LPA-induced suppression of periostin in human osteosarcoma cells is mediated by the LPA₁/Egr-1 axis

Abstract

Lysophosphatidic acid (LPA), a naturally occurring bioactive phospholipid, mediates a multitude of (patho)physiological events including activation of mitogen-activated protein kinases (MAPKs). As LPA may induce cellular reponses in human osteosarcoma, the present study aimed at investigating expression of various LPA receptors, LPA-mediated activation of MAPK via G-protein coupling, and expression of early response genes in a cellular model for human osteosarcoma. We show that MG-63 cells express three members of the endothelial differentiation gene (Edg) family of G-protein coupled receptor transcripts (LPA_1–3) but only two (LPA_4/5) out of three members of the non-Edg family LPA receptor transcripts. Stimulation of MG-63 cells with LPA or synthetic LPA receptor agonists resulted in p42/44 MAPK phosphorylation via LPA₁–LPA₃ receptors. Using pharmacological inhibitors, we show that LPA-mediated phosphorylation of p42/44 MAPK by LPA receptor engagement is transmitted by G_αi-dependent pathways through the Src family of tyrosine kinases. As a consequence, a rapid and transient upregulation of the zinc finger transcription factor early growth response-1 (Egr-1) was observed. Egr-1 expression was strictly mediated via G_αi/Src/p42/44 MAPK pathway; no involvement of the G_αq/11/PLC/PKC or the PLD/PI3 kinase/Akt pathways was found. LPA-induced expression of functional Egr-1 in MG-63 cells could be confirmed by electrophoretic mobility shift assay. LPA-induced Egr-1 upregulation was accompanied by a time-dependent decrease of periostin (previously called osteoblast-specific factor 2), a cell adhesion protein for pre-osteoblasts. Silencing of LPA₁ and/or Egr-1 in MG-63 cells reversed LPA-mediated suppression of periostin. We here demonstrate a crosslink between Egr-1 and periostin in cancer cells, in particular in human osteosarcoma.

Highlights

► LPA induces p42/44 MAPK activation in MG-63 cells via LPA₁-LPA₃ receptor. ► G_αi-dependent activation through the Src family of tyrosine kinases induces functional Egr-1 expression. ► LPA-induced Egr-1 upregulation was accompanied by a time-dependent decrease of periostin. ► Silencing of LPA₁ and/or Egr-1 in MG-63 cells reversed LPA-mediated suppression of periostin.

1. Introduction

Human osteosarcoma represents the most common nonhematologic malignant tumour of bone [1]. Osteosarcoma coincides with a period of rapid bone growth suggesting a correlation between bone growth and development of this tumour. Like all sarcomas in young patients, osteosarcoma begins to metastasize very early in its development. Recent investigations exhibited that human osteosarcomas are characterized by a variety of generic alterations resulting in inactivation of numerous tumour suppressor genes and overexpression of oncogenes [2]. Many osteosarcomas appear to harbour genetic lesions, such as INK4A deletion [3], that inactivate both, retinoblastoma and p53 pathways leading to dysfunction of cell cycle control. Oncogenes unrelated to retinoblastoma and p53 pathways, e.g. Myc [4] are overexpressed or activated in a proportion of osteosarcoma.

Lysophosphatidic acid (LPA), a naturally occurring phospholipid, mediates a multitude of (patho)physiological events [5]. LPA induces growth-factor-like responses, e.g. cell proliferation, survival and migration in most normal and transformed cell types which are concordant with many of the “hallmarks of cancer” [6]. At the cellular level, LPA-induced metabolic responses are mediated via G-protein coupled receptors and the broad spectrum of cellular and biological actions of LPA is achieved by engagement of LPA receptor subtypes 1–6 (LPA_1–6). While LPA₁–LPA₃ represent members of the endothelial differentiation gene (Edg) family of G-protein coupled receptors, LPA_4–6 are members of the non-Edg family of LPA receptors [7,8]. The most widely expressed LPA receptor subtype is LPA₁ [9] and functional importance has been demonstrated in lpa1^−/− mice, which show weight loss and neonatal mortality [10]. Compared to LPA₁, LPA₂ and LPA₃ exhibit restricted distribution but are suggested to contribute to the pathophysiology of cancer [11–15]. LPA₄ and LPA₅ are expressed at low levels in most tissues [16,17].

The major signalling cascades evoked by LPA are (i) G_q-mediated activation of phospholipase C (PLC) leading to Ca²⁺ mobilization, (ii) G_αi-mediated inhibition of adenylate cyclase, activation of mitogen-activated protein kinase (MAPK) pathway(s) and protein kinase B (PKB)/Akt pathway or (iii) G_α12/13-mediated activation of RhoA [18,19]. Activation of these pathway(s) defines specific cellular responses such as proliferation, migration and gene transcription, including expression of early growth response-1 (Egr-1) [20,21]. Egr-1 belongs to the group of early response genes, being expressed rapidly after stimulation with growth factors, hormones and neurotransmitters and has been shown to play important roles in bone formation [22]. Furthermore, Egr-1 is thought to be a convergence point to couple extracellular signals to long-term responses by altering gene expression of Egr-1 target genes such as c-Myc, c-Fos and growth factor receptors, respectively [23].

