An acidic microenvironment sets the humoral pattern recognition molecule PTX3 in a tissue repair mode

Role of PTX3 in tissue repair

We investigated the role of PTX3 in models of tissue damage, including skin wounding, liver, and lung sterile tissue injury. A model of full-thickness excision dorsal skin wounding was first used. Wounding was associated with induction of increased levels of PTX3 mRNA and protein in the skin and blood (Fig. 1, A and B). By confocal microscopy analysis, on day 2 PTX3 was found in the clot and mostly associated with aligned bundles of cells migrating toward the wound site, and epithelial outgrowth (Fig. 1, C and D). On day 7, PTX3 was scattered in the scab and damaged dermis (Fig. 1 C). At healing (day 14), PTX3 expression was confined to neo-formed tissue and residual scab (Fig. 1 C). In particular, on day 2 PTX3 expression was associated with bundles of remodeling cells that collectively invaded the wound site, including urokinase plasminogen activator (uPA) positive (Fig. 1 E) and uPA receptor (uPAR) positive (Fig. 1 F) cells, CD68⁺ macrophage-like cells and platelet-derived growth factor receptor (PDGFR)⁺ fibroblasts and/or mesenchymal cells (Fig. 1 G). On day 2, PTX3 was detected by flow cytometry in neutrophils (CD11b⁺Ly6G⁺), macrophages (CD11b⁺Ly6C⁺F4/80⁺), and stromal cells, in particular PDGFRα⁺ mesenchymal cells, PDGFRα⁺/fibroblast activation protein⁺ (FAP⁺) reactive fibroblasts, and α-smooth muscle actin⁺ (SMA⁺) smooth muscle cells or myofibroblasts, and at low level in vascular endothelium (CD31⁺; Fig. 1 H and Fig. S1). Expression of PTX3 was confirmed at the level of mRNA transcripts in sorted macrophages and mesenchymal cells (Fig. 1 I), but not in neutrophils that store preformed PTX3 in granules (Jaillon et al., 2007). On day 7, PTX3 was detected in PDGFRα⁺ and PDGFRα⁺FAP⁺ mesenchymal stromal cells but not in leukocytes (Fig. 1 H). Neutrophils, macrophages, and PDGFRα⁺ mesenchymal cells also had PTX3 associated on the cell membrane in the course of the skin repair (Fig. 1 H), although, as shown by confocal images (Fig. 1, D–G), PTX3 was mainly associated with the pericellular matrix. Indeed, PTX3 is generally rapidly released from the cells, in particular by neutrophils (Jaillon et al., 2007).

TLRs and IL-1 are potent inducers of PTX3 and are involved in the orchestration of tissue repair (Medzhitov, 2008; Dinarello, 2009). It was therefore important to assess the pathways upstream of PTX3 in skin wound healing. Induction of PTX3 was almost completely abolished in blood and at the wound site in MyD88-deficient mice, whereas a partial reduction was observed in Il1r^−/−, Tlr3^−/−, Irf3^−/−, and Ticam1^−/− mice (Fig. 1, J and K). These results suggest that induction of PTX3 is downstream of TLR sensing and IL-1 amplification, with the contribution of both MyD88- and TRIF-dependent pathways.

Early grossly evident acceleration of reduction of wound area was observed in Ptx3^−/− mice (Fig. 2, A and B). Histological analysis revealed that Ptx3^−/− mice had significantly thicker clots compared with controls (138.58 ± 25.8 µm and 78.59 ± 28.25 µm, in Ptx3^−/− and WT mice, respectively; P = 0.0001; n = 8) on day 2, and thicker scabs (1,112.08 ± 189.13 µm compared with 645.09 ± 87.16 µm in WT mice; P = 0.044; n = 8) on day 7, characterized by excessive accumulation of eosinophilic material (Fig. 2 C) and fibrin (Fig. 2, D and E) throughout the observation time. In addition, an augmented content of factors (CD62P, TGFβ, PDGF, serotonin, CXCL1, and CXCL2) possibly derived from platelets (Nurden, 2011; Semple et al., 2011; Fig. 2 F), and thrombin (Fig. 2 G) were measured in PTX3-deficient wounds on day 1. Higher expression of markers of differentiated and activated fibroblasts was observed at day 2 (Fig. 2, H and I). Some of these factors (e.g., PDGF, TGFβ, serotonin, and thrombin) acting on stromal cells may well be responsible for increased early wound contraction and reduction of wound area observed in Ptx3^−/− mice (Singer and Clark, 1999; Schäfer and Werner, 2008). Despite the accelerated reduction of wound area, PTX3 deficiency was associated with a defective reepithelialization at day 7 (the epithelial gap was 3,002 ± 432 µm and 1,306 ± 503 µm in Ptx3^−/− and WT mice, respectively; P = 0.026; n = 7) and epithelial hyperplasia (10.8 ± 0.8 epithelial layers in Ptx3^−/− vs. 6.0 ± 1.4 in WT mice; P = 0.0001; n = 8) and defective development of skin appendages at day 14, suggesting a defective mature tissue formation (Fig. 2 C). At later time points (25 d after wounding), the epithelial layers did not differ in the two groups (3.2 ± 1.1 epithelial layers in Ptx3^−/− skins compared with 3.4 ± 1.3 in Ptx3^+/+ skins; n = 5), or in unwounded skins (unpublished data), indicating that epithelial hyperplasia associated with altered skin wound healing in PTX3-deficient mice was a transient event. Vascular density was not affected by PTX3 deficiency (day 2, 11 ± 7 vs. 17 ± 11; day 7, 42 ± 16 vs. 47 ± 16; day 14, 49 ± 17 vs. 43 ± 16 CD31⁺ vessels/mm² in Ptx3^+/+ vs. Ptx3^−/−; n = 4–6; P = ns).

