different lanes from same gel; lanes 3, 5, and 6); human u-PAR from fibroblast lysates immunoprecipitated by a mixture of 2 monoclonal antibodies (American Diagnostica No. 3936 and 3937; lane 4) and detected by a polyclonal antibody against u-PAR (399R); and TX-100 lysates from 293 epithelial cells that do not express u-PAR (lane 7). Both Western and immunoprecipitated (IP) u-PAR migrated at 42 kDa.

Fig. 2.

Human lung fibroblasts express multiple integrins that recognize provisional matrix binding proteins. A: immunoprecipitation analysis of cell surface integrins in fibroblasts. Cells were surface labeled with biotin and immunoprecipitated using integrin-specific antibodies as indicated. Telative migration of the individual integrin subunits is indicated at left. Expression of integrins α_vβ₅ (lane 2) and α_vβ₃ (lane 3) was detected using heterodimeric-specific integrin antibodies. In the α_v-immunoprecipitate (lane 4), bands at 160, 120, and 100 kDa were detected consistent with the α_v-, β₁-, and β₃/β₅-subunits, respectively, indicating α_v pairs with β₁ and β₃/β₅ in these cells. In the anti-β₁-immunoprecipitate (lane 5), bands at 220, 160, and 120 kDa were detected, consistent with expression of the α₁-, α_v-, and β₁-subunits, respectively, indicating expression of α₁β₁ and α_vβ₁. No expression of α₈- or α₉-integrins was detected (lanes 6 and 7). B: flow cytometry detection of cell surface integrin expression in fibroblasts. Cell surface expression of integrins were detected by monoclonal antibodies against integrins u-PAR (No. 3936 American Diagnostica), α₁ (Mab 1973; Chemicon), α₂ (clone A2-IIE10), α₃ (clone P1B5), α₅ (clone SAM-1), β₁ (clone P4C10), α_vβ₃ (clone LM609), and α_vβ₅ (clone P1F6), and Mac-1 (Clone ICRF-44), respectively.

Table 1.

Integrin expression profile in human lung fibroblasts

Integrins Present	Integrins Absent
aVβ1	α8β1
αVβ3	α9β1
αVβ5
α1β1
α2β1
α3β1
α5β1

Human lung fibroblasts attach strongly to vitronectin, fibronectin, and collagen but weakly to fibrinogen.

To characterize the attachment of human lung fibroblasts to provisional alveolar matrix proteins, we first assessed their time and concentration dependencies. Fibroblasts attached rapidly to vitronectin-, fibronectin-, and collagen-coated microplate wells, reaching near-maximal attachment within 15 min (Fig. 3A). In sharp contrast, the attachment to fibrinogen was slower and less complete, reaching a maximal attachment of 68.2 ± 4.9% (mean ± SD; n = 4) of that seen with vitronectin after 60 min. Furthermore, the concentration dependence of attachment to vitronectin, fibronectin, and collagen (at 30 min) was similar (EC₅₀ ≈ 100 ng/ml), while the fibrinogen attachment (at 60 min) concentration curve was shifted to the right by more than two orders of magnitude (EC₅₀ ≈ 7,000 ng/ml; Fig. 3B). As these data indicate that fibroblast attachment to vitronectin, fibronectin, and collagen is stronger than that to fibrinogen, we focused on defining the mechanisms of attachment to vitronectin, fibronectin, and collagen.

Fig. 3.

Time-dependent and concentration-dependent attachment of fibroblasts to extracellular matrix proteins. A: time dependency of fibroblast attachment. Fibroblasts (10 × 10³ cells) attached to vitronectin (VN; 5 μg/ml), fibronectin (FN; 10 μg/ml), collagen (CL; 10 μg/ml), or fibrinogen (FGN; 10 μg/ml) in 96-well plates for 30 min in attachment assay buffer. After being washed at indicated times, remaining attached cells were counted. Value of cell attachment to fibrinogen at 60 min was 58.2 ± 4.9% (means ± SE; n = 4) of that to vitronectin at same time point. Results are expressed as percentage of maximal attachment to same matrix protein at 60 min. Values are means ± SE for n = 4. B: concentration dependency of fibroblast attachment. Cell attachment was assayed at either 20 min (for vitronectin, collagen, and fibronectin) or 60 min (for fibrinogen). Results are expressed as percentage of attachment to vitronectin at 20 min. Values are means ± SE for n = 4.

