Fetal Fibronectin Signaling Induces Matrix Metalloproteases and Cyclooxygenase-2 (COX-2) in Amnion Cells and Preterm Birth in Mice

Background: The function of fetal fibronectin (fFN) in the pathogenesis of preterm labor is not known.

Results: fFN activates MMP-1, MMP-9, and COX-2 in mesenchymal cells and causes preterm labor in mice.

Conclusion: fFN is biologically active and plays a significant role in the pathogenesis of preterm labor.

Significance: Signaling of fFN in fetal membranes is important in the pathophysiology of premature preterm rupture of the membranes.

Abstract

Fetal fibronectin (fFN) in cervical and vaginal secretions has been used as a predictor of preterm delivery. Here, we clarified the pathological function of fFN on cell type-specific matrix metalloproteinases (MMPs) and prostaglandin synthesis in fetal membranes. Treatment of amnion mesenchymal cells with fFN resulted in dramatic increases in MMP-1 and MMP-9 mRNA and enzymatic activity as well as COX-2 mRNA and prostaglandin E₂ synthesis, activating both NFκB and ERK1/2 signaling. Fetal FN-induced increases in MMPs and COX-2 were mediated through its extra domain A and Toll-like receptor 4 expressed in mesenchymal cells. Lipopolysaccharide and TNF-α increased the release of free FN in medium of amnion epithelial cells in culture. Finally, injection of fFN in pregnant mice resulted in preterm birth. Collectively, these results indicate that fFN is not only a marker of preterm delivery but also plays a significant role in the pathogenesis of preterm labor and premature rupture of fetal membranes.

Introduction

Preterm labor is the leading cause of perinatal morbidity and mortality (1). Preterm premature rupture of membrane (PPROM)² occurs in ∼1–3% of all pregnancies and is associated with 30–40% of preterm deliveries (2). PPROM represents abnormal disintegration of fetal membrane integrity for which there is no effective prevention or treatment.

Fetal membranes comprise several distinct layers of cells and extracellular matrix. In humans, by 18–20 weeks of gestation, the amnion adheres to the chorion, which melds with the underlying endometrium/decidua. By 36–38 weeks, membrane structure is disrupted in preparation for labor (3). The amnion in particular is targeted as it is the primary load-bearing structure of the fetal membranes (1). Mesenchymal cells that underlie a single layer of avascular epithelial cells are the primary source of collagen and matrix support of the amnion. Interstitial collagens (types I, III, and V) maintain the mechanical integrity of the amnion, whereas degradation of collagen is mediated primarily by matrix metalloproteinases (MMPs) (4–6). MMP-1 is increased in amniotic fluid obtained from women with PPROM (7), and increased amnion MMP-9 is observed in normal labor (8) as well as PPROM (9, 10). Cleavage of the triple helix of fibrillar collagen is primarily mediated by the interstitial collagenase MMP-1. Collagen fragments are then further degraded by the gelatinases MMP-2 and MMP-9.

Prostaglandins (PGs) together with collagenases also play a pivotal role in fetal membrane rupture (11). In human pregnancy, the principle source of PG is the amnion (12). Amniotic fluid concentrations of PG are increased in preterm labor associated with intraamniotic infection (13).

Fibronectin (FN) is a multidomain protein that binds to cell surface receptors, collagen, proteoglycans, and other FN molecules. Many of these domains and interactions are also involved in the assembly of FN dimers into a multimeric fibrillar matrix. Fetal fibronectin (fFN) is diffusely distributed in fetal membrane from amnion to decidua, providing structural support and adhesion of the fetal membranes to the uterine lining (14). fFN in cervicovaginal secretion has been used as a clinical marker of preterm labor (14), although it has limited positive predictive value (15). Many have hypothesized that fFN in cervicovaginal fluid signifies disruption of the fetal-maternal interface and release of fetal matrix molecules into the vagina. Others have suggested that proteolytic degradation of fFN leads to domain-specific activation of MMP-9 in immune cells that infiltrate the membranes at weak zones (16). The pathological function of fFN in preterm labor, however, is largely unknown.

In this study, we present evidence that fFN is more than a matrix support molecule. Rather, fFN has biological activity that increased MMP-1 and MMP-9 mRNA and enzymatic activity as well as prostaglandin endoperoxide synthase 2 (i.e. cyclooxygenase-2 (COX-2)) mRNA and PGE₂ biosynthesis in human amnion mesenchymal cells. Furthermore, fFN activated NFκB and ERK1/2 signaling pathways in these cells. Our data indicate that (i) fFN activates Toll-like receptor-4 (TLR4) in amnion mesenchymal cells, (ii) the extra domain A (EDA) of fFN is crucial for fFN-induced activation of MMPs and COX-2, and (iii) TNF-α and lipopolysaccharide (LPS) increase soluble fFN secretion in amnion epithelial cells. Collectively, the results suggest a pivotal role for fFN in the pathogenesis of preterm labor and PPROM.

EXPERIMENTAL PROCEDURES

Reagents

Sepharose 4B (17-0120-01) and gelatin-Sepharose 4B (17-0956-01) were purchased from GE Healthcare. Phenylmethanesulfonyl fluoride (PMSF; P7626) and lipopolysaccharides (rough strains) from Escherichia coli J5 (L-5014) were purchased from Sigma. DMEM/F-12 (Ham's, 11320), antibiotic-antimycotic solution (15240), and 10% Zymogram gelatin gel (EC61755) were purchased from Invitrogen. Rabbit anti-human fibronectin polyclonal antibody (AB1945) was purchased from Millipore (Billerica, MA). Goat anti-rabbit IgG (heavy + light)-HRP conjugate (170-6515) and goat anti-mouse IgG (heavy + light)-HRP conjugate (172-1011) were purchased from Bio-Rad. Recombinant human TNF-α (210-TA), polyclonal goat IgG (AB108-C), and anti-human TLR4 antibody (AF1478) were purchased from R&D Systems (Minneapolis, MN). The BCA (bicinchoninic acid) assay (23225) was purchased from Thermo Scientific (Waltham, MA). Mouse monoclonal antibody (IST-9) to fibronectin (AB6328) and anti-TATA-binding protein antibody (1TBP18, ab818) were purchased from Abcam (Cambridge, MA). PhosphoPlus MAPK antibody kits (9100) were purchased from New England Biolabs (Ipswich, MA). Phospho-NFκB p65 (Ser-536) (7F1) mouse mAb (3036) and GAPDH (14C10) rabbit mAb (2118) were purchased from Cell Signaling Technology (Beverly, MA). NFκB p65 (C-20, sc-372) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Protease inhibitor mixture tablets (Complete Mini, 04 693 124 001) and phosphatase inhibitor mixture tablets (PhosSTOP, 04 906 845 001) were purchased from Roche Applied Science.