At the beginning of new tissue formation, remodeling cells enter the wound, allowing the replacement of the fibrin matrix by mature granulation tissue rich in type I collagen, as well as other extracellular matrix (ECM) proteins (Romer et al., 1996; de Giorgio-Miller et al., 2005). PTX3 deficiency was associated with alterations in the organization of remodeling cells invading the injured area. As previously shown (Singer and Clark, 1999), in WT mice, CD68⁺ macrophages (Fig. 2 D) and PDGFRα⁺ mesenchymal cells (not depicted) collectively migrated toward the wound bed and the clot, creating patent passages within fibrin matrix (Fig. 2 D). In Ptx3^−/− mice, CD68⁺ macrophages and PDGFRα⁺ cells were scattered in the abundant fibrin-rich matrix and failed to create patent passages, suggesting defective directional migration and invasive phenotype (Fig. 2 D). Wound collagen was evaluated by collagen I and Sirius red staining. On day 7, collagen content was higher in Ptx3^−/− tissues after wounding compared with WT (Fig. 2 J). Treatment with recombinant PTX3 reversed the phenotype of Ptx3^−/− mice (Fig. 3, A–G) and had a significant effect on collagen deposition at one time point (day 7) in WT mice, suggesting that endogenous PTX3 is sufficient for gross biological function. The N-terminal domain, but not the C-terminal one, was necessary and sufficient to revert the phenotype (Fig. 3, A–G). Thus, after skin wounding, PTX3 deficiency was associated with increased clotting and excessive fibrin accumulation on day 2, followed by thicker scabs, augmented collagen deposition, and delayed reepithelialization on day 7, and epithelial hyperplasia and defective mature tissue formation on day 14, which are all signs of defective repair (Romer et al., 1996).

In an effort to extend these observations to models of sterile injury, we studied CCl₄-induced liver injury and acid-induced lung damage, in which genetic deficiencies causing prothrombotic responses are associated with impaired tissue remodeling and fibrosis (Pohl et al., 2001; Idell, 2003; Calvaruso et al., 2008). PTX3 plasma levels increased in the blood after acute liver injury (8–48 h), as well as chronic injury (6 wk), and returned toward basal levels at resolution of fibrosis (Fig. 4 A). PTX3 localized in necroinflammatory areas associated with GR1⁺ cells (8–48 h), and in the portal tracts associated with uPA⁺, PDGFR⁺, and CD68⁺ cells (6 wk; unpublished data). PTX3 deficiency did not affect necrosis (8–48 h; unpublished data); however, it was associated with an increased number of thrombosed centrolobular veins 8 h after CCl₄ injection (69.61 ± 17.16% in Ptx3^−/− mice compared with 20.80 ± 8.46% in WT mice; n = 4; P = 0.011; Fig. 4 B); augmented deposits of eosinophilic material rich in fibrin in necroinflammatory areas of lobular spaces close to the centrolobular vein at 24 and 48 h, and in the portal tracts after chronic injury (6 wk; Fig. 4, B and C); augmented fibrosis assessed by measurement of liver α-SMA⁺ immunoreactive area (Fig. 4 D); and OH-proline content (Fig. 4 E), Sirius red, and collagen I staining at portal septa (Fig. 4, F and G; 6 wk). Similarly, in the acid-induced lung injury model, we observed increased fibrin deposition at day 7 (Fig. 4 H) and increased collagen content at day 30 in Ptx3^−/− mice (Fig. 4 I). Finally, PTX3-deficient mice were more susceptible to arterial thrombosis in a model of FeCl₃-induced vascular injury (Koschnick et al., 2005; Fig. 5 A), and exogenous PTX3 reversed the phenotype in Ptx3^−/− mice and reduced the severity in Ptx3^+/+ mice (Fig. 5, B and C). These results suggest that in different models of tissue injury, PTX3 deficiency is associated with increased clotting and fibrin deposition and alteration of tissue repair, thus indicating that PTX3 has a protective role in tissue damage.

PTX3 binds P-selectin and has a negative regulatory function in selected models of inflammation (Bottazzi et al., 2010; Deban et al., 2010). Given the role of leukocytes in the orchestration of tissue repair, it was important to ascertain whether the altered wound healing observed in Ptx3^−/− mice was in fact the expression of deregulated leukocyte recruitment. The analysis of leukocyte recruitment by immunohistochemistry, flow cytometry and confocal microscopy in skin and liver, and myeloperoxidase (MPO) content in skin wound lysates showed no difference in the total number of recruited cell populations (polymorphonuclear cells and macrophages) in the two groups (Fig. 6, A–C and not depicted). Moreover, skin wound healing was performed in double-deficient Ptx3^−/− P-selectin^−/− mice. First P-selectin deficiency had no notable effect on skin healing (Fig. 6 D). Moreover, PTX3/P-selectin double-deficient mice showed a wound healing phenotype identical to that of Ptx3^−/− mice (Fig. 6 D), suggesting that interaction between PTX3 and P-selectin does not play a major role in the alterations of wound healing observed in PTX3 deficient hosts. Thus, altered P-selectin-dependent leukocyte recruitment is not responsible for the alterations in tissue repair observed in Ptx3^−/− mice. However, as stated above, in Ptx3^−/− mice CD68⁺ macrophages and PDGFRα⁺ cells failed to collectively migrate into the fibrin-rich skin wound site.