Multiple integrin subunits colocalize with u-PAR in cell membrane protrusions during fibroblast spreading.

To dissect potential interactions between u-PAR and integrin receptors during cell attachment and spreading, double label immunofluorescence analysis was performed on early phase spreading fibroblasts. This analysis reveals that u-PAR and the integrins α_v-, α₃-, and α₅ (and β₁, not shown)-subunits colocalize within filopodia/lamellipodia during the initial, but not later, phases of spreading cells (Fig. 4A). These data reveal a spatial and temporal interaction of u-PAR with several integrins in the protrusive structures in spreading fibroblasts.

Fig. 4.

Colocalization of relevant integrins with u-PAR in filopodia/lamellipodia of spreading fibroblasts. Fibroblasts were attached to indicated matrix-protein-coated glass coverslips for 15–30 min in attachment assay buffer. Cells were fixed (3% paraformaldehyde), blocked with (10% horse serum in HBSS), and then immunostained for u-PAR (green, Alexa Fluor 488-conjugated goat anti-mouse IgG secondary) and for indicated integrins (red, Alexa Fluor 594 secondary). Photomicrographs were performed (×1,000 original magnification), and basal 1-μm Z stack images were selected and merged. Yellow color indicates colocalization. Bottom right: immunofluorescent staining of u-PAR (red) and α₅-integrin (green) in spread cells on fibronectin. α₅-Integrin localizes to focal adhesions.

Attachment of fibroblasts to vitronectin, fibronectin, and collagen is dependent on u-PAR-integrin interactions.

To determine the integrins that mediate fibroblast attachment and the significance of u-PAR-integrin interactions in fibroblast attachment, we examined the attachment of the fibroblasts to matrix proteins present in normal and injured lungs interstitium (i.e., fibronectin, vitronectin, and collagen). To ferret out the relative importance of integrins and u-PAR-integrins interactions, attachment assays were performed in the presence or absence of function-blocking antibodies to individual integrins, function-blocking antibodies to u-PAR, and peptides that have been documented to selectively inhibit u-PAR-integrin interactions (52,67, 72). Function-blocking antibody studies revealed that attachment of fibroblasts to vitronectin was dependent on integrin α_vβ₅ but not α₃ (Fig. 5A) and that attachment to fibronectin was largely dependent on α₅β₁ but not α₃ (Fig. 5A). Integrin α_vβ₃ played a minor role in fibroblast attachment to fibronectin. Attachment to collagen was dependent on integrins α₁β₁, α₂β₁, and α₃β₁ but not α_vβ₃ (Fig. 5A).

Fig. 5.

Determination of molecules that mediate fibroblast attachment to vitronectin, fibronectin, and collagen. A: specific integrins that mediate fibroblast attachment. Fibroblasts were incubated with function-blocking antibodies against indicated integrins or serotype-matched IgGs at room temperature for 20 min with gentle rocking. Cell attachment to either vitronectin (5 μg/ml)-, fibronectin (10 μg/ml)-, or collagen (10 μg/ml)-coated tissue culture plastic microplate wells (15–30 min, 37°C, 5% CO₂). Results are expressed as percent attachment of IgG-pretreated cells. Values are means ± SD for n = 4/condition. *P < 0.05, compared with IgG control. B: role of u-PAR and u-PAR-integrin interactions in fibroblast attachment. Cells were incubated with indicated reagents (IgG control or u-PAR antibodies) or integrin-homologous peptides (α₃₂₅) or scrambled peptides [serum-containing medium (SCM)], and cell attachment was measured as described above. Results are expressed as percent attachment of either IgG- or vehicle (DMSO)-treated cells. Values are mean ± SD for n = 4/condition. *P < 0.05, compared with IgG or vehicle control. C: lack of effect of integrin-homologous peptides (α₃₂₅) on lung fibroblasts that are devoid of u-PAR. Murine lung fibroblast from u-PAR knockout mice were incubated with integrin-homologous peptides (α₃₂₅) or scrambled peptides (SCM) and assayed for their effects on attachment to vitronectin, fibronectin, and collagen, as in B. Values are means ± SD for n ≥ 4/condition.

