Newly deposited microfibrils strongly colocalise with fibronectin in primary fibroblasts. Microfibril formation is grossly inhibited by fibronectin depletion, but rescued by supplementation with exogenous cellular fibronectin. As integrin receptors are key determinants of fibronectin assembly, we investigated whether they also influenced microfibril deposition. Analysis of β1-integrin-receptor-null fibroblasts, blockage of cell surface integrin receptors that regulate fibronectin assembly and disruption of Rho kinase all result in suppressed deposition of both fibronectin and microfibrils. Antibody activation of β1 integrins in fibronectin-depleted cultures is insufficient to rescue microfibril assembly. In fibronectinRGE/RGE mutant mouse fibroblast cultures, which do not engage α5β1 integrin, extracellular assembly of both fibronectin and microfibrils is markedly reduced. Thus, pericellular microfibril assembly is regulated by fibronectin fibrillogenesis.
Fibrillin-1 (official protein symbol FBN1) is a large glycoprotein that assembles pericellularly into microfibrils that have a complex structural organisation and are widespread in elastic tissues such as skin, lung, arteries and ligaments (Hubmacher et al., 2006; Kielty, 2006). It contains 47 epidermal growth factor-like domains, 43 of which bind calcium, which are interspersed with eight cysteine-containing transforming growth factor-β (TGF-β) binding-protein-like (TB) modules (Pereira et al., 1993). The process of fibrillin-1 microfibril assembly remains poorly understood; however, as with other major extracellular matrix (ECM) assemblies such as fibronectin and collagen I fibrils, assembly might be driven by both self-assembly and cell-directed mechanisms. Fibrillin-1 N- and C-terminal interactions are thought to support linear assembly (Trask et al., 1999; Lin et al., 2002; Marson et al., 2005). Cellular control of assembly is apparent at the level of furin processing at N- and C-termini, which may be a prerequisite for interactions between fibrillin-1 termini during assembly (Wallis et al., 2003). However, the potential contribution of cell surface receptors to the process of microfibril assembly is unknown.
Evidence is emerging that fibronectin orchestrates the assembly of ECM (Velling et al., 2002; Dallas et al., 2005; Dallas et al., 2006; Sottile et al., 2007). Fibronectin fibrillogenesis occurs pericellularly and is dependent upon cell-surface-receptor engagement. Assembly is initiated by N-terminal fibronectin interactions with the cell surface, which induces α5β1 integrin clustering and translocation from focal contacts into fibrillar adhesions, followed by fibronectin self-association (Wierzbicka-Patynowski et al., 2003; Mao and Schwarzbauer, 2005; Pankov et al., 2000; Tomasini-Johansson et al., 2006). Integrin-activated RhoA/ROCK, FAK, Akt and protein kinase C pathways regulate cytoskeletal changes, and cellular ability to exert force on matrix leads to exposure of cryptic self-association sites that support higher-order fibronectin assembly (Wu et al., 1995; Zhong et al., 1998; Baneyx et al., 2002; Ilic et al., 2004; Wang et al., 2005; Féral et al., 2007; Somanath et al., 2007; Somers and Mosher, 1993; Yoneda et al., 2007). Fibronectin can also assemble without RGD-dependent α5β1 integrin engagement, in this case forming short thick pericellular filaments through αvβ3 integrin ligation of an N-terminal NGR motif (Takahashi et al., 2007).
We tested the hypothesis that fibronectin plays a central role in fibrillin microfibril assembly. Here we show that the deposition of extensive microfibril arrays is regulated by α5β1-integrin-mediated fibronectin fibrillogenesis.
Fibronectin is required for fibrillin-1 microfibril assembly
Confocal microscopy of primary human dermal fibroblasts (HDFs) was employed to determine whether fibronectin is an orchestrator of fibrillin-1 microfibril assembly (Fig. 1). In 24 hour cultures, fibronectin had formed short extracellular fibrils, whereas fibrillin-1 was mainly cell-associated with some pericellular staining. The 5 day cultures had assembled extensive organised, directionally aligned fibrillin-1 microfibrils and fibronectin arrays that strongly colocalised. RNAi was then used in HDFs to determine whether fibronectin influences fibrillin-1 microfibril formation.