Interaction of PTX3 with fibrin and plasminogen at acidic pH

The finding that excessive clot and thrombi formation and fibrin deposition and persistence were consistent phenotypes in tissue damage models in Ptx3^−/− mice and the role of fibrin as a provisional matrix protein guiding subsequent repair (Carmeliet et al., 1994; Bugge et al., 1996; Romer et al., 1996; Singer and Clark, 1999; Schäfer and Werner, 2008) prompted us to investigate the interaction of PTX3 with components of the coagulation and fibrinolytic cascade. We were also prompted in this direction by the finding that PTX3 interacts with ficolins via their fibrinogen domain (Ma et al., 2013). Previous studies indicating that PTX3 induced procoagulant activity in endothelial cells and monocytes (Napoleone et al., 2002) could not be reproduced on repeated attempts (unpublished data). PTX3 bound fibrinogen (FG) and plasminogen (Plg), but it did not bind other components of the coagulation cascade (listed in Materials and methods; unpublished data). When FG, fibrin, and Plg were immobilized on plastic wells, pH-dependent binding was observed, with optimal interaction at pH 5.5–6.5 (Fig. 7 A). In the same setting, C-reactive protein (CRP) did not bind FG, fibrin, or Plg (Fig. 7 B). Binding of PTX3 to fibrin and Plg was Ca²⁺ independent and unaffected by treatment with endoglycosidase F3 or/and neuraminidase (unpublished data). The apparent dissociation constants (K_d) evaluated by nonlinear fitting analysis at pH 6.0 were 31.2 ± 10.5 nM, 7.5 ± 0.9 nM and 66.7 ± 25.1 nM for interaction with FG, fibrin, and Plg, respectively (Fig. 7 C). The N-terminal domain bound FG, fibrin, and Plg, whereas the C-terminal pentraxin domain bound with low-affinity only Plg (Fig. 7 D). The analysis of binding was extended by surface plasmon resonance (Fig. 7, E–G), which showed higher affinity (K_d) of PTX3 or the N-terminal domain for Plg at pH 6.0 than at pH 7.4, due to a faster association rate (K_on) at acidic pH. Addition of mAb MNB4, directed against the PTX3 N-terminal domain, completely blocked PTX3 binding to fibrin but not to Plg. mAb MNB5, which recognizes the C-terminal domain, had no effect (Fig. 7 H). Moreover, no competition for binding to PTX3 was observed between fibrin and Plg. Actually, in the presence of PTX3, Plg bound to fibrin more efficiently; similarly, in the presence of Plg, PTX3 bound to fibrin more efficiently (Fig. 7 I). In competition experiments PTX3 binding to Plg was inhibited by l-lysine (Lys; ligand of Lys-binding sites in Plg Kringles [KRs] 1, 2, 3, and 5) and l-arginine (specific ligand of KR5; Law et al., 2012; not depicted). Moreover, PTX3 bound Plg KR3-KR5 fragment, but not KR1-KR3 fragment (Fig. 7 J). These data suggest that PTX3 interacts with Plg KR5, which triggers the Plg conformational change leading to the open form, and does not interact with Plg KR1 domain, which mediates the initial recruitment of closed Plg to the C-terminal Lys moieties of fibrin or cell surface (Law et al., 2012). All together, these data suggest that PTX3 interaction with fibrin and Plg occurred via different sites in the PTX3 N-terminal domain and that interaction of PTX3 with Plg did not interfere with fibrin binding to Plg, and occurred via a different site.

No pH-dependence was observed in the PTX3 interaction with microbes such as A. fumigatus conidia, P. aeruginosa and Escherichia coli, or OmpA from K. pneumonia (not depicted).

Functional relevance of PTX3 interaction with fibrin and plasminogen

In an effort to investigate the actual functional relevance of the interaction of PTX3 with FG/fibrin and Plg, an in vitro fibrinolysis assay was used. At pH 6.0 fibrin gel degradation mediated by plasmin and triggered by uPA and tissue plasminogen activator (tPA) was potentiated in the presence of PTX3, which had no activity on its own (Fig. 7 K). No potentiation was instead observed at neutral pH (pH 7.4, not depicted, and 7.6; Fig. 8).

The finding that PTX3 interacted with FG/fibrin and Plg, and facilitated fibrin dissolution at low pH raised the issue of the actual in vivo occurrence and functional relevance of this interaction during wound healing. After skin wounding, colocalization of PTX3 and fibrin and Plg was observed by confocal microscopy in the clot (Fig. 8 A) and in the pericellular matrix of invading cells or associated with Plg⁺ cells at day 2 (Fig. 8 B). Similarly, PTX3 colocalized with fibrin upon liver and lung injury (unpublished data). Treatment with dichloroacetate (DCA) to revert local acidosis inhibited the colocalization between PTX3 and fibrin or Plg, but not between fibrin and Plg (Fig. 8, C and D). Two-photon microscopy was then used to simultaneously measure the pH using a pH-sensitive fluorescent probe (BCECF) and to visualize PTX3 and fibrin and Plg colocalization in the wounded skin area in vivo (Supplemental material). The surface of the skin normally has an acidic extracellular pH with values ranging from 4.5 to 6, due to the hydrolysis of skin lipids to free fatty acids, whereas lower epidermis and dermis have a neutral-alkaline extracellular pH in the range of 7–9 (Hanson et al., 2002). Accordingly, in the dermis (depth 10–20 µm) of unwounded skin, the pH measured by two-photon microscopy was 7.63 ± 0.5 (mean ± SEM; range 6.53–8.73; mean ± σ). In contrast, in the proximity of the damaged dermis (the analysis of the pH in the wound itself was technically impossible because of the endogenous autofluorescence of the clot) the pH was acidic (6.77 ± 0.5; P = 0.011; range, 5.67–7.87; Fig. 8 E and Fig. S2), which suggests cellular metabolic adaptation to trauma-induced tissue hypoperfusion and hypoxia (Schneider et al., 2007). In the z-projection (14 optical sections of 10 µm) of the dermis between the perilesion area and the wound, a significantly higher colocalization of PTX3 and plasminogen was measured in a representative lower pH region than in a representative higher pH region (Fig. 8, F and G) of the wound site, with Pearson’s coefficients of 0.28 ± 0.03 and 0.15 ± 0.03, respectively (P = 0.007). Similarly, a significantly higher colocalization of PTX3 and fibrin was observed in the lower pH area (Pearson’s coefficients: 0.83 ± 0.05 and 0.63 ± 0.07 in lower pH and higher pH regions, respectively, P = 0.005). Binding of PTX3 to fibrin and Plg (Fig. 8 H), and amplification of cell free fibrinolytic activity by PTX3 (Fig. 8, I and J) were confirmed at the mean pH measured by two-photon microscopy in wounded skin (pH 6.77). Moreover, PTX3-deficient wounds showed decreased Plg deposition and plasmin formation (Fig. 9, A and B), and lower levels of D-dimer compared with WT lesions at day 2 after wounding (Fig. 9 C).