Blocking studies of u-PAR function with anti-u-PAR antibody, previously shown to inhibit u-PAR function, demonstrate that u-PAR plays a direct role in attachment to vitronectin, as published in other cell types, whereas u-PAR played no direct role in cell attachment to fibronectin and collagen (Fig. 5B). To determine the involvement of u-PAR-integrin interactions in the cell attachment process, we studied the effect of a previously characterized 17-mer peptide homologous to the putative u-PAR binding site on the W4 extracellular region of integrin α-chains (52, 67, 72). This peptide (α₃₂₅) inhibited fibroblast attachment to vitronectin, fibronectin, and collagen (Fig. 5B). The α₃₂₅-peptide inhibited attachment to all matrix ligands (vitronectin shown) in a concentration-dependent manner (EC₅₀ approximately = 25 μM) with a maximal inhibition of 78 ± 3%. Furthermore, this inhibition was sequence specific as a scrambled version of the 17-mer peptide (SCM) or the corresponding α_v-integrin homologous peptide at the same concentration had no effect on cell attachment to all three matrix proteins (Fig. 6A). This latter result is notable in light of the fact that α_vβ₅ is the predominant integrin involved in fibroblast attachment to vitronectin. The results in Figs. 5 and 6 indicate that fibroblasts attach to vitronectin through 1) integrin ligation by vitronectin, 2) u-PAR-integrin interactions, and 3) likely direct u-PAR ligation to vitronectin. In contrast, u-PAR-integrin interactions are uniquely important for fibroblast attachment to fibronectin and collagen, as u-PAR antibody blocking studies do not demonstrate a role for direct u-PAR ligation in attachment to fibronectin and collagen (Fig. 5B, grey and open bars). To demonstrate that α₃₂₅ had no direct effect on integrin function or other confounding effects, we examined the effect of the α₃₂₅-peptide of attachment of cells lacking u-PAR (primary isolates of lung fibroblasts from mice genetically devoid of u-PAR). The number of attached u-PAR-devoid fibroblasts to vitronectin, fibronectin, or collagen in the presence of the α₃₂₅-peptide was 99 ± 6, 89.5 ± 19, and 115 ± 27% (n = 4/condition, all P = NS) of an equivalent concentration (10 μM) of the scrambled peptide, respectively (Fig. 5C). Thus the peptide had no effect on attachment in the u-PAR null fibroblasts. The data regarding cell attachment, when taken together with the immunofluorescence colocalization data, indicate that u-PAR interacts with and modulates the attachment function of several different integrins, specifically, integrin α_vβ₅ in cells attaching to vitronectin, integrin α₅β₁ in cells attaching to fibronectin, and one or more integrins (i.e., α₁β₁, α₂β₁, or α₃β₁) in cells that are attaching to collagen.

Fig. 6.

Preference of u-PAR association with α-integrins. A: specificity of integrin α-chain homologous peptides. Lung fibroblasts were preincubated with indicated concentrations of the peptide homologous regions of either α₃- or α_v-integrins, or a scrambled peptide control (SCM), and tested for attachment to vitronectin. Data are plotted as cells attached compared with vehicle alone (DMSO; n ≥ 4/condition). *P < 0.05, compared with SCM peptide by ANOVA. B: α₃-integrin chain peptide effects on attachment to individual α-chain integrin antibodies. Lung fibroblasts were preincubated with 30 μM α₃₂₅ or SCM control peptides and then allowed to attach to indicated integrin α-chain-specific antibodies. Data are plotted relative to attachment after incubation with S-325 to each integrin α-chain antibody (n ≥ 5/condition). *P < 0.05, greater inhibition compared with other integrin antibodies; **P < 0.05, less inhibition compared with other integrin antibodies.

u-PAR exhibits preference for functional association with the α₃β₁-integrin.