The aim of the present study was to investigate expression of various LPA receptors as well as LPA-mediated G-protein coupled downstream signalling in human MG-63 osteosarcoma cells. MAPK cascades are key signalling pathways involved in the regulation of normal cell proliferation, and aberrant regulation of the MAPK cascade contributes to induction of early response genes in cancer [24]. MAPKs include three different pathways where the extracellular signal regulated protein kinases (also known as p42/44 MAPK) have been the subject of intense research scrutiny [25]. We were interested whether LPA-induced signal transduction pathway(s) occurs via the MAPK cascade and may lead to functional expression of Egr-1 as a basis for its involvement in bone biology. We demonstrate that LPA stimulates rapid and transient synthesis of Egr-1 on mRNA and protein level via G-proteins of the G_αi subtype. This activation is dependent on p42/44 MAPK while no alterations of certain growth factor receptors were found [26]. Finally, we show that Egr-1 inhibits the expression of the extracellular matrix protein periostin in human MG-63 osteosarcoma cells in response to LPA-induced signalling cascades.

2. Materials and methods

2.1. Reagents

Foetal calf serum (FCS) and trypsin/EDTA were from PAA (Linz, Austria). Gentamycin, α-minimum essential medium (α-MEM), NuPAGE^® 10% Bis-Tris Gels, NuPAGE^® MES SDS running buffer, NuPAGE^® LDS sample buffer, NuPAGE^® sample reducing agent, SeeBlue^® Plus prestained standard, nitrocellulose membranes (0.45 μm pore size) and DNase-I were from Gibco Invitrogen (Lofer, Austria). LPA (1-oleoyl-2-hydroxy-sn-glycero-3-phosphate) was from Avanti^® Polar Lipids (Alabaster, AL, USA). Pertussis toxin (PTX), manumycin A, (S)-1-oleoyl-2-O-methyl-glycerophosphothionate triethylammonium salt (OMPT), N-palmitoyl-l-serine phosphoric acid (l-NASPA), and Ki16425 were from Sigma–Aldrich (Vienna, Austria). Gö6983, LY294002, PD98059, SQ22536, SB203580, JNK inhibitor II, PP2, Akt inhibitor, tyrphostins (AG1295, AG183, AG1478) and toxin B were from Calbiochem (La Jolla, CA, USA). U73122, propranolol, dodecylphosphate (DDP) and rp-cAMPS were from Biomol (Hamburg, Germany). Cell permeable Rho inhibitor exoenzyme C3 transferase was from Cytoskeleton (Denver, CO, USA). BCA™ protein assay, SuperSignal^® West Pico chemiluminescent substrate and NE-PER^® nuclear and cytoplasmic protein extraction kit were from Pierce Biotechnology (Rockford, IL, USA). Complete Mini protease inhibitor cocktail tablets were from Roche Applied Science (Vienna, Austria). Hyperfilm™ MP and [γ-³²P]ATP (3000 Ci/mmol) were from GE Healthcare (Vienna, Austria). QIAGEN^® OneStep RT-PCR kit and RNeasy^® RNA purification kit were from QIAGEN (Hilden, Germany). All primer pairs (Supplement, Table I) were from TibMolBiol (Berlin, Germany).

2.2. Cell culture

The human osteosarcoma cell line MG-63 (ATCC, Manassas, VA, USA) was maintained in α-MEM, supplemented with 5% (v/v) FCS, 50 μg/ml (w/v) gentamycin and 2 mM l-glutamine (37 °C, humidified atmosphere of 5% CO₂). Cells were used between 10th and 20th passage counted from the day of receipt. Cells were cultured in 75 cm² tissue culture flasks and seeded prior to treatment in six-well plates (Corning Incorporated, New York, NY, USA) at a seeding density of 10,000 cells/cm² in a final volume of 4 ml α-MEM. After 3 days, the medium was changed to 4 ml of fresh α-MEM (with FCS) and experiments were performed on day 6.