Macrophages are key players in tissue repair, orchestrate pericellular fibrinolysis in vitro and in vivo (Carmeliet et al., 1994; Cook et al., 2006), and are major sources of PTX3 (Martinez et al., 2009). Because of these factors, and in light of the defective invasive phenotype of macrophages observed in Ptx3^−/− mice (Fig. 2 D), we assessed pericellular fibrinolysis mediated by Ptx3^−/− macrophages. As shown in Fig. 9 D, Ptx3^−/− peritoneal macrophages showed defective fibrinolytic activity compared with WT cells. Similar results were obtained with PTX3-deficient and -sufficient embryonic fibroblasts (Fig. 9, E and F).

In an effort to assess whether an interaction with key components of the coagulation/fibrinolysis system actually underlies the role of PTX3 in wound healing, pharmacologic inhibitors of fibrin deposition with different molecular targets were used. These included antithrombin III (ATIII; an inhibitor of thrombin and Factor Xa), Argatroban (thrombin inhibitor), and Batroxobin (which depletes fibrinogen). In the presence of these inhibitors of the coagulation cascade, the differences in skin wound healing (accelerated reduction of wound area and collagen accumulation) between PTX3-competent and -incompetent mice were abolished (Fig. 10, A–F).

Reagents

The following reagents were used: pyrogen-free saline (S. A. L.; Bergamo); RPMI 1640 medium (Biochrom); 200 mM _l-glutamine (Biochrom); aseptically collected FCS (Hyclone Laboratories); thrombin and recombinant urokinase-type Plg activator (uPA; R&D Systems); human plasma Plg-depleted FG, Glu-Plg, tissue Plg activator (tPA), and CRP (EMD Millipore); Cascade Blue trisodium salt and 2′,7′-Bis-(2-Carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF free acid; Invitrogen); Coumarin 314 (Sigma-Aldrich); plasma antithrombin III and Argatroban monohydrate (Sigma-Aldrich); Batroxobin (Pentapharm); recombinant Plg fragments KR1-KR3 (#ab73645; AbCam) and KR3-KR5 (#P00747; Gln274∼Cys560; USCN Life Science Inc.); Factor II/Thrombin, Factor III, Factor VII, Factor XIV, αVβ3, and PAI-1 (R&D Systems); and recombinant human Annexin II (AbD Serotec). Recombinant human and mouse PTX3, and the PTX3 C-terminal (C domain) and N-terminal (N domain) fragments were purified under endotoxin-free conditions by immunoaffinity, ion-exchange chromatography, and gel filtration, respectively, from the supernatants of stably transfected CHO cells as previously described (Bottazzi et al., 1997; Deban et al., 2010). Purified OmpA from Klebsiella pneumonia was prepared as previously described (Jeannin et al., 2005). Rabbit polyclonal antibody to PTX3 and rat monoclonal antibodies anti-PTX3 MNB4 (IgG2a) and MNB5 (IgG1) were obtained as previously described (Camozzi et al., 2006). The rat isotype control was from AbCam (IgG1; #RTK2071). All proteins were at least 95% pure as assessed by SDS-PAGE followed by silver staining.

Animals

PTX3-deficient mice were generated as previously described (Garlanda et al., 2002) and were used on a 129/SvPas or C57BL/6J inbred (backcrossed for 11 generations) genetic background. Wild-type mice were obtained from Charles River or were cohoused littermates (e.g., 2 out of 15 experiments for skin wound healing). Il1r1^−/−, P-selectin^−/−, Tlr3^−/−, and Ticam1^−/− (all on a C57BL/6J genetic background) mice were from The Jackson Laboratory. Irf3^−/− and Myd88^−/− mice were provided by K.A. Fitzgerald (University of Massachusetts Medical School, Worcester, MA) and S. Akira (Osaka University, Osaka, Japan), respectively. Ptx3/P-selectin-double-deficient mice were generated by crossing single-deficient mice. Mice were housed in the specific pathogen–free animal facility of the Humanitas Clinical and Research Center in individually ventilated cages and were used between 8 and 12 wk of age. Procedures involving animals handling and care were conformed to protocols approved by the Humanitas Clinical and Research Center (Rozzano, Milan, Italy) in compliance with national (4D.L. N.116, G.U., suppl. 40, 18–2-1992; D.L. N. 26, G.U. 4–3-2014) and international law and policies (EEC Council Directive 2010/63/EU, OJ L 276/33, 22–09-2010; National Institutes of Health Guide for the Care and Use of Laboratory Animals, US National Research Council, 2011). The study was approved by the Italian Ministry of Health (approval n. 71/2012-B, issued on the 09/03/2012). All efforts were made to minimize the number of animals used and their suffering.

