Dysregulated secretion and extracellular activation of TGF-β1 stimulates myofibroblasts to accumulate disordered and stiff extracellular matrix (ECM) leading to fibrosis. Fibronectin immobilizes latent TGF-β-binding protein-1 (LTBP-1) and thus stores TGF-β1 in the ECM. Because the ED-A fibronectin splice variant is prominently expressed during fibrosis and supports myofibroblast activation, we investigated whether ED-A promotes LTBP-1–fibronectin interactions. Using stiffness-tuneable substrates for human dermal fibroblast cultures, we showed that high ECM stiffness promotes expression and colocalization of LTBP-1 and ED-A-containing fibronectin. When rescuing fibronectin-depleted fibroblasts with specific fibronectin splice variants, LTBP-1 bound more efficiently to ED-A-containing fibronectin than to ED-B-containing fibronectin and fibronectin lacking splice domains. Function blocking of the ED-A domain using antibodies and competitive peptides resulted in reduced LTBP-1 binding to ED-A-containing fibronectin, reduced LTBP-1 incorporation into the fibroblast ECM and reduced TGF-β1 activation. Similar results were obtained by blocking the heparin-binding stretch FNIII12-13-14 (HepII), adjacent to the ED-A domain in fibronectin. Collectively, our results suggest that the ED-A domain enhances association of the latent TGF-β1 by promoting weak direct binding to LTBP-1 and by enhancing heparin-mediated protein interactions through HepII in fibronectin.
Tissue fibrosis manifests as severe deformities in the skin and leads to reduced function and/or failure of vital organs like lung, heart, liver and kidney (Wynn and Ramalingam, 2012). Myofibroblasts are responsible for the irreversible accumulation and excessive remodeling of collagenous extracellular matrix (ECM) that characterizes fibrosis (Hinz, 2016; Klingberg et al., 2013; Tomasek et al., 2002). Three key conditions coordinate myofibroblast activation from a variety of different precursor cells: (1) the presence of a mechanically resistant ECM (Arora et al., 1999; Li et al., 2017), (2) biologically active TGF-β1 (Desmoulière et al., 1993) and (3) extradomain-A (ED-A)-containing fibronectin (FN; also known as FN1) (hereafter denoted ED-A FN) (Serini et al., 1998). However, it is still unclear how TGF-β1 and mechanical stress collaborate with ED-A FN to promote myofibroblast activation.
Fibroblasts cultured from different organs and species have been shown to activate latent TGF-β1 from stores in the ECM by a process that requires cell contraction and a sufficiently stressed ECM (Hinz, 2015; Sarrazy et al., 2014; Wipff et al., 2007). In the soluble latent form, TGF-β1 is non-covalently bound to its pro-peptide form, known as latency-associated peptide (LAP) (Robertson et al., 2015). Covalent binding of the LAP portion of latent TGF-β1 to the latent TGF-β binding protein-1 (LTBP-1) intracellularly forms a large latent complex that is incorporated into the ECM upon secretion. LTBP-1 has been shown to mainly interact with two ECM proteins, FN and fibrillin-1 (Hynes, 2009; Ramirez and Rifkin, 2009; Zilberberg et al., 2012) with FN acting as master template for the initial LTBP-1 incorporation into the maturating ECM (Dallas et al., 2005; Klingberg et al., 2014; Koli et al., 2005). FN exists in two principal forms that are generated by alternative splicing from one single gene: (1) plasma FN that is secreted by hepatocytes into the circulation, and (2) insoluble cellular FN that is secreted by a variety of different cells, including fibroblasts (Klingberg et al., 2013; Pankov and Yamada, 2002; Singh et al., 2010; White et al., 2008; Zollinger and Smith, 2017). Cellular, but not plasma FN, contains the alternatively spliced FN type III (FNIII) extradomains ED-A (also known as EIIIA or FNIII EDA) and/or ED-B, which are transiently expressed during embryogenesis (Astrof and Hynes, 2009; Peters and Hynes, 1996). Under normal conditions, ED-A FN, ED-B FN and ED-A/B FN (i.e. containing both ED-A and ED-B) are typically not expressed in adult connective tissue, but become re-expressed as ‘oncofetal FNs’ during the ECM remodeling associated with wound repair, fibrosis and tumor development. The presence of ED-A FN is characteristic for tissue repair/healing and fibrosis, whereas ED-B FN is most frequently associated with tumor development and angiogenesis (Astrof et al., 2004; Bhattacharyya et al., 2014; Jarnagin et al., 1994; Kelsh et al., 2015; Kumra and Reinhardt, 2015; Sackey-Aboagye et al., 2016; Serini and Gabbiani, 1999; White et al., 2008).
