NADPH oxidase-mediated induction of reactive oxygen species and extracellular matrix deposition by insulin-like growth factor binding protein-5

Abstract

Insulin-like growth factor binding protein-5 (IGFBP-5) induces production of the extracellular matrix (ECM) components collagen and fibronectin both in vitro and in vivo and is overexpressed in patients with fibrosing lung diseases, such as idiopathic pulmonary fibrosis (IPF) and systemic sclerosis (SSc). However, the mechanism by which IGFBP-5 exerts its fibrotic effect is incompletely understood. Recent reports have shown a substantial role of reactive oxygen species (ROS) in fibrosis; thus we hypothesized that IGFBP-5 induces production of ROS to mediate the profibrotic process. In vitro analyses revealed that ROS production was induced by recombinant and adenoviral vector-mediated IGFBP-5 (AdBP5) in a dose- and time-dependent manner, regulated through MEK/ERK and JNK signaling, and primarily mediated by NADPH oxidase (Nox). Silencing IGFBP-5 in SSc and IPF fibroblasts reduced ROS production. The antioxidants diphenyleneiodonium and N-acetylcysteine blocked IGFBP-5-stimulated ECM production in normal, SSc, and IPF human primary lung fibroblasts. In murine fibroblasts lacking critical components of the Nox machinery, AdBP5-stimulated ROS production and fibronectin expression were reduced compared with wild-type fibroblasts. IGFBP-5 stimulated transcriptional expression of Nox3 in human fibroblasts while selective knockdown of Nox3 reduced ROS production by IGFBP-5. Thus IGFBP-5 mediates fibrosis through production of ROS in a Nox-dependent manner.

INTRODUCTION

Fibrosing lung diseases represent a large clinical burden with significant morbidity and mortality worldwide (24, 42, 75). In these diseases, fibrosis often involves tissue remodeling with histopathological features including characteristic excess production and/or accumulation of extracellular matrix (ECM) components, often leading to functional impairment of organs either locally, such as affecting the lungs in idiopathic pulmonary fibrosis (IPF), or systemically, affecting skin, lungs, and a myriad other organs in systemic sclerosis (SSc; scleroderma) (3, 21, 62).

Mediating a variety of processes including growth and development, insulin-like growth factor binding protein (IGFBP)-5 is a member of the insulin-like growth factor (IGF) axis, which includes six IGFBPs (IGFBP1–6), two peptides (IGF-1 and IGF-II), and receptors (IGF1R, IGF2R, and insulin receptor) (7). Traditionally, the function of IGFBPs was thought only to encompass modulating the bioavailability of the IGFs, but recent evidence has shown that IGFBP-5 can mediate cellular processes independently of the IGFs, including cellular migration, proliferation, differentiation, programmed cell death, and ECM production and serve as a cancer biomarker for certain malignancies (16, 18, 57, 60, 67, 74).

We previously reported a link between fibrosing lung diseases and IGFBP-5 wherein transcription of IGFBP-5 was significantly increased in primary early passage dermal fibroblasts cultured from the affected skin of patients with SSc compared with fibroblasts from unaffected areas or from healthy controls (12). Further studies have shown increased IGFBP-5 transcript expression in and secretion by IPF fibroblasts, as well as higher IGFBP-5 expression in IPF lungs, with IGFBP-5 localized to fibroblasts and epithelial cells (50). A murine dermal fibrosis model optimized in our laboratory utilizes overexpression of IGFBP-5 to promote a profibrotic environment leading to fibroblast activation, myofibroblast conversion, and increased deposition of ECM components (57, 68). Using this model, we have shown that IGFBP-5 exerts its profibrotic effects in an IGF-independent manner (57). Although IGFBP-5 has been associated with fibrosis, the exact mechanism of IGFBP-5-mediated fibrosis has not been completed delineated.

Reactive oxygen species (ROS), such as superoxide anion and hydroxyl radical, participate in cell signaling or injury repair as a part of normal cellular physiology but have been implicated as mediators in the pathophysiology of fibrosis (8, 33, 40, 58, 59). Enhanced oxidative burden, due to pro-oxidant and antioxidant redox imbalance caused by increased ROS or decreased antioxidant capacity, has been linked to lung dysfunction in IPF, SSc, and related pulmonary fibroses (25, 26, 37, 39). The levels of oxidized proteins in bronchoalveolar lavage (BAL) fluid and biomarkers of oxidative stress in serum and expired breath condensates are increased in IPF patients (30, 32, 51). Biomarkers of lipid peroxidation in the urine, serum, and BAL fluid of SSc patients are increased and strongly correlate with lung involvement, indicating redox imbalance and oxidative stress in SSc (37, 45, 56, 63). Clinical trials with antioxidants in IPF have been met with varied results (4, 11, 19); thus it is critical to enhance our understanding of the role of oxidative stress in the fibrotic disease process.

Due to the correlative association of ROS and IGFBP-5 in IPF and SSc patients, we hypothesized that IGFBP-5 would induce the production of ROS, stimulating the secretion of ECM components and contributing to the fibrotic process. In this study, we demonstrate that IGFBP-5-induced ECM deposition was indeed dependent on the production of ROS.

MATERIALS AND METHODS

Reagents.