Animal models

RNA and protein

Real-time PCR experiments were performed using the SybrGreen PCR Master Mix (Applied Biosystems) in a 7900HT fast real-time PCRsystem (Applied Biosystems) in accordance to the manufacturer’s instructions (Applied Biosystems) and the following primers: Ptx3, 5′-ATTGCTGTTTCACAACCTGC-3′, 5′-CGCTGTGCTGGAGGAACT-3′; or 5′-ACGAAATAGACAATGGACTTC-3′, 5′-AGTCGCATGGCGTGGG-3′ (for sorted cells); TF, 5′-CCGACGAGATTGTGAAGGATGT-3′, 5′-AGAGGCTCCCCAGCAGAAC-3′. Data were normalized by β-actin or GAPDH expression, analyzed with the Δ²CT method, and expressed as fold change in respect to unwounded skin or absolute values for isolated cells.

Whole wounds were collected at 1, 2, and 7 d after wounding and homogenized in 50 mM Tris-HCl, pH 7.5, containing 2 mM EGTA, 1 mM PMSF, 100 KU Aprotinin, 1% Triton X-100 (all from Sigma-Aldrich), and complete protease inhibitor cocktail (Roche). Total proteins were measured by DC Protein Assay, according to manufacturer’s instructions (Bio-Rad Laboratories). Mouse PTX3 was measured by ELISA based on the monoclonal antibodies MnmE1 (IgG1) and MnmE2 (IgG1), which were obtained by immunization of Ptx3^−/− mice. D-dimer (AMS Biotechnology), PDGF (Bio-Medical Assay Co.), 5-HT (LDN Labor Diagnostika Nord), TGFβ, CD62P, CXCL1, CXCL2, and myeloperoxidase (MPO; R&D Systems) were measured by ELISA in accordance with the manufacturer’s instructions. Western blot analysis for Plg/plasmin and Factor II/Thrombin was performed after loading 50 or 20 µg of wound protein extracts on SDS-PAGE (10% acrylamide). The goat polyclonal anti-Plg Kringle 5 (0.5 µg/ml; R&D Systems) and anti-Factor II/thrombin (0.2 µg/ml; R&D Systems) were used. The monoclonal anti-vinculin (0.2 µg/ml; hVIN-1; Sigma-Aldrich) was used as loading control. Plg, plasmin, thrombin, or fibrin bands were quantified by Fiji-ImageJ (ImageJ; NIH) as a ratio of mean gray intensity values of each protein relative to vinculin bands, or as absolute values.

Microtiter binding assay

Binding assays were performed as previously described (Sahni and Francis, 2000; Jeannin et al., 2005). In brief, 96-well plates (Nunc) were coated overnight with 100 ng/50 µl of purified FG, Glu-Plg, Plg fragments, Factor II/Thrombin, Factor III, Factor VII, Factor XIV, αVβ3, Annexin II, PAI-1, or OmpA diluted in PBS²⁺ (containing 130 mg/l CaCl₂ and 100 mg/l MgCl₂; Lonza). For fibrin-coated wells, 1 U/ml of thrombin (Sigma-Aldrich) was added to 1 mg/ml FG diluted in PBS²⁺. The solution was mixed, rapidly poured in 96-well plates and, after 45 s, before polymerization, aspirated, leaving a thin film of fibrin coating on the surface. Nonspecific binding to the plates was blocked by incubation for 2 h at 25°C with 0.5% dry milk in PBS²⁺ before incubation (1 h at 25°C) with 100 µl of PTX3, biotin-conjugated C and N domain or biotin-conjugated PTX3 (b-PTX3; 3.5–222 nM) in PBS²⁺ (at indicated pH) containing 0.05% (vol/vol) Tween 20. After washing, PTX3, C and N domain, or bPTX3 were detected by incubation with 100 µl of purified polyclonal rabbit antibody (500 ng/ml) followed by horseradish peroxidase-linked secondary antibody (GE Healthcare) or with 100 µl of horseradish peroxidase-linked avidin (BioSpa). Absorbance values were measured at 450 nm after addition of tetramethylbenzidine substrate (TMB; Sigma-Aldrich). A similar protocol was used for CRP (EMD Millipore) and Plg (EMD Millipore) binding to fibrin. A rabbit polyclonal antibody anti-CRP (EMD Millipore) and a mouse monoclonal anti-Plg (IgG1; AbCam) were used, respectively. In different series of binding experiments, Plg and/or PTX3, CRP, and BSA (Sigma-Aldrich) were, respectively co-incubated with PTX3 or Plg.

Before binding experiments, CRP and the PTX3 C-terminal domain were cross-linked with BS3 (Bis[sulfosuccinimidyl]suberate; Thermo Fisher Scientific) according to the manufacturer’s instructions (Bottazzi et al., 1997).

PTX3 binding to conidia of Aspergillus fumigatus was performed as previously described (Garlanda et al., 2002) at different pHs.

