Abstract
Using a cytochemical approach, we examined the role of tissue transglutaminase (tTgase, Type II) in the incorporation of latent TGF-β binding protein-1 (LTBP-1) in the extracellular matrix of Swiss 3T3 fibroblasts in which tTgase expression can be modulated through a tetracycline-controlled promoter. Increased tTgase expression led to an increased rate of LTBP-1 deposition in the matrix, which was accompanied by an increased pool of deoxycholate-insoluble fibronectin. Matrix deposition of LTBP-1 could also be reduced by the competitive amine substrate putrescine. Immunolocalization at the fluorescence and electron microscopic level showed that extracellular tTgase is located at the basal and apical surfaces of cells and at cell-cell contacts, and that LTBP-1 is co-distributed with cell surface tTgase, suggesting an early contribution of tTgase to the binding of LTBP-1 to matrix proteins. LTPB-1 was also found to co-localize with both intracellular and extracellular fibronectin, and increased immunoreactivity for LTBP-1 and fibronectin was found in large molecular weight polymers in the deoxycholate-insoluble matrix of fibroblasts overexpressing tTgase. We conclude that regulation of tTgase expression is important for controlling matrix storage of latent TGF-β1 complexes and that fibronectin may be one extracellular component to which LTBP-1 is crosslinked when LTBP-1 and tTgase interact at the cell surface.
Keywords
IN MOST CELL TYPES, transforming growth factor-β (TGF-β) is secreted in a biologically inactive form, known as large latent TGF-β complex, which consists of latent TGF-β binding protein (LTBP, 120–190 kD) disulfide-linked to the latent TGF-β precursor, also called small latent TGF-β complex. Latent TGF-β precursor consists of the mature TGF-β homodimer (25 kD) noncovalently associated with the amino terminal pro-peptide homodimer (LAP, latency associated peptide; 75 kD) (reviewed in Miyazono et al. 1993). The association of mature TGF-β with LAP renders the growth factor latent and therefore incapable of binding cell surface receptors (Massague 1990). LTBP is not required for latency of the growth factor because it does not bind and inactivate mature TGF-β (Kanzaki et al. 1990). However, accumulating evidence indicates that LTBP plays an important role in the regulation of stored latent TGF-β, which is associated with the extracellular matrix (ECM) (Taipale et al. 1994). Moreover, LTBP appears to be involved in the activation of TGF-β (Flaumenhaft et al. 1993), which results in the dissociation of LAP from the mature growth factor through a mechanism that still needs to be elucidated in vivo.
Four members of the LTBP family have been described thus far (reviewed in Sinha et al. 1998) that display characteristic EGF-like repeats and eight cysteine repeats and form complexes with TGF-β precursor, presumably in specific tissues. The overall structure of LTBPs is similar to the structure of the microfibrillar proteins fibrillin-1 and fibrillin-2, which are compo nents of elastic fibers (Rosenbloom et al. 1993). LTBP Type 1 (LTBP-1) is associated with latent TGF-β1 in the ECM of fibroblasts (Taipale et al. 1996), fetal calvarial cells (Dallas et al. 1995), human dermis (Karonen et al. 1997), and developing embryonic heart (Nakajima et al. 1997). In addition to its role in matrix storage of latent TGF-β, LTBP-1 may also act as an important structural protein (Dallas et al. 1995).
Given the major role of TGF-β in the regulation of ECM synthesis and degradation and its potent effect in controlling proliferation and differentiation (Massague 1990), it might be expected that the amount stored in the ECM and its subsequent activation would be tightly regulated, such that the growth factor is readily available in active form when required. Early in the proposed mechanism of TGF-β1 activation, an important role is played by LTBP-1-mediated targeting of latent TGF-β1 to the ECM, from which both LTBP-1 and latent TGF-β1 are subsequently co-released by proteolysis (Taipale et al. 1994). By utilizing a cell-free system, Nunes et al. (1997) have recently reported that recombinant LTBP-1 is an in vitro crosslinking substrate for tissue transglutaminase and that the enzyme appears to be responsible for matrix association of LTBP-1. Tissue transglutaminase (tTgase Type II; EC2.3.2.13) belongs to a family of Ca2+-dependent crosslinking enzymes, which catalyze the post-translational modification of proteins by forming intermolecular e (γ-glutamyl)lysine bridges. Well-characterized family members include the plasma transglutaminase (Factor XIIIa), which is activated by thrombin during wounding, and the keratinocyte transglutaminase, involved in terminally differentiating keratinocytes (reviewed in Lorand and Conrad 1984; Aeschlimann and Paulsson 1994). Tissue transglutaminase has been implicated in a diverse range of biological functions. However, its role in the crosslinking and stabilization of ECM proteins has been consolidated by a growing amount of evidence (Aeschlimann et al. 1995; Raghunath et al. 1996) and has also been the subject of our investigations (Johnson et al. 1994,1997; Jones et al. 1997).
In a recent study using transfected Swiss 3T3 fibroblasts (Verderio et al. 1998), we showed that inducible expression of tTgase led to an increase in extracellular enzyme. Therefore, increased expression of tTgase may be an important regulatory factor in matrix storage of latent TGF-β1 complex before activation. However, this relationship has not been investigated directly in a cellular model and without using tTgase inhibitors/enhancers, whose action is likely to be associated with secondary tTgase-unrelated effects. Moreover, whereas LTBP-1 and latent TGF-β1 precursor have been co-localized in vivo and in various cell systems and were found to be associated with extracellular microfibrils, the precise extracellular localization of tTgase has not been fully elucidated because cytochemical detection of tTgase in the ECM is difficult. This may be the result of occlusion of recognizable epitopes when tTgase is engaged in extracellular crosslinking and/or bound to ECM proteins, or it may depend on changes in the tTgase protein after exposure to conventional fixatives (Gentile et al. 1992; Verderio et al. 1998). This has limited immunomicroscopic studies on tTgase and LTBP-1 that might clarify the functional interactions between the two proteins and facilitate identification of the matrix protein(s) to which LTBP-1 is crosslinked. In this study we have investigated these interactions by tracking extracellular tTgase in stably transfected Swiss 3T3 fibroblasts, in which tTgase expression can be specifically modulated by a tetracycline-regulated promoter (Verderio et al. 1998), and by localizing the extracellular enzyme with optimized immunocytochemical methods. We report that specific induction of tTgase expression leads to an increased pool of enzyme located at the basal and apical surfaces of cells and at cell-cell contacts. This increased tTgase pool leads to an increased rate of LTBP-1 deposition into the ECM, which is accompanied by a higher level of fibronectin in the deoxycholate-insoluble matrix. We also show that LTBP-1 is co-distributed with both extracellular tTgase and fibronectin and that the interaction of LTBP-1 with tTgase appears to be at its optimum at the cell surface, where tTgase is more concentrated.