Preparation of fFN and Plasma Fibronectin (pFN)

Fetal membranes were obtained at the time of elective repeat cesarean sections at term, and plasma was obtained from volunteers under protocols approved by the Institutional Review Board at the University of Texas Southwestern Medical Center. pFN was purified from plasma and fFN was purified from human amnion by gelatin affinity chromatography according to the methods of Retta et al. (17) with modification. Human amnion was washed extensively with PBS to completely remove blood, minced, and homogenized with TBS (25 mm Tris-HCl, 150 mm NaCl, 2 mm KCl, pH 7.4) including 1 mm PMSF and 2 m urea. Homogenates were stirred overnight at 4 °C. Samples were then centrifuged at 25,000 × g for 20 min at 4 °C, and the supernatant was applied to Sepharose 4B and passed through at a 2 ml/min flow rate at room temperature. The flow-through material was diluted 20-fold and applied to gelatin-Sepharose (2 ml/min flow rate at room temperature). For pFN purification, 50 ml of whole blood with 0.1% EDTA was centrifuged for 2000 × g for 30 min at 4 °C. The supernatant (plasma) was brought to 1 mm PMSF and centrifuged again at 10,000 × g for 15 min at 4 °C. The obtained supernatant was applied to a Sepharose 4B column at a 2 ml/min flow rate at room temperature. The flow-through material was applied to gelatin-Sepharose (2 ml/min flow rate at room temperature). Gelatin-Sepharose columns were first washed with 2 volumes of 10 mm Tris-HCl, pH 7.4 containing 0.5 m NaCl and then with 3 volumes of TBS, pH 7.4. Bound FN was eluted with 8 m urea in TBS. Fractions were pooled and dialyzed against TBS, pH 7.4 at 4 °C. After filter sterilization, the final concentration of FN was measured by BCA assay, and FN was concentrated using a speed vacuum concentrator. Lyophilized FN was reconstituted in sterile TBS, and aliquots were stored at −80 °C.

Isolation and Culture of Amnion Epithelial and Mesenchymal Cells

Separation and isolation of amnion epithelial and mesenchymal cells were performed as described previously (18). Briefly, amnion tissue was separated by blunt dissection. The amnion tissue was minced, and cells were dispersed by enzymatic digestion. Isolated amnion cells were suspended in DMEM/F-12 that contained fetal bovine serum (10%, v/v) and antibiotic-antimycotic solution (1%, v/v). Cells were plated in plastic culture dishes, maintained at 37 °C in a humidified atmosphere of 5% CO₂ in air, and allowed to replicate in a monolayer to confluence.

Quantitative Real Time PCR

Quantitative RT-PCR was used to determine relative levels of gene expression as described (19). Primer sequences for amplification are shown in Table 1. Gene expression was normalized to that of GAPDH.

TABLE 1.

Sequences of primers used for quantitative RT-PCR

FAM, carboxyfluorescein; TAMRA, tetramethylrhodamine.

Gene	Primers
GAPDH
Forward	5′-GGA GTC AAC GGA TTT GGT CGT A-3′
Reverse	5′-CAA CAA TAT CCA CTT TAC CAG AGT TA-3′
MMP-1
Forward	5′-AGA TGA AAG GTG GAC CAA CAA TTT-3′
Reverse	5′-CCA AGA GAA TGG CCG AGT TC-3′
MMP-2
Forward	5′-TTG ATG GCA TCG CTC AGA TC-3′
Reverse	5′-TGT CAC GTG GCG TCA CAG T-3′
MMP-9
Forward	5′-CCA CCA CAA CAT CAC CTA TTG G-3′
Reverse	5′-GCA AAG GCG TCG TCA ATC A-3′
COX-2
Forward	5′-GAA TCA TTC ACC AGG CAA ATT G-3′
Reverse	5′-TCT GTA CTG CGG GTG GAA CA-3′
Probe	6-FAM-TCC TAC CAC CAG CAA CCC TGC CA-6-TAMRA
COL1A1
Forward	5′-ACG AAG ACA TCC CAC CAA TCA-3′
Reverse	5′-CGT TGT CGC AGA CGC AGA T-3′
COL1A2
Forward	5′-TGA GAC TCA GCC ACC CAG AGT-3′
Reverse	5′-TGG CTT CCA TAG TGC ATC CTT-3′
COL3A1
Forward	5′-TCT TGG TCA GTC CTA TGC GGA TA-3′
Reverse	5′-CGG ATC CTG AGT CAC AGA CAC A-3′
Probe	6-FAM-AGA TGT CTG GAA GCC AGA ACC ATG CC-6-TAMRA
LOX
Forward	5′-GCG GCG GAG GAA AAC TGT-3′
Reverse	5′-AGC AGC ACC CTG TGA TCA TAA TC-3′
TIMP1
Forward	5′-TGT TGG CTG TGA GGA ATG CA-3′
Reverse	5′-GGT CCG TCC ACA AGC AAT G-3′
TIMP2
Forward	5′-CAC CCA GAA GAA GAG CCT GAA-3′
Reverse	5′-GGC AGC GCG TGA TCT TG-3′
FN1
Forward	5′-CAC GGG AGC CTC GAA GAG-3′
Reverse	5′-ACA ACC GGG CTT GCT TTG-3′
FN-EDA
Forward	5′-AGT AAC CAA CAT TGA TCG CCC TAA-3′
Reverse	5′-TTC CCA AGC AAT TTT GAT GGA-3′
FN-non-EDA
Forward	5′-GTG GTT AGT GTC TAT GCT CAG AAT CC-3′
Reverse	5′-CAG TTG GTG CAG GAA TAG TGG TT-3′

