Factor Xa and thrombin, but not factor VIIa, elicit specific cellular responses in dermal fibroblasts


  • 1

    Present addresses: Department of Medicine, University Hospital of Zurich, Raemistrasse 100, Zurich 8091, Switzerland, 2Department of Surgery, University of Western Australia, Fremantle Hospital, PO Box 480, Fremantle, Western Australia 6959.

    Drs Bachli and Pech contributed equally to this work.

and reprints: Dr John McVey, Haemostasis Research, MRC Clinical Sciences Centre, The Faculty of Medicine, Imperial College, Hammersmith Campus, Du Cane Rd, London W12 0NN, UK.
Tel.: +44 20 8383 8253; fax: +44 870 131 3540; e-mail: john.mcvey@csc.mrc.ac.uk


Summary.  Coagulation factors (F)VIIa, FXa and thrombin are implicated in cellular responses in vascular, mesenchymal and inflammatory cells. Fibroblasts are the most abundant cells in connective tissue, and damage to blood vessels places coagulation factors in contact with these and other cell types. Objectives: To investigate cellular responses of primary dermal fibroblasts to FVIIa, FXa and thrombin by following changes in expression of candidate proteins: monocyte chemotactic protein-1 (MCP-1), interleukin-8 (IL-8), interleukin-6 (IL-6) and vascular endothelial growth factor (VEGF), and to determine the expression of receptors implicated in signaling by these coagulation factors. Methods: Steady-state mRNA levels were quantified by RNase protection assay, and protein secretion by ELISA. PAR gene expression was assessed by ribonuclease protection assay and conventional and quantitative reverse-transcription–polymerase chain reaction. Results: FVIIa did not induce the candidate genes. In contrast, FXa and thrombin induced MCP-1 mRNA and protein secretion strongly, IL-8 moderately, and IL-6 weakly. Neither FXa nor thrombin induced VEGF mRNA or protein secretion, although FXa induced VEGF protein secretion in lung fibroblasts. Comparison of the presence of candidate receptors in the two fibroblast subtypes demonstrated higher levels of PAR-1 and PAR-3 in lung fibroblasts relative to their dermal counterparts and the additional expression of PAR-2. Conclusions: FXa and thrombin induce expression of MCP-1, IL-8 and IL-6, and distribution and expression of PARs on dermal fibroblasts is reduced relative to their lung counterparts. Tissue origin may influence the cellular response of fibroblasts to coagulation proteases.


The proteolytic activity of factor (F)VIIa, FXa and thrombin has been implicated in cellular responses in vascular, mesenchymal and inflammatory cells. FVIIa and FXa activate mitogen-activated protein (MAP) kinases, and induce intracellular calcium mobilization and expression of genes encoding inflammatory cytokines, transcriptional regulators and growth factors [1–9]. Furthermore, FVIIa initiates expression of genes encoding extracellular matrix signaling proteins and those involved in cell migration [7,10], while FXa additionally stimulates nitric oxide release [11], and up-regulates expression of adhesion molecules [4] and tissue factor (TF), the initiator of coagulation [12]. Thrombin exerts diverse effects, leading to platelet aggregation, endothelial cell activation, proliferation and cytokine release by smooth muscle cells and fibroblasts, and calcium signaling in T lymphocytes [13].

Many of the actions of thrombin are mediated through protease-activated receptors (PARs), a subfamily of G-protein-coupled receptors. Following limited proteolytic cleavage of the PAR extracellular domain, a new amino terminus is exposed, which binds intramolecularly and generates an intracellular signal [13]. To date, four members of the PAR family have been identified (PAR-1 to PAR-4). All are expressed in a cell-type specific manner. PAR-1 mRNA has been detected in platelets, mononuclear cells (MNC) and endothelial cells (EC); PAR-2 in neutrophils, MNC and EC; PAR-3 in MNC and EC and PAR-4 in platelets and MNC [14,15]. PAR-1, PAR-3 and PAR-4 are cleaved and activated by thrombin, and in human platelets, thrombin signaling is mediated at low concentrations by PAR-1 and at high concentrations by PAR-4 [14]. PAR-2 can be activated by trypsin, mast-cell tryptase and sperm acrosin [16–18]in vivo, and by FVIIa and FXa in keratinocytes and FXa in EC in vitro[7]. It is not clear whether these are physiological activators; however, activation of PAR-2 by FXa implicates this PAR as a candidate receptor for FXa signaling effects.

Several mechanisms for FVIIa signaling have been described. FVIIa elicits specific cellular responses in a TF-dependent manner [2,10,19] and also through TF-dependent FXa generation [8,19]. Furthermore, investigation of the nature of this TF contribution demonstrates a requirement for the TF cytoplasmic domain in some in vitro models [5,20–22]. However, others have shown no such requirement for this domain, suggesting the role of TF is to localize FVIIa to the membrane surface and to act as a cofactor in the activation of FVIIa [7,23].