Surface plasmon resonance

The ProteOn XPR36 Protein Interaction Array system (Bio-Rad Laboratories), based on surface plasmon resonance (SPR) technology, was used for binding studies. Plg was immobilized onto one cell surface of a ProteOn GLC sensor chip (Bio-Rad Laboratories), using amine-coupling chemistry. In brief, the surface was activated for 5 min with a mixture of EDC (0.2 M) and sulfo-NHS (0.05 M), and then Plg was diluted to 30 µg/ml in acetate buffer pH 5.0 and flowed for 5 min at a rate of 30 µl/min. The remaining activated groups were blocked with a 5-min injection of 1 M ethanolamine. The immobilization level was ∼6,500–7,000 resonance units (RU; 1 RU = 1 pg protein/mm²). Full-length PTX3, its C and N domain were injected over immobilized Plg at a rate of 30 µl/min for 10 min (association phase). To exclude mass transport artifacts, the experiment was also performed changing the flow rate to 100 µl/min. Dissociation phase was measured in the following 15 min. The analyte running buffer was PBS²⁺ at pH 6.0 or pH 7.4, containing 150 mM NaCl and 0.005% (vol/vol) Tween 20. Alternatively, PTX3 and its domains were covalently immobilized as previously described (Deban et al., 2010). In brief, after the activation step, the different ligands were diluted to a final concentration of 30 µg/ml in acetate buffer (pH 2.5–3.0 for PTX3 and N domain PTX3; 4.0 for C domain PTX3). The amounts of PTX3 and C and N domain of PTX3 covalently immobilized onto the surface were ∼3,000, 1,000, and 2,000 RU, respectively. After 90 degree rotation of the fluidic system, Plg was diluted in running buffer (PBS²⁺ pH 6.0) and injected over immobilized ligands at a rate of 30 µl/min, for 5 min (association phase), whereas dissociation was followed for 10 min. The sensorgrams (time course of the SPR signal in RU) were normalized to a baseline value of 0. The resulting sensorgrams were fitted using the simplest 1:1 interaction model (Langmuir model; ProteOn analysis software) to obtain the corresponding association and dissociation rate constants (K_on and K_off). To determine the affinity of a single molecule of Plg to a single binding site of PTX3, the experiment was also performed flowing Plg (monomeric) onto immobilized PTX3 (octameric) and the N domain (tetramer; Inforzato et al., 2010). Binding constants were calculated considering PTX3 as an octamer, the N domain as a tetramer, and the C domain as a monomer (Inforzato et al., 2010).

Collagen

The collagen content of damaged liver and lung was measured with a conventional OHpro method (Tager et al., 2004). In brief, organs were homogenized in 1 ml of PBS²⁺, and a 0.5 ml aliquot was hydrolyzed in 6 N HCl at 110°C for 12 h. 25-µl aliquots were added to 0.5 ml of 1.4% chloramine T (Sigma-Aldrich), 10% n-propanol, and 0.5 M sodium acetate. After 20 min of incubation at room temperature, 0.5 ml of Erlich’s solution (1 M p-dimethylaminobenzaldehyde [Sigma-Aldrich] in 70% n-propanol, 20% perchloric acid) was added and a 15-min incubation at 65°C was performed. Absorbance was measured at 550 nm and the amount of OH-pro was determined against a standard curve. Sirius red staining is described in the Histology section.

In vitro fibrinolysis models

For cell-free studies, as previously described (Hiraoka et al., 1998), fibrin gels (50 µg each) previously prepared by mixing FG (1 mg/ml) and thrombin (1 U/ml), were incubated in 50 mM Tris, 0.01% (vol/vol) Tween 20, pH 6.0, with the following purified proteins, and combinations thereof: Plg (1 µg/100 µl), uPA (1 ng/100 µl), tPA (1 ng/100 µl), and PTX3 (100 ng/100 µl). After a 16-h incubation at 37°C, fibrin gels were suspended in reducing buffer and proteolysis was assessed in 2 µl of sample by Western blot (see below).

For pericellular fibrinolysis studies, WT and Ptx3^−/− peritoneal macrophages and mouse embryonic fibroblasts (MEF) were embedded in fibrin gels (3 × 10³/ml) as previously described (Carmeliet et al., 1994; Hiraoka et al., 1998). In brief, cells were dissociated with trypsin and centrifuged at 1,000 rpm for 5 min and cell pellet resuspended in a mixture of 2.5 mg/ml of Plg-free FG. Thrombin (1 U/ml) was then added and the suspension was immediately poured into a 24-multiwell plate (300 µl/well) on rounded glasses. The plate was incubated for 30 min at 37°C to allow fibrin polymerization. After washing with PBS²⁺, fibrin cultures were incubated with 1 ml/well of complete RPMI 1640 in triplicate at 37°C. Cell fibrinolytic activity was evaluated by quantifying fibrinolysis-derived products released in the medium by ELISA, Western blot and by microscopy (see Confocal and two-photon microscopy section). In brief, 96-well plates were coated overnight with the supernatants diluted 1:10⁴ in PBS²⁺. Nonspecific binding was blocked by incubation for 2 h at 25°C with 0.5% dry milk in PBS²⁺ before incubation (1 h at 25°C) with 100 µl of a purified rabbit polyclonal antibody anti-fibrin(ogen) (500 ng/ml; Dako or EMD Millipore) followed by horseradish peroxidase-linked secondary antibody (GE Healthcare) and TMB substrate (Sigma-Aldrich). Values are expressed as mean O.D. ± SD of triplicates. Soluble fibrinolysis products (3 µl of each sample) were also resolved under reducing conditions on 4–12% SDS-PAGE Criterion XT Precast gels (Bio-Rad Laboratories) and detected with polyclonal antibodies anti-fibrin(ogen).

Histology and assessment of tissue sections

For histological analysis, 5-µm-thick paraffin-embedded skin, liver and carotid artery sections were stained with hematoxylin and eosin (H&E) and with Sirius red. Specimens were imaged with an Olympus BX51 microscope equipped with a 2× (N. A. 0.06), 4× (N. A. 0.10), 20× (N. A. 0.4) objectives and with an Olympus XC50 camera. The thickness of wounded skin clot and scab, epithelial gap segment and necroinflammatory area of injured livers were measured in different regions of interest of each single image as segments or free-hand area drawing and quantified using ImageJ (NIH). Count of recruited cells was performed in 3–5 fields (0.05 mm²) per section. An Olympus dotSlide BX virtual slide scanner equipped with a 10× N.A. 0.25 objective and with a Peltier cooled camera (1376 × 1032 pixels) was used to acquire the entire injured liver specimens. The OlyVIA (version 3.5; Olympus) image software was used to count occluded central veins in the entire section. For quantitative analysis of Sirius red-stained collagen and α-smooth muscle actin; (α-SMA) in skin and liver, 2–10 serial images from 3–10 mice per group were digitally captured at 100× magnification. Collagen staining, α-SMA⁺ and fibrin in lung and liver immunoreactive areas were measured using the computer-assisted digital image processing software Image-Pro Plus (version 7.0; MediaCybernetics). The total stained area was automatically selected on the basis of RGB color segmentation and results are expressed as mean percentage of the immunoreactive area ± SD.