Calculation of EDA-containing FN mRNA (EDA-FN)

To differentiate EDA-FN from non-EDA-FN, EDA-specific primers were used. As EDA is located between exon III₁₁ and III₁₂, EDA-FN forward primer was designed to span the junction of exon III₁₁ and EDA (Fig. 6C and Table 1). Non-EDA-FN reverse primer was designed to span the junction of exon III₁₁ and III₁₂ (Fig. 6C and Table 1). To compute mRNA amounts, C_T values were converted to relative RNA using standard curves. EDA-FN (%) was calculated as (EDA-FN amount/total FN mRNA amount (EDA-FN + non-EDA-FN)) × 100.

FIGURE 6.

Up-regulation of MMPs and COX-2 is mediated via Toll-like receptor-4. Amnion mesenchymal cells were pretreated with serum-free medium containing with 10 μg/ml neutralizing antibodies to human TLR4 or control IgG for 1 h and then stimulated with either 10 μg/ml BSA or fFN (A) or recombinant EDA (B) for 24 (MMP-1 and COX-2) or 48 h (MMP-9). Data represent mean ± S.D. (error bars) of relative mRNA levels normalized to GAPDH. *, p < 0.05; **, p < 0.01, one-way analysis of variance. Ab, antibody.

MMP-1 Activity Assay

MMP-1 activity of the conditioned medium was assayed using a Sensolyte® Plus 520 MMP-1 Assay kit (72012, Anaspec, Fremont, CA) according to the manufacturer's instructions. The proteolytic activity of MMP-1 was quantified using FRET peptide.

Gelatin Zymography

Gelatin zymography was performed as described previously (20). Briefly, conditioned medium was concentrated using Amicon Ultra-4 Centrifugal Filter Units (3-kDa cutoff; UFC800324, Millipore) at 3200 × g for 50 min at 4 °C. Concentrated samples with added protease inhibitor mixture were used for immunoblotting and zymography. Concentrated medium was applied to 10% gelatin polyacrylamide minigels in standard sodium dodecyl sulfate (SDS) loading buffer containing 0.1% SDS. After electrophoresis, gels were soaked in renaturing buffer (2.7% Triton X-100) with gentle shaking for 30 min with one change after 30 min to remove SDS. Gels were soaked in assay buffer (50 mm Tris, 200 mm NaCl, 10 mm CaCl₂, 0.05% Brij 35, pH 7.5) for 16 h at 34 °C and stained with Coomassie Brilliant Blue R-250 in 50% methanol and 10% acetic acid and destained with 20% methanol and 10% acetic acid. Areas of lysis were analyzed using a Fuji LAS 3000 image analysis system (Fujifilm Life Science, Tokyo, Japan).

Prostaglandin E₂ ELISA

Prostaglandin E₂ concentration in the condition medium was assayed by Parameter PGE₂ immunoassay (KGE004B, R&D Systems) according to the manufacturer's directions.

Urea Extraction

Protein extraction with urea buffer was performed as described previously (21). Briefly, amnion cells were washed with PBS and resuspended in urea buffer (6 m urea, 16 mm potassium phosphate, pH 7.8, 0.12 m NaCl) containing protease inhibitor mixture. Collected samples were rocked overnight at 4 °C. Thereafter, samples were centrifuged at 10,000 × g for 30 min, and supernatants were used.

Nuclear Protein Extraction

Nuclear extraction from amnion cells was performed as described previously (22). Cells were harvested in 2 ml of cold PBS containing 0.6 mm EDTA and centrifuged at 5000 × g for 30 s. Cell pellets were gently resuspended in 800 μl of cold Buffer A (10 mm Tris-HCl, pH 7.9, 10 mm KCl, 0.1 mm EDTA, 0.1 mm EGTA, 1 mm dithiothreitol) containing protease and phosphatase inhibitor mixture and allowed to swell on ice for 30 min. Then 25 μl of a 10% Nonidet P-40 solution was added to the cell suspensions and vortexed for 5 s. After centrifugation at 12,000 × g for 30 s, nuclear pellets were resuspended in 150 μl of cold Buffer B (20 mm Tris-HCl, pH 7.9, 400 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1 mm DTT) containing protease and phosphatase inhibitor mixture and rocked for 1 h at 4 °C. After removal of nuclear debris by centrifugation at 12,000 × g for 5 min at 4 °C, nuclear extracts were stored at −80 °C.

Immunoblotting

Total protein (30 or 40 μg/lane) was applied to 6 or 10% polyacrylamide gels, separated by electrophoresis, and transferred to nitrocellulose membranes overnight at 4 °C. The membranes were blocked with TBS containing 2 or 5% nonfat powdered milk for 1 h at 37 °C. For immunoblots of FN, a membrane was incubated with primary antibody (rabbit anti-human fibronectin polyclonal antibody (AB1945) or mouse monoclonal antibody (IST-9)) for 1 h at 37 °C. For the nuclear extracted proteins, the membranes were incubated with primary antibody (phospho-NFκB p65 (Ser-536), NFκB p65, phospho-MAPK, total MAPK, TATA-binding protein, or GAPDH antibody) overnight at 4 °C. Thereafter, blots were incubated with secondary antibody (goat anti-rabbit IgG-HRP conjugate or goat anti-mouse IgG-HRP conjugate at 1:10,000) at room temperature for 1 h. Immunoreactive signals were detected by chemiluminescence using Amersham Biosciences ECL Plus Western Blotting Detection kit (RPN2132, GE Healthcare).