Dulbecco’s modified Eagle’s medium (DMEM) was obtained from Mediatech (Manassas, VA). MEK inhibitor U0126, phosphoinositide 3-kinase (PI3K) inhibitor LY294002, and antibodies for phospho-Raf, phospho-MEK1/2, phospho-ERK, phospho-mitogen-activated protein kinase kinase 4, phospho-JNK, phospho-c-Jun, phospho-activating transcription factor 2 (ATF2), and total JNK were from Cell Signaling (Danvers, MA). Antibodies for fibronectin, total MEK1/2, total ERK, caveolin-1, and GAPDH were from Santa Cruz Biotechnology (Santa Cruz, CA). N-acetylcysteine (NAC) and protease inhibitor cocktail were from Sigma-Aldrich (St. Louis, MO). JNK inhibitor II, p38 inhibitor SB203580, and diphenyleneiodonium (DPI) were from EMD Chemicals (Billerica, MA). Recombinant human IGFBP-5 (rBP5) and IGFBP-5 antibody were from GroPep Bioreagents (Thebarton SA 5031). Transforming growth factor-β1 (TGF-β1) was acquired from R&D Systems (Minneapolis, MN). Antibody for type I collagen was from BD Biosciences (Billerica, MA). Chemiluminescence reagents were purchased from PerkinElmer (Waltham, MA). Fluorescence-conjugated or biotinylated secondary antibodies and VECTASHIELD antifade mounting medium with DAPI were from Vector Laboratories (Burlingame, CA). Chloromethyl (CM)-H₂DCFDA, penicillin/streptomycin, and antimycotic were from Invitrogen (Carlsbad, CA). Lipofectamine 2000 was from ThermoFisher Scientific (Waltham, MA).

Mice.

Murine experiments were carried out with the authorization of the Institutional Animal Care and Use Committee of the University of Pittsburgh using approved protocols. Primary lung fibroblasts, a generous gift from Dr. Augustine Choi, were cultured from lung tissues of C57BL/6J wild-type (WT; Jackson Laboratory, Bar Harbor, ME), gp91^phox−/−, and p47^phox−/− null mice (Taconic Biosciences, Germantown, NY).

Primary fibroblast culture.

Human primary lung fibroblasts were cultured under a protocol approved by the University of Pittsburgh Institutional Review Board from the explanted lungs of normal organ donors whose lungs were not used for transplantation and from lung tissues of patients with SSc or IPF who underwent lung transplantation (50). For both human and mouse lungs, primary lung fibroblasts were cultured via an outgrowth method as previously described (50). Briefly, ~2-cm pieces of peripheral lung were minced and adhered to plastic. Fibroblasts were cultured from the minced lung pieces in DMEM supplemented with 10% fetal bovine serum, penicillin/streptomycin, and antimycotic until confluent or ready for experimentation. For experiments, 2.0 × 10⁵ primary fibroblasts were plated and cultured in 35-mm dishes. Fibroblasts were treated with 10 ng/ml TGF-β1, 500 ng/ml recombinant human IGFBP-5, or hydrochloric acid as a vehicle control in DMEM supplemented with 0.5% fetal bovine serum, penicillin/streptomycin, and antimycotic for the duration of the experiment. Fibroblasts were infected with cAd or Ad-BP5 at a multiplicity of infection of 50 for adenoviral vector-mediated IGFBP-5 experiments. For experiments involving inhibitors (DPI, NAC, MEK inhibitor, JNK inhibitor, p38 inhibitor, or PI3K inhibitor), fibroblasts were preincubated with inhibitor for 1 h, unless otherwise specified. Experiments utilized fibroblasts between passages 4 and 7.

Adenovirus construct preparation.

Adenovirus constructs were designed and prepared as previously reported (50). Briefly, full-length cDNAs of IGFBP-5 obtained by RT-PCR using total RNA extracted from primary fibroblasts were subcloned into the shuttle vector pAdlox and used for the preparation of replication-deficient adenovirus serotype 5-expressing IGFBP-5 (Ad-BP5) or empty vector containing no cDNA (cAd) by the Vector Core Facility at the University of Pittsburgh (Pittsburgh, PA). Fibroblasts infected with Ad-BP5 or cAd were treated and harvested 72 h postinfection unless otherwise indicated.

Sucrose-gradient lipid raft fraction.

Sucrose-gradient raft fractions were separated as described previously (22). Briefly, fibroblasts were lysed in ice-cold MBS buffer [150 mM NaCl and 25 mM 2-(N-morpholino)-ethanesulfonic acid pH 6.5] containing 1% Triton X-100 and protease inhibitors. Homogenates were mixed 1:1 with 80% sucrose in MBS in an ultracentrifuge tube and then overlaid with 35% then 5% discontinuous sucrose gradient layers in MBS buffer without detergent. Samples were centrifuged at 39,000 rpm for 18 h and fractionated into 12 subfractions.

Detection of ROS production.

Detection of ROS using CM-H₂DCFDA was performed as described previously (41). Briefly, 4.0 × 10⁴ primary lung fibroblasts were cultured on coverslips coated with type I collagen. After infection (with cAd or Ad-BP5) or administration of rBP5 (500–1,000 ng/ml), fibroblasts were preincubated with 10 mM CM-H₂DCFDA for 30 min and images were taken on an Olympus Fluoview 1000 confocal microscope (Olympus America, Melville, NY) using identical camera settings for each experiment. Images were taken at 0.5-μm intervals in z-axis from top to bottom of whole cells; fluorescence detected in the same pixel was added and reconstructed into three-dimensional images using MetaMorph software to calculate the ROS production in whole cells. Total fluorescent signal in whole cells was shown at ×1,200 magnification and reported as “relative fluorescence intensity” on the ordinate axis. Detection of ROS in live cells was further confirmed using two independent high-throughput commercially available ROS detection kits based on different fluorescence technologies (BioVision, Milpitas, CA and Abcam, Cambridge, MA) per manufacturer’s instructions and reported on the ordinate axis as relative fluorescence units (RFU).