Expression of ED-A FN precedes and is necessary for myofibroblast activation from various different precursor cells and in different fibrotic conditions (Arslan et al., 2011; Booth et al., 2012; Hirshoren et al., 2013; Kohan et al., 2011, 2010). The myofibroblast-permissive action of ED-A FN is inhibited by the ED-A FN function-blocking antibody IST-9 and recombinant ED-A peptides (Hinz et al., 2001; Serini et al., 1998). Consistent with this, ED-A FN-null (ED-A FN−/−) mice display abnormal healing of skin wounds (Muro et al., 2003) and are protected against bleomycin-induced lung fibrosis (Muro et al., 2008). ED-A FN−/− fibroblasts exhibit reduced responsiveness to active TGF-β1, and fail to activate latent TGF-β1 for yet unknown reasons (Muro et al., 2008). Collectively, these findings led us to test the hypothesis that ED-A FN is controlling the storage and/or activation of latent TGF-β1 in the ECM. Our data show that: (1) ED-A FN and LTBP-1 expression are co-upregulated during myofibroblast activation by ECM stiffness and colocalize in the myofibroblast ECM; (2) LTBP-1 incorporation into fibroblast ECM is enhanced by ED-A FN compared to what is seen with ED-B or plasma FN; (3) binding of purified LTBP-1 to recombinant ED-A domain peptides and full-length FNs without the ED-A domain is lower than with full-length ED-A FN; (4) LTBP-1 association with ED-A FN is inhibited by antibodies directed against either ED-A or the adjacent heparin-binding domains FNIII12-13-14 (denoted HepII). Incorporation of LTBP-1 into ED-A FN-containing ECM is blocked by heparan sulfate (HS). These findings suggest that ED-A plays a dual role in guiding LTBP-1 to FN by promoting specific, but low-affinity, interactions and enhancing the availability of the HepII heparin-binding sites in FN. Because reducing incorporation of LTBP-1 into the ECM by means of ED-A blocking antibodies also results in reduced TGF-β1 activation by fibroblast cultures, we propose that ED-A FN presents a potential target for anti-fibrosis strategies.
ECM stiffness regulates expression of ED-A FN and LTBP-1
Mechanical stress arising from ECM stiffening during remodeling is pivotal for myofibroblast activation (Hinz, 2010, 2015). To test whether expression of ED-A FN and LTBP-1 and secretion into the myofibroblast ECM are controlled by mechanical factors, we cultured primary human dermal fibroblasts (hDfs) on differently compliant silicone substrates for 7 days. A Young's modulus of 3 kPa was chosen to simulate the mechanical conditions of normal skin (Achterberg et al., 2014), 100 kPa substrates were used to simulate fibrotic tissue stiffness (Li et al., 2017) and intermediate stiffnesses (10 kPa and 25 kPa) for ECM in tissue under remodeling. Substrates of 3000 kPa provided mechanical growth conditions comparable to those on non-physiologically stiff conventional tissue plastic culture. Immunofluorescence staining demonstrated increasing expression of both ED-A FN and LTBP-1 with increasing substrate stiffness. Both proteins colocalized in the ECM, which was most pronounced in hDf cultures on 100 kPa and 3000 kPa stiff substrates (Fig. 1A). The pattern was similar when co-staining for LTBP-1 and total FN. Because corresponding expression levels of ED-B FN were overall low (Fig. 1A, insets), ED-A FN appears to be the predominant FN splice variant in hDf cultures. Western blotting of lysates of cells plus ECM confirmed increasing ED-A FN and LTBP-1 expression with increasing ECM stiffness, as was the case for expression of the myofibroblast marker α-smooth muscle actin (α-SMA) (Fig. 1B). To support the finding that LTBP-1 and ED-A FN colocalize in the ECM, we performed western blotting in non-reducing conditions (Fig. 1C). hDf cultures were differentially fractionated into conditioned medium supernatants (SN, 10× concentrated) and NH4OH-extraced ECM (ECM) after 7 days growth on plastic. Cell lysates were used as controls; these were obtained from a dish of cells grown in parallel by performing gentle trypsinization, followed by resuspension in lysis buffer (Fig. 1C, ‘cells’, controlled by GAPDH and α-tubulin). ED-A FN and total FN were predominantly present in the conditioned supernatants and ECM fractions and migrated at an apparent molecular mass of ∼220 kDa whereas LTBP-1 (∼180 kDa) mainly associated with the ECM and cell fractions (Fig. 1C).