Materials and Methods
Cell Culture and tTgase Induction Protocol
Transfected Swiss 3T3 fibroblasts, displaying inducible expression of tTgase (clone TG3 and TG27), were cultured as described by Verderio et al. (1998). To induce tTgase expression before experiments were undertaken, cells were trypsinized and reseeded in medium without tetracycline (which normally was 2 μg/ml) for 72 hr, with one change of medium after 48 hr.
Indirect Immunofluorescence and Image Analysis
Exponentially growing cells were seeded in 8-well glass chamber slides to obtain 105 cells/well (confluent cultures) at the time of the experiment. The cell seeding density was adjusted considering that the doubling time for clone TG3 and clone TG27 is, respectively, 23 hr and 27 hr, when calculated from cells exponentially grown for 2 days. Cells were stained in culture in serum-free Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich, Poole, Dorset, UK) for 2 hr with the respective antibody or in normal serum-containing medium (deprived of selection reagents). The presence of serum made little difference to the results but prevented cells from dying and releasing tTgase when they were deprived of serum factors. After incubation, the antibody was removed with the medium and cells were fixed in 3.7% (w/v) paraformaldehyde in PBS for 15 min at room temperature (RT). Fixed cells were blocked in 3% (w/v) BSA in PBS for 30 min at RT and incubated with fluorochrome-conjugated secondary antibodies in blocking buffer for 2 hr at RT. To localize LTBP-1, cells were incubated with polyclonal antibody to human platelet LTBP-1 (Ab-39) (Pharmingen; San Diego, CA) diluted 1:100 in culture medium. Secondary antibody was an FITC-conjugated swine anti-rabbit IgG (Dako; High Wycombe, Buckinghamshire, UK). Indirect immunofluorescence of fibronectin and tTgase was performed using mouse monoclonal antibodies against, respectively, the fourth Type III repeat of plasma fibronectin (IST-3) (Sigma-Aldrich) diluted 1:100 and the active site of tTgase (Cub7402) (Neomarkers; Freemont, CA), 2 μg/ml. Bound antibodies were detected with an FITC-conjugated rabbit anti-mouse IgG (Dako). Double labeling of LTBP-1 and fibronectin and LTBP-1 and tTgase was performed by using Ab-39 and IST-3, and Ab-39 and Cub7402, respectively. Secondary antibodies were FITC-conjugated swine anti-rabbit IgG (Dako) and TRITC-conjugated goat anti-mouse IgG (Sigma-Aldrich). The concentration of secondary antibodies was approximately 20 μg/ml. Nonimmune rabbit and mouse IgGs (2 μg/ml) were used as controls instead of primary antibodies. Coverslips were mounted with Vectashield (Vector Laboratories; Peterborough, UK) and observed by confocal fluorescent microscopy using a Leica TCSNT confocal laser microscope system (Leica Lasertechnik; Heidelberg, Germany) equipped with an argon/krypton laser adjusted at 488 and 560 nm for fluorescein and rhodamine excitation. For comparison, confocal images of induced and noninduced transfected cells were obtained at constant microscope settings and corresponded to the central section of cells. Fluorescence intensity measurements of LTBP-1 were performed with the Leica TCSNT (version 1.5-451) image processing menu on fixed cells. Between 100 and 300 induced and noninduced cells in random fields were compared in each independent experiment and fluorescence values were expressed per cell number after staining of nuclei with propidium iodide. Image analysis of double immunofluorescence staining was undertaken using the software Image Pro Plus (Media Cybernetics; Silver Spring, MD).
Detection of In Situ tTgase Activity
Cells were seeded into 8-well glass chamber slides as described above, allowed to settle for 4 hr and then incubated in the presence of 0.5 mM fluorescein-cadaverine (Molecular Probes; Eugene, OR) in normal serum-containing medium for a total of 15 hr. Cells were fixed in methanol at −20C for 10 min and mounted (Verderio et al. 1998).
Electron Microscopy
Cells were cultured to confluency on 0.5-cm squares of Melinex, previously conditioned by overnight incubation in serum-containing DMEM. For immunostaining of extracellular tTgase, cells were stained live in culture before fixation, using the method previously detailed for immunofluorescence. Cells were fixed in 1% (w/v) paraformaldehyde and 0.05% (w/v) glutaraldehyde in PBS, dehydrated through increasing concentrations of ethanol, and placed in hydrophilic resin [LR Gold resin and glycolmethacrylate (low acid) (6:4), plus 0.1% bezoinethylether (Taab; Berks UK)]. After several changes of resin, the samples were embedded in plastic trays and the resin polymerized using UV light (360 nm) under nitrogen gas for 24 hr at RT. The Melinex was removed and some of the resin-embedded samples were re-embedded, to allow both en face and vertical sectioning of the cells. Ultrathin sections (60-90 nm) were collected on collodion [2% (w/v) in amyl acetate]-coated nickel grids. Before immunolabeling, sections were blocked with 0.5% (w/v) BSA in TBS (20 mM Tris, 225 mM NaCl, pH 7.6) and exposed to the primary antibody [mouse anti-fibronectin (IST-3, 1:2000), rabbit anti-LTBP-1 (Ab-39, 1:2500) or, for tTgase intracellular staining, mouse anti-tTgase (Cub7402, 1:500) antibody] in blocking buffer containing 0.1% (v/v) Tween-20. Grids were then incubated with the respective colloidal gold-conjugated secondary antibodies in blocking buffer without Tween-20. Secondary antibodies (BioCell; Cardiff, UK) included goat anti-rabbit antibodies (15-nm and 30-nm gold conjugate, diluted 1:60 and 1:30, respectively) and goat anti-mouse antibodies (5-nm and 15-nm gold conjugate, diluted 1:200 and 1:60, respectively). Some grids were silver-enhanced (BioCell) before counterstaining, with 2% aqueous uranyl acetate and alkaline lead citrate. Samples were viewed on a JEOL transmission electron microscope (100 CX-II).