Western blot analysis.

Cellular lysates were obtained from cultured primary fibroblasts as described previously (50). Equal amounts of protein were separated using 10–12% SDS-PAGE gels. Proteins were transferred to nitrocellulose membranes using wet methanol transfer. Membranes were blocked for 1 h in 5% nonfat dry milk, probed with primary antibody, washed in Tris-buffered saline with Tween-20, and then probed with horseradish peroxidase-conjugated secondary antibody. Signals were detected via chemiluminescence and densitometry analyzed with ImageJ software.

siRNA-mediated knockdown.

Fibroblasts grown to 60–70% confluence were subjected to Lipofectamine 2000-mediated knockdown using 100 nM nontargeting scrambled or targeted siRNA (Dharmacon, Lafayette, CO) per manufacturer’s instructions.

Real-time PCR.

RNA was isolated from fibroblast lysates using the TRIzol isolation method (Invitrogen, Carlsbad, CA) per manufacturer’s instructions and quantified using a NanoDrop Lite spectrophotometer (ThermoFisher, Waltham, MA). One microgram of RNA was used to prepare a 20-μl cDNA reaction with SuperScript IV First-Strand Synthesis components (ThermoFisher) and run on a C1000 Touch Thermal Cycler (Bio-Rad Laboratories, Hercules, CA). PCR reaction was performed with 1 μl cDNA in a 10-μl reaction utilizing TaqMan gene expression master mix and best-coverage primers [NADPH oxidase 1 (Nox1): Hs00246589_m1; Nox2: Hs00166163_m1; Nox3: Hs00210462_m1; Nox4: Hs00418356_m1; Noxo2: Hs00165362_m1; and B2M: Hs00187842_m1] on an Applied Biosystems StepOne Real-Time system (ThermoFisher). Relative gene expression was calculated using the ΔΔCt method.

Statistical analysis.

Data are represented as means (SD). Statistical comparisons were performed using the Mann-Whitney U-test as appropriate. Differences were considered statistically significant if P < 0.05 (noted in figures as *P < 0.05, **P < 0.01, and ***P < 0.001).

RESULTS

IGFBP-5 induces production of ROS.

A primary cell type identified as contributing significantly to fibrosis is the fibroblast (13); therefore, we chose to use early passage human lung fibroblasts to carry out these experiments. To examine the effect of IGFBP-5 on production of ROS, human lung fibroblasts stimulated by adenoviral vector-mediated or recombinant IGFBP-5 were analyzed for intracellular ROS production using CM-H₂DCFDA, a chemical indicator that is converted into a fluorescent adduct in the presence of ROS. Adenoviral infection of IGFBP-5 (Ad-BP5) or control (cAd) into early passage human lung fibroblasts was used to assess the effect of IGFBP-5 on ROS production. As shown in Fig. 1, A and B, untreated fibroblasts exhibited a steady-state level of ROS, indicative of ROS participation in normal cellular physiological processes; this level was unchanged following infection with cAd. Ad-BP5 significantly increased the amount of detectable ROS in fibroblasts compared with cAd (3.4 SD 0.8 vs. 1.0 SD 0.2, P < 0.002). In comparison, the known ROS-inducing cytokine TGF-β used as a positive control for ROS production (14, 73) significantly increased production of ROS in primary human lung fibroblasts, similar to Ad-BP5.

Fig. 1.

Adenoviral vector-mediated insulin-like growth factor binding protein-5 (IGFBP-5) triggers the production of reactive oxygen species (ROS). ROS in primary human lung fibroblasts infected with adenovirus expressing IGFBP-5 (Ad-BP5) or control (cAd) was detected by chloromethyl (CM)-H₂DCFDA (A) and fluorescent signal represented with data relative to control (cAd) (B) with n ≥ 6. Fibroblasts stimulated with transforming growth factor-β1 (TGF-β1; 10 ng/ml) or vehicle (Veh) for 24 h are shown as positive and negative controls, respectively. The total fluorescent signal in whole cells is shown at ×1,200 magnification with a bar length of 10 μm. ROS was also detected in human lung fibroblasts treated with Ad-BP5 or cAd for 24, 48, or 72 h (C) with histogram of fluorescent signal relative to 24 h cAd (D); results represent n > 10. A representative immunoblot of IGFBP-5 and GAPDH in cellular lysates of lung fibroblasts treated with Ad-BP5 or cAd for 24, 48, or 72 h is shown (E). **P < 0.01, and ***P < 0.001 by Mann-Whitney U-test.

Furthermore, ROS production in lung fibroblasts treated with Ad-BP5 gradually increased over time (Fig. 1, C and D). Ad-BP5-stimulated ROS production was 2.1-fold at 24 h, 3.6-fold at 48 h, and 6.2-fold at 72 h compared with cAd. Increased adenovirally mediated production of IGFBP-5 was confirmed using immunoblotting of fibroblast lysates (Fig. 1E). Thus adenoviral vector-mediated overexpression of IGFBP-5 was sufficient to cause a significant increase in ROS production in primary human lung fibroblasts.