ED-A FN is more potent in guiding LTBP-1 to fibroblast ECM than other FN splice variants
FN is a master regulator of the fibroblast ECM, and loss of FN expression has been previously shown to abolish LTBP-1 incorporation into the ECM of fibroblasts and osteoblasts (Dallas et al., 2005). Similarly, knockdown of FN expression in the human fibroblast cell line MRC-5 using human-specific small interfering (si)RNA directed against total FN almost completely abolished formation of FN fibrils in the ECM (Fig. 2A). FN knockdown efficiency was ∼90% and resulted in co-downregulation of the expression of LTBP-1 and fibrillin-1, another ECM protein dependent on FN (Sabatier et al., 2009) in lysates of cells plus ECM (Fig. 2B). The remaining low levels of LTBP-1 mainly accumulated in concentrated culture supernatants (SN) whereas LTBP-1 was virtually absent from the ECM fraction after FN knockdown (Fig. 2C). As a result, the ratio between the LTBP-1 level after FN knockdown and the LTBP-1 level in control conditions was higher in supernatant than in ECM (Fig. 2C).
To investigate whether different splice variants of FN differentially affect expression and ECM organization of LTBP-1, we used 6xHis-tagged full-length rat FN constructs, containing: (1) only the ED-A extradomain (ED-A FN), (2) only the ED-B extradomain (ED-B FN), (3) both, ED-A and ED-B domains (ED-A/B FN), and (4) no extradomains (FN0) (Fig. 3A,B). Expression and purification from human embryonic kidney-293 cells (HEK293) confirmed correct molecular masses and the presence of ED-A in the ED-A FN splice variant (Fig. 3B). When stably overexpressed in wild-type MRC-5 cells, all full-length FN constructs incorporated into the ECM, colocalizing with endogenous FN and LTBP-1 after 7 days culture (Fig. 3C,D).
MRC-5 cell lines, selected to express equal levels of His-tagged rat FN versions (Fig. 3E), were then knocked down for human FN (Fig. 3F–I). The ECM organization of His-tagged rat FN in human FN-depleted MRC-5 (Fig. 3E) was similar to that of FN in wild-type MRC-5 cells (Fig. 3C,D). Rescue of knocked down endogenous human FN with rat ED-A and ED-A/B FN restored the LTBP-1 to ∼40% of that measured in control (not siRNA treated) MRC-5 cultures as assed by quantifying the LTBP-1 signal intensity from immunofluorescence images (Fig. 3G,H) and western blots of cell plus ECM lysates (Fig. 3I). Notably, rescuing human FN-deficient MRC-5 cells with rat ED-A and ED-A/B FN resulted in ∼2-fold higher LTBP-1 incorporation into the ECM (Fig. 3G,H) and expression of LTBP-1 (Fig. 3I) compared with what was seen with ED-B FN and FN0 at similar expression levels (Fig. 3F–I). LTBP-1 levels were ∼4-fold higher than in non-rescued FN-knockdown cells (Fig. 3I). Hence, although all FN variants were able to recruit LTBP-1, ED-A FN was 2-fold more efficient at doing so.
Function blocking of the FN ED-A domain inhibits incorporation of LTBP-1 into fibroblast ECM
We next investigated whether function blocking of the ED-A domain in FN with competitive peptides affected the capability of fibroblasts to incorporate LTBP-1 into the ECM in a potential therapeutic setting. We recombinantly produced His-tagged short peptide fragments of rat FN domains in E. coli: (1) ED-A and ED-A with flanking domains (11-ED-A-12) as ‘active’ peptides, and (2) the flanking FNIII domains of ED-A alone (11-12) and only the FNIII11 domain (11) as specificity controls (Fig. 4A). Purity and the correct size of the peptides were confirmed on Coomassie-stained SDS gels (Fig. 4B) and by anti-6xHis western blotting (Fig. 4C). hDfs were then cultured for 7 days in the presence of FN domain peptides to compete with the different endogenous FN domains (Fig. 4D). LTBP-1 incorporation into the fibroblast ECM was reduced in the presence of domain peptides ED-A and 11-ED-A-12 but not by domain peptides FNIII11 or FNIII11-12, shown by performing immunofluorescence staining (Fig. 4E). FN domain peptides did not interfere with the organization and incorporation of endogenous ED-A FN (Fig. 4D). These results were confirmed by western blotting analysis showing an ∼5-fold reduced LTBP-1 level in ECM fractions after treatment with 11-ED-A-12 and ∼3-fold reduced LTBP-1 levels after treatment with ED-A domain peptides compared to the values in controls (Fig. 4E). Addition of FN domain peptides did not affect overall LTBP-1 production but resulted in the accumulation of LTBP-1 in hDf culture supernatants (Fig. 4E). Soluble LTBP-1 was increased by 6-fold after treating hDfs with 11-ED-A-12 and by 3.5-fold after ED-A domain peptide treatment compared to the values in controls (Fig. 4E). These data show that LTBP-1 incorporation into the fibroblast ECM can be competitively inhibited by soluble ED-A-containing peptides.