Fibroblasts are the most abundant cells in connective tissue, and are characterized by functional differences that reflect the specialized role of the organ in which they are located [24]. They constitutively express TF, contributing to the hemostatic envelope that ensures FVII/FVIIa in blood is exposed to cells expressing TF following disruption of vascular integrity, thus leading to the initiation of blood coagulation. Human lung fibroblasts respond to coagulation proteases [8,10,25], as do human adventitial fibroblasts [26]; however, little is known of their effect on primary human dermal fibroblasts. In this study, we examined the response of primary dermal fibroblasts to FVIIa, FXa and thrombin by analyzing the expression of candidate genes MCP-1, interleukin (IL)-8, IL-6 and VEGF. We demonstrated a differential response to the coagulation proteases. Moreover, we observed differences in the response of dermal vs. lung fibroblasts to FXa. Finally, dermal fibroblasts expressed quantifiably lower levels of PAR-1 and PAR-3 compared with lung fibroblasts, which additionally expressed PAR-2.

Materials and methods


Unless stated otherwise, all chemicals were of the highest available analytical grade and were obtained from Sigma Chemical Co. or BDH (Poole, Dorset, UK).

Cell culture

Primary human dermal fibroblasts (CRL-2091, CCD-1070Sk; ATCC, Manassas, VA, USA) and primary human lung fibroblasts (CCD-11Lu; ATCC) were cultured in complete medium: MEM supplemented with 2 mmol L−1 glutamine, 100 U mL−1 penicillin and 0.1 mg mL−1 streptomycin (all Sigma), 0.1 mmol L−1 sodium pyruvate (Gibco BRL, Paisley, Renfrewshire, UK) and 10% heat-inactivated fetal calf serum (FCS; Labtech International, Uckfield, W. Sussex, UK). For all experiments fibroblasts from passage 8–12 were seeded at 3 × 104 cells cm−1[2], grown until 60% confluent (referred to hereafter as subconfluent), washed twice with phosphate-buffered saline (PBS) A (137 mmol L−1 NaCl, 2.7 mmol L−1 KCl, 1.5 mmol L−1 KH2PO4, 8.1 mmol L−1 Na2HPO4, pH 7.4) and serum-starved for 48 h in FCS-free complete medium.

Fibroblasts were incubated with human coagulation proteases diluted in serum-free medium supplemented with 1% albumin (Bio Products Laboratory, Elstree, Herts, UK; Delta Biotechnology Ltd, Nottingham, Notts, UK) and CaCl2, to a final concentration of 2.4 mmol L−1 (hereafter referred to as basal medium) for specified incubation periods. Final concentrations of individual coagulation proteases were 100 nmol L−1 FVIIa, 80 nmol L−1 FXa and 10 nmol L−1 thrombin.

Pooled human umbilical vein endothelial cells (HUVEC, passage 4; Clonetics™ TCS Biologicals Ltd, Buckingham, Bucks, UK) were cultured on 0.1% gelatin (Sigma) in M199 medium (Gibco BRL) supplemented with 20% heat-inactivated FCS, 10 ng mL−1 human recombinant acidic fibroblast growth factor (R & D Systems Europe, Abingdon, Oxon, UK) and 16.1 U mL−1 porcine intestinal mucosa heparin (H-3149; Sigma).

Coagulation factors

Recombinant human FVII was produced in CHO-K1 cells using protein-free expression medium [27], purified and auto-activated to give FVIIa. This material has been extensively characterized by polyacrylamide gel electrophoresis, N-terminal sequence analysis, functional analyzes in coagulation and chromogenic assays, binding assays for TF using surface plasmon resonance, and structural studies leading to the X-ray crystal structure of FVIIa [27–31]. To prepare FXa, plasma-derived human FX concentrate (2 mg; kind gift from Dr P. Feldman, Bio Products Laboratory) was dissolved in Tris-buffered saline (TBS; 50 mmol L−1 Tris, 150 mmol L−1 NaCl, pH 7.4) and dialyzed against TBS. We then added 2 mL settled volume of benzamidine Sepharose® 6B (Amersham Pharmacia Biotech, Little Chalfont, Bucks, UK) and 5 mmol L−1 CaCl2 (final concentration) was added. Activation of FX to FXa was achieved by coincubation with the purified FX activator derived from Russell's viper venom (RVV-X; Hematologic Technologies Inc., Essex Junction, VT, USA) at a ratio of 10 : 1 w/w FX to RVV-X. The reaction mixture was washed extensively with TBS, and FXa eluted with 50 mmol L−1 benzamidine in TBS. Benzamidine was removed by passing the FXa eluate through a Sepharose G25 Super Fine Column (Amersham Pharmacia Biotech). Human α-thrombin and FX (for assessment of TF expression on fibroblasts) was purchased from Hematologic Technologies Inc. FVIIa, FXa and thrombin were active site-inhibited with 1,5-dansyl-Glu-Gly-Arg chloro-methyl ketone (DEGR-CK, 1 : 1 w/w; Calbiochem, Nottingham, Notts, UK) overnight at room temperature, and buffer was exchanged to TBS as described previously for FVIIai preparation [29,30]. Homogeneity of the protein preparations was demonstrated under reduced and non-reduced conditions following electrophoresis on 10% Bis-Tris-HCl pH 6.4 polyacrylamide precast gels (NOVEX®. NuPAGE Electrophoresis System, Frankfurt am Main, Germany) at 200 V (data not shown). Activity of coagulation proteases was verified by chromogenic assay (see below). Recombinant human TF1-219 for measurement of FVIIa amidolytic activity was expressed in Escherichia coli and purified as described [29].