Fibronectin knockdown disrupts fibrillin-1 microfibril assembly
Fibronectin was depleted in HDFs using either of two siRNA oligonucleotide sequences, designated FN1 and FN2 (Fig. 2). Control HDFs were transfected with a scrambled version of the siRNA oligonucleotide FN2, or were untransfected. In all experiments, cells were cultured in medium containing serum from which plasma fibronectin had been stripped. Fibronectin knockdown by FN1 or FN2 oligonucleotides, assessed at the mRNA level by RT-PCR, was significant compared with levels in control cells (P=0.0002). mRNA expression levels of fibrillin-1, and also collagen I as a control ECM molecule, were not significantly altered by fibronectin RNAi (Fig. 2A). Western blotting showed drastically reduced levels of fibronectin in the cell lysates and conditioned medium of 24 hour and 5 day FN1 and FN2 siRNA cultures compared with control cells (Fig. 2B). However, fibrillin-1 protein levels were similar in control and fibronectin siRNA cultures (Fig. 2B). Subsequent knockdown experiments utilised FN1 oligonucleotides.
Confocal microscopy of HDF cultures from 24 hours to 5 days after fibronectin knockdown revealed that, as expected, fibronectin deposition was greatly reduced in the fibronectin-depleted cultures. From 24 hours to 3 days in the siRNA cultures, there was very little fibronectin staining, but by 5-7 days, a few fibrils were detected (Fig. 2C). Fibronectin RNAi also profoundly disrupted fibrillin-1 microfibril assembly. At 24 hours, fibrillin-1 staining was mostly cell-associated, as seen in control cells (see Fig. 1); however, in the 5 day fibronectin siRNA cells, fibronectin and microfibril deposition was severely ablated compared with that in control cells (Fig. 2D).
Exogenous fibronectin rescues microfibril assembly in fibronectin siRNA cultures
To investigate whether exogenous fibronectin rescues fibrillin-1 microfibril assembly, fibronectin siRNA cultures were supplemented with human plasma or cellular fibronectin. RT-PCR confirmed that addition of exogenous fibronectin to the fibronectin siRNA cultures did not alter the depleted fibronectin mRNA levels or the fibrillin-1 and collagen I control mRNA levels (not shown). Western blotting also confirmed that addition of fibronectin did not alter fibrillin-1 levels in the lysates or conditioned medium of fibronectin siRNA or control cultures (not shown). Confocal microscopy of the fibronectin-depleted HDF cultures at 5 days showed limited rescue of fibronectin and microfibril assembly by plasma fibronectin (not shown) but strong rescue and colocalisation of both assemblies by cellular fibronectin (Fig. 2D). These knockdown and rescue experiments show that fibronectin is required for fibrillin-1 microfibril assembly.
Microfibril assembly also requires α5β1-integrin-mediated fibronectin assembly
We investigated whether integrin α5β1 and αvβ3, which are key regulators of extracellular fibronectin assembly (Mao and Schwarzbauer, 2005; Takahashi et al., 2007), also influence fibrillin-1 assembly. Flow cytometry confirmed the expression by HDFs of both α5β1 and αvβ3 integrin (Fig. 3A).
Blocking integrin α5β1 disrupts fibronectin and microfibril assembly
Immunoblotting of cell lysates and conditioned medium from the 24 hour and 5 day cultures confirmed that protein levels of fibrillin-1 were similar in the presence or absence of integrin-blocking antibodies (Fig. 3B). Confocal microscopy of 24 hour and 5 day cultures, treated with an anti-β1-integrin function-blocking antibody (mAb13) and an anti-α5-integrin function-blocking antibody (mAb16), showed severely ablated fibronectin and microfibril deposition compared with untreated cultures (Fig. 3C) or cultures treated with the non-function-blocking α5-integrin mAb11 (not shown). In the cultures treated with blocking antibodies, fibrillin-1 still colocalised with fibronectin, although not all fibronectin was associated with fibrillin-1. Thus, inhibiting integrin α5β1 grossly disrupted the extracellular assembly of both fibronectin and fibrillin-1 microfibrils.
Integrin-β1-null murine embryonic fibroblasts disrupt fibronectin and microfibril assembly
To confirm that β1 integrins are needed for fibronectin and microfibril assembly, wild-type and β1-integrin-null mouse embryonic fibroblasts (MEFs) were analysed at 24 hours and 5 days. Flow cytometry confirmed that integrin subunits β1 and α5 were expressed by the wild-type MEFs but not by β1-integrin-knockout MEFs (Fig. 4A). Confocal microscopy revealed that the wild-type cells began to assemble extracellular fibronectin fibrils by 24 hours, whereas fibrillin-1 staining was mostly cell-associated (not shown). By 3 days, wild-type MEFs had deposited dense colocalising networks of fibronectin and fibrillin-1 (Fig. 4B). By contrast, the β1-integrin-knockout MEFs had deposited markedly reduced fibronectin and fibrillin-1 microfibrils by 3 days, although colocalisation was still apparent (Fig. 4B). These data confirm that β1 integrin plays a critical role in fibronectin assembly, and indicate that the deposition of microfibrils follows fibronectin assembly (Fig. 4B). As β1-integrin-null murine cells express αvβ3 (Retta et al., 2001), the low levels of fibronectin might have assembled through αvβ3 integrin (Takahashi et al., 2007).