Next, we assessed whether the ED-A domain-specific blocking antibody IST-9 (Carnemolla et al., 1987; Serini et al., 1998) had similar effects in inhibiting LTBP-1 anchoring with the ECM and can thus be used to prevent TGF-β activation from the large latent complex. Immunofluorescence analysis showed that treating hDfs for 7 days with IST-9 reduced integration of LTBP-1 into the ECM of hDfs and colocalization with ED-A FN compared to what was seen upon treatment with control antibody BC-1 (binds to FN when ED-B is present) and IgG (Fig. 5A). The amount and organization of ED-A FN, and lower expression of ED-B FN remained unaffected by IST-9 or control antibodies (Fig. 5A). Treatment of hDf cultures with IST-9 but not with control antibodies increased levels of total TGF-β but decreased active TGF-β in the supernatants, as measured by using TGF-β-reporter cells (Fig. 5B). Collectively, these data show that blocking the ED-A domain in FN with specific peptides and antibodies substantially decreases the ability of fibroblasts to immobilize LTBP-1 in the ECM. Loss of LTBP-1 from the ED-A FN ECM reduces the ability of fibroblasts to store, and thus activate, latent TGF-β1.
The ED-A domain plays a potential dual role in mediating the association of LTBP-1 with FN
To test whether the ED-A domain directly promotes LTBP-1 binding, we immobilized purified 6xHis-tagged FN domain peptides and full-length FN splice variants on enzyme-linked immunosorbent assay (ELISA) plates. The coated plates were subsequently incubated with purified LTBP-1 (Buscemi et al., 2011; Klingberg et al., 2014), which was then immunolocalized and quantified (Fig. 6). Plates coated with LTBP-1 at the same concentration served as the standard for maximal signal. LTBP-1 did not bind to FNIII domain peptides 11 and 11-12; binding was significantly higher to ED-A (1.4-fold) and 11-ED-A-12 domain peptides (1.7-fold over controls) (Fig. 6A). LTBP-1 binding to ED-A and 11-ED-A-12 was completely abolished in the presence of the ED-A-blocking antibody IST-9 as an additional binding specificity control (Fig. 6B). Despite being specific, LTBP-1 binding to ED-A-containing peptides was low and reached only 14–17% of the maximal signal (Fig. 6A,B).
In contrast, binding of LTBP-1 to full-length ED-A FN and ED-A/B FN was as high as 50% of the maximal LTBP-1-binding signal (Fig. 6C). Binding of LTBP-1 was ∼2-fold higher when full-length FN contained the ED-A domain than for FN0 and ED-B FN, both lacking ED-A (Fig. 6C). Presence of ED-B did not further enhance binding of LTBP-1 to full-length ED-A FN (ED-A/B FN) and ED-B alone had no enhancing effect (ED-B FN) (Fig. 6C). Blocking ED-A with IST-9 reduced LTBP-1 binding to ED-A FN to the level of that seen with FN0 (Fig. 6D). Collectively, the solid-state ELISA binding studies showed that LTBP-1 binds to all full-length FN constructs, which was enhanced by 2-fold when the ED-A domain was present. Binding of LTBP-1 to ED-A domain peptides was specific but low, suggesting an additional mechanism modulates how ED-A enhances LTBP-1 binding in the context of the FN molecule.
Thus, we tested whether the ED-A domain can potentiate LTBP-1 binding by modulating the availability of other binding domains in full-length FN. LTBP-1 binding to FN has been shown to be mediated by heparin (Chen et al., 2007; Massam-Wu et al., 2010), and heparin binding occurs in a stretch directly adjacent to ED-A in the FNIII domains 12-13-14 (HepII) (Clark et al., 2003; Mostafavi-Pour et al., 2001). Consequently, preincubation of LTBP-1 with heparan sulfate (HS) but not control chondroitin sulfate (CS) reduced binding to ED-A FN (and FN0) in ELISA assays with purified FN (Fig. 7A). Adding HS to hDf during the 7 day culture period resulted in almost complete inhibition of LTBP-1 incorporation into the ECM as shown on western blots with ECM fractions (Fig. 7B) and by immunofluorescence studies of hDF cultures (Fig. 7C). In addition, adding the HepII-blocking antibody A32 (Underwood et al., 1992) to hDf cultures reduced the ECM contents of LTBP-1 compared to what was seen with control antibody treatment, with moderate effects on FN organization (Fig. 7D). Blocking HepII also reduced the binding of purified LTBP-1 to full-length FN0 and ED-A FN splice variants (Fig. 7E) but not to ED-A-containing domain peptides lacking the HepII domain stretch (Fig. 7F). Collectively, these data suggest that the ED-A domain may increase the accessibility of the adjacent heparin-binding stretch FNIII12-13-14 (HepII) for LTBP-1 binding, in addition to weakly, but specifically, interacting with LTBP-1 in a direct fashion.