Detection of Fibronectin and tTgase in the ECM of Cell Cultures Using a Modified ELISA
Appropriate cell numbers as defined earlier were seeded in 96-well plates and left to grow for the time required, either 48, 24, or 6 hr. For analysis of deoxycholate (DOC)-insoluble fibronectin, cells were washed twice with PBS, pH 7.4, and solubilized in 0.1% (w/v) sodium deoxycholate containing 2 mM EDTA for 10 min at RT (Jones et al. 1997). The DOC-soluble fraction was collected and proteins were precipitated using trichloroacetic acid to a final concentration of 10% (v/v). Proteins were measured using the bicinchoninic acid method (Brown et al. 1989) to normalize ECM fibronectin levels to 100 mg of DOC-soluble proteins. The remaining DOC-insoluble ECM was washed three times in PBS, pH 7.4, and blocked for 30 min at RT in blocking buffer (5% dry skimmed milk in PBS, pH 7.4). Fibronectin was detected by incubating the ECM with the monoclonal antibody IST-3, diluted 1:1000 in blocking buffer, for 2 hr at RT, followed by three 5-min washes in blocking buffer and incubation with peroxidase-labeled anti-mouse IgG diluted 1:1000, for 2 hr at RT. After another set of washes at the end of the incubation period and one rinse in PBS, pH 7.4, the substrate tetramethyl benzidine was added for color development and the reaction stopped by addition of 2.5 N H2SO4. Absorbances were read at 450 nm. For measurement of extracellular tTgase, the monoclonal antibody Cub7402 (diluted 1:1000) was added directly to cells in live culture. In this way the antibody binds to the extracellular enzyme only. After 3-hr incubation, cells were washed in PBS, pH 7.4, and fixed in methanol at RT for 20 min. The antigen-antibody complex was detected as described for fibronectin. For protein quantification, identical cell numbers were grown in parallel, extracted in 0.1% (w/v) DOC, and proteins assayed in this fraction as described for fibronectin. Data were normalized to 100 mg of DOC-soluble proteins as described for the assay of fibronectin.
Methods were first optimized for antibody concentration shown to give linear responses when 102-105 cells per well were used. Nonspecific binding of primary antibody was obtained using a monoclonal antibody raised to wheat gliadins. Values (typically in the range of A450nm = 0.01-0.05) were subtracted from sample data before inclusion (Table 1).
Isolation of DOC-insoluble ECMs and Immunoblotting for Fibronectin and LTBP-1
Transfected Swiss 3T3 cells, previously induced or repressed for overexpression of tTgase, were reseeded in 60-mm Petri dishes (106 cells) and cultured for 24 hr in serum-containing medium, at which time they reached confluency. Cells were extracted in PBS containing 1% (w/v) DOC and 2 mM EDTA using a modification of the method of McKeown-Longo and Mosher (1983). Any DOC-insoluble material, which is representative of the fibrillar ECM, was pelleted by centrifugation and solubilized in SDS-containing sample buffer. SDS-PAGE of the DOC-insoluble fraction was undertaken in 7.5% (w/v) polyacrylamide resolving gel and 2.5% (w/v) stacking gel according to Laemmli (1970) under nonreducing conditions. Samples were electroblotted using an LKB semi-dry blot system and immunoprobed with primary antibody for either fibronectin (IST-3, 1:1000) or LTBP-1 (Ab-39, 1:3000). Proteins were revealed by incubation with the respective HRP-coupled secondary antibody, using enhanced chemiluminescence. The Western blot film was scanned using a BioRad 9000 volume densitometer (BioRad; Hemel Hempstead, UK) and peak areas were expressed as OD units/μm2.
Statistics
Differences between datasets (shown as mean ± SD) were determined by the Student's t-test or Mann-Whitney test (analysis of modified ELISA) at a significance level of p < 0.05.
Relative levels of extracellular tTgase and DOC-insoluble fibronectin at increasing culture times on specific regulation of tTgase expression a
aDOC-insoluble fibronectin (FN) and extracellular tTgase (TG) were detected in cells induced (-tet) and noninduced (+tet) to overexpress tTgase at increasing culture periods by optimized ELISA performed as described in Materials and Methods. Each value represents the mean ± SD (n = 5); data are expressed as absorbances at 450-nm/100 mg of DOC-soluble proteins. Background absorbance (in the range of 0.01-0.05) was subtracted from data. All values from induced cells (-tet) are significantly different (p < 0.05) from those of noninduced cells (+ tet).
bValues at 48-hr culture significantly different (p < 0.05) from corresponding data at 6-hr culture.
Results
Correlation of tTgase Expression with LTBP-1 Deposition in the ECM of Swiss 3T3 Fibroblasts
After modulation of tTgase expression in Swiss 3T3 fibroblasts by tetracycline, LTBP-1 was localized in the cell lines TG3 and TG27 (Verderio et al. 1998) using a polyclonal antiserum specific for LTBP-1 (Ab-39), which recognizes mouse LTBP-1 (Nakajima et al. 1997). Bound antibody, revealed by indirect immunofluorescence, indicated that LTBP-1 was localized in extracellular fibrillar structures. In a 15-hr culture, LTBP-1 appeared to be more pronounced and organized in cells induced to overexpress tTgase (Figures 1A and 1E) than in cells expressing low endogenous levels of tTgase (Figures 1C and 1G). Of the two cell lines considered in this study, clone TG3, which displayed the highest induction level of tTgase (Verderio et al. 1998), also showed more abundant LTBP-1 fibers (Figure 1A) than clone TG27 (Figure 1E). Similarly, clone TG27, which has little endogenous tTgase when cultured in the presence of tetracycline, showed very little LTBP-1 deposition in the matrix under these conditions (Figure 1G) compared to clone TG3 (Figure 1C). The LTBP-1 fibrillar staining was abolished in both clones by treating the cells with plasmin (Figures 1B-1H), which releases TGF-β complexes from the matrix (Taipale et al. 1994). Quantitative fluorescence measurement of LTBP-1 present in the matrix of clone TG3, induced and noninduced to express tTgase and grown for 15 hr, undertaken in four independent experiments, revealed a significantly increased (39 ± 9%; p < 0.0002) immunolabeling of LTBP-1 in cells overexpressing tTgase compared to cells expressing endogenous level of enzyme. These data confirm the observed correlation between expression of tTgase and LTBP-1 deposition and indicate that the association of LTBP-1 with the ECM is dependent on tTgase-mediated isopeptide crosslinking. To confirm this, clone TG3 cells induced for tTgase overexpression were grown in the presence of the competitive amine substrate putrescine (10 mM), which competes with endogenous amine donors for reactive glutamines, thus blocking protein crosslinking. After 15-hr incubation with putrescine, which we previously showed to reduce extracellular tTgase activity (Verderio et al. 1998), cells had grown to confluency similarly to parallel untreated cultures, without any evident sign of cell toxicity. However, when LTBP-1 was immunolocalized, the typical LTBP-1 staining pattern found in cells overexpressing tTgase (Figure 1I) was considerably reduced in cells incubated with putrescine, as shown in Figure 1L, thus confirming that the incorporation of LTBP-1 into the ECM is mediated by tTgase, via formation of intermolecular e (γ-glutamyl)lysine crosslinks.