Blocking integrin-induced cytoskeletal signals disrupts fibronectin and microfibril assembly
Integrins link the ECM to the actin cytoskeleton, regulating cell contractility and generating tension that allows fibronectin self-association. Integrin α5β1 activates RhoA and downstream Rho kinase (ROCKI and ROCKII) (Yoneda et al., 2007), and Rho kinase activity is necessary for fibronectin assembly (Baneyx et al., 2002; Wu et al., 1995). We therefore investigated whether this signalling pathway is also necessary for fibrillin-1 microfibril deposition. The selective ROCK inhibitors, H1152 and Y27632, at concentrations of 10 μM and 30 μM, respectively, caused grossly different fibrillar organisation and reduced deposition of both fibronectin and fibrillin-1, compared with control cells either supplemented with 0.2% (v/v) DMSO or unsupplemented cultures (Fig. 5). Immunoblotting of cell lysates and conditioned medium at 5 days confirmed that fibronectin and fibrillin-1 protein levels were unaffected by ROCK inhibition (not shown). Thus, integrin-regulated cytoskeletal changes suppress fibronectin and microfibril assembly.
Integrin αvβ3 is not essential for microfibril assembly
Although integrin α5β1 is a primary regulator of extracellular fibronectin assembly (Mao and Schwarzbauer, 2005), αvβ3 integrin also supports the assembly of short pericellular fibronectin fibrils (Takahashi et al., 2007). Incubation with cilengitide (0.1 μg/ml), a specific cyclic RGD peptide inhibitor of αvβ3 integrin, did not disrupt either fibronectin or microfibril assembly (Mitjans et al., 1995) (not shown). Cells incubated with the αvβ3-integrin function-blocking antibody LM609 also assembled well-organised networks of both molecules (not shown). Incubation with mAb 17E6, which inhibits and internalises αv integrins (Castell et al., 2001), also did not ablate either fibronectin or microfibril assembly (not shown). Immunoblotting of cell lysates and medium from the LM609 antibody experiments showed that there was no alteration in mRNA or protein levels of either molecule (not shown). Together, these data show that αvβ3 integrin is not a major regulator of fibronectin or microfibril assembly in primary fibroblasts.
Activation of β1 integrins in fibronectin-depleted cultures does not rescue microfibril assembly
Having established that the assembly of extensive microfibril arrays is suppressed whenever α5β1-integrin-mediated fibronectin fibrillogenesis is disrupted, we investigated whether fibronectin-independent activation of β1 integrins can directly rescue microfibril assembly, using fibronectin-depleted cultures incubated with the integrin-β1-activating antibody TS2/16 at a concentration (10 μg/ml) that activates β1-integrin-mediated signalling (Lomas et al., 2007), or with mouse IgG as a control. RT-PCR results indicated that fibronectin was significantly reduced by RNAi (P=0.0027) and neither TS2/16 nor a mouse IgG control stimulated fibronectin expression (not shown). mRNA levels of fibrillin-1 and a control matrix molecule, collagen type I, were not affected by fibronectin RNAi or by supplementation with either antibody (not shown). Immunoblotting of the 24 hour and 5 day cell lysates and media showed that TS2/16 supplementation had no effect on protein levels of fibronectin or fibrillin-1 (not shown). Confocal microscopy of fibronectin-depleted or control cell cultures at 24 hours and 5 days showed that microfibril assembly was not rescued by TS2/16 (not shown). Thus, activation of β1 integrins in fibronectin-depleted cells is not sufficient to support microfibril assembly.
α5β1-integrin-mediated fibronectin assembly is required for deposition of extensive microfibril arrays
Studies on fibronectinRGE/RGE mice and cells have shown that, in the absence of RGD-dependent α5β1 integrin interactions, fibronectin can assemble into short thick pericellular fibrils through αvβ3 integrin (Takahashi et al., 2007). We used confocal immunomicroscopy of nonpermeabilised cells (N-19 pAb) to examine microfibril deposition in wild-type and fibronectinRGE/RGE murine cells, in order to establish whether microfibril deposition is dependent on α5β1-integrin-mediated fibronectin assembly. In the wild-type MEFs, abundant extracellular microfibril networks were deposited and strongly colocalised with fibronectin (Fig. 6). In the fibronectinRGE/RGE cultures, markedly reduced levels of extracellular fibronectin were detected and there were very few extracellular microfibrils (Fig. 6). These experiments show that the deposition of extensive microfibril arrays requires α5β1-integrin-mediated fibronectin fibrillogenesis.