The effect of recombinant IGFBP-5 (rBP5) administration in primary human lung fibroblasts was also assessed. Concentrations of rBP5 corresponding to physiological levels detected in human donor serum (6, 10, 36) were used. As shown in Fig. 2, A and B, rBP5 significantly induced ROS production in a dose-dependent manner (1.7 SD 0.4 at 500 ng/ml and 2.5 SD 0.5 at 1,000 ng/ml) compared with vehicle-treated lung fibroblasts. TGF-β-induced ROS production was comparable to that of rBP5 at 500 ng/ml.

Fig. 2.

Recombinant insulin-like growth factor binding protein-5 (IGFBP-5) induces the production of reactive oxygen species (ROS). ROS in primary human lung fibroblasts treated with recombinant human IGFBP-5 (rBP5; 500 and 1,000 ng/ml) was detected by chloromethyl (CM)-H₂DCFDA (A), and fluorescent signal is represented with data relative to vehicle (B) at 72 h with n ≥ 8. Fibroblasts stimulated with transforming growth factor-β (TGF-β; 10 ng/ml) or vehicle (Veh) were shown as positive and negative controls, respectively. The total fluorescent signal in whole cells is shown at ×1,200 magnification with a bar length of 10 μm. ROS was also detected in primary human lung fibroblasts treated with rBP5 (500 ng/ml) for 2, 6, and 24 h (C), and fluorescent signal is represented graphically with data relative to 0 h (D) with n = 12. Lung fibroblasts stimulated with vehicle, rBP5 (500 and 1,000 ng/ml), and TGF-β (10 ng/ml) were analyzed for ROS production using a high-throughput assay at 17, 24, 48, and 72 h (E) with n = 4 donors, each run with 4–12 technical replicates. RFU, relative fluorescence units. Lung fibroblasts stimulated with vehicle, rBP5 (1,000 ng/ml), and TGF-β (10 ng/ml) for 24 h were analyzed for ROS/reactive nitrogen species production with green oxidative stress and orange superoxide detection reagents (F) with n = 4 donors, each run with 8 technical replicates. *P < 0.05, **P < 0.01, and ***P < 0.001 by Mann-Whitney U-test.

Moreover, as shown in Fig. 2, C and D, recombinant administration of IGFBP-5 also induced ROS production in a time-dependent manner. Although unchanged after only 2 h, ROS production by rBP5 was significantly increased at 6 h and further increased after 24 h of stimulation compared with vehicle-treated fibroblasts.

To verify ROS induction by IGFBP-5, intracellular ROS production was additionally measured over time (17–72 h) in lung fibroblasts using a commercially available assay with a more sensitive high-throughput readout capable of detecting hydroxyl, peroxyl, and other reactive oxygen species (Fig. 2E). IGFBP-5 (1,000 ng/ml) and TGF-β significantly increased ROS production at all time points examined, while rBP5 (500 ng/ml) caused a significant production of ROS from 24 h–72 h compared with vehicle. ROS was additionally detected using two specific fluorescent probes: green oxidative stress detection reagent and orange superoxide detection reagent. The green probe reacts in the presence of various reactive ROS and reactive nitrogen species, with efficient detection of hydrogen peroxide, hydroxyl radicals, peroxynitrite, and peroxyl radicals. The orange probe specifically detects the presence of superoxide. After a 24-h stimulation, both rBP5 and TGF-β significantly increased both green and orange fluorescence compared with vehicle, indicating significant production of various ROS species, including superoxide (Fig. 2F). Thus administration of IGFBP-5, whether in an adenoviral vector-mediated or recombinant manner, triggered production of intracellular ROS in primary human lung fibroblasts in a dose- and time-dependent manner in three different assays.

IGFBP-5-induced ROS production involves activation of JNK and MEK/ERK signaling.

Previous reports have shown IGFBP-5 activation of mitogen-activated protein kinases, like ERK and p38, occurs in osteoblasts and mouse smooth muscle cells (1, 28). We have shown that IGFBP-5 activates mitogen-activated protein kinase and downstream targets Egr1 and DOK5 in primary human fibroblasts (67, 69). To examine the specific protein kinase pathway involved in the regulation of ROS production induced by IGFBP-5, primary human lung fibroblasts were infected with Ad-BP5 or cAd in the presence or absence of pathway inhibitors for JNK, MEK, p38, or PI3K. As shown in Fig. 3, inhibitors of JNK, MEK, p38, and PI3K had no significant effect on cAd-infected ROS production. Ad-BP5 caused a significant 5.6-fold increase in ROS production compared with cAd, which was significantly reduced to 1.7-fold and 1.2-fold by use of JNK or MEK inhibitors, respectively, completely negating the effect of adenoviral vector-mediated IGFBP-5-stimulated ROS production. The p38 and PI3K inhibitors did not have an effect on IGFBP-5-stimulated ROS production as they did not cause a significant reduction in ROS compared with Ad-BP5, and thus p38 and PI3K appear not to play a critical role in mediating ROS production by IGFBP-5.

Fig. 3.