Preparation of Recombinant FN EDA

Preparation of recombinant FN EDA was performed as described previously with modifications (23). A cDNA library was prepared using RNA from human amnion epithelial cells by reverse transcription. cDNA coding for FN EDA and III₁₁ regions were PCR-amplified by Platinum Pfx DNA polymerase (11708-013, Invitrogen). The primers used to amplify EDA region were sense primer EDA-s (5′-GGAATTCCATATGAACATTGATCGCCCTAAAGGACT-3′) and antisense primer EDA-a (5′-ATAAGAATGCGGCCGCTGTGGACTGGGTTCCAATCAGGGG-3′). The primers used to amplify III₁₁ region were sense primer III₁₁-s (5′-GGAATTCCATATGGAAATTGACAAACCATCCCA-3′) and antisense primer III₁₁-a (5′-ATAAGAATGCGGCCGCGGTTACTGCAGTCTGAACCA-3′). (The NdeI site in the sense primers and the NotI site in the antisense primers are underlined.) PCR-amplified DNA was digested with NdeI and NotI restriction enzymes, purified by gel extraction, and then subcloned into the bacterial expression vector pET-28a (Novagen, Madison, WI), which enables expression of fusion proteins carrying six additional histidine residues (His₆ tags) at the carboxyl terminus. Clones were confirmed by sequencing. Individual EDA or III₁₁ protein with His₆ tags was expressed in E. coli BL21(DE3) and purified by nickel-nitrilotriacetic acid-agarose (Qiagen, Venlo, Netherlands). EDA and III₁₁ proteins were dialyzed with sterile PBS, pH 7.4 with three changes, then made endotoxin-free using Detoxi-Gel endotoxin-removing gel (20344, Thermo Scientific, Rockford, IL), and filter-sterilized.

Mice

C57/Bl6 mice with a gestational age time of 19 days were maintained in a non-pathogen-free environment and kept on a 12/12-h light/dark cycle. Matings occurred for 5 h (from 10:00 to 15:00). Pregnant animals with 6–10 pups were anesthetized with tribromoethanol (250 μg/g of body weight, intraperitoneal) at 19:00 on day 17 postcoitum. A midline incision was made, and 4 μl of PBS or fFN was injected in the interface between the fetal membranes and uterine lining of six pups (i.e. 4.8 μg total for the 200 μg/ml dose). The maternal abdomen was closed with 5-0 silk. Animal experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee of the University of Texas Southwestern Medical Center.

Immunofluorescence

Epithelial and mesenchymal cells were plated in tissue culture chambers mounted on glass microslides (4808, Lab-Tek) and treated with PBS, TNF-α, or LPS. Post-treatment cells were washed with PBS and fixed with 4% formalin in PBS for 15 min at room temperature. Formalin was removed, and cells were washed three times with PBS for 5 min each. Cells were covered with 100% methanol for 10 min at −20 °C and rinsed with PBS for 5 min. After blocking with 5% normal goat serum (Santa Cruz Biotechnology) in PBS with 0.3% Triton X-100 for 1 h, cells were incubated in primary antibody (2.5 μg/ml anti-human fibronectin polyclonal antibody (AB1945, Millipore) or monoclonal IST-9 to fibronectin EDA (AB6328) in PBS with 0.3% Triton X-100 overnight at 4 °C. After washing three times with PBS for 5 min each, cells were incubated with secondary antibody (1:300; Alexa Fluor 488 goat anti-rabbit IgG (A11034) or goat anti-mouse IgG (A11029), Molecular Probes by Invitrogen) with 0.3% Triton X-100 for 2 h without light, washed with PBS three times for 5 min each, and mounted with DAPI (ProLong Gold antifade reagent with DAPI, P36395, Molecular Probes by Invitrogen), coverslipped, and imaged using a Leica TCS SP5 confocal microscope with or without wavelengths for FITC or DAPI. Different treatments were conducted on the same slide, and images were captured with identical settings.

Statistical Analysis

Data were analyzed by unpaired Student's t test or one-way analysis of variance followed by the Student-Newman-Keuls test for analysis of the TLR4 antibody blocking experiments. χ² analysis (200 μg/ml) or Fisher's exact (100 μg/ml) tests were used to analyze preterm delivery rates. p values ≤0.05 were considered significant.

RESULTS

Effect of fFN on MMP-1, -2, and -9 in Amnion Cells

Gelatin affinity chromatography was used to purify FN from human amnion and human blood. By SDS-PAGE, the molecular size of fFN was increased relative to pFN (∼250 kDa; Fig. 1A) likely due to its higher carbohydrate content (24). Immunoblotting confirmed that these proteins were FN (Fig. 1A). To assess the effect of fFN on MMP-1, MMP-2, or MMP-9, primary amnion cells were treated with fFN (0–20 μg/ml) for 4–48 h, and mRNA and activity levels were quantified using quantitative PCR, FRET, or quantitative zymography (Fig. 1). Whereas MMP-1 mRNA levels were low and unchanged in amnion epithelial cells, fFN treatment resulted in dramatic increases in MMP-1 mRNA in a dose- and time-dependent manner in amnion mesenchymal cells (Fig. 1, B, panel a, and E, panel a). Base-line MMP-1 activity in mesenchymal cells was 3–4-fold that of epithelial cells, and although not regulated in epithelial cells, fFN induced significant (>100-fold) increases in MMP-1 activity in a dose- and time-dependent manner in mesenchymal cells (Fig. 1, C and F).

FIGURE 1.