Despite intense research, fibrillin microfibril assembly remains a poorly understood process. It is widely considered that microfibrils assemble pericellularly, following furin processing, through high-affinity N- and C-terminal interactions (Hubmacher et al., 2006; Kielty, 2006). However, assembly of other ECM polymers, such as fibronectin and collagen fibrils, are driven by both cell-directed and self-assembly processes (Mao and Schwarzbauer, 2005; Velling et al., 2002; Dallas et al., 2006; Sottile et al., 2007). In this study, we have investigated how the cell-matrix interface influences fibrillin-1 assembly, by testing the hypothesis that integrin-mediated fibronectin assembly is critical for fibrillin-1 microfibril assembly (Fig. 7). Our novel findings are that the deposition of microfibrillar arrays is dependent upon the assembly of fibronectin through α5β1 integrin.
Immunofluorescence microscopy showed that extracellular fibronectin and fibrillin-1 microfibrils strongly colocalise (although fibronectin is more abundant, and not all fibronectin is associated with microfibrils), and that siRNA depletion of fibronectin grossly reduced both fibronectin and fibrillin-1 microfibril deposition. Thus, there is a critical assembly relationship between these ECM polymers. The deposition of fibronectin and microfibrils was more strongly rescued by cellular fibronectin, which contains EDA and EDB domains flanking the central cell-binding region (Pankov and Yamada, 2002), than by soluble plasma fibronectin. Thus, fibronectin isoforms may differentially influence microfibril assembly in vivo.
Fibronectin fibrillogenesis generally requires α5β1 integrin (Clark et al., 2005), although αvβ3 integrin can support reduced assembly of short thick pericellular fibronectin fibrils (Takahashi et al., 2007). Using integrin antibody-blocking and inhibitor experiments, and β1-integrin-null murine cells, it was clear that α5β1 integrin is also critical for the deposition of fibrillin-1 microfibril arrays because both fibronectin and microfibril deposition is greatly reduced in these conditions. The cosuppression of fibronectin and fibrillin-1 assembly in the antibody-blocking and mouse mutant cultures supports a model in which microfibril assembly is dependent upon fibronectin assembly. Furthermore, since antibody activation of β1 integrins failed to rescue microfibril deposition in fibronectin-depleted cells, activated integrin alone is insufficient to drive microfibril assembly. These data, which indicated that microfibril assembly is dependent on fibronectin fibril formation through α5β1 integrin, were reinforced using fibronectinRGE/RGE cultures, which cannot assemble fibronectin through α5β1 integrin (Takahashi et al., 2007), because almost no extracellular fibrillin-1 microfibrils were apparent. In addition, although HDFs can adhere to fibrillin-1 molecules and microfibrils in an RGD-dependent manner through α5β1 integrin (Bax et al., 2003; Bax et al., 2007), we found that supplementation with fibrillin-1 RGD fragments or a fibrillin-1-specific RGD antibody (Lee et al., 2004) has no disruptive effect on microfibril assembly (our unpublished data), so microfibril assembly does not require direct α5β1 integrin interactions. Fibrillin-1 interactions with cells may be physiologically important in the form of assembled microfibrils, as seen in blood vessels (Bunton et al., 2001).
Integrin α5β1 ligation of fibronectin at the cell surface activates cell-signalling pathways that regulate the cytoskeleton, including RhoA/ROCK (Yoneda et al., 2007; Féral et al., 2007), and tensional forces may then expose cryptic self-assembly sites (Zhong et al., 1998). We confirmed that cytoskeletal disruption due to ROCK inhibition suppresses fibronectin assembly, and the concurrent disruption to microfibril deposition supports the dependence of fibrillin-1 assembly on α5β1-integrin-mediated fibronectin assembly. It seems likely that the fibronectin interaction with α5β1 integrin, which induces exposure of cryptic fibronectin self-association sites, results in a pericellular form of fibronectin that acts as a suitable template to support the assembly of extensive microfibril arrays. Although fibrillin-1 can colocalise with fibronectin that is not assembled through α5β1 integrin, microfibril deposition is greatly reduced in such conditions (see Fig. 4B, Fig. 6). Using in vitro binding assays, we detected molecular interactions between fibrillin-1 and specific recombinant fragments of fibronectin, including the N-terminal region of fibronectin that is particularly important in assembly (supplementary material Fig. S1); these interactions may support transient cell surface associations. Only weak interactions were detected between plasma fibronectin and fibrillin-1, implicating cryptic fibronectin sites that may be exposed through integrin interactions. It is also possible that cell surface heparan sulphate, which binds both molecules, could mediate their association.