As human lung fibroblasts utilize multiple and unique integrins to attach to the individual provisional matrix proteins, we analyzed the effect of the inhibitory peptide (α₃₂₅) on u-PAR-integrin interactions in fibroblast attachment to specific immobilized anti-integin antibodies. The α₃₂₅-peptide maximally inhibited attachment to immobilized monoclonal anti-integrin α₃β₁ (Fig. 6B). This observation is not a consequence of differential expression of integrins as FACS analysis reveals roughly similar cell surface expression levels of α₁-, α₂-, α₃-, α₅-, and α_v-integrins (Fig. 2B). The relative order of inhibition of attachment to immobilized monoclonal anti-integrin antibodies using the α₃₂₅-integrin peptide was integrins α₃β₁ > α₅β₁, α₂β₁, α₁β₁ > α_v (Fig. 6B; P < 0.05 by ANOVA). Concordantly, a preference for u-PAR association with integrin α₃β₁ was noted compared with other integrins, as the α_v-homologous peptide was ineffective as an inhibitor of fibroblast attachment to any matrix ligand (Fig. 6A). This is notable given the requirement for integrin α_vβ₅ in fibroblast attachment to vitronectin and for α₅β₁ for fibroblast attachment to fibronectin. The α₅-integrin homologous peptide inhibited attachment to vitronectin and fibronectin in an intermediate manner but had no effect on fibroblast attachment to collagen (data not shown). In summary, the α₃₂₅-peptide inhibited attachment to all matrix ligands and to all anti-integrin antibodies, despite the lack of α₃β₁-integrin dependence of attachment to vitronectin or fibronectin. Our data, taken together with prior work demonstrating that the α₃-homologous peptide directly inhibits binding of soluble u-PAR to immobilized integrins, suggest that the α₃₂₅-peptide has more favorable kinetics of association with u-PAR than the α_v- or α₅-homologous peptides.

Fibroblast spreading depends on u-PAR-integrin interactions.

Cell spreading requires a spatial-temporal coordinated action of integrin ligation to matrix, as well as integrin clustering, integrin-mediated intracellular signaling, and cytoskeletal reorganization. Initially, human lung fibroblasts spread circumferentially on purified vitronectin, fibronectin, or collagen, reaching a t_1/2max for spreading at 42, 28, and 32 min after plating, respectively (data not shown). To study the effect of interactions of u-PAR with integrins on cell spreading, the effect of either α₃₂₅-peptide (75 μM), scrambled peptide (S-325; 75 μM), or vehicle alone (equivalent; DMSO) on cell area was measured in fibroblasts plated and spreading on matrix. Incubation of matrix-attached fibroblasts with the α₃₂₅-peptide (75 μM) reduced the maximal extent of cell spreading by 85 ± 8% (P < 0.05 by repeated measures ANOVA) compared with scrambled peptide, DMSO vehicle, or no additives (Fig. 7A). Furthermore, the poorly spread fibroblasts incubated with α₃₂₅-peptide exhibited a lower density of actin stress fiber than that of the controls (Fig. 7B). In contrast, cell spreading of control, u-PAR-devoid fibroblasts on collagen, fibronectin, and vitronectin in the presence of the α₃₂₅-peptide was 93 ± 17, 92 ± 23, and 104 ± 9% (n = 100 cells/condition, all P = NS) that of an equivalent concentration of the scrambled peptide, respectively. Thus the peptide had no effect on spreading in the u-PAR null fibroblasts.

Fig. 7.

Inhibition of fibroblast spreading by peptide α₃. Fibroblasts attached to vitronectin-coated plates as described above. Scrambled (S-325) or active peptides (α₃₂₅, 75 μM) were added and the cells were observed for an additional 140 min. Photomicrographs were taken at indicated times thereafter and cell area was measured using Compix software (>100 cells/time point/condition). A: cell area measurement. Values are means ± SD for all cells in at least 2 different pictures. *P < 0.05, compared with scrambled peptide (SCM) by repeated measures ANOVA. B: representative photomicrograph (at 160 min). Fibroblasts were incubated with indicated reagents for 160 min, fixed, and stained with phalliodin (green) and Hoescht (blue).