Fetal FN dose- and time-dependently increases MMP-1 and MMP-9 mRNA and enzymatic activity in amnion mesenchymal, but not epithelial, cells. A, purification of fetal fibronectin from human amnion. Left, SDS-PAGE (6%) of human pFN and fFN from human amnion stained with Coomassie Blue. Right, immunoblots of FN using anti-human fibronectin polyclonal antibody. B, amnion cells were treated with different doses of fFN for 24 h and analyzed for MMP-1 (panel a), MMP-2 (panel b), or MMP-9 (panel c) mRNA. 0 μg/ml fFN represents treatment with 10 μg/ml BSA. C, MMP-1 enzymatic activity in the conditioned medium from cells treated with either 20 μg/ml BSA or fFN for 24 h. D, gelatin zymography of conditioned medium from cells treated with different doses of fFN for 24 h. n = 3 in each group. E, fFN (20 μg/ml)-induced increases in MMP-1 (panel a) and MMP-9 (panel b) mRNA as a function of time. F, MMP-1 enzymatic activity in conditioned medium from amnion mesenchymal cells treated with 20 μg/ml fFN for various times. G, gelatin zymography of conditioned medium from amnion mesenchymal cells. Cells were stimulated with 20 μg/ml fFN for various times. Data represent mean ± S.D. (error bars). n = 3 in each group. *, p < 0.05; **, p < 0.01.

Basal expression of MMP-2 was 6–8-fold in mesenchymal cells compared with epithelial cells (Fig. 1B, panel b), and treatment of mesenchymal cells with fFN resulted in modest, but significant, increases in MMP-2 gene expression and MMP-2 in conditioned medium (Fig. 1D). Similarly, MMP-9 mRNA transcripts were increased in mesenchymal cells relative to epithelial cells, and fFN treatment resulted in 3-fold increases in MMP-9 mRNA and proportionate time-dependent increases in pro-MMP-9 in conditioned medium (Fig. 1, B, panel c, D, E, panel b, and G). These data indicate that fFN has significant effects on interstitial collagenase, MMP-1, and the gelatinases MMP-2 and MMP-9 in mesenchymal, but not epithelial, cells of human amnion.

Gene Expression Related to Collagen Synthesis or MMP Inhibition

To determine whether fFN-induced increases in protease activity were balanced by expression of genes that limit or oppose collagen degradation, the effect of fFN on collagen and lysyl oxidase mRNA was analyzed. Fetal FN did not alter collagen type I α1 or α2 (COL1A1 or COL1A2) mRNA in either cell type of the amnion. Similarly, collagen type III (COL3A1) gene expression was not regulated by fFN (data not shown). Furthermore, fFN did not alter mRNA levels of the major collagen cross-linking enzyme, lysyl oxidase (LOX), in epithelial cells but resulted in decreased LOX expression in mesenchymal cells (∼20%; data not shown). Moreover, tissue inhibitors of MMP-1 and -2 (TIMP1 and TIMP2), which are known to inhibit MMP activity, were unchanged by fFN treatment. These results suggest that fFN-induced increases in MMP-1 and MMP-9 were not restricted by increased collagen synthesis or MMP inhibition and that fFN-induced decreases in LOX may contribute to, rather than oppose, fFN-induced loss of membrane integrity.

fFN Increases COX-2 mRNA and PGE₂ Synthesis in Amnion Mesenchymal Cells

PGs are believed to be involved in PPROM, initiation of myometrial contractions of labor, and cervical ripening during parturition. Thus, we quantified the effect of fFN on PG synthesis in primary amnion cells. In epithelial cells, COX-2 mRNA was unchanged, whereas fFN increased COX-2 mRNA and PGE₂ 10-fold in mesenchymal cells (Fig. 2).

FIGURE 2.

fFN increases COX-2 mRNA and PGE₂ synthesis in amnion mesenchymal cells. Primary amnion epithelial or mesenchymal cells were treated with fFN for the indicated dose or time. A, COX-2 mRNA level in amnion cells treated with different doses of fFN for 24 h. 0 μg/ml fFN indicates the treatment with 10 μg/ml BSA. B, PGE₂ levels in conditioned medium. Amnion cells were treated with 20 μg/ml BSA or fFN for 24 h. C, COX-2 mRNA in amnion mesenchymal cells treated with 20 μg/ml fFN as a function of time. D, PGE₂ in conditioned medium from amnion mesenchymal cells treated with 20 μg/ml fFN for the indicated time. Data represent mean ± S.D. (error bars). n = 3 in each group. *, p < 0.05; **, p < 0.01.

Differential Effects of Plasma and Fetal FN

Intrauterine or vaginal bleeding is a strong risk factor of preterm labor and PPROM (25, 26). Because human plasma contains 300–400 μg/ml FN (27), we compared effects of fetal and plasma FN on MMP-1, MMP-9, and COX-2 mRNA (Fig. 3). In contrast with the dramatic effects of fFN, pFN did not alter expression of these genes in mesenchymal cells (Fig. 3A, panels a–c). We considered the possibility that one unique domain of fFN termed EDA (28) may be important in the differential effects of fetal FN and pFN. Immunoblot analysis with an antibody specific for the EDA demonstrated strong immunoreactivity in fFN that was absent in pFN (Fig. 3B). To determine whether this domain is limited to amnion or differentially expressed in epithelial and mesenchymal cells, EDA-specific primers were used to estimate the relative proportion of EDA-containing FN expressed in human fetal membranes (illustrated in Fig. 3C). Fetal tissues (i.e. amnion tissues from both rupture and non-rupture sites, primary cultured amnion epithelial and mesenchymal cells, and chorion) were compared with maternal decidua. Using quantitative PCR, we estimated that fetal tissues contained ∼40% EDA-FN compared with only 10% in maternal decidua (Fig. 3D).

FIGURE 3.