The ED-A splice variant of FN is an important element controlling myofibroblast activation during wound healing and development of fibrosis (Klingberg et al., 2013; Shinde et al., 2015; White et al., 2008). Part of this action appears to be mediated by binding of ED-A FN-specific integrins. Integrins α9β1 and α4β1 recognize the EDGIHEL motif in ED-A FN (Liao et al., 2002; Shinde et al., 2008), and blocking antibodies against α4-integrins reduce the extent of bleomycin-induced lung fibrosis in mice (Gailit et al., 1993; Wang et al., 2000). This effect is likely caused by affecting inflammatory cells that most prominently express α4β1 integrin. The ED-A FN-binding integrin α4β7 has been directly implicated in myofibroblast differentiation of murine lung fibroblasts (Kohan et al., 2011, 2010). ED-A FN−/− mice are protected against bleomycin-induced skin and lung fibrosis (Muro et al., 2003, 2008) but not from experimentally induced liver fibrosis (Olsen et al., 2012). Notably, ED-A FN−/− liver hepatic stellate cells are able to become myofibroblasts in vivo and in vitro, suggesting that lacking ED-A FN can be compensated for by other myofibroblast-inducing factors. Myofibroblast activation further depends on the presence of active TGF-β1 and mechanical stress arising from the stiff ECM of the scar tissue. Although previous studies have shown that ED-A FN−/− mice and fibroblasts exhibit reduced levels of total and active TGF-β1 in conditions of lung fibrosis (Muro et al., 2008), the mechanistic link between ED-A FN and TGF-β1 activation or storage remained elusive. TGF-β1 activation by αv integrins depends on LTBP-1 binding to the ECM, which provides physical resistance against the cell pulling that is required to induce a conformational change in the latent complex (Annes et al., 2004; Buscemi et al., 2011; Shi et al., 2011; Wipff et al., 2007). We hypothesized that ED-A FN stores latent TGF-β1 in the myofibroblast ECM particularly efficiently. Our central finding is that the presence of the ED-A domain enhances the capacity of FN to recruit LTBP-1 to the ECM, which is required for the sequestration and subsequent activation of latent TGF-β1. Consistently, inhibition of the ED-A domain in our fibroblast cultures reduced levels of active TGF-β1 and increased the release of soluble LTBP-1 and latent TGF-β1 into culture medium.
Binding to the ECM is mediated through the N-terminus of LTBP-1 (Dallas et al., 2000; Nunes et al., 1997; Unsold et al., 2001), and the minimal ECM-binding sequence that allows integrin-mediated TGF-β1 activation comprises amino acids 402–449 in the N-terminal hinge domain in LTBP-1 (Annes et al., 2004; Fontana et al., 2005). LTBP-1 binds to different proteins of the ECM, including FNs, fibrillins and fibulins, in a carefully orchestrated sequence of changing binding partners (Todorovic and Rifkin, 2012). Initial ECM targeting of LTBP-1 is dependent on FN, but a transfer to fibrillin-1-containing microfibrils was shown to subsequently occur in cell culture models (Chaudhry et al., 2007; Dallas et al., 2000, 2005; Isogai et al., 2003; Ono et al., 2009; Sabatier et al., 2009, 2013). The partner domains in either fibrillin or FN that bind the hinge region of LTBP-1 have not yet been identified.
Our results show that presence of the ED-A domain in FN enhances LTBP-1 incorporation into the ECM of fibroblast cultures over FN lacking the ED-A domain as follows: (1) inhibition of ED-A using specific antibodies and competitive peptides prevents LTBP-1 targeting to the ECM and instead leads to its release into the cell culture supernatant; (2) expression of ED-A FN in FN-depleted fibroblasts rescue LTBP-1 targeting to the ECM more efficiently than ED-B FN or FN without extradomains; and (3) full-length purified FN binds purified LTBP-1 more efficiently when the ED-A domain is present. Although recombinant ED-A domain peptides display a low potential to directly bind purified LTBP-1 in vitro, they potently blocked targeting of LTBP-1 to the fibroblast ECM. We thus consider that ED-A creates a favorable FN conformation for LTBP-1 binding, in addition to being a direct binding partner.