The dependence of fibrillin-1 assembly on fibronectin and α5β1 integrin has pathological implications. α5β1-integrin-null mice are embryonic lethal (Bouvard et al., 2001), as are fibronectin-null mice, which suffer from severe cardiovascular defects (George et al., 1997); these pathologies may thus reflect loss of fibronectin and also consequent disruption to the assembly of fibrillin-1. Fibronectin also regulates the assembly of other ECM molecules, including the latent TGFβ binding protein (LTBP1), which controls TGFβ bioavailability. LTBP1, which also contains cbEGF and TB domains, is a member of the fibrillin superfamily of ECM glycoproteins and interacts directly with the N-terminal region of fibrillin-1 (Isogai et al., 2003; Chaudhry et al., 2007). It is of note that, although the matrix deposition of LTBP-1 requires fibronectin (Dallas et al., 2005), LTBP1 binds fibrillin-1 but does not interact directly with fibronectin, so its deposition might in fact be dependent on fibronectin-mediated microfibril assembly. In summary, we have shown that the pericellular assembly of fibrillin-1 microfibrils is dependent on α5β1-integrin-mediated fibronectin assembly.
Materials and Methods
Primary antibodies used were: rabbit anti-human fibronectin polyclonal antibody (pAb) (plasma and cellular) (Sigma-Aldrich); fibronectin-3E2 mouse anti-cellular fibronectin monoclonal antibody (mAb) (Sigma-Aldrich); MAB1919/clone 11C1.3 mouse anti-fibrillin-1 mAb (Chemicon); rabbit anti-fibrillin-1 proline-rich region (raised to a recombinant peptide) (here designated `PRO') pAb (gift from P. A. Handford, Oxford, UK); rabbit anti-fibrillin-1 proline-rich region (raised to a synthetic peptide) pAb (raised to a synthetic peptide by Eurogentec); rabbit pAb raised (Bethyl Laboratories) to the fibrillin-1 RGD and flanking sequence (CYLDIRPRGDNGTA), as previously reported (Sakamoto et al., 1996); goat anti-fibrillin-1 N-terminal region (N-19) pAb (Santa Cruz, sc-7541), anti-human integrin α5 mAbs (blocking,mAb16; non-inhibitory, mAb11); mouse anti-human integrin α5 mAb (12G10) and rat anti-mouse integrin α5 mAb (5H10) (gifts from M. J. Humphries, Manchester, UK); TS2/16 mouse anti-human integrin β1 mAb (Pierce Biotechnology); KMI6 rat anti-mouse integrin β1 mAb (BD Biosciences); LM609 mouse anti-human integrin αVβ3 mAb (Chemicon); 17E6 mouse anti-human integrin αV blocking mAb (kindly supplied by S. Goodman, Merck KGaA, Germany); 23C6 mouse anti-human αvβ3 (Abcam); AC-74 mouse anti-β-actin (Sigma-Aldrich). Secondary antibodies were donkey anti-mouse (FITC), donkey anti-rabbit (FITC), donkey anti-rabbit (TRITC), donkey anti-rat (FITC), donkey anti-rat (TRITC), donkey anti-sheep (FITC) and donkey anti-sheep Cy3 (Jackson ImmunoResearch Laboratories); rabbit anti-mouse HRP, goat anti-rabbit HRP, and goat anti-sheep HRP (DAKO). Control immunoglobulins (IgGs) from murine serum, rabbit serum, rat serum and sheep serum were from Sigma. F-actin was stained using Texas Red-X phalloidin (Molecular Probes). A cyclic RGD peptide inhibitor of αv integrin (cilengitide) was kindly provided by S. Goodman, Merck KGaA (Germany) and used at 10 μM concentration.
Cells and cell culture
HDFs (Invitrogen) were cultured in Modified Eagle's Medium (MEM) plus Earle's plus L-glutamine (Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS; BioWhittaker), 1 mM sodium pyruvate, non-essential amino acids (100×), MEM vitamins (100×), 2 mM L-glutamine, 10 μg/ml penicillin, 5 μg/ml streptomycin sulphate (all from Gibco) at 37°C, 5% CO2 up to passage 8. FBS was stripped of plasma fibronectin by passing the serum over a gelatin Sepharose™ 4B column (Amersham Biosciences), as described (McKeown-Longo et al., 1985). Where appropriate, medium was supplemented with either 20 μg/ml human plasma fibronectin (Chemicon) or with 10 μg/ml cellular fibronectin isolated from human foreskin fibroblasts (Sigma). In some experiments, medium was supplemented with inhibitory or stimulatory antibodies or control immunoglobulins (IgGs), at a concentration of 10 μg/ml, or with either of two well-characterised inhibitors of ROCKI and ROCKII, H1152 and Y-27632 (Calbiochem), respectively, at working concentrations of 10 μM and 30 μM, or with recombinant fibrillin-1 RGD fragments (Bax et al., 2007).