Fetal, but not plasma, FN up-regulated MMPs and COX-2, indicating a potential role of the EDA. A, primary amnion mesenchymal cells were treated with different doses of pFN or fFN for 24 h and analyzed for MMP-1 (panel a), MMP-9 (panel b), or COX-2 (panel c) mRNA. 0 μg/ml fFN indicates treatment with 10 μg/ml BSA. B, SDS-PAGE of purified pFN and fFN (left panel). 1 μg of protein was loaded in each lane and stained with Coomassie Blue. Left lane, fFN; right lane, pFN. Right panel, immunoblot analysis of EDA in purified pFN and fFN (1 μg each). Left lane, fFN; right lane, pFN. C, schema of FN-EDA, FN without EDA (FN-non-EDA), and primers (arrows) designed with specificity for FN-EDA or FN without EDA. fw, forward primer; rev, reverse primer. D, the percentage of EDA-containing FN mRNA in various human tissues and cells. RS, rupture site; NRS, non-rupture site; Epi, primary epithelial cells; Mes, primary mesenchymal cells; Dec, decidua. Data represent mean ± S.D. (error bars). n = 3 in each group. **, p < 0.01 compared with pFN (A) or decidua (D).

Recombinant EDA Up-regulates MMPs and COX-2

Recombinant EDA and its adjacent type III repeat were purified (Fig. 4A). Treatment of amnion mesenchymal cells with recombinant EDA resulted in dramatic increases in MMP-1 mRNA and enzymatic activity (Fig. 4, B, panel a, and C). In contrast, recombinant III₁₁ had no effect (Fig. 4, B, panel a, and C). Effects of EDA and III₁₁ on MMP-2 and MMP-9 are also shown in Fig. 4B, panels b and c. Although treatment of mesenchymal cells with EDA resulted in modest increases in MMP-2 mRNA, increases in pro- or active MMP-2 were not detected by zymography (Fig. 4D). On the other hand, EDA, but not III₁₁, resulted in significant increases in MMP-9 mRNA and pro-MMP-9. EDA also up-regulated COX-2 gene expression and PGE₂ synthesis in mesenchymal, but not epithelial, cells (Fig. 4, B, panel d, and E). Taken together, these results suggest that the EDA of fFN is crucial for fFN-induced increases in MMPs and COX-2.

FIGURE 4.

Recombinant EDA up-regulates MMPs and COX-2. A, Coomassie Blue staining of SDS-PAGE of 2 μg/lane recombinant EDA (left lane) and III₁₁ (right lane). B, epithelial or mesenchymal cells were treated with different doses of EDA (solid bars) or III11 (open bars) for 24 h and analyzed for MMP-1 (panel a), MMP-2 (panel b), MMP-9 (panel c), or COX-2 (panel d) mRNA. Data represent mean ± S.D. (error bars) of relative mRNA levels normalized to GAPDH. C, MMP-1 enzymatic activity in the conditioned medium from cells treated with either 200 nm III₁₁ or EDA for 24 h. D, gelatin zymography of conditioned medium from cells treated with 100 nm III₁₁ or EDA for 48 h. E, PGE₂ levels in conditioned medium. Amnion cells were treated with 200 nm III₁₁ or EDA for 24 h. Data represent mean ± S.D. (error bars). n = 3 in each group. ctl, control, no reagents in the medium. *, p < 0.05; **, p < 0.01.

Up-regulation of MMPs and COX-2 Is Mediated via Toll-like Receptor-4

The mechanisms by which fFN mediates intracellular signal pathways in amnion mesenchymal cells are not known. Because FN activates integrins and its functional element is RGD sequences, we tested the possibility that integrins may serve as FN signal transducers in mesenchymal cells. Neutralizing antibodies to β1 (10 μg/ml; mAb13, P4C10, and 6S6) or α5 integrins (10 μg/ml; P1D6) did not inhibit fFN-induced up-regulation of MMPs and COX-2 (Fig. 5). It has been reported that the EDA of fFN activates TLR4 in amnion epithelial cells, but the cells required ectopic expression of the TLR4 accessory protein MD-2 (16). Epithelial and mesenchymal amnion cells were treated with LPS, and mRNA levels of MMPs and COX-2 were quantified. The results confirmed that mesenchymal, but not epithelial, cells were responsive to LPS in terms of activation of these genes and that MD-2 was not expressed in epithelial cells (data not shown). Because the TLR4·MD-2 complex is crucial for mediating LPS signals, we determined whether TLR4 was involved in mediating fFN signals in mesenchymal cells. Cells were treated with fFN and EDA in the presence or absence of TLR4-neutralizing antibodies (Fig. 6). Neutralization of TLR4 inhibited fFN-induced increases of MMP-1, MMP-9, and COX-2 mRNA by 65, 43, and 86%, respectively (Fig. 6A). EDA-induced increases in MMP-1, MMP-9, and COX-2 mRNA were also inhibited significantly by TLR4-neutralizing antibodies (Fig. 6B). TLR2-neutralizing antibody on the other hand did not block these effects (data not shown).

FIGURE 5.

α5 and β1 integrin-neutralizing antibodies do not inhibit fFN-induced up-regulation of MMPs and COX-2. Amnion mesenchymal cells were pretreated with serum-free medium containing control IgG, β1 integrin- (mAb13, p4C10, or 6S6), or α5 integrin (P1D6)-neutralizing antibodies (10 μg/ml) for 1 h and then stimulated with either BSA (10 μg/ml) or fFN (10 μg/ml) for 24 (MMP-1 and COX-2) or 48 h (MMP-9) h. Data represent mean ± S.E. (error bars) of mRNA levels normalized to GAPDH.

fFN Activates NFκB and ERK1/2 Signaling

The effects of fFN on NFκB activation were determined using phosphorylation and nuclear localization of its p65 subunit (29, 30). In mesenchymal cells, fFN treatment resulted in increased phosphorylation of p65 and increased nuclear localization of total p65 after 60 min (Fig. 7). Phospho-p65 was not detected in nuclear fractions (Fig. 7A) because phosphorylated p65 is rapidly dephosphorylated in the nucleus (30).