In vitro wound closure of fibroblast monolayers is dependent on u-PAR-integrin interactions.

To determine whether u-PAR-integrin interactions are important to fibroblast migration, the effect of peptide α₃₂₅ on fibroblast migration was investigated using an in vitro monolayer “wound closure” assay. As this is a several day assay, fibroblast migration in this system occurs on a complex matrix derived from both the initial serum and that synthesized by the fibroblast themselves. Nonetheless, such migration is highly dependent on fibronectin-α₅β₁-integrin interaction (92 ± 7% inhibition with α₅-integrin blocking Abs; P < 0.05) and less so on collagen-integrin α₃β₁ interaction (42 ± 9% inhibition; P < 0.05) and vitronectin-integrin α_v (35 ± 11% inhibition; P < 0.05) function (Fig. 8A). Wound closure was delayed (time to 1/2 maximum closure, 53 vs. 38 h) and less complete (by 30 ± 6%; P < 0.001 by repeated measures ANOVA) upon incubation of the monolayer with the α₃₂₅-peptide (100 μM) compared with the scrambled peptide (Fig. 8, B and D). Furthermore, u-PAR-integrin interactions can modulate wound closure, even in the presence of the multiple promotility factors in serum (Fig. 8C). In contrast, the scrambled peptide (SCM) did not affect wound healing in human lung fibroblasts. The α₃₂₅-peptide wound healing was 101 ± 21% (n = 4/condition, P = NS) of that of an equivalent concentration of the scrambled peptide in fibroblasts that lack u-PAR. Thus the peptide had no effect on migration in the u-PAR null fibroblasts. Similar conclusions are reached if wound closure is measured as migrating cell number or migrating cell area (data not shown). Our observations are not a consequence of peptide effects on proliferation, as serum-induced DNA synthesis (bromodeoxyuridine incorporation) was not different among fibroblasts incubated with α₃₂₅-peptide, scrambled peptide (SCM), or vehicle (DMSO) (data not shown) and it occurred in the absence of proliferation-requiring serum. Taken together, these data indicate that u-PAR modulates the integrin-mediated cell functions of attachment and that integrin-initiated signaling drives cell spreading and migration.

Fig. 8.

Inhibition of fibroblast wound healing by α₃-integrin homologous peptide. Confluent fibroblast monolayers were quiescent in EMEM medium containing 0.4% FBS for 48 h and wounded by pipette tip. Cell monolayers were then fed with EMEM medium containing either 0, 0.4, or 10% FBS with or without indicated function-blocking integrin antibodies or with or without α₃₂₅- or S-325 peptides. Photomicrographs of same wound area were taken at 15 min post-“wounding” (time 0) and every 6–12 h thereafter. Wound areas were quantified using Compix software and compared with that noted at time 0. A: α₅β₁ is the predominant integrin involved in fibroblast wound closure assay. The indicated function-blocking integrin antibodies (10 μg/ml) were added immediately after wounding. Data are plotted as means ± SD percent wound area closed relative to time 0. Control IgG (10 μg/ml) wounds underwent 57 ± 4% closure at 24 h. *^,+P < 0.05 either decreased or increased compared with IgG by ANOVA (n > 4/condition). B: inhibition of fibroblast wound healing by α₃-peptide. Peptide α₃ (α₃₂₅) and scrambled peptide (S-325) control (SCM) were added after the “wound.” Data are plotted as means ± SD wound area as percentage of initial value over time. *P < 0.05, compared with SCM peptide by repeates measures ANOVA (n ≥ 4/condition). C: inhibition of fibroblast wound healing by α₃ peptide is independent of serum. Peptide α₃ and scrambled peptide control (SCM) were added after the “wound” in EMEM medium containing indicated concentrations of serum. Results are expressed as means ± SD (n = 6/condition) of wound area remaining (at 72 h) relative to maximal closing at the same time. *P < 0.05, compared with SCM peptide under the same conditions by ANOVA (n ≥ 4/condition). D: representative photomicrographs of inhibitory effect of peptide wound closure. As above, with serum-free EMEM, conditions indicated on each photomicrograph.