Flow cytometry

A single-cell suspension of wounded skin was generated by cutting samples into small pieces and digestion in PBS (pH 7.4) supplemented with collagenase type I (1 mg/ml; Sigma-Aldrich) and hyaluronidase (125 U/ml; Sigma-Aldrich) at 37°C for 1 h twice. Finally, a single-cell suspension was obtained by mechanical separation followed by 100 µm filtration and erythrocytes lysis (0.15 M NH₄Cl, 10 nM KHCO₃, 1 nM Na₄EDTA, pH 7.2). The following fluorophore-conjugated antibodies were used for immunophenotyping and intracellular staining: anti–mouse CD45 BV605^TM (#30-F11; BD) or anti–mouse CD45 BV650 (#A20; BioLegend), anti–mouse CD11b BV421 (#M1/70; BioLegend), anti–mouse Ly6G PE-CF594^TM (#1A8; BD), anti–mouse Ly6C FITC^TM (#AL-21; BD), anti–mouse F4/80 PE/Cy7^TM (#BM8; BioLegend), anti–mouse CD3 PerCP-Cy5.5^TM (#BM8; BD), anti–mouse CD140A (PDGFRα) BV421 (#APA5; BD), anti–mouse CD31 PerCP-Cy^TM5.5 (#MEC 13.3; BD), rabbit anti-FAP Pacific Orange (AbCam), anti–mouse α-SMA FITC (#1A4; Sigma-Aldrich), goat anti–mouse PTX3 Alexa Fluor 647 (R&D Systems), and isotype-matched control antibodies. Cells were fixed with 2% PFA in PBS for 10 min, and when indicated permeabilized with BD Cytofix/Cytoperm kit (BD) according with manufacturer’s instructions. Cells viability before the fixation and permeabilization was determined with LIVE/DEAD Fixable Aqua Dead Cell Stain kit (Molecular Probes). Flow cytometry was performed using BD LSRFortessa instrument and BD FACSDiva Version 8.0.1 software (BD). Cell sorting was conducted using BD FACSAria III system. The purity of isolated cells was ≥ 98%.

Confocal and two-photon microscopy

5-µm cryostat sections were incubated in 5% of normal goat (Dako) or donkey (Sigma-Aldrich) serum, 2% BSA, 0.1% Triton X-100 (Sigma-Aldrich) in PBS²⁺, pH 7.4, (blocking buffer) for 1 h at room temperature. Sections were incubated with the following primary antibodies for 2 h at room temperature: affinity-purified rabbit polyclonal IgG anti-PTX3 (1 µg/ml) or affinity purified goat IgG anti-PTX3/TSG-14 (2 µg/ml; R&D Systems); rat monoclonal (IgG2, 2 µg/ml; Cell Sciences) or rabbit polyclonal (prediluted; AbCam) anti-PDGFRα; rat monoclonal anti–mouse CD68 (IgG2a, 1 µg/ml; Hycult Biotech); goat polyclonal anti-uPAR (5 µg/ml; AbCam); goat polyclonal anti-uPA (2 µg/ml; R&D Systems); rabbit polyclonal anti-fibrin(ogen) (5 µg/ml; Dako); rat monoclonal anti-Plg (IgG1, 5 µg/ml; Cell Sciences); rabbit polyclonal anti-collagen I (4 µg/ml; AbCam); mouse monoclonal anti–α-SMA (2 µg/ml; Sigma-Aldrich); goat anti-proteinase activated receptor-1 (PAR1; 1 µg/ml; R&D Systems); rabbit anti-fibroblast activation protein (FAP; 5 µg/ml; AbCam); rat anti-CD45 allophycocyanin (APC; BD); and rat anti-Ly6C FITC (BD). Sections were then incubated for 1 h with Alexa Fluor (488, 594, 647)–conjugated species-specific cross-adsorbed detection antibodies (Molecular Probes). For DNA detection, DAPI (300 nM; Invitrogen) was used. After each step, sections were washed with PBS²⁺, pH 7.4, containing 0.01% (vol/vol) Tween 20 (washing buffer). Tissues from Ptx3^−/− mice were used as negative control for PTX3 immunostaining. Sections were mounted with the antifade medium FluorPreserve Reagent (EMD Millipore) and analyzed with an Olympus Fluoview FV1000 laser scanning confocal microscope with 10× (N.A. 0.4) or oil immersion lens 40× (N.A. 1.3). Immunoreactivity of α-SMA, FAP, PAR1, CD45, Ly6C, collagen I, and fibrin in the wounded skin was evaluated as mean ± SD of the intensity grayscale levels of confocal images (1024 × 1024) obtained with 10×, 20×, or 40× lens, analyzing the whole field of view by FV10-ASW software (version 4.1a; Olympus). For fibrin, 2–5 10× and for Plg, 2–5 40× images/mouse were acquired. Colocalization analysis was performed on confocal images acquired with oil immersion PLAPO 60× lens (N.A. 1.35) at resolution of 1 AU. Overlay RGB images (1024 × 1024; 4–6 images from 5 wounded skins) were imported to Imaris software assembled and colocalization was evaluated by Coloc module by Imaris (Bitplane AG). In all images, the same threshold fluorescence intensity for each channel was fixed, a binary mask of mask colocalization was created and the percentage of colocalization material and the Pearson’s index of linear correlation were analyzed. MEF fibrin cultures were fixed in 4% paraformaldehyde, and incubated with polyclonal antibodies anti-fibrin(ogen) and PTX3 and fluorophore-linked secondary antibodies (Invitrogen). Specimens were analyzed with an Olympus Fluoview FV1000 laser scanning confocal microscope in a channel mode with 405, 488, 559, and 635 nm excitations. The resulting fluorescence emissions were collected using a 425–475 nm (for DAPI), 500–545 nm (for Alexa Fluor 488), 595–620nm (for Alexa Fluor 594), and 655–755 nm (for Alexa Fluor 647) band-pass filters. Samples were imaged with oil immersion lens 60× (N. A. 1.35) in xy optical sections (800 × 800 pixels) of 0.75 µm. For fibrin cultures, the z-stacks were acquired at a resolution of 1 AU to allow three-dimensional reconstruction and presented as three-dimensional projections by Imaris software (X64, version 7.2.1; Bitplane AG).