FIGURE 7.

fFN activates NFκB and ERK1/2 signaling. Primary amnion mesenchymal cells were treated with 20 μg/ml fFN for the indicated time. A, 40 μg of cytoplasmic (left panel) or nuclear (right panel) extracts were immunoblotted with anti-phospho-p65 (upper panel), anti-p65 (middle panel), and anti-TATA-binding protein (TBP) (as loading control; lower panel) antibodies. B, 40 μg of cytoplasmic extracts were immunoblotted with anti-phospho-p44/p42 (Erk1/2) (upper panel), anti-p44/p42 (Erk1/2) (middle panel), and anti-GAPDH (loading control; lower panel) antibodies.

MAPK/ERK signaling was examined because this pathway is a major intracellular mechanism by which COX-2 and PGE₂ synthesis is regulated in amnion (31, 32). Whereas total p44 and p42 were not changed by fFN, both phospho-p44 (Erk1) and phospho-p42 (Erk2) were up-regulated after 120 min of treatment and rapidly returned to basal levels (Fig. 7B). Taken together, fFN treatment activated NFκB and ERK1/2 signaling pathways, both of which are considered major signal transduction pathways in the activation of MMPs and PGE₂ synthesis, in mesenchymal cells.

Effect of Endotoxin and TNF-α on fFN Expression in Amnion Cells

In physiological resting conditions, tissue FN exists in its fibrillar form tightly bound to the matrix. The conformation of free FN, generated in response to injury, differs dramatically from its fibrillar form. To understand the biology of fFN signaling, we investigated the regulation of release of fFN from the extracellular matrix of amnion epithelial and mesenchymal cells. Specifically, LPS and TNF-α were considered as possible regulators because infection and inflammation play significant roles in the pathogenesis of preterm labor (33). In epithelial, but not mesenchymal, cells, both LPS and TNF-α resulted in increased fibronectin-1 (FN1) mRNA, which includes both EDA- and non-EDA-FN (Fig. 8, A and B). Immunofluorescent staining of epithelial cells using antibodies to full-length or EDA-FN revealed that under control conditions both total FN and EDA-FN were distributed in fibrillar adhesions to the cell membrane and extracellular FN fibrils (Fig. 8C). Although total FN appeared slightly increased after LPS treatment, EDA-FN was increased dramatically and ensheathed the cells. Treatment with TNF-α also resulted in an increased EDA-FN staining intensity around the cell membrane (Fig. 8C). To determine whether FN was differentially increased in matrix-bound or soluble fractions, immunoblot analysis was conducted in matricellular urea extracts or medium from epithelial and mesenchymal cells treated with LPS or TNF-α (Fig. 8, D and E). LPS and TNF-α resulted in increased amounts of non-cross-linked matrix-bound FN in epithelial cells. Interestingly, in mesenchymal cells, FN was predominantly matrix-bound and not altered appreciably by LPS and TNF-α. In epithelial cells, however, FN was expressed as both soluble and matrix-bound forms, and LPS and TNF-α treatment resulted in marked increases in both forms. Overall, these results indicate that, although epithelial cells were not responsive to the effects of fFN on MMPs and PGE₂ synthesis, endotoxin and inflammatory cytokines increased matrix-bound and free fFN in epithelial, but not mesenchymal, cells.

FIGURE 8.

Effect of endotoxin and TNF-α on fFN expression in amnion cells. Primary human amnion epithelial or mesenchymal cells were treated with the indicated doses of LPS or TNF-α for 24 h. A and B, FN1 mRNA in LPS- (A) or TNF-α (B)-treated amnion cells. *, p < 0.05; **, p < 0.01. C, amnion epithelial cells were treated with PBS (Ctl), LPS (1 μg/ml), or TNF-α (10 ng/ml) for 24 h. After fixation, cells were immunostained with antibodies to full-length FN (upper panel) or EDA-specific FN (lower panel). Images represent results obtained in duplicate from two different cell preparations. Immunostaining was absent in negative controls without primary antibody (not shown). D, immunoblot analysis of matrix FN in urea extracts (30 μg/lane) from epithelial and mesenchymal cells treated with various doses of LPS or TNF-α for 24 h. E, immunoblot analysis of soluble FN in conditioned medium (30 μg/lane) from epithelial and mesenchymal cells treated with various doses of LPS or TNF-α for 24 h.

Effect of fFN on Gestation Length in Vivo

Finally, the effect of fFN on gestational length was determined using a pregnant mouse model. Control (PBS) or fFN (100–200 μg/ml) was injected between the uterine wall and fetal membranes of precisely timed pregnant mice on day 17 postcoitum (Fig. 9A). Fetal FN resulted in preterm delivery in 11 of 14 pregnant mice (p < 0.05 compared with control. In all cases of preterm delivery (defined as <42 h after injection), pups were premature, and fetal survival was impaired (Fig. 9B). The mean time to delivery was 58 ± 5 h in PBS controls, 53 ± 11 h for pFN (200 μg/ml), and 38 ± 7 h for fFN (200 μg/ml).

FIGURE 9.

Effect of fFN on preterm birth in mice. Pregnant mice (day 17) were injected with PBS or fFN and monitored for time of delivery. A, animals delivering at term are noted by solid symbols. Open symbols denote preterm birth. χ² analysis was performed. B, fetal survival rates in animals injected with PBS or fFN. Data represent mean ± S.E. (error bars). *, p < 0.01.

DISCUSSION

In this study, we demonstrated that fFN is not only a fetal membrane matrix molecule that serves as a clinical marker of preterm labor but also bioactive in amnion mesenchymal cells and may thereby play a significant role in the pathogenesis of preterm rupture of the fetal membranes and preterm labor. Our data indicate that (i) although epithelial cells release soluble fFN the primary fFN-responsive cell type is the mesenchymal cell, and (ii) fFN-induced up-regulation of MMPs and COX-2 is mediated predominantly by mesenchymal TLR4.