Chromogenic assays for FVIIa, FXa and thrombin

Amidolytic activities of FVIIa, FXa and thrombin were measured using chromogenic substrates S-2288, S-2765 and S-2238, respectively (Chromogenix Quadratech, Epsom, Derbyshire, UK). FVIIa amidolytic activity was quantified following addition to 2.5 nmol L−1 TF1-219, 6 mmol L−1 CaCl2 and 1 mmol L−1 S-2288. FXa and thrombin activities were quantified in the presence of 6 mmol L−1 CaCl2 and 1 mmol L−1 S-2765 or S-2238, respectively. Standard curves were generated using purified FVIIa, FXa and thrombin in 1% HSA (Bio Products Laboratory) in TBS, pH 7.4. All measurements were determined kinetically at 405 nm on a Thermomax plate reader (Molecular Devices, Sunnyvale, CA, USA) using SoftmaxPro V. 1.20 analysis software (Molecular Devices).

RNA isolation

Cells were washed twice with PBS and lyzed in a solution containing 4 mol L−1 guanidinium isothiocyanate, 20 mmol L−1 sodium acetate pH 5.2, 0.1 mmol L−1 DTT and 0.5% w/v N-lauryl-sarcosine. Total RNA was isolated by centrifugation through a solution of 5.7 mol L−1 cesium chloride and 100 mmol L−1 EDTA [32].

Ribonuclease protection assays

Riboprobe labeling and ribonuclease protection assays (RPAs) were performed with custom cDNA templates (MCP-1, IL-8, IL-6 and VEGF) or the hAngio-2 Human Angiogenesis Multi-Probe template set (PARs 1–3), and reagents according to the manufacturer's protocols (PharMingen Becton Dickinson, Oxford, Oxon, UK). MCP-1 and IL-8 cDNAs comprised one template set; VEGF and IL-6 cDNAs were each in separate template sets. cDNA of the housekeeping gene, L32, was included in all template sets for normalization. Protected fragments were electrophoresed on 6% acrylamide, 7 mol L−1 urea gels (NOVEX™ QuickPoint™), visualized with a Molecular Dynamics Phosphorimager 445 SI, and quantified using ImageQuant V. 3.1 software (both Amersham Pharmacia Biotech).

PAR-4 reverse-transcription–polymerase chain reaction

PAR-4 cDNA was synthesized from total RNA with a specific antisense primer, 5′-GTG CTG GGG TTC AGC CTG TC-3′ (nucleotides 33–52; GenBank Accession No. AF080214), and oligo dT. RNA (fibroblasts, 1 µg; platelets, 0.1 µg) and 10 pmol of each oligonucleotide were mixed and denatured for 10 min at 70 °C. dNTPs (0.7 mmol L−1) and 200 U Superscript II reverse transcriptase (Gibco BRL) were added and incubated for 60 min at 42 °C, then 5 min at 95 °C. Reactions without reverse transcriptase were prepared as negative controls. 1/7.5 (13%) of the reaction product volume was amplified using 300 ng of a PAR-4 specific sense primer, 5′-CAG CAG TCG CGA GGT TCA TC-3′ (nucleotides 366–347), the antisense primer and Red Hot Thermostable DNA polymerase (ABgene, Epsom, UK). Initial DNA denaturation (95 °C, 5 min) was followed by 30 cycles of denaturation (94 °C, 1 min), annealing (57 °C, 2 min) and extension (72 °C, 3 min). A final extension (72 °C, 10 min) completed the reaction. Polymerase chain reaction (PCR) products were separated on 1% agarose and verified by Southern blot analysis using a 330-bp PAR-4 PCR product (prepared from platelet RNA, subcloned into TOPO TA [Invitrogen, Groningen, The Netherlands] and sequence-verified).