For ex vivo imaging experiments of pH quantification in the skin, 2-Photon microscopy and the pH-sensitive fluorescent probe 2’, 7’-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein (BCECF; Molecular Probes) were used as described (Hanson et al., 2002). BCECF reports the local pH through a shift in the excitation spectrum and change in the intensity/spectrum of emission as the probe passes from an acid (protonated) to a basic (unprotonated) form. These spectral changes are accompanied by a change in the fluorescence lifetime. 8-wk-old 129/Sv female mice were shaved on the back using a mechanical trimmer and depilatory cream. Two days after wounding with a 6-mm-diam biopsy punch, animals were anesthetized with i.p. injection of 2.5% Avertin and BCECF (acid; 80 µM in ethanol) was applied topically, as previously described (Hanson et al., 2002). After 3 applications within 1 h, whole back skin was removed and stretched on a Petri dish. Images were acquired from the epidermis to the dermis with a LaVision TriM Scope II (LaVision BioTec) two-photon laser-scanning microscope equipped with a 20× (XLUMPFL N.A. 1.0; Olympus) water immersion lens and FLIM X16 TCSPC detector (LaVision BioTec). Excitation was performed using a Chameleon Ultra II (Coherent) laser at 820 nm output. Acquisitions were 400 × 400 µm, 0.78 µm resolution x-y and 5 µm voxel depth, 10 slices. For each mouse, 2–5 acquisitions of both unwounded and wounded areas were performed. Values of pH in dermis were obtained in the depth range of 10–20 µm (Hanson et al., 2002). To exclude possible artifacts due to autofluorescence from hemoglobin present in the clot, acquisitions were performed as indicated in between perilesion area and the ulcer. FLIM z-stacks were imported into Imspector Pro (LaVision BioTec) and fitted to lifetime images. An off-peak time gate was set, to reduce the contribution from SHG. Lifetimes were corrected for instrumentation effects and skin refractive index and calculated as previously described (Hanson et al., 2002). Mean refractive index at all acquisition depths was calculated by measuring the mean lifetime of Coumarin 314 (Sigma-Aldrich). This gave a mean 1.1-fold correction for lifetimes measured in skin, in accordance with the literature (Hanson et al., 2002). The pH at each depth was calculated as a fraction of contributions from lifetime of HBCECF and BCECF.

Specific antibodies and correspondent irrelevant IgGs, were conjugated with Cascade Blue dye and Alexa Fluor fluorophores with Alexa Fluor antibody labeling kits (Invitrogen-Molecular Probes). 8-wk-old 129/Sv female mice were s.c. injected 4 h before acquisition with 200 µl/mouse of Cascade Blue-conjugated rabbit ant-fibrin(ogen; 50 µg/ml), Alexa Fluor 568-conjugated rat anti-Plg (75 µg/ml) and Alexa Fluor 647-conjugated goat anti-Ptx3 (75 µg/ml), or with irrelevant controls. Ptx3^−/− mice were further used a negative control. BCECF (80 µM in ethanol) was applied topically and skin collected as described before. To avoid the interference of endogenous autofluorescence of the epidermis, the skin sample was then reversed, stretched on a Petri dish and fat and blood were removed with synthetic absorption triangles (Fine Science Tools). Multifluorescent images were acquired on reversed skin samples from the panniculus carnosus to the dermis, excluding the epidermis. Images were acquired using LaVision TriM Scope II microscope with the excitation laser at wavelength 800 nm. Fluorescence emission was collected with the photomultipliers H6780-20, H6780-01, and H7422 (Hamamatsu). Filter configuration was as follows: 420/50, LP493, 525/50, LP593, 609/54, LP625, and 655/40 (Chroma Technologies). Image acquisition, fitting lifetimes, and conversion to pH images were performed using Imspector Pro (LaVision BioTec) and a python algorithm derived from Hanson et al. (2002). Raw image stacks (200 × 200 µm, 0.48 µm pixel resolution in x–y plane and 10 µm voxel depth, 20–50 slices) were imported into Fiji-ImageJ (ImageJ) and 3D mosaics were assembled with Image Stitching plugin. Pearson’s r coefficients in the different regions of the reconstructed images were then calculated with Coloc module by Imaris (Bitplane). Statistical analysis was performed with Prism 5 (GraphPad).

Statistical analysis

Results are expressed as mean ± SD or SEM, as specified. Statistical significance between groups was assessed by unpaired two-tailed Mann-Whitney test or Student’s t test with Prism software (version 4.00 for Windows; GraphPad).

Online supplemental material

Fig. S1 showing the gating strategy used in flow cytometry analysis for immunophenotyping and sorting leukocytes and stromal cells in the course of skin repair, and Fig. S2 showing the method used to measure the pH in wounded skin by two-photon microscopy. Online supplemental material is available at http://www.jem.org/cgi/content/full/jem.20141268/DC1.