Localization of LTBP-1 by immunofluorescence in cultures of transfected Swiss 3T3 fibroblasts induced to overexpress tTgase (-tet) and expressing endogenous levels of tTgase (+tet). After tTgase induction, cells were grown into chamber slides in serum-containing medium to reach confluency after 15-h culture. Cells were then washed in serum-free medium and recultured either in serum-free medium alone (A,C,E,G) or medium containing bovine plasmin (0.1 U/ml) (B,D,F,H) for 1 hr. After removal of plasmin and further washing, cells were stained in culture with anti-LTBP-1 antibody, which was revealed by an FITC-conjugated secondary antibody and detected by laser confocal microscopy. Induced cells were also plated in serum-containing medium in the presence (L) or absence (I) of the competitive amine substrate putrescine (10 mM). LTBP-1 was immunolocalized as described above after the cells were allowed to grow for 15 hr. (A-D,I-L) Clone TG3; (E-H) clone TG27. (A,B,E,F,I,L) Cells induced to overexpress tTgase by tetracycline withdrawal. (C,D,G,H) Cells maintained in medium containing tetracycline. Bar = 10 μm.
Time-dependent deposition of LTBP-1 after regulation of tTgase expression in transfected Swiss 3T3 fibroblasts: comparison with fibronectin and tTgase deposition. (A) Cells (clone TG3) previously induced (-tet) for increased expression of tTgase or expressing endogenous level of tTgase (+tet) were reseeded at adjusted cell densities to obtain confluent cultures at 6, 24, and 48 hr (105, 5 × 104, 2.5 × 104 cells, respectively). At each indicated time interval, cells were immunostained directly in live culture with anti-LTBP-1 antibody (a-f), anti-fibronectin antibody (g-n), or anti-tTgase antibody (o-t), which were revealed by FITC-conjugated secondary antibodies and visualized soon after by laser confocal microscopy. (a-c,g-i,o-q) Cells overexpressing tTgase; (d-f,l-n,r-t) cells expressing low endogenous levels of tTgase. Bar = 10 μm. (B) Quantification by image analysis of the time-dependent deposition of LTBP-1. Fluorescence intensity measurements were performed as outlined in Materials and Methods by comparing a minimum of 300 cells, induced for increased expression of tTgase (-tet) and noninduced (+tet), in five random fields at each time interval. Mean values (±SD) are expressed per cell number after staining of nuclei with propidium iodide. Results shown are from a representative experiment undertaken on a single slide.
To follow the time-dependent binding of LTBP-1 to the matrix, cells were plated and stained for LTBP-1 at 6, 24, and 48 hr of culture. The number of cells seeded was adjusted so that the same number of confluent cells was present at each time interval. Consistency in cell density of both induced and noninduced clones was monitored by transmitted light microscopy (not shown) before acquisition of each image. After 6 hr of culture, hardly any LTBP-1 fibrils were detectable in the ECM of the culture exhibiting low expression of tTgase (Figure 2Ad), whereas very thin initial LTBP-1 fibrils were detected in the corresponding culture induced to express high levels of tTgase (Figure 2Aa). Within 24 hr, cells overexpressing tTgase formed a well-defined network of LTBP-1 fibers (Figure 2Ab). Similar fibers were also observed in the ECM of cells expressing background levels of tTgase, but these were shorter and appeared to be at an earlier stage of maturation, as exemplified in Figure 2Ae. At 2-day culture, LTBP-1 was localized in thicker and longer extracellular fibers in both cultures (Figures 2Ac and 2Af), although confocal microscope depth profiles indicated a more organized distribution of LTBP-1 throughout the layers of induced cells (not shown). These findings strongly suggest that an increased expression of tTgase contributes to a more rapid accumulation of LTBP-1 in the ECM of fibroblasts, leading to a more complex arrangement of LTBP-1 fibrillar structures throughout the culture. Quantification by image analysis of the time-dependent deposition of LTBP-1 confirmed these observations that an increased tTgase expression affects the rate of LTBP-1 deposition at 6 and 24 hr rather than the final level accumulated in a 2-day-old matrix (Figure 2B). In parallel to LTBP-1 staining, fibronectin deposition was also followed by immunostaining for fibronectin. At 6 hr from cell plating, fibronectin fibrils were easily detectable in both the induced and the noninduced cells (Figures 2Ag and 2Al), suggesting that fibronectin fibers appear before LTBP-1. At 24 hr, fibronectin fibrils became progressively coarser and thoroughly distributed in the ECM of both cultures (Figures 2Ah and 2Am). The immunolabeling also appeared to suggest that the appearance of the fibronectin fibers was slightly more dense in cells overexpressing tTgase (Figures 2Ag and 2Ah) at these early time periods. Fibronectin staining eventually assumed an intense meshwork pattern after 48 hr in both cultures (Figures 2Ai and 2An). Immunolocalization of tTgase at the same time intervals revealed that in cells induced to overexpress the enzyme, extracellular tTgase was detectable starting from 6-hr culture (Figures 2Ao-2Aq). However, unlike fibronectin and LTBP-1, tTgase was not found uniformly distributed in the monolayer of 2-day cultured cells but was often present in dense patches localized between cells in areas of high cell confluency (Figure 2Aq). The endogenous level of tTgase expressed by the noninduced cells was hardly detectable at each time interval using immunofluorescence (Figures 2Ar-2At), although we have previously confirmed the presence of extracellular tTgase in noninduced Swiss 3T3 fibroblasts by developing cell surface-specific tTgase activity assays (Jones et al. 1997; Verderio et al. 1998).