PAR quantitative reverse-transcription–PCR

cDNA was generated from 1 to 2 µg RNA with the First Strand cDNA Synthesis Kit (AMV; Roche Applied Science, East Sussex, UK) using random primers as described by the manufacturer. Forward and reverse primers (Table 1) for PARs 1–4, GAPDH and beta-glucoronidase (GUSβ) were designed using Primer Express (Applied Biosystems, Warrington, Cheshire, UK) and Oligo software (MBI, Cascade, CO, USA). Quantitative reverse-transcription–PCR (RT-PCR) was performed on the MJ Opticon (Genetic Research Instrumentation, Braintree, Essex, UK) in a 20 µL reaction volume with 0.9 µL cDNA, primer concentrations indicated in Table 1, and 10 µL SYBR green PCR Master Mix (Applied Biosystems). Initial denaturation at 95 °C for 10 min was followed by 40 cycles of denaturation (95 °C, 15 s), annealing (60 °C, 1 min) and acquisition of fluorescence after each annealing step. Amplification specificity of PCR products was confirmed by melting curve analysis and agarose gel electrophoresis.

Table 1.  Primer pairs for quantitative RT-PCR
mRNA targetOligonucleotides (5′(r)3′)aProduct
size (bp)
(nmol L−1)
  • a

    F, R indicate forward an reverse primers, respectively; numbers indicate sequence position.

  • b

    This PAR-4 primer pair does not span an intron–exon boundary.

PAR-4F255: GTC TAC GAC GAG AGC GG89/∼200900

Data was analyzed with MJ Opticon Monitor software V. 1.4 (Genetic Research Instrumentation). Standard curves were constructed from serial dilution of the cloned amplicon of interest. Raw abundance values (copy number) for each gene of interest from dermal and lung fibroblasts were normalized to raw abundance values for GUSβ or GAPDH.These calculated ratios are expressed relative to HUVECs (arbitrarily designated 1.0) with an accuracy of one decimal place.

MCP-1, IL-8, IL-6 and VEGF ELISAs

For VEGF ELISAs, cells were seeded in 96-well plates. Each test and control condition was replicated in nine wells, and after the specified period, supernatants from three wells were pooled; thus all samples were replicated (n = 3). For MCP-1, IL-8 and IL-6 ELISAs, tissue culture supernatants from the RPA experiment were assayed. MCP-1, IL-8, IL-6 and VEGF were measured in duplicate according to the manufacturer's instructions (R & D Systems Europe), with the exception that absorbance was measured kinetically at 650 nm on a Thermomax plate reader. The detection limits of the assays are 15.6 pg mL−1 for MCP-1 and IL-8, 2.3 pg mL−1 for IL-6 and 5 pg mL−1 for VEGF.

The specificity of the thrombin and FXa responses was assessed by preincubation with hirudin (200 nmol L−1; Refludan®, Hoechst Marion Roussel, Frankfurt am Main, Germany) at room temperature for 30 min. Dermal fibroblasts seeded in 6-well plates were incubated with the inhibited coagulation protease preparations or the prothrombin activator Ecarin (0.5 U mL−1; Sigma) for 24 h, or hirudin for 30 min, followed by FXa for 24 h. Supernatants from each of six replicates were analyzed for MCP-1, IL-8 and IL-6. For these experiments, IL-8 was quantified using the Biotrak ELISA system (Amersham Pharmacia Biotech; detection limit 25.6 pg mL−1), which demonstrated increased intra-assay reproducibility compared with the R & D Systems IL-8 ELISA.

Estimation of cell number

For normalization of secreted protein to cell number, cells were fixed with 1.25% glutaraldehyde and stained with 0.1% w/v crystal violet in 200 mmol L−1 2-[N-morpholino]ethanesulfonic acid pH 6.0 for 30 min at room temperature. Absorbance of the acetic acid eluate in each well was measured at 590 nm with a Thermomax plate reader as previously described [33]. Absorbance is directly proportional to cell number and was verified for dermal fibroblasts (data not shown).

TF expression in human dermal and lung fibroblasts

Following two washes with PBS, cells were scraped from plates and centrifuged at 720 × g for 3 min at 4 °C. Cell pellets were lyzed with 1% Triton X-100 (Sigma) in TBS pH 8.4 for 12 h at 4°C. TF content of cell lysates was quantified in triplicate with IMUBIND® Tissue Factor ELISA Kit (American Diagnostica, Greenwich, USA; detection limit 10 pg mL−1). Protein content of lysates was quantified by BCA protein assay (Pierce & Warriner, Chester, UK).

TF functional activity was assessed by measuring the amount of FXa generation in the presence of recombinant FVIIa and purified FX. Cells were washed twice with HBSS (Sigma) and incubated with 20 nmol L−1 FVIIa, 2.5 mmol L−1 CaCl2, 1% HSA and 100 nmol L−1 FX in HBSS at room temperature with continuous shaking. Aliquots of supernatants were taken at 10 min and added to 10 mmol L−1 EDTA. After adding 4 mmol L−1 S-2765, color generation was followed kinetically at 405 nm using a Thermomax plate reader [29].

TF assay specificity was confirmed by replacement of FVIIa with equimolar FVIIai, or by pretreatment of cells with 3.5 µg mL−1 polyclonal affinity-purified rabbit antihuman TF antibody, both of which abolished color generation. Pre-treatment with an irrelevant antibody, affinity-purified rabbit antihuman FVIII, did not change color generation. Affinity-purified antibodies were prepared from rabbits immunized with recombinant human TF1-219 or FVIII (a kind gift of Dr G. Vehar, Genentech, San Francisco CA, USA).