Insulin-like growth factor binding protein-5 (IGFBP-5)-induced reactive oxygen species (ROS) production is JNK and MEK/ERK dependent. ROS in primary human lung fibroblasts infected with adenovirus expressing IGFBP-5 (Ad-BP5) or control (cAd) for 72 h concurrently treated with inhibitors against JNK (JNKi, JNK inhibitor II; 10 μM), MEK (MEKi, U0126; 10 μM), p38 (p38i, SB203580, 10 μM), or phosphoinositide 3-kinase (PI3K) (PI3Ki, LY294002; 10 μM) was detected by chloromethyl (CM)-H₂DCFDA (A), and fluorescent signal is graphed with data relative to vehicle (Veh) cAd (B), representing results from 12 experiments. The total fluorescent signal in whole cells is shown at ×1,200 magnification with a bar length of 10 μm. ***P < 0.001 by Mann-Whitney U-test.

To examine the role of JNK and MEK/ERK signaling further, upstream and downstream members of each pathway were analyzed for activation by IGFBP-5. As shown in Fig. 4A, recombinant IGFBP-5 (500 ng/ml) activated MKK4 in a transient manner from 5–30 min, after which time the activated signal was lost and returned to baseline level. Similarly, JNK activation was sustained for 5–60 min, while total JNK levels did not change over time. ATF2 was activated at 5 min and remained activated through 60 min in the presence of rBP5. Lastly, c-Jun activation initially decreased in the presence of rBP5 at 5 min and then increased 10–30 min before trending back toward the baseline phosphorylation level at 60 min. The effect of recombinant rBP5 on activation of members of the MEK/ERK signaling pathway is shown in Fig. 4B. With rBP5, Raf was activated at 5–10 min before signal activation began to wane. Both MEK1/2 and ERK1/2 followed the same pattern of activation in the presence of rBP5, showing activation from 5–15 min, while total MEK1/2 and total ERK1/2 levels did not change. Thus the JNK and MEK/ERK pathways were both activated by the addition of recombinant IGFBP-5.

Fig. 4.

Insulin-like growth factor binding protein-5 (IGFBP-5) activates JNK and MEK/ERK signaling pathways. Representative immunoblots of JNK pathway activation (A), MEK/ERK pathway activation (B), and JNK-MEK/ERK pathway cross talk (C and D) in primary human lung fibroblasts treated with recombinant human IGFBP-5 (rBP5; 500 ng/ml) for 0–60 min. In B, data shown for phospho-Mek and total Mek are from the exact same samples that were run on separate gels. ATF2, activating transcription factor 2. Fibroblasts were pretreated for 1 h with MEKi (U0126; 10 μM) and then stimulated with rBP5 (500 ng/ml) for 10 min and probed for JNK pathway activation (C) or pretreated 1 h with JNKi (JNK inhibitor II; 10 μM) and then stimulated with rBP5 (500 ng/ml) for 10 min and probed for ERK pathway activation (D). Samples in D are from nonadjacent lanes of the same gel and were reordered for clarity. Veh, vehicle. Lipid raft fractions from IGFBP-5-stimulated fibroblasts were probed for IGFBP-5, MEK activation, and caveolin-1 colocalization (E).

Cross talk between signaling pathways has been widely reported (27, 38) and was scrutinized for its role in IGFBP-5-mediated cellular effects. IGFBP-5-stimulated fibroblasts pretreated with a MEK inhibitor were probed for activation of JNK (Fig. 4C), while cells pretreated with a JNK inhibitor were probed for activation of ERK (Fig. 4D). Blockade of the MEK pathway not only prevented basal activation of JNK but also blocked IGFBP-5-stimulated activation of JNK without affecting overall levels of total JNK (Fig. 4C). To ensure pathway inhibition by MEKi, ERK activation was probed. Similar to JNK, inhibition of MEK not only blocked IGFBP-5-induced ERK activation but also obliterated basal levels of active ERK phosphorylation. Alternatively, inhibition of JNK led to decreased basal activation of ERK1/2 (Veh-DMSO vs. Veh-JNK inhibitor) and partially reduced the IGFBP-5-stimulated phosphorylation of ERK1/2, while not affecting total ERK1/2 levels (Fig. 4D). JNKi not only blocked phosphorylation of ATF2 under basal conditions but also prevented its phosphorylation in the presence of IGFBP-5. Taken together, these results suggest that MEK/ERK signaling is indispensable for activation of the JNK pathway and JNK signaling has a synergistic effect for MEK/ERK activation.

Lipid rafts have been reported to be important in various signal transduction pathways (53). Empty vector (cAd) or adenoviral vector-mediated IGFBP-5-expressing fibroblasts were subjected to lipid raft fractionation and examined for localization with MEK signal transduction components (Fig. 4E). Caveolin-1 was associated with lipid raft subfraction #5 in cAd-infected fibroblasts. IGFBP-5, activated MEK, and caveolin-1 all localized to lipid rafts in the same subfraction in Ad-BP5-treated fibroblasts, suggesting that lipid rafts might be an important platform for IGFBP-5-induced signal transduction via MEK/ERK in primary human lung fibroblasts.

IGFBP-5-induced ROS and ECM production are blocked by antioxidants, and IGFBP-5 is required for ROS production in IPF and SSc lung fibroblasts.

We previously reported that IGFBP-5 induces ECM production in vitro and in vivo (50, 68, 69, 71). To determine if IGFBP-5-induced ROS production is critical to the induction of ECM deposition, two antioxidants were utilized to block ROS production in primary human lung fibroblasts: a general antioxidant, NAC, which is a precursor of the natural antioxidant glutathione, and a more specific antioxidant, DPI, a flavoenzyme and NADPH oxidase inhibitor (Fig. 5). Ad-BP5-stimulated production of ROS was significantly reduced by DPI at 1 μM and completely ablated by the addition of 10 mM NAC and 2 μM DPI (Fig. 5, A and B). Adenoviral vector-mediated expression of IGFBP-5 significantly stimulated fibronectin and type 1 collagen production as we previously reported (Fig. 5, C and D) (50). Addition of either NAC (10 mM) or DPI (1 µM) significantly blocked IGFBP-5-stimulated production of both ECM proteins (Fig. 5, C and D) in normal lung (NL) fibroblasts.