Detection of DOC-insoluble Fibronectin and Extracellular tTgase by Modified ELISA
Changes in matrix fibronectin on induction of tTgase expression, which were apparent from immunofluorescent stainings, were examined in more depth using a procedure designed to detect more specifically ECM-associated fibronectin. DOC-insoluble fibronectin was quantified using a modified immunoassay based on ELISA. Cells were first removed from the matrix by extraction with 0.1% DOC containing 2 mM EDTA (to inhibit tTgase-mediated crosslinking of fibronectin) (Jones et al. 1997) before addition of primary antibody. Measurement of DOC-insoluble fibronectin (Table 1, FN) indicated a steady increase in deposition with time, which was significantly different between 6-hr and 48-hr cultures in the cells expressing increased amount of tTgase, whereas in the noninduced cells the apparent increase in fibronectin deposition was not statistically significant. Interestingly, comparison of the induced and noninduced cells indicated that the amount of fibronectin deposited was significantly different in cells with an increased level of tTgase at each time of culture, thus confirming the suggestion obtained by immunofluorescence that the density of fibronectin fibrils present was generally greater in the induced Swiss 3T3 fibroblasts.
Because there is a general degree of difficulty in detecting externalized tTgase by immunocytochemical staining, we sought other methods to quantify the amount of tTgase deposited in the extracellular space over the time periods the cells were cultured. To this end, we developed a modified ELISA optimized for detection of extracellular tTgase. In keeping with the method used for immunofluorescence studies, primary antibody was added to cells in live culture, thus preventing measurement of intracellular enzyme that might bind to the ECM during cell lysis, given its high affinity for fibronectin (Jeong et al. 1995). This modified ELISA (Table 1, TG) not only demonstrated that tTgase deposition was significantly greater in the induced cells at each time of culture, confirming the findings obtained by immunofluorescence microscopy, but was also able to detect the presence of extracellular tTgase in noninduced Swiss 3T3 fibroblasts starting from 6-hr culture. This finding suggests that there is endogenous tTgase in the extracellular space of noninduced Swiss 3T3 fibroblasts, which is able to accumulate LTBP-1 starting from the 24-hr culture time. In the induced cells, the level of extracellular tTgase detected by ELISA significantly increased between 6 and 48 hr, in agreement with immunofluorescence data (Figures 2Ao-2Aq). However, this method did not show any significant increase in the amount of extracellular tTgase deposited in the noninduced cells at the same time interval.
LTBP-1 Co-localizes with In Situ tTgase Activity
In previous reports, tTgase activity was visualized in situ by incorporation of the competitive amine substrate fluorescein-cadaverine into endogenous substrates (Lajemi et al. 1997; Verderio et al. 1998). When used at low concentrations (0.5 mM), this reporter amine is not toxic to cells and can be used to localize some of the tTgase-reactive peptide-bound γ-glutamyl residues that are used by the enzyme during protein crosslinking. In cells induced to overexpress tTgase, the fibrillar nature of LTBP-1 staining, which was revealed by immunofluorescence in a 15-hr culture (Figure 3B), was found to co-localize with in situ tTgase activity (Figure 3A). Fluorescein-cadaverine was incorporated into a network of fibrils that appeared to cover the surface of adherent cells, indicating that under normal conditions tTgase is predominantly active extracellularly. However, the labeling was more prominent in certain fibers likely to correspond to areas of higher cell density, where LTBP-1 was also found to be co-localized (see arrows in Figures 3A and 3B). Interestingly, the fluorescence was almost completely abolished on treatment of monolayer with plasmin (0.1 U/ml) for 1 hr at 37C (not shown), indicating that fluorescein-cadaverine was incorporated into proteins that are plasmin-sensitive, such as LTBP-1 (see also Figures 1B-1H). When measured by image analysis, 63.6 ± 20.8% (n = 4) of the total LTBP-1 was found to co-localize with the fluorescein-cadaverine labeling (the high standard deviation was caused by occasional spots of insoluble fluorescent substrates that remained on the specimen after methanol washing). The highly reduced level of immunoreactive LTBP-1 in the confluent monolayer of noninduced cells expressing low endogenous levels of tTgase (Figure 3D) suggests that fluorescein-cadaverine led to some inhibition of LTBP-1 deposition in the matrix (compare the LTBP-1 immunostaining in Figures 3D and 1C) in a manner comparable to that of high concentrations of putrescine in the high tTgase-expressing cells (Figures 1I-1L).