Fibroblasts preferentially attach to fibroproliferative lesions.

To extend the physiological relevance to pulmonary fibrosis, we evaluated the capacity of the fibroblasts to adhere to either normal areas or fibroproliferative lesions from a well-established murine model of experimentally induced pulmonary fibrosis. Fibroblasts preferentially attached to fibroproliferative lesional areas by fivefold, compared with normal areas of lung, measuring absolute numbers, and approximately threefold (2.7 ± 0.3; P < 0.001) when measured as the number of cells per unit area of lung tissue (Fig. 9A –C). Furthermore, attachment of fibroblasts to lung tissue was reduced by 35 ± 7% in the presence of α₃₂₅-peptide (10 μM) compared with the density of fibroblasts attached in the presence of the scrambled peptide (10 μM; Fig. 9D). Taken together these data suggest that u-PAR-intergrin interactions mediate fibroblast attachment to remodeling lung lesional matrix, indicating the physiological relevance of our observations.

Fig. 9.

Preferential attachment of fibroblasts to fibroproliferative lesions of fibrotic lungs. Lungs were harvested from mice 21 days after intravenous instillation of bleomycin, inflated, embedded with optimum cutting temperature compound, and frozen. Tissue sections were blocked and incubated with labeled human lung fibroblasts for 45 min (37°C in 5% CO₂). Slides were exhaustively washed, and attached cells were photographed (×10 original magnification) under phase contrast and fluorescent conditions. Cell number was counted and tissue area was measured using Compix software. A: fibroblast attachment to normal lung areas. Representative dark field/fluorescent photomicrograph (×10 original magnification). Fibroblasts are red. B: fibroblast attachment to fibroproliferative lesional areas. Representative dark field/fluorescent photomicrograph (×10 original magnification). Fibroblasts are red. C: quantitative comparison of cell attachment to normal and lesional areas. Data are reported as cell number per 166 mm² lung tissue. P value as indicated, by t-test. (n ≥ 4 in duplicate/condition). D: effect of α₃₂₅-peptide on fibroblast attachment to fibroproliferative lesions. *P < 0.05, compared with no peptide value; ⁺P < 0.05 compared with scrambled peptide (SCM) value (n ≥ 4/condition in duplicate).

DISCUSSION

The key findings in this study are that u-PAR mediates human lung fibroblast attachment to vitronectin, fibronectin, and collagen through a functional modulation of different integrins and, further, that this interaction extends its influence to cell spreading, to stress fiber formation on a purified matrix proteins, and to closure of an in vitro wound on a complex, serum-derived/fibroblast-synthesized matrix. The direct relevance to lung disease was demonstrable through preferential adhesion of normal lung fibroblasts to areas of fibroproliferative lesions from experimentally induced fibrotic lungs. The association of u-PAR and integrins was demonstrable spatiotemporally using immunofluorescence in which we show co-concentration in the lamellipodia/filopodia of early spreading fibroblasts. The physiological relevance of this association was further supported by the ability of the integrin α-chain homologous peptide to inhibit adhesion, spreading, and migration on multiple purified matrices and to inhibit attachment to experimentally induced fibrotic lung tissue.

The similar kinetics and concentration dependency of attachment of fibroblasts to vitronectin, fibronectin, and collagen and their inhibition by a u-PAR-integrin peptide demonstrate promiscuity of u-PAR-integrin interactions. While promiscuous, we detected a clear preference of u-PAR modulation of integrin function for integrins α₃ > α₅ > α_v, despite similar integrin expression levels. First, the α₃-integrin-homologous peptide inhibited attachment to multiple immobilized anti-integrin-antibodies and the inhibitory effects were greater on fibroblast attachment to immobilized antibodies directed toward intregins α₃ > α₅ > α_v. Second, the α₃-integrin chain homologous peptide, but not the α_v-chain-derived peptide, partially blocked α_vβ₅-mediated attachment to vitronectin, α₅β₁-mediated attachment to fibronectin, and α₁-, α₂-, and α₃-integrin-mediated attachment to collagen. Taken together, these data are most consistent with a model whereby the α₃-peptide binds to u-PAR integrin-association site with high affinity and thereby competes with u-PAR binding to other integrins. Further support comes from the greater sequence homology (60 vs. 40%) of the α₃-sequence-derived peptide than the αv sequence-derived peptide to the potent α_M-sequence-derived peptide (52, 66). Complementary findings (66) of other investigators in cell free systems further support this model as soluble u-PAR bound to immobilized α₃β₁ in a saturable manner that could not be competed by α₅- or α_v-integrin homologous peptides (500-fold molar excess), and soluble u-PAR binds to immobilized α₃β₁ with higher affinity than to α₅β₁.