Fig. 5.

Insulin-like growth factor binding protein-5 (IGFBP-5)-induced reactive oxygen species (ROS) and extracellular matrix (ECM) production are blocked by antioxidants and ROS production by idiopathic pulmonary fibrosis (IPF) and systemic sclerosis (SSc) pulmonary fibroblasts is IGFBP-5 dependent. ROS in primary human lung fibroblasts infected with adenovirus expressing IGFBP-5 (Ad-BP5) or control (cAd) treated for 1 h with antioxidants N-acetylcysteine (NAC; 10 mM) or diphenyleneiodonium (DPI; 1–2 μM) were detected by chloromethyl (CM)-H₂DCFDA (A), and fluorescent signal is graphed with data relative to vehicle (Veh) cAd (B) with n ≥ 6. The total fluorescent signal in whole cells is shown at ×1,200 magnification with a bar length of 10 μm. Representative immunoblots of the ECM components fibronectin and type 1 collagen in Ad-BP5 or cAd infected primary human lung fibroblasts treated with 10 mM NAC or 1 μM DPI (C) and histogram of densitometry normalized to GAPDH and relative to Veh cAd (D) are shown. Representative immunoblots of IPF fibroblasts treated for 1 h with 10 mM NAC or 1 μM DPI (E) and histogram of densitometry normalized to GAPDH and relative to respective vehicle (F). Unt, untreated. Representative immunoblots of SSc fibroblasts treated for 1 h with 10 mM NAC or 1 μM DPI (G) and histogram of densitometry normalized to GAPDH and relative to respective vehicle (H) with n = 3. Steady-state ROS production in normal lung (NL), IPF, and SSc fibroblasts (I) and following siRNA-mediated knockdown of IGFBP-5 in IPF and SSc (J) with n = 3. RFU, relative fluorescence units. *P < 0.05, **P < 0.01, and ***P < 0.001 by Mann-Whitney U-test.

Elevated levels of IGFBP-5 have been reported in two fibrotic diseases: IPF and SSc (12). Lung fibroblasts derived from IPF and SSc patients were treated with NAC and DPI and then examined for fibronectin and collagen protein production (Fig. 5, E–H). Fibronectin levels were significantly reduced with 1 h treatment of 10 mM NAC and 1 µM DPI in IPF fibroblasts, whereas collagen was significantly reduced following treatment with DPI (Fig. 5, E and F). Both NAC and DPI significantly reduced both fibronectin and collagen production in SSc lung fibroblasts (Fig. 5, G and H). Steady-state ROS production was analyzed in NL, IPF, and SSc fibroblasts (Fig. 5I). The unstimulated level of ROS production was significantly elevated in both IPF and SSc compared with NL fibroblasts plated at the same density and cultured for the same duration. Subsequently, knockdown of IGFBP-5 significantly reduced ROS production in both IPF and SSc fibroblasts (Fig. 5J), suggesting IGFBP-5 mediates ROS production not only in normal fibroblasts but also in IPF and SSc fibroblasts.

IGFBP-5-induced ROS and ECM production are NADPH oxidase dependent.

NADPH oxidase represents a major source of endogenous oxygen free radicals and is a primary source of ROS generation in the vascular system (9, 39). Previous reports have shown that ROS production in SSc dermal fibroblasts occurs via NADPH oxidase (52) and that IGFBP-5 is upregulated in primary lung fibroblasts from SSc patients (50). Therefore, to study the role of NADPH oxidase in IGFBP-5-induced ROS and ECM production, we utilized both in vivo and in vitro approaches. Murine primary lung fibroblasts established from the lung tissues of mice lacking one of two critical NADPH oxidase complex components (gp91^phox−/− or p47^phox−/−) were analyzed for IGFBP-5-stimulated ROS production and compared with WT littermates (Fig. 6, A and B). In cAd-infected cells, there was no change in steady-state ROS production in gp91^phox−/− or p47^phox−/− fibroblasts compared with those from WT mice, suggesting that the adenoviral infection process does not impact ROS levels. Adenoviral vector-mediated IGFBP-5 induced a significant 3.4-fold increase in ROS production in fibroblasts from WT mice compared with cAd-infected fibroblasts. However, compared with Ad-BP5-stimulated WT fibroblasts, ROS production was significantly decreased and returned to baseline levels in Ad-BP5-stimulated fibroblasts from gp91^phox−/− and p47^phox−/− mice, suggesting that NADPH oxidase is a critical mediator of IGFBP-5-induced production of ROS.

Fig. 6.