Co-distribution of LTBP-1 with Extracellular tTgase and Fibronectin
To further evaluate the role of tTgase and fibronectin in the matrix deposition of LTBP-1, studies were undertaken to localize LTBP-1 with respect to extracellular tTgase and fibronectin by double immunolabeling analysis at the fluorescence and electron microscopic levels. Immunolocalizations were undertaken on confluent Swiss 3T3 fibroblasts after 15-hr culture. To facilitate the detection of extracellular tTgase, cells induced for increased expression of the transfected enzyme were used. Double immunofluorescent labeling of LTBP-1 and tTgase revealed that the fibrillar nature of LTBP-1 staining (Figure 4A) partly overlapped with the staining of extracellular tTgase (Figure 4B). Unlike LTBP-1, tTgase appeared to be preferentially concentrated at the cell surface at areas of cell-cell contact (see arrows in Figure 4B) where LTBP-1 and tTgase occurred together, as clearly shown by the yellow-orange fluorescence observed by superimposition of the two fluorochromes (see arrows in Figure 4C). The use of image analysis to quantify the percentage of total tTgase staining co-localized with that of LTBP-1 indicated that 67.3 ± 9.8% (n = 10) of the extracellular tTgase was associated with LTBP-1. At the ultrastructural level, immunogold labeling of tTgase revealed very dense labeling of cells on their apical and basal surfaces and at cell-cell contacts (Figure 5A), thus showing agreement with the immunofluorescence data shown in Figure 4B. The co-localization of LTBP-1 and tTgase was further confirmed by double immunogold labeling of induced Swiss 3T3 cells using antibodies against the two proteins, which were revealed by secondary antibodies combined with different-sized gold particles. A close association between LTBP-1 and tTgase (represented by 30-nm and 15-nm gold particles, respectively) was found in proximity of the cell surface (Figure 5B) and at cell junctions (Figure 5C, arrows), but tTgase and LTBP-1 were found to be less co-distributed in the extracellular matrix (see arrow in Figure 5D). This finding was consistent with the immunofluorescence data (Figures 4A-4C) and suggests that the initial association of LTBP-1 with tTgase takes place close to the cell surface and at cell-cell contacts, where the enzyme is often found in dense clusters (see Figure 5A). When cells were stained for LTBP-1 and fibronectin by immunofluorescence, the fibrillar staining pattern of LTBP-1 (Figure 4D) and fibronectin (Figure 4E) also demonstrated co-localization (Figure 4F). By using image analysis, we found that 66.1 ± 4.9% (n = 3) of the total fibronectin staining co-localized with that of LTBP-1. When double immunogold labeling experiments were performed to demonstrate this association at the electron microscopic level, LTBP-1 and fibronectin, represented by the 15-nm and 5-nm gold particle, respectively, showed extensive co-localization (Figure 6), once again agreeing with immunofluorescence data. Interestingly, LTBP-1 and fibronectin, both of which use the ER/Golgi secretory pathway, were also detected in close proximity to one another in the intracellular environment (see arrowheads in Figure 6), suggesting that they may be present in the same secretory vesicle during their passage to the outside of the cell.
Immunofluorescent staining of LTBP-1 and in situ tTgase activity detected by fluorescein-cadaverine incorporation in transfected Swiss 3T3 fibroblasts. Cells (clone TG3) previously induced (-tet; A,B) or noninduced (+tet; C,D) for expression of transfected tTgase were cultured in the presence of fluorescein-cadaverine for detection of tTgase activity in situ as described in Materials and Methods. The primary antibody to LTBP-1 was added to the cells in fresh culture medium without fluorescein-cadaverine 2 hr before fixation of cells and was revealed with a TRITC-labeled secondary antibody after fixation in methanol. (A,C) Visualization of fluorescein-cadaverine labeling. (B,D) Visualization of LTBP-1 immunolocalization. Arrows point to areas of high tTgase activity, where most LTBP-1 was found (B). Background staining allows appreciation of comparable high cell density in the counterpart of the clone displaying low endogenous tTgase activity (C,D). Bar = 10 μm.
Co-distribution of LTBP-1 with extracellular tTgase and fibronectin by immunofluorescence. Transfected Swiss 3T3 fibroblasts (clone TG3), previously induced for increased expression of tTgase, were grown into chamber slides and cultured for a further 15 hr at confluency. Cells were immunostained live for LTBP-1 and tTgase (A,B, respectively) and LTBP-1 and fibronectin (D,E, respectively), as described in Materials and Methods. LTBP-1 was detected with an FITC-labeled secondary antibody, and tTgase and fibronectin with a TRITC-labeled secondary antibody. C and F represent superimposition of the different immunolabels. Arrows point to areas of concentration of extracellular tTgase (B), which are also areas of co-localization with LTBP-1 (C). Bar = 10 μm.
Co-distribution of LTBP-1 with extracellular tTgase revealed by immunogold electron microscopy. Transfected Swiss 3T3 cells (clone TG3) were induced and cultured under comparable conditions to those shown in Figure 4, except that cells were processed for immunogold electron microscopy according to Materials and Methods. For tTgase staining, the primary antibody was added to the live cell culture before fixation and processing of cells to reveal the presence of tTgase in the ECM. The large particles (30-nm gold) indicate LTBP-1 and the smaller particles (15-nm gold) indicate tTgase. (A) Localization of extracellular tTgase at the cell surface (arrow), cell-cell contacts (arrowheads), and cell-matrix (short arrows). (B) Co-distribution of LTBP-1 and tTgase at the cell surface (arrow). (C) Association of LTBP-1 and tTgase at cell junctions (arrows). (D) Distribution of the two proteins in the ECM (arrow). Bars = 150 nm.
Analysis of DOC-insoluble ECM Proteins by SDS-PAGE
DOC-insoluble ECM proteins were prepared from previously induced and noninduced Swiss 3T3 cells (clone TG3), which were cultured for a further 24 hr to reach confluency under the same conditions shown in Figure 2 and then analyzed by immunoblotting. Proteins were fractionated by nonreducing SDS-PAGE, Western-blotted, and immunoprobed with antibodies to fibronectin and LTBP-1 consecutively, with stripping of the blot between application of each antibody. By probing with anti-fibronectin antibody, increased amounts of fibronectin polymers, consisting of both disulfide-linked and e (γ-glutamyl)lysine-linked proteins, were found at the top of both the resolving and stacking gel of the induced fibroblasts (Figure 7B) compared to noninduced cells (Figure 7A). Given that polymer formation is related to fibronectin fibril assembly, this result is in agreement with a significantly increased level of fibronectin measured by ELISA in the DOC-insoluble ECM of induced fibroblasts compared to noninduced fibroblasts (Table 1, FN). Similarly, the immunoprobing of the same blot for LTBP-1, which is reactive in all its forms with Ab-39 only in a nonreduced state (Taipale et al. 1994), indicated the presence of high molecular weight polymers at the top of the stacking gel of the induced cells, as shown for fibronectin (Figure 7D). These immunoreactive aggregates were detected with some difficulty, and it cannot be ruled out that they are also present in the noninduced cells, but the smaller amount is beyond the detection limits. Low levels of free LTBP-1 and large latent TGF-β1 complex, likely to correspond to the 120-140-kD and 220-240-kD immunoreactive fragments, respectively (Figure 7C,D), were detectable in the DOC-insoluble fraction in the absence of plasmin release of LTBP-1 from the ECM, in agreement with previous work (Taipale et al. 1994). However, a densitometric scan of the Western blot film indicated a two- to threefold increase of all peak areas of LTBP-1 immunoreactivity in cells overexpressing tTgase. The finding that cells induced to overexpress tTgase have more of the high molecular weight LTBP-1-containing polymers and LTBP-1 fragments indicates that the matrix LTBP-1 fraction is directly related to the increased presence of extracellular tTgase activity.