FXa and thrombin but not FVIIa induce MCP-1, IL-8 and IL-6 gene expression and secretion from dermal fibroblasts

Steady-state levels of transcripts encoding MCP-1, IL-8, IL-6 and VEGF in fibroblasts exposed to basal medium with and without added FVIIa (100 nmol L−1), FXa (80 nmol L−1) and thrombin (10 nmol L−1) were quantified by RPA and compared with those of serum-starved fibroblasts (Fig. 1 and data not shown). In comparison with basal medium, FVIIa did not significantly induce any of the mRNAs. In contrast, FXa and thrombin induced MCP-1 strongly (2.6- and 3.3-fold, respectively, at 12 h), IL-8 moderately (2.4- and 2.9-fold, respectively, at 12 h) and IL-6 weakly (1.7-fold at 24 h). Note that these values indicating fold increase take into account increases due to basal medium. Neither FXa nor thrombin had an effect on VEGF gene expression (data not shown).

Figure 1.

Effect of coagulation proteases on MCP-1, IL-8 and IL-6 mRNA steady state levels and protein secretion in dermal fibroblasts. Dermal fibroblasts serum-starved for 48 h and incubated with basal medium (BM; ●) or additionally with 100 nmol L−1 FVIIa (▪), 80 nmol L−1 FXa (▴) or 10 nmol L−1 thrombin (▾) were analyzed for (a) MCP-1 (b) IL-8 and (c) IL-6 gene expression by RPA. Total RNA (4–10 µg) isolated from cells following incubation with agonists for 2 h 15 min, 6 h, 9 h, 12 h and 24 h was hybridized to template sets and analyzed as described. Intensity of protected fragments was quantified, normalized to L32 and expressed as fold increase above values quantified from fibroblasts serum-starved for 48 h. A value of 1 on the y-axis represents steady-state levels in serum-starved fibroblasts. Tissue culture supernatants from cells following incubation with agonists for 6 h, 9 h, 12 h and 24 h were analyzed for protein secretion. Quantities of secreted (d) MCP-1 (e) IL-8 and (f) IL-6 (pmol) were determined from duplicate aliquots of supernatants from 15-cm diameter tissue culture dishes for each test condition.

To quantify protein secretion, tissue-culture supernatants from serum-starved dermal fibroblasts incubated with FVIIa, FXa or thrombin in basal medium were aspirated at 6 h, 9 h, 12 h and 24 h. The conditioned media were assayed for MCP-1, IL-8, IL-6 and VEGF by ELISA. MCP-1, IL-8 and IL-6 proteins were secreted in culture supernatants in response to FXa and thrombin, whereas FVIIa failed to stimulate secretion of any of these chemokines above levels quantified in basal medium (Fig. 1). FXa and thrombin increased MCP-1 secretion weakly (1.6-fold above levels in basal medium), IL-8 secretion strongly (4.4- and 4.5-fold, respectively) and IL-6 secretion moderately (2.3- and 2.1-fold, respectively). No significant VEGF secretion in response to any of the proteases was detectable above levels in basal medium (data not shown).

The requirement for a functional active site for FXa and thrombin induction of MCP-1, IL-8 and IL-6 protein secretion was investigated in serum-starved dermal fibroblasts using the covalent active site-inhibited proteases FXai and FIIai. Following incubation for 24 h with FXai or FIIai in basal medium, secretion of MCP-1, IL-8 and IL-6 was equivalent to levels quantified in basal medium alone (Fig. 2a–c). The specificity of MCP-1, IL-8 and IL-6 protein secretion induced by FXa and thrombin was then investigated using the thrombin-specific inhibitor hirudin. Importantly, preincubation of FXa with a 20-fold molar excess of hirudin did not alter protein secretion relative to FXa alone, thus the FXa response could not be attributed to trace amounts of thrombin in the FXa preparation (Fig. 2d–f). Pre-incubation of thrombin with hirudin had the anticipated inhibitory effect on protein secretion. To test the possibility that the FXa was activating cell surface-associated prothrombin, cells were incubated with Ecarin, a snake venom protein that directly activates prothrombin. This did not induce protein secretion, indicating the absence of prothrombin on the surface of these dermal fibroblasts, and confirming the specificity of the FXa response (Fig. 2d–f). Similarly, preincubation of dermal fibroblasts with hirudin for 30 min to inhibit any cell-associated thrombin did not inhibit subsequent protein secretion by FXa.

Figure 2.