Previous findings support the idea that the presence of ED-A primes the FN structure for LTBP-1 binding. LTBP-1 contains a sensitive proline-rich hinge region with a heparin-binding consensus sequence, and LTBP-1 binding to both fibrillin-1 and FN has been shown to be mediated by heparin (Chen et al., 2007; Massam-Wu et al., 2010). In addition to heparin-binding domains in proximity to the N-terminus of FN, a major heparin-binding domain – HepII – is located in the FNIII domain stretch 12-13-14, adjacent to ED-A (Clark et al., 2003; Mostafavi-Pour et al., 2001). HepII plays a crucial role in growth factor binding to FN (Mitsi et al., 2008; Wan et al., 2013). Moreover, ligation of αv integrins to the RGD binding site in the FNIII domain 10, and α5β1 integrin to the additional consensus sequence in FNIII 9 has been shown to enhance cell responses to various growth factors that promiscuously bind to FNIII domains 12-13-14 (Martino and Hubbell, 2010). Our own results demonstrate that LTBP-1 binding to ED-A FN is abolished by saturating heparin-binding sites with HS and upon treatment with blocking antibodies directed against HepII. It is thus conceivable that the presence of ED-A places the HepII heparin-binding site into a favorable position for FN interaction with LTBP-1. ED-A is a cryptic FN domain subject to regulatory cell processes (Julier et al., 2015; Klein et al., 2003), and mechanical stress has recently been shown to collaborate with heparin binding to FNIII 11-12-13 to determine FN structure (Hubbard et al., 2014; Zollinger and Smith, 2017). It remains to be shown whether mechanical stress can also modulate the function of ED-A in LTBP-1 binding to FN as suggested by increasing colocalization of both proteins in ECM with increasing stiffness in our fibroblast cultures.
In summary, we found that the mechanical conditions that fibroblasts encounter in advanced stages of wound healing and fibrosis stimulate ED-A FN and LTBP-1 co-expression, and their interaction in the ECM. The ED-A domain plays a supporting role in promoting FN interactions with LTBP-1, and the HepII domain stretch adjacent to ED-A in FN appears to enhance this binding. We propose that blocking the interaction of LTBP-1 with ED-A FN by using competitive domain peptides or specific antibodies is a potential strategy to specifically reduce TGF-β1 storage in the myofibroblast-associated ECM and ultimately development of fibrosis.
MATERIALS AND METHODS
Normal fibroblasts were explanted from human dermal tissue samples (n=5), received from Dr Benjamin A. Alman (Sick Kids Hospital, Toronto, ON) as described previously (Klingberg et al., 2014). Written consent for the use of human biopsy material was obtained from patients and procedures approved by the Institutional Review Board of the Hospital for Sick Children (Toronto, Canada). HDfs between passages (P)2–P5 and lineage MRC-5 fibroblasts (ATCC, Manassass, VA) were maintained in standard cell culture (Dulbecco's modified Eagle's medium, DMEM; Life Technologies, Burlington, ON), supplemented with 10% fetal bovine serum (Sigma-Aldrich, Oakville, ON) and penicillin-streptomycin (Life Technologies). HEK293 cells, stably expressing LTBP-1–EGFP or different rat FN constructs were selected and maintained in Zeocin™ and G418 (Life Technologies) (Klingberg et al., 2014). In select experiments, mechanical growth conditions for fibroblasts were controlled by using deformable silicone (polydimethylsiloxane) substrates with a Young's modulus of 3, 10, 25, 100 and 3000 kPa (Excellness Biotech SA, Lausanne, Switzerland) that were coated with 2 µg/cm2 gelatin (Sigma-Aldrich) (Li et al., 2017).