Wild-type and fibronectinRGE/RGE mutant mouse embryonic fibroblasts (MEFs), a gift from Reinhardt Fässler (Munich, Germany) (Takahashi et al., 2007), were cultured for up to 14 days in Medium 231 with serum-free growth supplement (Cascade Biologics) at 37°C, 5% CO2. Immortalised wild-type and β1-integrin-knockout MEFs (from M. J. Humphries, Manchester, UK) were cultured at 33°C in Dulbecco's MEM supplemented with L-glutamine (BioWhittaker), 10% (v/v) FBS (BioWhittaker) precleared of fibronectin, 1 mM sodium pyruvate (Gibco), 2 mM L-glutamine (Gibco) and 20 U/ml interferon gamma (Sigma).
Cell surface receptors were analysed by flow cytometry, using established protocols. Briefly, 100 μl of each cell suspension (1×106 cells) was mixed with 100 μl primary antibody (for MEFs: mAbs 5H10, KM16; for HDFs: mAbs mAb13, mAb16, 17E6, 23C6) diluted to 20 μg/ml in phosphate-buffered saline (PBS), and incubated on ice for 45 minutes. Cells were harvested by centrifugation for 5 minutes at 800 g and washed three times in 600 μl PBS containing 1% (v/v) FBS. Cell pellets were resuspended in 50 μl FITC-conjugated secondary antibody diluted 1:50 in PBS containing 10% (v/v) human serum (Sigma-Aldrich) and incubated on ice for 45 minutes in the dark. Cells were collected by spinning for 5 minutes at 800 g and washed twice in 600 μl PBS containing 1% (v/v) FBS. Cells were washed in 300 μl PBS and resuspended in 400 μl PBS for analysis. For each sample, 20,000 cells were counted using a Cyan flow cytometer (Dako) at a flow rate of less than 200 events/second and an excitation wavelength of 488 nm.
Cells were immunostained using established protocols. Briefly, cells were cultured for various lengths of time before being fixed with 3% (v/v) formaldehyde in PBS for 20 minutes at room temperature. Cells were washed three times with PBS and formaldehyde groups were quenched with 0.2 M glycine in PBS for 20 minutes at room temperature. Cells were washed three times with PBS before being permeabilised in 0.5% (v/v) Triton X-100 in PBS for 4 minutes at 20°C. For the `nonpermeabilised' fibronectinRGE/RGE MEF experiments, cells were washed in PBS but were not treated with Triton X-100. After further PBS washes, cells were blocked with 3% (w/v) bovine serum albumin (BSA) for 1 hour at room temperature. Cells were incubated with primary antibodies diluted in 3% (w/v) BSA for 1 hour at room temperature, and were washed three times with PBS before incubation with fluorophore-conjugated secondary antibody (Jackson ImmunoResearch Laboratories) diluted in 3% BSA, for 45 minutes at room temperature in the dark. For HDFs, the primary antibodies were anti-fibrillin-1 PRO pAb (1:200) or anti-fibronectin (FN-3E2) mAb (1:200). For the β1-integrin-null and wild-type MEFs, the primary antibodies were anti-fibrillin-1 PRO pAb (1:200) or anti-fibronectin (FN-3E2) mAb (1:200). For the fibronectinRGE/RGE and wild-type MEFs, the primary antibodies were anti-fibrillin-1 N-19 pAb (1:50) or anti-fibronectin (FN-3E2) mAb (1:200). Cells were washed three times with PBS before incubating with 300 nM 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI; Molecular Probes) for five minutes at room temperature in the dark. Coverslips were mounted onto slides using Prolong® antifade mounting medium (Molecular Probes) and sealed with nail varnish. Slides were stored in the dark at 4°C until examination. Slides were examined either by conventional epifluorescence microscopy using an Olympus BX51 microscope and a Photometrics Coolsnap HQ camera driven by Metamorph software, or by confocal microscopy using a Leica SP2 AOBS (acusto-optical beam splitter) confocal microscope. Images were compiled and analysed using ImageJ software (National Institute of Health). Imaging of FITC or TRITC was always conducted at nonoverlapping wavelengths.
RNA interference was performed with siRNA oligonucleotides that were synthesised, purified and duplexed by Dharmacon RNA Technologies. siRNA duplexes were designed against the target sequence using Dharmacon siDESIGN® Center (two fibronectin sequences targeted: FN1, 5′-AACAAATCTCCTGCCTGGTAC-3′; FN2, 5′-AAGTGGTCCTGTCGAAGTATT-3′. The order of nucleotides in the second fibronectin sequence targeted was scrambled to generate a control sequence: FN2 scrambled, 5′-AACTAGGCGATAACACTCAAC-3′). HDFs were transfected with 100 nM siRNA duplex using an Amaxa Nucleofector™ device and Nucleofector™ kit optimised for HDFs (Amaxa Biosystems) according to the manufacturer's instructions. Cells were cultured for up to 7 days, and analysed by RT-PCR, immunofluorescence and western blotting of culture medium and cell lysates. RNAi knockdown efficacy was analysed using unpaired Student's t-tests (GraphPad Prism 2.0). Results were statistically significant when P<0.05 (*P<0.05, **P<0.001 and ***P<0.0001).