Active site requirement and specificity of induction of MCP-1, IL-8 and IL-6 protein secretion by FXa and thrombin. (a–c) Dermal fibroblasts serum-starved for 48 h and incubated with basal medium (BM) alone or additionally with 80 nmol L−1 FXa or FXai, or 10 nmol L−1 thrombin or FIIai for 24 h were assayed by ELISA for secretion of (a) MCP-1 (b) IL-8 and (c) IL-6 proteins. The amount of secreted protein was normalized to estimated cell number and expressed as fmol 10−4 cells with standard deviations calculated from replicate samples. Data from one of two experiments is presented (n = 3). (d–f) FXa and thrombin were preincubated with 200 nmol L−1 hirudin for 30 min at room temperature. Serum-starved dermal fibroblasts seeded in 6-well plates were incubated with the inhibited coagulation protease preparations or the prothrombin activator Ecarin (0.5 U mL−1) for 24 h, or hirudin for 30 min followed by FXa for 24 h. Supernatants from each of six replicates were analyzed by ELISA for (d) MCP-1 (e) IL-8 and (f) IL-6. The amount of secreted protein is expressed as fmol mL−1 per 104 cells. Data from one of three experiments is presented (n = 6).

Differential VEGF secretion in dermal and lung fibroblasts

VEGF was not expressed or secreted by dermal fibroblasts in response to FVIIa, FXa or thrombin (data not shown). Failure to detect secreted VEGF or transcripts in response to FXa and thrombin contrasts with a recent report investigating the effect of these coagulation proteases on VEGF expression in primary lung fibroblasts [8]. Therefore, VEGF secretion in response to FVIIa, FXa and FCS was directly compared in dermal and lung fibroblasts by ELISA (Fig. 3). As previously observed (data not shown), following serum-starvation and incubation with either 100 nmol L−1 FVIIa or 80 nmol L−1 FXa in basal medium for 24 h, VEGF secretion from dermal fibroblasts remained at or below levels quantified in basal medium. VEGF secretion in basal medium by lung fibroblasts was 6.5-fold that of dermal fibroblasts. In addition, lung fibroblasts secreted VEGF in response to FXa 2-fold above the quantity in basal medium, but not in response to FVIIa. In contrast to the differential response of the two cell types to FXa, FCS induced VEGF secretion to similar high levels in dermal and lung fibroblasts. VEGF was not detectable in FCS-supplemented medium prior to incubation with cells (data not shown).

Figure 3.

Comparison of VEGF secretion in dermal and lung fibroblasts in response to FVIIa, FXa and serum. Dermal and lung fibroblasts serum-starved for 48 h were incubated with basal medium alone (BM) or additionally with 100 nmol L−1 FVIIa or 80 nmol L−1 FXa or with complete medium (FCS) for 24 h. Secreted VEGF protein was assayed by ELISA in three replicate samples. The amount of secreted protein was normalized to estimated cell number and expressed as fmol per 1000 cells with standard deviations calculated from replicate samples.

Dermal and lung fibroblasts constitutively express TF

TF is the only known cellular receptor for FVIIa. To determine whether the lack of responsiveness to FVIIa could be explained by the absence of TF, TF antigen was assessed on the surface of dermal fibroblasts by ELISA, and as a surrogate measure of TF activity, FXa generation was measured by chromogenic assay. Serum-starved dermal fibroblasts and those subsequently incubated with complete medium expressed 6.2 ± 0.2 and 11.4 ± 0.1 pmol of TF antigen per mg total cellular protein, respectively (mean ± SD). In the chromogenic assay, the rate of FXa generation under these conditions was 0.3 and 0.7 nmol min−1, respectively.

For the purposes of a direct comparison, equivalent numbers of serum-starved dermal and lung fibroblasts were incubated for 12 h with complete medium. Serum-starved dermal and lung fibroblasts generated 0.15 and 0.13 nmol min−1 FXa, respectively; this increased two-fold to 0.28 and 0.26 nmol min−1, respectively following exposure to complete medium. These data indicate that serum-starved dermal and lung fibroblasts express functional TF constitutively at comparable levels, with similar increases (approximately 2-fold) on serum exposure.

PAR expression

Expression of the four known PARs was investigated in dermal and lung fibroblasts, since these are candidate receptors for coagulation protease signaling to cells. The presence of PAR-1, PAR-2 and PAR-3 transcripts was studied by RPA using total RNA isolated from serum-starved dermal and lung fibroblasts (Fig. 4a). Proliferating HUVECs were included as a positive control for PAR-1 and PAR-2 [34]. Transfer RNA was included as a negative control for the RPA, and as expected, no protected fragments were visible (data not shown). Relative quantities of PAR-1, PAR-2 and PAR-3 steady-state levels were determined following normalization to the housekeeping gene product, L32 (data not shown). PAR-1 transcripts were more abundant than PAR-2 or PAR-3 transcripts in all cell types examined. In addition, PAR-1 transcripts were more abundant in serum-starved lung fibroblasts than in their dermal counterparts, as apparent in Fig. 4(a). PAR-2 transcripts were detectable in HUVECs as reported [34] but not in either fibroblast subtype. PAR-3 transcripts were detectable in serum-starved lung fibroblasts and HUVECs, and following normalization to L32, were calculated to be less than 1% of PAR-1 levels. PAR-3 transcripts were not detectable in serum-starved dermal fibroblasts by this method.