Reagents and antibodies
siRNA constructs directed against human FN were designed and purchased from Thermo Fisher Scientific (Burlington, ON). All cell transfections were performed according to the manufacturer's specifications by using an electroporation device (NEON, Life Technologies), with two pulses at 1150 V and 20 ms duration. Primary antibodies used in this study were directed against: α-SMA (clone SM1, a kind gift of Giulio Gabbiani, University of Geneva, Switzerland, 1:100), (total) FN (rbAb, Sigma-Aldrich, F3648, 1:100), ED-A FN (mAb, clone IST-9, Santa Cruz Biotech, Dallas, TX, sc-59826, 1:100), ED-B FN (mAb, clone BC-1, Abcam, Cambridge, MA, ab154210, 1:100), FN HepII (mIgG1, clone Ab32, Thermo Fisher CSI 005-32-02, 1:50) LTBP-1 (mAb, R&D Systems, Minneapolis, MN, MAB388, 1:100; and rbAb39, a very generous gift from Carl-Hendrik Heldin, Uppsala University, Sweden, 1:200), HIS (mAb A00186 and rbAb A00174, Genscript, Piscataway, NJ, 1:200), fibrillin-1 (rF6H ID 157, a kind gift from D. Reinhardt, McGill University, Montreal, Canada, 1:50), and vimentin (mAb, Dako, Burlington, ON, M0725, 1:400). Secondary antibodies used were: goat anti-mouse-IgG conjugated to Alexa Fluor 568 (Life Technologies, A-11004, 1:250), goat anti-mouse-IgG1 conjugated to FITC (Southern Biotechnology, Birmingham, AL, 1070-02, 1:100), goat anti-mouse-IgG2a conjugated to TRITC (Southern Biotechnology, 1080-03, 1:100), goat anti-mouse-IgG2b conjugated to TRITC (Southern Biotechnology, 1090-03, 1:100), and goat anti-rabbit-IgG conjugated to TRITC and FITC (Sigma-Aldrich, F9887, 1:100). To stain DNA and nuclei, 4,6-diamidino-2-phenylindole dihydrochloride (DAPI, Sigma-Aldrich, D9542) was used (1:50). For blocking experiments, hDfs were cultured in standard conditions for 2 days and then incubated with anti-ED-A FN (mAb clone IST-9, sodium azide-free, Abcam, ab6328) or anti-HepII (Ab32; Underwood et al., 1992) at 100 µg/ml (Hinz et al., 2001; Serini et al., 1998) or purified FN peptide fragments (50 µg/ml) for another 5 days with daily replenishments. Controls were 10% FBS, BC-1 (Abcam) and human IgG1 antibodies (Sigma-Aldrich, I4506), all at 100 µg/ml. Heparan sulfate and control chondroitin sulfate were added at 500 µg/ml with the same time course and replenishments.
Cell fractionation and western blotting
hDfs were cultured to confluency for 7 days in all experiments. For regular western blots (‘whole lysates’), culture medium was removed, and cultures were washed with PBS and scraped into standard lysis buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 0.1% Bromophenol Blue and 10% glycerol). To blot proteins from conditioned medium, culture supernatants were harvested, concentrated 10-fold with size exclusion centrifugation filters, and dialyzed against RIPA buffer (150 mM NaCl, 1 mM EDTA, 25 mM Tris-HCl pH 7.4 and 1% Triton X-100), supplemented with protease inhibitor cocktail (1:50 dilution, Sigma P8340) and 1 mM sodium orthovanadate and thoroughly sonicated. After removing supernatants, the same cultures were then used to either prepare cell fractions or ECM fractions. To obtain ECM fractions (‘ECM’), cells were removed using decellularization buffer (20 mM NH4OH and 0.5% Triton X-100 in PBS), rinsed twice in the same buffer, washed three times with PBS, and then scraped into RIPA buffer. To obtain cell fractions (‘cells’), cells were gently trypsinized (0.25% trypsin, 5 min), centrifuged (800 g for 5 min) and the pellet lysed in RIPA buffer. Remaining material from these dishes was not used to produce ECM fractions. Western blotting was performed in reducing conditions (α-SMA and vimentin) or non-reducing conditions (all ECM proteins) on 8% and 10% SDS-PAGE gels. Proteins were transferred onto nitrocellulose membranes by using a wet transfer technique. Protein membranes were blocked with 5% skim milk, and primary antibodies were detected with fluorescently labeled anti-mouse-IgG or anti-rabbit-IgG conjugated to 680 nm 800 nm IRDye®, respectively, secondary antibodies (1:10,000, LICOR Biosciences, LIC-926-68020 and LIC-926-32211). Signals were detected and quantified with a LICOR Fx imaging system (LI-COR Biosciences, Lincoln, NE).
Rat FN domain peptides were expressed in E. coli, purified, and characterized as published in detail previously (Kohan et al., 2010). Full-length FN constructs were produced by cloning the entire sequences of the rat FN splice variants (Schwarzbauer et al., 1987) into pcDNA3.1 using the Invitrogen TOPO TA cloning method (Invitrogen) (Sackey-Aboagye et al., 2016). All 6xHis-tagged proteins were purified from serum-free conditioned medium from transfected HEK293 cells. In brief, conditioned medium was collected and dialyzed against phosphate-buffered saline (PBS, Life Technologies) before it was run through an ion metal affinity chromatography column with HIS-Select® Nickel Affinity Gel (Sigma-Aldrich). Columns were washed with PBS buffer containing 0, 10 or 15 mM imidazole (Sigma-Aldrich). Fractions containing LTBP-1 were eluted with 250 mM imidazole. Full-length protein constructs were detected by western blotting using anti-His antibodies.