Cell layer extractions
Cells were cultured in six-well plates until confluent. The medium was removed and protease inhibitor cocktail (Sigma), diluted 1:2000 was added with urea, to a final concentration of 2 M. After washing twice with PBS, cells were lysed with 25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS for 30 minutes, in the presence of a protease inhibitor cocktail (1:2000), and scraped from the tissue culture flask. Debris was pelleted at 20,000 g for 5 minutes. Supernatants were removed and analysed by SDS-PAGE and immunoblotting.
Samples were mixed with 4× NuPAGE lithium dodecyl sulphate (LDS) sample buffer (Invitrogen) for analysis by SDS-PAGE. Where appropriate, samples were reduced in 10× NuPAGE reducing agent (Invitrogen) at 57°C for 30 minutes. Samples were run on SDS-PAGE gels in reducing conditions using the NuPAGE system (Invitrogen). Gels were silver stained or immunoblotted. For the immunoblotting, transferred membranes were blocked in 5% (w/v) dried milk in TTBS (0.15 M NaCl, 25 mM Tris-HCl, 0.5% (v/v) Tween-20, pH 8.0) for 1 hour at room temperature before incubation with primary antibodies diluted in 2% (w/v) milk in TTBS for 1 hour at 20°C. Membranes were washed in TTBS before incubation with HRP-conjugated secondary antibodies diluted in 2% (w/v) milk in TTBS for 1 hour at 20°C. Membranes were washed in TTBS prior to a final wash in TBS (0.15 M NaCl, 25 mM Tris-HCl, pH 8.0). Membranes were incubated with a 1:1 mixture of stable peroxide: enhancer ECL reagents (Amersham Biosciences) for 1 minute at 20°C then exposed to Kodak Biomax MR film. The nitrocellulose membrane was reduced in 4 M urea, 10 mM Tris-HCl pH 8.0, 10 mM DTT for 30 minutes at 20°C. Membranes were then blocked and probed, as described above. Fibrillin-1 was detected using the rabbit anti-fibrillin-1 proline-rich region pAb, and fibronectin using FN-3E2 mouse anti-cellular fibronectin, in all blots.
Reverse transcriptase polymerase chain reaction (RT-PCR)
RNA was isolated from cells using an SV Total RNA Isolation Kit (Promega Corporation). For first strand cDNA synthesis, 200 ng random primers (Promega Corporation), 500 ng polyA+ RNA and DNase, RNase-free UltraPURE water (Gibco) up to a volume of 22.5 μl were mixed in an RNase-free tube and incubated at 70°C for 10 minutes. Tubes were chilled on ice, and 8 μl of 5× reverse transcriptase reaction buffer (Roche), 4 μl 0.1M dithiothreitol (DTT; Promega Corporation), 4 μl of 10 mM dNTPs (Bioline) and 0.5 μl of 40 U/μl RNase inhibitor (Roche) were added. Mixtures were incubated at 37°C for 2 minutes prior to the addition of 1 μl AMV reverse transcriptase (25 U/μl; Roche), then incubated at 37°C for a further hour. Reactions were stopped by incubation at 95°C for 5 minutes.