The physiological importance of u-PAR-integrin interactions is further demonstrable in that they play a dominant regulatory role in fibroblast spreading, they blunt α₅β₁-dependent monolayer wound closure, and they inhibit fibroblast attachment to fibroproliferative lesional areas of fibrotic lung. These matrices are more representative of those seen in vivo during tissue repair. Our interest in studying these interactions in native lung fibroblasts under normal/disease-relevant conditions precluded the use of alternate methods, such as transfection technologies designed to alter the u-PAR and integrin levels or structure. We and others (17, 66) have demonstrated the capacity and specificity of the peptides in both solid phase, solution phase, and cell-based systems. Nonetheless, we cautiously interpret these findings using the peptide to suggest that u-PAR-integrin interactions mediate the cell adhesive and migratory processes necessary for normal tissue repair after injury and may potentially be manipulated to blunt fibrotic reactions in vivo.

Prior work in cells that express high levels of u-PAR, either de novo in neoplastic cells or experimentally induced, demonstrates that putative u-PAR-interacting partners, physiological consequences on cell physiology, and activation of intracellular signaling pathways vary depending on the levels of u-PAR and with the cell type (13, 32, 46, 51, 65). These studies also show that u-PAR will bind the somatomedin B (SMB) domain of vitronectin directly (K_d < 20 nM), an interaction that is enhanced upon u-PAR ligation with u-PA (49, 69). Despite this direct u-PAR-vitronectin (SMB domain) interaction, RGD-dependent integrin binding to vitronectin did contribute to weak cell adhesion to vitronectin, and this two-step process was required for downstream changes in cell morphology (lamellipodial formation) in u-PAR overexpressing renal epithelial cells (34). We similarly found that the u-PAR antibody was capable on blocking adhesion selectively to vitronectin, thereby demonstrating the importance of u-PAR direct mediation of adhesion to vitronectin. However, u-PAR mutants with combined substitution of amino acids in the published integrin binding sites in u-PAR were functionally similar to wild-type u-PAR, suggesting that u-PAR integrin interactions are not required for vitronectin adhesion, signaling, and actin cytoskeletal remodeling (34). Similarly, u-PAR overexpression in murine embryo fibroblasts has shown that u-PAR can induce protrusions and motility, as well as enhance nondirectional motility on vitronectin in a Rac-dependent manner but in a u-PA-independent manner (30). This phenotypic change, while involving integrins, was surprisingly independent of integrin-vitronectin ligation (30). These studies have led to important mechanistic insights into u-PAR biology; however, u-PAR interactions and physiology may be dramatically different in native cells that express u-PAR at a low level. Our data showing a reduction in fibroblast attachment to vitronectin with neutralizing antibodies to either u-PAR or integrin α_vβ₅ documents the cooperative role of u-PAR and α_vβ₅ and extends these findings to the low level u-PAR expressing normal human lung fibroblasts. Further, the ability of the α-chain integrin peptide to reduce adhesion to vitronectin and to immobilized anti-integrin α_vβ₅ monoclonal antibody suggests that u-PAR-integrin interactions have physiological consequences, even in cells expressing low levels of u-PAR.