Discussion
The storage and subsequent release of matrix-bound latent TGF-β is now recognized as a key mechanism in matrix remodeling that has important consequences for tissue repair, wound healing, and development (Border and Ruohslati 1992; Nakajima et al. 1997). However, the in vivo events regulating this activation process are still not well understood. Recent evidence suggests that the association of large latent TGF-β1 complex with the ECM may be an important factor in both the storage of the molecule and its subsequent proteolytic activation, and that this association is mediated by transglutaminase (Kojima et al. 1993; Nunes et al. 1997). Regulation of tTgase expression in cells, particularly the amount of extracellular tTgase, may therefore be a key factor in localizing and concentrating latent TGF-β1 complexes before activation during matrix remodeling. To address this question and to identify the potential substrates to which LTBP-1 is bound in the matrix, we used a novel model cell system in which tTgase expression is under the control of the tetracycline-regulated promoter (Verderio et al. 1998). Using this experimental model, we demonstrate directly, for the first time by immunocytochemistry, that the deposition of LTBP-1 in the ECM correlates with tTgase expression and externalization in two separate Swiss 3T3 cell lines (TG27 and TG3) in which tTgase was specifically induced at the transcriptional level by a tetracycline-controlled promoter. Further confirmation that changes in LTBP-1 deposition were due to increased tTgase crosslinking activity was gained by inclusion of the competitive amine substrate putrescine in the medium, which reduced the amount of LTBP-1 found in the matrix during a 15-h culture. In addition, when the amine substrate fluorescein-cadaverine was added to the culture medium at low concentrations to visualize tTgase activity in situ, it became incorporated into a cell surface and fibrillar protein that co-localized with LTBP-1 and was plasmin-sensitive, suggesting LTBP-1 to be an extracellular tTgase substrate, containing reactive γ-glutamyl residues, in these cells. In a time-dependent study in which induced and noninduced cells were grown to confluency for 6, 24, and 48 hr, we demonstrate that cells overexpressing tTgase accumulated LTBP-1 in the matrix more rapidly than those carrying endogenous levels of tTgase. Moreover, the distribution pattern, particularly at 24-hr culture, appeared more fibrillar and complex, thus confirming the importance of extracellular tTgase in LTBP-1 deposition. However, the small differences in LTBP-1 immunostaining shown between the 48-hr cultures of induced and noninduced cells indicate that the total amount of LTBP-1 bound by tTgase in a long-term culture is likely to be the same in fibroblasts with endogenous and increased level of tTgase. We do not know the reason for this, but our findings (not shown) rule out a saturation of the immunolocalization assay at the antibody concentration employed. A possible explanation is that the matrix becomes saturated with LTBP-1 at 48 hr, which is in keeping with any enzyme reaction given that there is only one variable, the enzyme concentration. Parallel staining of fibronectin at similar time intervals indicated the appearance of a fibronectin network before the deposition of LTBP-1, suggesting that fibronectin may be essential for LTBP-1 deposition to take place. The density of the fibronectin network appeared slightly greater in cells overexpressing tTgase. This microscopic observation was confirmed by quantifying fibronectin with a modified ELISA technique. This procedure, which was specifically designed to detect DOC-insoluble, ECM-associated fibronectin, showed that cells displaying increased tTgase accumulated significantly more fibronectin fibrils at each time of culture. Interestingly, differences measured in the pool of ECM fibronectin were greater than predicted from immunofluorescence cell staining of the total pool of extracellular fibronectin, thus showing that modulation of tTgase affects the insoluble fibronectin fraction. In the high tTgase-expressing cells, tTgase accumulation paralleled that of LTBP-1, whereas in the cells expressing endogenous tTgase, deposition of extracellular tTgase was clearly detectable by immunofluorescence cell staining only after 2-day culture. However, as previously reported (Verderio et al. 1998), there is a shortage of data showing the immunocytochemical detection of tTgase in the extracellular space, because tTgase is not easily accessible once associated with ECM proteins. However, when a modified ELISA was used, which is a more sensitive method optimized to analyze externalized tTgase, the presence of the enzyme was clearly detectable in both induced (high-expressing tTgase) and noninduced (expressing endogenous tTgase) TG3 fibroblasts starting from an early time of culture.
Co-distribution of LTBP-1 with fibronectin revealed by immunogold electron microscopy. Cells were prepared as described in Figure 5, except that staining was undertaken with a secondary antibody conjugated to 5-nm gold particles for fibronectin and 15-nm gold particles for LTBP-1. Arrowheads point to intracellular LTBP-fibronectin aggregates. Bar = 150 nm.
SDS-PAGE analysis of fibronectin and LTBP-1 in the DOC-insoluble matrix of Swiss 3T3 cells expressing high and low amounts of tTgase. Cells previously induced (-tet; B,D) or noninduced (+tet; A,C) for tTgase expression were reseeded and cultured for a further 24 hr in serum-containing medium. Cells were extracted with 1.0% (w/v) DOC-containing buffer, and any insoluble material collected by centrifugation was solubilized in nonreducing SDS-containing sample buffer and electrophoresed according to Materials and Methods. After Western blotting, the blots were probed consecutively for fibronectin (FN; A,B) and LTBP-1 (LTBP; C,D), with a stripping procedure between each antibody. Mr markers are shown; large arrows denote the top of the resolving and stacking gels and arrowhead denotes dimeric fibronectin.