PCR was performed using 2.5 μl first strand cDNA, mixed with 2.5 μl 10× PCR reaction buffer with MgCl2 (Roche), 0.25 μl of 25 mM dNTPs (Bioline), forward and reverse primers to a final concentration of 0.5 μM, 1.25 U Taq DNA polymerase (Roche) and DNase, RNase-free UltraPURE water (Gibco) to a final volume of 25 μl. Reaction mixtures were incubated for 2 minutes at 94°C, followed by 30 seconds at 94°C, 30 seconds at 55°C and 30 seconds at 72°C for 25 cycles, then a 10 minute incubation at 72°C. PCR was performed using a GeneAmp PCR system (Applied Biosystems). Oligonucleotide primers for PCR were designed using Primer3 software (Rozen and Skaletsky, 2000), using the same parameters resulting in similar Tm values and product lengths as shown. Primers used (MWG Biotech or Eurogentec) were: Fibrillin-1, 100 bp; forward primer 5′-CCTGTTCCGCTGTGAGTG-3′,Tm 58.90; reverse primer, 5′-ACTGATGCACGTGGTTGG-3′, Tm 59.04 (Ball et al., 2004). Collagen type I α1 chain, 86bp; forward primer, 5′-GGATTGACCCCAACCAAG-3′, Tm 58.69; reverse primer, 5′-AGTGGGGTACACGCAGGT-3′, Tm 58.97 (Ball et al., 2004). Fibronectin EDa splice variant, 82bp; forward primer, 5′-ACTCGAGCCCTGAGGATG-3′, Tm 58.86; reverse primer, 5′-CTGAGGCCTTGCAGCTCT-3′, Tm 59.39 (Ball et al., 2004). GAPDH, 93bp; forward primer, 5′-GAAGGCTGGGGCTCATTT-3′, Tm 60.16; reverse primer, 5′-TGGTTCACACCCATGACG-3′, Tm 59.92. PCR products were resolved by 2.5% UltraPURE agarose gels (Gibco) run in TAE buffer (40 mM Tris-acetate, 1 mM EDTA), with either a 1 kb DNA ladder (NewEngland Biolabs) or hyperladder V (Bioline Ltd). DNA was visualised under UV after by staining with 0.5 μg/ml ethidium bromide in TAE for 30 minutes.
Fibrillin-1 interaction with fibronectin
Plasma fibronectin and recombinant fibronectin fragments (N-terminal fragments 30 kDa and 45 kDa) were purchased from the Sigma Chemical Company (Poole, Dorset, UK). The cell binding region 50 kDa, and the H120 fragment (Schofield et al., 1998) were obtained from M. J. Humphries (Manchester, UK). The recombinant fibrillin-1 fragments were generated as previously described (Rock et al., 2004; Marson et al., 2005; Bax et al., 2003; Bax et al., 2007). Domain structures of these fragments and an SDS-PAGE gel of the FN fragments are shown in supplementary material Fig. S1A.
Plasma fibronectin or recombinant fibronectin fragments were immobilised onto 96-well plates (Immulon 4) at 10 μg/ml (100 μl/well) overnight at 4°C. Unbound FN was removed and the wells blocked with 200 μl 5% BSA (w/v) in Tris-buffered saline (TBS) for 2-3 hours at 20°C. Wells were washed three times with 200 μl wash solution (TBS + 0.1% BSA), then 100 μl of each recombinant fibrillin-1 fragment (His6-tagged) was added to the wells and incubated overnight at 4°C. Unbound fibrillin-1 fragments were removed, and the wells washed three times with 200 μl wash solution. Bound ligand was detected by incubation with anti-His6 antibody diluted 1:1000 in wash buffer for 3 hours, followed by anti-mouse horse radish peroxidase (HRP) secondary antibody diluted 1:2000 in wash buffer for 1 hour. Wells were washed four times with 200 μl wash solution then 100 μl detection reagent (11 mg 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS) dissolved in 0.5 ml H2O + 10 ml NaOAc (0.1 M)/NaH2PO4 (0.05 M) pH 5.0 +1 μl 30% H2O2) was added to the wells. Absorbance was read at 405 nm using a Dynex plate reader. Binding was analysed using unpaired Student's t-tests (GraphPad Prism 2.0). Results were statistically significant when P<0.05 (*P<0.05, **P<0.001 and ***P<0.0001).
To establish whether fibronectin could act as a direct template for cell surface fibrillin-1 assembly, we investigated potential molecular interactions using recombinant fibrillin-1 and fibronectin fragments in solid-phase binding assays. Only weak interactions were detected between plasma fibronectin and fibrillin-1 (not shown). However, several significant interactions were detected between specific recombinant human fibrillin-1 and FN fragments (supplementary material Fig. S1B). Fibrillin-1 overlapping fragments PF5 and PF7 bound to the fibronectin 30 kDa N-terminal fragment but not to the adjacent fibronectin 45 kDa fragment. PF5 and PF7 also bound the FN H120 fragment. N-terminal overlapping fibrillin-1 fragments PF1 and PF2 both bound to the FN 50 kDa central cell-binding region. These interactions may align fibrillin-1 molecules transiently on fibronectin during assembly.
This work was funded by the Wellcome Trust (R.K.), the Medical Research Council (G0200246) and the European Union (LSHM-CT-2005-018960). C.M.K. is a Royal Society Wolfson Research Merit Award holder. We thank Reinhardt Fässler (Max-Planck Institute of Biochemistry, Department of Molecular Medicine, Martinsried, Germany) for kindly supplying the fibronectin RGE mutant cells.
Supplementary material available online at http://jcs.biologists.org/cgi/content/full/121/16/2696/DC1
- Accepted June 3, 2008.
- © The Company of Biologists Limited 2008