In support of the importance of integrin u-PAR interactions to cell adhesion/motility dynamics is the hard evidence that 1) mutagenesis studies have mapped the u-PAR-interacting site in integrins [W4 BC loop outside the laminin-5 binding region in α3 (AA 224–232 or Ser 227) in the β₁-chain of α₅β₁], and the areas of u-PAR critical for its interaction with integrins [domain 2 of u-PAR (AA 130–142) of u-PAR for α_vβ₃ and α₅β₁ and domain 3 (AA 240–248) for interaction with α₅β₁ (13, 14, 65, 66, 72)]; 2) su-PAR binding to immobilized integrins and u-PAR-integrin coimmunoprecipitates from u-PAR expressing cell lysates can be blocked by integrin chain peptides (65, 66, 68, 72); and 3) even in u-PAR overexpressing 293 cells, cell morphologic changes, lamellipodial formation, and signal transduction (ERK 1/2 activation) was dependent on integrins (34).

u-PAR has been shown to directly bind integrin α₅β₁ in HT-1080 fibrosarcoma cells, or in purified protein systems (mapped to AA 224–232 in the β₁-chain), thereby stabilizing integrin α₅β₁ in a unique conformation. This conformational change modulates the α₅β₁-function, such that it becomes an efficient mediator of RGD-independent cell attachment to fibronectin (attachment to fibronectin type III repeats 12–14) and enhances fibronectin matrix assembly, while not affecting the RGD integrin site-dependent binding to fibronectin (type III repeats 9–11; Ref. 65). The level of u-PAR expression was again critical in that low u-PAR expressing cancer cells (MCF-7 and T47D) exhibited only RGD-dependent attachment to fibronectin (65). Similarly, colon carcinoma cell proliferation, ERK activation, and u-PAR-integrin α₅β₁ interaction were dependent on the level of u-PAR expression (2). The peptide blocking studies in our lung fibroblasts demonstrate that u-PAR interaction with integrin α₅β₁ decreases cell adhesion to fibronectin and to immobilized anti-integrin α₅β₁ antibody. We would suggest that u-PAR interacts with integrin α₅β₁ in a manner that alters its RGD-dependent binding to the fibronectin type III repeat 9–11 (i.e., different from that observed in cells expressing high levels of u-PAR). These data further support that u-PAR modulates cell processes in a manner dependent on its expression level. Thus the use of cells that have endogenous low levels of u-PAR supports the determination of mechanisms that do not depend on vast molar excesses of interacting components and thereby demonstrates interactions that have physiological relevance to fibroproliferative disorders. The use of a single, unaltered, nontransformed, fibroblast cell strain is a strength, as we have fully characterized the expression levels and physiology of the relevant receptors. However, further testing, in future work, of several fibroblast cell lines with varying receptor levels will surmount the limitations and support greater generalizibility of our results.

The consequences of u-PAR on fibroblast biological events important to the fibroproliferative process extend beyond adhesion and migration. First, u-PAR expression and function is upregulated in fibroblasts from patients with idiopathic pulmonary fibrosis (48). u-PAR can regulate fibronectin matrix fibrillar organization on the cell surface (35). As fibronectin has been shown to guide collagen fibrillar deposition, then u-PAR could regulate dysfunctional fibrillar collagen organization in fibrotic tissue (31, 62). u-PAR has also been shown to modulate TGFβ-induced myofibroblast differentiation (6). u-PAR can regulate plasminogen-induced apoptosis of fibroblasts through enhancement of cell surface plasminogen activation upon ligation with u-PA, an effect blunted by TGFβ-mediated induction of PAI-1 (26). We have shown that repletion of glutathione, a potent antioxidant, in TGFβ-treated fibroblasts restores the TGFβ-induced collagen degradation defect (61) and, furthermore, that the effect of glutathione on overall collagen degradation in TGFβ-treated fibroblasts is a consequence of regulation of plasminogen activation (61).

In conclusion, u-PAR-integrin interactions mediate attachment to vitronectin, fibronectin, and collagen through different integrins in low u-PAR expressing, human lung fibroblasts. Furthermore, this interaction extends its influence to cell spreading, stress fiber formation, closure of an in vitro wound, and attachment to the physiologically relevant matrix from fibrotic lungs. These observations suggest that u-PAR-integrin interactions mediate the cell adhesive and migratory processes necessary for normal tissue repair after injury and may be manipulated to control fibrotic reactions in the lung and other organs.

GRANTS

We thank Janice Buffett for secretarial assistance.