Insulin-like growth factor binding protein-5 (IGFBP-5)-induced reactive oxygen species (ROS) and extracellular matrix (ECM) production are NADPH oxidase-dependent. ROS in murine primary fibroblasts from lung tissues of gp91^phox−/− mice, p47^phox−/− mice, or wild-type (WT) littermates infected with adenovirus expressing IGFBP-5 (Ad-BP5) or control (cAd) was detected by chloromethyl (CM)-H₂DCFDA (A), and fluorescent signal is graphed with data relative to WT cAd (B), representative of 10 experiments. The total fluorescent signal in whole cells is shown at ×1,200 magnification with a bar length of 10 μm. Representative immunoblot of fibronectin in primary mouse lung fibroblasts from gp91^phox−/− mice, p47^phox−/− mice, or WT littermates infected with Ad-BP5 or cAd (C) and histogram of densitometry normalized to GAPDH and relative to WT cAd (D) are shown, representative of 4 experiments. Selective knockdown of NADPH oxidase 3 (Nox3) reduced IGFBP-5- and TGF-β-stimulated ROS production in NL fibroblasts (E) from 4 different donors and 8 technical replicates. Transcriptional expression of Nox3 in recombinant human IGFBP-5 (rBP5) stimulated NL fibroblasts (F). RFU, relative fluorescence units; Unt, untreated. Relative expression was compared with vehicle (Veh)-treated fibroblasts at each time point unless otherwise specified; n > 3; *P < 0.05, **P < 0.01, and ***P < 0.001 by Mann-Whitney U-test.

Furthermore, NADPH oxidase was analyzed for its role in IGFBP-5-stimulated ECM production using fibroblasts from gp91^phox−/− mice, p47^phox−/− mice, and WT littermates. As shown in Fig. 6, C and D, the IGFBP-5-induced upregulation of fibronectin observed in Ad-BP5-infected WT fibroblasts was absent in Ad-BP5-stimulated fibroblasts from gp91^phox−/− and p47^phox−/− mice.

To confirm these findings in primary human cells, NL fibroblasts were subjected to knockdown of select components of the NADPH oxidase machinery (including Nox1, Nox2, Nox3, Nox4, and Noxo2) and then stimulated with rBP5 for 48 h. IGFBP-5- and TGF-β-stimulated ROS production in NL fibroblasts treated with Nox3 siRNA was significantly decreased compared with scrambled siRNA-treated control (Fig. 6E). Knockdown of Nox1, Nox2, Nox4, and Noxo2 did not lead to significant reduction of IGFBP-5-induced ROS production (data not shown), although Nox4 and Noxo2 silencing reduced TGF-β-mediated ROS production as reported (2, 17, 65). We additionally sought to determine whether IGFBP-5 regulated Nox enzyme expression at the transcriptional level (Fig. 6F, data not shown). IGFBP-5 significantly increased Nox3 transcription at 48 h (Fig. 6F). Earlier changes were noted in Nox1 (significantly increased at 24 h) and Noxo2 (decreased at 1–3 h; data not shown) expression levels.

DISCUSSION

To date, this is the first report linking IGFBP-5 to ROS and ECM production. Our findings show that IGFBP-5 mediates increased ECM deposition through generation of ROS in primary human lung fibroblasts, predominantly through NADPH oxidase, with the signal transduced by two pathways, JNK and MEK/ERK, that may act independently or in a coordinated manner via cross talk between the two pathways. Transduction of the IGFBP-5 signal has been studied in other systems; the profibrotic effect of IGFBP-5 was found to be promulgated through ERK signaling in cardiac fibroblasts (55), and similar to our findings, Wang et al. (64) reported that IGFBP-5 functions through the JNK and MEK/ERK pathways to mediate the differentiation of two populations of mesenchymal stem cells. We have also shown that the Raf/MEK/ERK pathway mediates IGFBP-5-induced migration of peripheral blood mononuclear cells and monocytes (70). Cross talk between the JNK and MEK/ERK pathways has been reported previously in a wide variety of cell types, including dual activation or feedback inhibition of either pathway, as well as independent parallel activation leading to the phosphorylation of c-Jun from both cascades (34, 35, 72). However, in human primary lung fibroblasts, the IGFBP-5-stimulated cross talk between the JNK and MEK/ERK pathways appears to be upstream of c-Jun activation, as inhibition with a MEK inhibitor affected phosphorylation of JNK.

In our findings, IGFBP-5 was associated in the same lipid fraction with caveolin-1 and activated MEK. Lipid rafts are composed of cholesterol and saturated lipids with the ability to move laterally within the larger plasma membrane of unsaturated phospholipids, affecting the overall fluidity of the cellular membrane (48, 53). Lipid rafts are distributed across the surface of a fibroblast with no apparent polarity and have been reported to be important in signal transduction of both the JNK and MEK/ERK pathways (23, 46). Caveolae are specialized lipid rafts composed of polymerized caveolin-1. Caveolae-mediated endocytosis represents an important mechanism for component uptake into a cell (5, 29); as such, caveolae have been reported as indispensable in the cellular uptake of IGFBP-5 (66). Additionally, H-Ras, neuronal nitric oxide synthase, endothelial nitric oxide sythase, and MEK have been shown to be associated with lipid rafts (53), suggesting the involvement of H-Ras upstream of Raf, which was activated within 5 min of IGFBP-5 addition to fibroblasts.