Our results strongly suggest that tTgase and LTBP-1 interact close to the cell surface. Superimposition of the different staining patterns for tTgase and LTBP-1 obtained by double immunofluorescent labeling suggests that co-localization is always at its greatest around the immediate vicinity of the cell. This observation was confirmed at the electron microscopic level by use of double immunogold labeling of tTgase and LTBP-1, which showed a close co-distribution of the two proteins, particularly at the cell surface and at areas between cells. This finding was also supported by immunogold staining of tTgase, which showed a preferential localization of the enzyme in dense clusters close to the cell surface and at areas of cell-cell contact. It is noteworthy that these data provide the first evidence at the electron microscopic level for the extracellular location of tTgase antigen obtained in cells in culture (Figure 5A). The degree of partial co-localization of tTgase with LTBP-1 that was measured at the fluorescent level is consistent with the association of tTgase with other proteins at the cell surface.
When the distribution of LTBP-1 and fibronectin was analyzed by double immunogold labeling, LTBP-1 was found to be associated with fibronectin in the extracellular and intracellular environment, suggesting that these two proteins share a common secretory vesicle. This is the first evidence to indicate that LTBP-1 and fibronectin may share a common secretory pathway. However, it should be noted that other workers have shown co-localization of LTBP-1 with fibrillin (Raghunath et al. 1998). It is possible that this association occurs later in the cycle of events involving secretion and deposition of these matrix proteins. In contrast to LTBP-1 and fibronectin, tTgase secretion from the cells is unlikely to follow the Golgi/ER pathway because tTgase does not possess a leader sequence.
It is known that purified tTgase has high affinity for fibronectin (Jeong et al. 1995) and that fibronectin is a crosslinking substrate for tTgase in Swiss 3T3 fibroblasts under physiological conditions. Moreover, tTgase is closely associated with fibronectin in the extracellular environment on the surface of fibroblasts (Verderio et al. 1998). We show here that there is also a preferential co-localization of tTgase and LTBP-1 at the cell surface. This suggests that the involvement of tTgase in LTBP-1 crosslinking is likely to occur when the two proteins interact at this site. Our studies suggest that fibronectin would be a good candidate to which LTBP-1 is immobilized by tTgase, because both proteins are tTgase substrates, they both co-localize with tTgase at the cell surface, and they are both present in greater quantities in large molecular weight polymers from cells induced to overexpress tTgase when ECM extracts are analyzed by Western blotting. Two possible reasons spring to mind why cells overexpressing tTgase are able to deposit more LTBP-1 and ECM fibronectin. Increased extracellular tTgase activity increases the efficiency of the cell in laying down the fibrillar matrix as a result of increased stabilization. Our data indicate that cells overexpressing tTgase show an increased pool of matrix fibronectin compared to their noninduced counterparts. This agrees with studies in which exogenously added activated plasma transglutaminase (Factor XIIIa) was observed to increase the rate of fibronectin matrix assembly by cultured fibroblasts (Barry and Mosher 1989). We have previously demonstrated that Swiss 3T3 cells overexpressing tTgase contain more ∊ (γ-glutamyl)lysine crosslinks than their noninduced counterparts and also display nonreducible high molecular weight polymers of cellular fibronectin (Verderio et al. 1998). On the other hand, increased deposition of LTBP-1 in the ECM through tTgase-mediated crosslinking may also result in increased activation of TGF-β, leading to autocrine stimulation (Bernasconi et al. 1995) whereby cells are stimulated to increase the amount of ECM proteins that they secrete. Such an idea is compatible with recent findings showing that the binding of antibody to LTBP-1 in the matrix of cultured rat mesangial cells leads to inhibition of ECM expression induced by stretch (Hori et al. 1998), thus confirming the importance of LTBP-1 in the activation of TGF-β1.
The binding of large latent TGF-β1 complex to the matrix has been suggested as another level of control for TGF-β, whereby this important cytokine can exert its effects on cells at a later date after release from the matrix and subsequent activation (Taipale et al. 1994; Nunes et al. 1997). Our data indicate that binding of LTBP-1 to fibronectin is regulated via the expression and externalization of tissue transglutaminase. Therefore, the amount of stored TGF-β, which is dependent on extracellular tTgase activity, may play a key role in determining whether wound healing will be normal, leading to minimal scarring, or whether it becomes a pathological event, as in fibrosis. Both increased expression of tTgase and excess TGF-β have been implicated in a number of fibrotic disease states (Griffin et al. 1979; Mirza et al. 1997; Johnson et al. 1997). Local activation of TGF-β has also been recently shown to be induced by αvβ6 integrin during injury (Munger et al. 1999). Active TGF-β has been shown to induce tTgase in a number of cell types (George et al. 1990), which may be explained by the recent finding of a TGF-β response element in the tTgase promoter (Ritter and Davies 1998). Induction of tTgase in these disease states, which generally involve the interaction of different cell types, could therefore result in both the increased storage and activation of further TGF-β, leading to a net positive cooperative cycle. Regulation of the amount of stored LTBP-1 may therefore be of considerable clinical importance in these diseases. Progressive induction of tTgase expression, recently observed in areas such as the anterior necrotic zone and the interdigital web during limb development (Nagy et al. 1998), may also be the consequence of a tTgase and TGF-β induction/activation cycle, thus suggesting an important role for the enzyme in tissue remodeling.
In conclusion, this study is the first to demonstrate that the specific regulation of tTgase expression leads to changes in stored LTBP-1 in the matrix when detected directly by immunofluorescence. Using a novel modified ELISA, we also demonstrate for the first time that increased expression of tTgase leads to an increased deposition of fibronectin in the matrix. By modulating tTgase expression in cells, we also provide novel data on the co-localization of tTgase and LTBP-1 and provide evidence strongly suggesting that the crosslinking of LTBP-1 by tTgase might occur when both proteins are in close association with each other at the cell surface. This study has also demonstrated the close association of LTBP-1 with fibronectin, both intracellularly and in the extracellular matrix, and given that both these proteins are found at greater levels in high molecular weight protein polymers in cells overexpressing tTgase, our data strongly suggest that fibronectin is one of the proteins to which LTBP-1 is linked via tTgase.
Footnotes
Acknowledgements
Supported by a grant from the EPSRC (Ref GR/L43688).
We wish to thank P.J.A. Davies (University of Texas Health Center, Houston) for interesting discussions on TGF-β/tTgase induction in fetal development, M. Armitage and J. Adams (Smith & Nephew; York, UK) for providing assistance in image analysis, and R. Jones, A. Mezzogiorno, and A. Hargreaves for helpful advice in this study.