Immunofluorescence, microscopy and quantitative image analysis
Samples were treated in sequence for immunostaining: fixation with 3% paraformaldehyde for 10 min, permeabilization with 0.2% Triton X-100 (Sigma-Aldrich), incubation with primary antibodies for 1 h, and labeling with secondary antibodies for 1 h, all at room temperature. Fluorescence microscopy images were acquired with an Axio Imager upright microscope equipped with an AxioCam HRm camera, Apotome 2 structured illumination and ZEN software (Zeiss, Oberkochem, Germany). Plan-Apochromat objectives were used (Zeiss, 40×, NA 1.2, and Zeiss, 63×, NA 1.4, Oil-DIC) in addition to a Fluar objective (Zeiss, 20×, NA 0.75). Confocal images were acquired at the Centre for Microfluidics Systems, University of Toronto, using a Nikon Eclipse Ti microscope system and Apo 60× objective. Quantitative image analysis was performed using ImageJ (http://imagej.nih.gov/ij/) using customized macros (available upon request). Figures were assembled in Adobe Photoshop CS5 (Adobe Systems, San Jose, CA).
To study protein–protein interactions, ELISA with fluorescent detection was established by coating black clear bottom 96-well plates with 10 µg of full-length FN splice variants, domain peptides of FN or 10 µg LTBP-1 (control) overnight at 4°C. Wells were then incubated with 0.5% BSA and 10 µg/ml heparin in PBS for 1 h. After three washes with PBS, 10 µg of LTBP-1 was added to the wells for 2 h at 4°C, with the exception of LTBP-1 control wells. In select experiments, LTBP-1 was pre-incubated with 0.5 mg/ml BSA or chondroitin sulfate (controls) or HS, to block heparin-binding sites, before adding the whole solution to full-length FNs for binding assays. Subsequently, wells were washed with PBS, stained for LTBP-1 and signals detected with fluorescent antibodies in a LICOR Fx imaging system (LI-COR Biosciences).
Active and total TGF-β were quantified by using transformed reporter mink lung epithelial cells (TMLCs), producing luciferase under the control of the PAI-1 promoter in response to TGF-β (Abe et al., 1994). After treatment with ED-A domain blocking antibodies and controls, supernatants of hDf cultures were collected. To assess TGF-β levels, TMLCs (60,000 cells/cm2) were adhered for 4 h before being subjected to either native conditioned medium (active TGF-β) or heat-activated medium for 10 min at 80°C (total TGF-β) for an additional 16 h. All results were corrected for TMLC baseline luciferase production in non-conditioned culture medium.
When applicable, data are presented as means±s.d. Differences between groups were assessed with a one-way analysis of variance (ANOVA) followed by a post-hoc Dunnett's multiple comparison test and the significance level set at P≤0.05. Statistical analyses and data plots were performed using Prism software (GraphPad, San Diego, CA). *P≤0.05; **P≤0.01; ***P≤0.005.
We thank Dr Pierre Jurdic (Université de Lyon, France) for providing Flp-In™ 293 cells, Drs Christine Chaponnier and Giulio Gabbiani (University of Geneva, Switzerland) for providing antibodies directed against α-SMA, Dr Carl-Hendrik Heldin (Uppsala University, Sweden) for anti-LTBP-1 (Ab39) and Dr Dieter Reinhardt for anti-fibrillin-1 (McGill University, Canada). Dr Christopher A. G. McCulloch and Dr Craig Simmons (University of Toronto, Canada) are acknowledged for technical support and advice.
Conceptualization: F.K., R.W., E.W., B.H.; Methodology: F.K., G.C., S.B., M.C., A.K., A.O., M.I., R.W., E.W.; Validation: F.K., G.C., M.C., A.K., M.L., B.H.; Investigation: F.K., G.C., M.W., S.B., M.C., A.K., M.I., B.H.; Resources: B.H.; Data curation: F.K., G.C., M.W., S.B., M.C., A.K., A.O., M.I., M.L.; Writing - original draft: F.K., G.C., M.C., A.K., B.H.; Writing - review & editing: R.W., E.W., B.H.; Supervision: B.H.; Project administration: B.H.; Funding acquisition: B.H.
This research was supported by the Canadian Institutes of Health Research (CIHR; grants #375597, #210820, #286920 and #286720), the Collaborative Health Research Programme (CIHR and Natural Sciences and Engineering Research Council of Canada grant #413783), the Canada Foundation for Innovation and Ontario Research Fund (CFI/ORF grants #26653 and 36050), the E-Rare Joint Transnational Program ‘Development of Innovative Therapeutic Approaches for Rare Diseases’ (Grant # ERL-138395) (all to B.H.), and from the US National Institutes of Health (PO1 grants DK-058123 to R.G.W., and HL085083 to E.S.W.). Data presented herein was further receiving support from the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement #237946, the CIHR Cell Signals Training program (F.K.) and an NIH training grant to A.L.O. (F30-DK081265-01). M.W. was supported by post-doc fellowships from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (Netherlands Organization for Scientific Research) and the CIHR. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.