Figure 4.

PAR expression. (a) Total RNA isolated from 48 h serum-starved dermal and lung fibroblasts and proliferating HUVECs (positive control) was analyzed by RPA for expression of PAR-1, PAR-2 and PAR-3. Total RNA (20 µg) isolated from cells was hybridized to the hAngio-2 Human Angiogenesis Multi-Probe Template (Pharmingen) and analyzed as described for Fig. 1. (b) Total RNA isolated from serum-starved dermal and lung fibroblasts and from platelets prepared from PRP (positive control) was subjected to RT-PCR and Southern blot analysis using a PAR-4 cDNA probe. Data from one of two experiments is shown. Dermal ss, serum-starved dermal fibroblasts; Lung ss, serum-starved lung fibroblasts; HUVEC, human umbilical vein endothelial cells.

Expression of PAR-4 mRNA was studied separately by RT-PCR analysis in serum-starved dermal and lung fibroblasts with platelet RNA included as positive control [35]. Southern blot analysis of PCR products yielded hybridization to a fragment of the expected size (330 bp) in control platelet RNA, indicating the presence of PAR-4 (Fig. 4b). PAR-4 was not detected in serum-starved dermal and lung fibroblast RNA.

PAR gene expression was further verified by quantitative RT-PCR in two replicate experiments (Table 2). These data confirm the higher relative abundance of PAR-1 in serum-starved lung fibroblasts compared with their dermal counterparts, and the expression of PAR-3 in lung fibroblasts. Significantly, PAR-2 was detectable at low levels in serum-starved lung but not dermal fibroblasts. This method additionally enabled detection of low levels of PAR-3 in serum-starved dermal fibroblasts. Finally, the absence of PAR-4 in dermal and lung fibroblasts demonstrated by conventional RT-PCR (Fig. 4b) is confirmed by these quantitative RT-PCR experiments. Thus, for dermal fibroblasts the relative abundance of PAR gene expression is PAR-1 > PAR-3 ≫ PAR-2, PAR-4 and for lung fibroblasts, PAR-1 > PAR-3 > PAR-2 ≫ PAR-4. In contrast to an earlier report [15], PAR-4 was detectable in these HUVECs by conventional and quantitative RT-PCR (Table 2 and data not shown).

Table 2.  PAR expression in dermal and lung fibroblasts determined by quantitative RT-PCR. Primer pairs listed in Table 1 were used to determine relative abundance of PARs 1–4 by quantitative RT-PCR. Raw abundance values (copy number) for each gene of interest from dermal and lung fibroblasts were normalized to raw abundance values for GAPDH (a) or GUS? (b). These calculated ratios are expressed relative to HUVECs (arbitrarily designated 1.0) with an accuracy of one decimal place. Data from duplicate experiments (‘a’ and ‘b’) are shown.
Dermal fibroblasts, serum-starved0.30.1000.
Lung fibroblasts, serum-starved2.


Vascular damage following physical injury or resulting from pathophysiological states is a complex condition in which the cellular response to coagulation proteases may be important in regulating inflammation, angiogenesis and wound-healing. In this context, fibroblasts comprise the most abundant cell type in well-vascularized connective tissue, and consequently are frequently exposed to coagulation proteases following vascular damage. Serum, which contains coagulation proteases, has been shown by cDNA microarray analysis to induce specific temporal expression patterns of many genes in human dermal fibroblasts, and a high proportion of these have known roles in processes relevant to the physiology of wound healing [36]. In this study, we examined the ability of the coagulation proteases FVIIa, FXa and thrombin to induce expression, in primary dermal fibroblasts, of four proteins known to be involved in inflammation or angiogenesis: MCP-1, IL-8, IL-6 and VEGF.

Exposure of cells to FVIIa did not increase accumulation of mRNA or secretion of any of the proteins examined in primary dermal fibroblasts. Chromogenic assay for FVIIa confirmed the added protease was proteolytically active in conditioned medium and remained active after exposure to the cells for 48 h (data not shown). The presence of TF on the cell surface is an absolute requirement for FVIIa-induced intracellular signals and gene expression in a variety of cell types, including embryonic lung fibroblast-, pancreatic cancer-, melanoma- and keratinocyte-derived cell lines [2,6,10,19,22,37], macrophages [5] and primary adult lung fibroblasts [38]. The primary dermal fibroblasts used in these experiments constitutively expressed functional TF (6.2 pmol TF antigen mg−1 total cellular protein). One explanation for the lack of response may be that FVIIa does not induce a cellular response directly in these cells; rather, the TF-FVIIa complex generates FXa and/or subsequently thrombin that then induces cellular responses through protease-activated receptors.