Differential Responses in Amnion Cell Types

In amnion epithelial cells, TLR4 signaling is impaired due to the absence of the TLR4 obligatory accessory protein MD-2. Okamura et al. (16) demonstrated that, although the EDA of fFN activated NFκB signaling in THP-1 cells, amnion epithelial cells required introduction of MD-2 for EDA-induced activation of NFκB. Full-length fFN was not effective even with ectopic expression of MD-2 (16). Thus, it was speculated that proteolytic processing of fFN was required. Our data support these findings in that fFN-mediated up-regulation of MMPs and COX-2 was absent in epithelial cells. Here, we found that full-length fFN (with no evidence of significant proteolysis) was effective in inducing ERK1/2 and NFκB signaling and activation of MMP-1, -2, -9, and COX-2 in amnion mesenchymal cells. Recombinant EDA (100 nm), however, was more potent than fFN (20 μg/ml = ∼80 nm), suggesting that proteolysis of fFN into EDA-containing fragments may amplify these signaling pathways. These findings are in agreement with those of Saito et al. (23) who suggested that domains in the FN protein distal to the EDA may suppress FN-induced activation of proinflammatory pathways. We did not observe LPS- or TNF-α-induced proteolysis of fFN in preparations of cells or medium, suggesting that intact soluble fFN mediates cell signaling in mesenchymal cells.

Soluble fFN from Epithelial Cells

Chorioamnionitis plays a major role in the pathophysiology of spontaneous preterm labor. Intrauterine infection and release of endotoxin from bacteria activate fetal membranes and decidua to produce proinflammatory cytokines such as TNF-α, interleukin-1α (IL-1α), IL-1β, IL-6, and IL-8 (34). It has been reported that LPS and these proinflammatory cytokines increase the release of fFN in conditioned medium of amnion tissues (35). In this study, we clarified that amnion epithelial, not mesenchymal, cells respond to LPS and TNF-α with increased soluble fFN. The mechanisms by which amnion epithelial cells function as a sensor to LPS and proinflammatory cytokines are unknown. Fibroblasts, smooth muscle cells, and adherent platelets promote assembly of globular soluble FN into insoluble fibrillar FN that maintains adhesion, growth, and cell migration (36). This process of FN assembly involves FN-integrin interactions and elongation of the circulating globular form of pFN (37, 38). Little is known, however, regarding the reverse process in which fibrillar FN is solubilized to a conformation that facilitates FN-induced cell signaling. Our data indicate that although LPS and TNFα do not activate MMPs or COX-2 in epithelial cells both resulted in the release of soluble fFN into the medium. It is possible that newly synthesized fFN may remain soluble due to LPS-induced interruption of FN fibrillar assembly (39). FN fragments containing the fibrin-binding domains of FN also block FN assembly (40–42). Thus, the complex intimate relationship between epithelial and mesenchymal cells may augment LPS- or cytokine-induced effects on membrane integrity. TNF-α may further augment this effect by increasing the proportion of EDA-containing FN released from epithelial cells. MMPs secreted from mesenchymal cells not only degrade collagen and stimulate the release of proinflammatory cytokines such as TNF-α (43, 44) but also may induce FN fragmentation (45) to inhibit fibrillar FN assembly in epithelial cells. This model of mechanisms by which fFN may induce PPROM and preterm labor is illustrated in Fig. 10.

FIGURE 10.

Model of fFN signaling in the pathogenesis of PPROM and preterm labor. LPS and proinflammatory cytokines such as TNF-α effect increased amounts of soluble fFN from amnion epithelial cells (which do not express functional TLR4 complexes) either through release from its fibrillar form or through increased production of non-assembled fFN. Soluble fFN activates MD-2·TLR4 receptor complexes on mesenchymal cells through its EDA. Activation of TLR4 leads to intracellular signaling through NFκB and ERK1/2 to induce expression of COX-2 and MMPs, thereby leading to cervical ripening, uterine contractions, and collagenolytic degradation of the fetal membranes. MMPs may also fragment fFN, which inhibits FN fibrillar assembly in epithelial cells, further amplifying EDA signaling.

Fetal FN induced preterm labor in our in vivo mouse model of preterm birth (11 of 14 mice). This proportion of preterm birth is less than that observed with RU486 (46) or LPS (47–50) in which preterm delivery rates approach 100% possibly because fFN does not activate luteolysis or uterine contractility. Nonetheless, our experimental findings in vitro and in vivo indicate that fFN is not only a matrix molecule but is biologically active in vivo.

Mechanisms of fFN Signaling

Our data indicate that fFN activates TLR4 and NFκB in amnion mesenchymal cells. As NFκB regulates transcription of MMP-1, MMP-9, and COX-2 (51), up-regulation of these gene products is likely mediated by NFκB signaling. The ERK1/2 pathway was also activated by fFN in amnion mesenchymal cells. ERK1/2 activates activator protein-1, which binds to the proximal promoter regions of MMP-1 and MMP-9 (52). For example, in skin cancer cells, FN increases MMP-9 expression via the ERK-MAPK pathway (53), and ERK1/2 also regulates COX-2 expression in these cells (54). Thus, NFκB and the ERK-MAPK pathway are thought to be involved in fFN-induced up-regulation of MMP-1, -9, and COX-2.

In other cells and tissues, FN-induced cell signaling is mediated by α5β1 integrin, a primary receptor for FN (55). For example, FN activates integrin signaling through its RGD sequences to induce MMPs (56). Our data do not support a role for α5β1 integrin signaling pathways in fFN-induced activation of MMPs or COX-2. Although both RGD integrin binding sequences are conserved in pFN and fFN, pFN did not induce MMPs or COX-2 in amnion cells. Rather, a unique alternatively spliced exon encoding EDA in fFN, but not pFN (28, 57), was crucial for fFN-induced signaling. EDA has been shown to activate MMP-1, MMP-3, and MMP-9 gene expression in synovial cells (23) and MMP-9 in THP-1 cells (16). Here, we confirmed that EDA-induced activation of TLR4, not integrins, was important in mediating these effects in amnion mesenchymals cells. This finding is clinically important because TLR4 antagonists are undergoing clinical trials for the treatment of septic shock (58). Neutralization of fFN-EDA interactions or antagonism of TLR4, therefore, may have therapeutic potential for preterm labor and PPROM.