Seven identified members of the NADPH oxidase family exist within the cell, localized to the plasma membrane, mitochondrial membrane, endoplasmic reticulum, and nuclear membrane (54). NADPH oxidase enzymes are composed of both membrane and cytosolic components that function together to mediate production of several ROS species, including hydroxyl radicals, peroxynitrite, peroxide, superoxide, nitric oxide, hydroxyl, and other radicals (9), whose generation can lead to beneficial or pathophysiological effects (47). In a murine bleomycin-induced fibrosis model, lung fibrosis was attenuated in p47^phox null mice compared with WT littermates (33). Our studies identified various components of the NADPH oxidase machinery that when reduced (via siRNA) or knocked out mitigated IGFBP-5-stimulated ROS production, suggesting the primary source of ROS contributing to IGFBP-5-mediated ECM production and fibrosis was indeed through NADPH oxidase components, although the specific species of ROS involved in this mechanism is yet to be determined. In our study, ROS production by IGFBP-5 appears to be mainly regulated by Nox3 in primary human lung fibroblasts but may be mediated via multiple NADPH oxidase subunit components in mouse fibroblasts. This may be due to the fact that use of siRNA does not achieve 100% silencing efficiency in primary fibroblasts, whereas use of fibroblasts from mice null for the gene of interest achieves complete loss of expression. Furthermore, species-specific differences might explain why IGFBP-5 mediates ROS production via different NADPH oxidase enzymes in mouse compared with human cells. An additional consideration is that IGFBP-5 may regulate mitochondrial ROS, which is the focus of future studies. Steady-state levels of ROS appeared to be unaffected in cAd-infected fibroblasts of gp91^phox and p47^phox mice compared with WT and may be due to ROS contribution of other enzymes (mitochondrial enzymes; lipo-, mono-, and cyclo-oxygenases; xanthine oxidases; myeloperoxidases; CYP450s; endothelial nitric oxide synthase; etc.) or nonenzymatic mechanisms. Although we detected a transcriptional induction of Nox3 and Nox1 by IGFBP-5, regulation of global ROS can alternately be affected by many intracellular processes, including changes in antioxidant status, which will require further study. In addition to our findings implicating Nox3 in IGFBP-5-stimulated ROS production, recent studies have also implicated Nox4 as a significant contributor in maintaining a profibrotic phenotype in IPF fibroblasts and shown that blockage of Nox4 promoted resolution of fibrosis in mice (17). Further study into the redox flux contributing to IGFBP-5-mediated induction of ROS and ECM in fibrosis could delineate potential therapeutic strategies for fibrosing diseases.

NADPH oxidase activity has been found to be dependent on and controlled by lipid rafts/caveolae (49). Physiological ROS is beneficial to cells, yet excess ROS production leads to cellular oxidative stress, contributing to fibrosis, cancer, aging, and other detrimental conditions (43, 61). Cells contain natural antioxidant mechanisms to maintain redox balance, including mitochondrial manganese superoxide dismutase, peroxisomal and cytosolic catalase, and mitochondrial and cytosolic glutathione peroxidase (31, 43). Addition of two different cellular antioxidant inhibitors allowed restoration of cellular redox and blocked ROS-mediated ECM production by IGFBP-5 stimulation. The mechanism of ECM production through upregulation of ROS remains to be completely elucidated. ROS has been reported to mediate matrix metalloproteinase transcript and enzymatic activity, as well as enhance TGF-β signaling, contributing to the selective activation of myofibroblasts and development of a profibrotic milieu, with ROS and TGF-β reciprocally regulating each other (20, 31, 43). Our findings herein correlate well with previous studies utilizing fibroblasts from the skin of SSc patients, which showed that dermal SSc fibroblasts release more NADPH oxidase-sourced ROS than fibroblasts from healthy subjects and that treatment of these cells with NAC blocked ROS production and downregulated collagen transcript levels (52). The putative profibrotic mediator TGF-β has the ability to stimulate ROS production and, in a positive feedback loop, can then actuate more TGF-β through a twofold mechanism involving pro-oxidant stimulation of ROS and downregulation of cellular antioxidant defenses, suggesting the possibility that IGFBP-5-stimulated ROS could potentially trigger downstream activation of TGF-β (15, 31), in addition to increasing its own production (44) in a parallel positive feedback loop. Interestingly, incidental findings in normal fibroblasts have shown that stimulation with TGF-β is not sufficient to promote increased secretion of IGFBP-5 (50). Therefore, the effect of IGFBP-5 on cellular antioxidant defenses and transduction of the fibrotic signal triggering the deposition of ECM and contributing to lung fibrosis remains to be completely elucidated.

GRANTS

This work was supported in part by NIH Grants R01-AR-050840 (to C. A. Feghali-Bostwick), K24-AR-060297 (to C. A. Feghali-Bostwick), 5T32-AR-050958 (to S. M. Garrett), TL1-TR-001451 (to X. Nguyen), and UL1-TR-001450 (X. Nguyen), American Lung Association Dalsemer Research Scholar Award (to C. A. Feghali-Bostwick), American Heart Association Pennsylvania/Delaware Affiliate (to C. A. Feghali-Bostwick and H. Yasuoka), Uehara Memorial Foundation (to H. Yasuoka), and SmartState and Kitty Trask Holt Endowment (to C. A. Feghali-Bostwick).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.Y. and C.A.F.-B. conceived and designed research; H.Y., S.M.G., X.-X.N., and C.M.A. performed experiments; H.Y., S.M.G., C.M.A., and C.A.F.-B. analyzed data; H.Y., S.M.G., C.M.A., and C.A.F.-B. interpreted results of experiments; H.Y., S.M.G., and C.A.F.-B. prepared figures; H.Y., S.M.G., and C.A.F.-B. drafted manuscript; H.Y., S.M.G., X.-X.N., C.M.A., and C.A.F.-B. edited and revised manuscript; H.Y., S.M.G., X.-X.N., C.M.A., and C.A.F.-B. approved final version of manuscript.