FXa and thrombin induced accumulation of MCP-1, IL-8 and IL-6 mRNAs (Fig. 1). Furthermore, these increases in mRNA were accompanied by secretion of the corresponding proteins. MCP-1, IL-8 and IL-6 protein secretion remained close to levels in basal medium when the fibroblasts were incubated with FXai or FIIai (Fig. 2), consistent with data from a number of laboratories demonstrating that the active site is required to elicit cellular responses to coagulation proteases [1,4,6,23]. Furthermore, induction of the chemokines by FXa and thrombin was shown to be specific, since preincubation of each of the coagulation proteases with the thrombin-specific inhibitor, hirudin, abolished only the thrombin response. The FXa response is therefore not due to contaminating thrombin in the FXa. Furthermore, preincubation of cells with hirudin or Ecarin excluded the possibility that the FXa was acting through activation of prothrombin on the surface of these dermal fibroblasts and thus inducing a thrombin response.

Induction of VEGF mRNA or secretion of the corresponding protein was not observed in dermal fibroblasts with any of the coagulation proteases. This contrasts with a recent report describing increased expression of VEGF by lung fibroblasts in response to FXa and thrombin [8]. Direct comparison of these two fibroblast subtypes in the current study demonstrates there is indeed a differential response to FXa: lung but not dermal fibroblasts secrete VEGF (Fig. 3). Furthermore, dermal fibroblasts secrete less VEGF in basal medium controls than lung fibroblasts. Further evidence of a differential response is provided by the reduced proliferative capacity of human dermal fibroblasts in response to FXa relative to their lung counterparts [39]. The differential capacity of the two fibroblast types to secrete detectable VEGF is probably due to the tissue origin and emphasizes the specialized functions of this ubiquitous cell type in diverse tissues [24].

The protease-activated receptors mediate cell signaling by coagulation proteases. Thrombin signaling occurs through activation of PAR-1, PAR-3 and PAR-4 [13]. FXa induces signaling via PAR-1 and PAR-2 [7,40–42] and TF-dependent FVIIa signaling is mediated via PAR-1 and PAR-2 [7]. In addition, effector cell protease receptor-1 (EPR-1) is reportedly involved in some FXa-mediated responses [11,39,40,43] but not others [4,6,44], although the existence of this receptor has recently been questioned [45].

Expression of the PAR family of receptors on dermal and lung fibroblasts was therefore characterized. Both fibroblast subtypes expressed the thrombin receptor PAR-1, with notably higher steady state mRNA levels quantified in lung fibroblasts relative to dermal fibroblasts (Fig. 4 and Table 2). Of the remaining two thrombin receptors, PAR-3 was expressed at low levels in dermal fibroblasts and moderate levels in lung fibroblasts, and PAR-4 was not detectable in either fibroblast subtype. PAR-2, recently shown to be the predominant PAR mediating FXa effects in human and murine endothelial cells in vitro[7,40,42] and isolated rat aorta [44], was detectable at low levels in lung fibroblasts only. These data demonstrate the presence of known receptors for thrombin (PARs 1 and 3, respectively) on dermal fibroblasts. Furthermore, the data show that lung fibroblasts express higher levels of two thrombin receptors, PAR-1 and PAR-3, than their dermal counterparts and additionally express the FXa receptor, PAR-2. Notably, the expression of PAR-1 and PAR-3 only in dermal fibroblasts concurs with the PAR expression pattern observed in human gingival fibroblasts [46].

Our observations suggest that upregulation of VEGF by FXa in lung fibroblasts requires a signal through PAR-2, and that this pathway is inactive in dermal fibroblasts due to the absence of this receptor. Upregulation of MCP-1, IL-8 and IL-6 by FXa in dermal fibroblasts, however, may occur via cleavage of PAR-1. Indeed, comparable FXa concentrations have previously been shown to elicit cellular responses via PAR-1 in HeLa cells, which express only this PAR [41]. Moreover, these FXa cellular responses are the same as those observed for thrombin.

In summary, the coagulation proteases FXa and thrombin elicit concordant responses in MCP-1, IL-8 and IL-6 secretion and mRNA accumulation while FVIIa is not effective. Interestingly, none of the coagulation proteases substantially induce secretion of VEGF in dermal fibroblasts; however, in lung fibroblasts, VEGF is secreted in response to FXa. Relative to dermal fibroblasts, lung fibroblasts express increased levels of PAR-1 and PAR-3 transcripts, and additionally, express PAR-2. This comparison of some characteristics of dermal and lung fibroblasts suggests tissue origin may influence the cellular response of fibroblasts to coagulation proteases. Our verification of human dermal fibroblasts as an appropriate model system will facilitate further characterization of the cellular response to coagulation proteases.


The Medical Research Council of the United Kingdom and the Swiss Science Foundation supported this work. We thank Dr Geoff Kemball-Cook for critical reading of the manuscript and many helpful discussions. Dr Pech is the recipient of a University of Western Australia Medical Research Fellowship.