2.3. Cell culture experiments

For time-dependent activation of the p42/44 MAPK pathway, cells were incubated up to 45 min in medium containing indicated LPA concentrations. Time-dependent expression of early response genes was performed by stimulation of cells up to 3 h (mRNA expression) and 6 h (protein expression). For concentration-dependent activation of p42/44 MAPK, the cells were incubated in medium containing either subtype-specific LPA receptor agonists (l-NASPA, DDP, OMPT) or LPA at indicated concentrations for 5 min. Investigating signal transduction pathways, cells were incubated for 30 min in medium containing the LPA receptor antagonist Ki16425 (10 μM) and/or various inhibitors of signal transduction: manumycin A (10 μM), Gö6983 (500 nM), PD98059 (25 μM), SB203580 (10 μM), JNK inhibitor II (1 μM) PP2 (10 μM), U73122 (10 μM), propranolol (100 μM), SQ22536 (25 μM), rp-cAMPS (25 μM), LY294002 (25 μM), Akt inhibitor (25 μM), tyrphostins AG1478, AG183, and AG1295 (all at 10 μM); PTX (100 ng/ml), cell permeable C3 transferase (5 μg/ml) and toxin B (100 ng/ml) were used as inhibitors of G-protein signalling. All signal transduction inhibitors [27] were dissolved according to manufacturer's suggestions.

2.4. Protein isolation and Western blot analysis

Confluent cells were treated as indicated by addition of various pharmacological agonists or antagonists directly into the medium. Experiments were terminated by aspiration of the medium and cells were washed twice with chilled PBS (pH 7.4). Cell lysis was performed on ice for 10 min with 100 μl lysis buffer (HEPES 50 mM, NaCl 150 mM, EDTA 1 mM, Na₄P₂O₇ 10 mM, Na₃VO₄ 2 mM, NaF 10 mM, Triton X-100 1% (v/v), glycerol 10% (v/v), Complete Mini protease inhibitor; pH 7.4). Insoluble cell debris was removed by centrifugation at 11,000 g (4 °C; 10 min). Protein content of cell lysates was determined using the BCA™ protein assay according to the manufacturer's suggestions. Aliquots of cell lysates (25–50 μg protein) were diluted with an equal volume of NuPAGE^® LDS sample buffer and supplemented with NuPAGE^® sample reducing agent (5% [v/v]). After heating for 10 min at 70 °C samples were subjected to electrophoresis on 4–12% gradient SDS-PAGE gels. Proteins were transferred to nitrocellulose membranes (2 h, 80 V). The following primary antibodies were used: (i) mouse monoclonal anti-phosphorylated p42/44 (pp42/44) MAPK (dilution 1:1000, Cell Signaling Technology, Beverly, MA, USA); (ii) rabbit polyclonal anti-Egr-1 (clone C-19, dilution 1:1000, Santa Cruz Biotechnology, Santa-Cruz, CA, USA); (iii) rabbit polyclonal anti-periostin (clone H-300, dilution 1:2000, Pierce Biotechnology, Rockford, IL, USA); (iv) rabbit polyclonal anti-epidermal growth factor receptor (EGFR, clone 1005, dilution 1:1000, Santa Cruz Biotechnology); (v) rabbit polyclonal anti-platelet derived growth factor receptor α (PDGFRα, clone C-20, dilution 1:1000, Santa Cruz Biotechnology); (vi) and rabbit polyclonal anti-PDGFRβ (clone P-20, dilution 1:1000, Santa Cruz Biotechnology). Detection of immunoreactive bands was performed with horse-radish-peroxidase (HRP)-conjugated goat anti-mouse IgG (dilution 1:10,000, Rockland, Gilbertsville, PA, USA) or HRP-conjugated goat anti-rabbit IgG (dilution 1:10,000, Jackson ImmunoResearch Laboratories, Inc., Austria) followed by incubation with the SuperSignal^® West Pico chemiluminescent substrate. For normalization, the membranes were stripped (using an antibody stripping buffer [Gene Bio Application, Kfar Hanagid, Israel]) and reprobed with rabbit polyclonal anti-p42/44 MAPK (anti-ERK-1, clone K-23, dilution 1:1000) or mouse monoclonal anti-β-actin (clone C4, dilution 1:1000, Santa Cruz Biotechnology) as primary antibodies [28].

2.5. RNA isolation and reverse transcriptase polymerase chain reaction (RT-PCR)

Cells were cultured in 6-well plates and no medium exchange was performed within 60 h prior to the experiments. After reaching confluence, cells were treated as indicated and washed twice with PBS (pH 7.4). Total RNA was extracted using RNeasy^® RNA purification kit, treated with DNase-I and RNA content was determined by measuring the absorbance at 260 nm. RT-PCR was performed with QIAGEN^® OneStep RT-PCR kit according to the manufacturer's suggestions including 100 ng template RNA and gene-specific primers at final concentrations of 0.6 μM. The reverse transcription reaction was performed at 42 °C for 1 h, followed by an initial PCR activation step for 15 min at 95 °C, 10 s at 94 °C, 30 s at indicated annealing temperature (Supplement, Table I), 30 s at 72 °C and a final elongation at 72 °C for 7 min. Primer sequences, annealing temperatures, number of amplification cycles and amplicon size for each RT-PCR are listed in Supplement, Table I. RT-PCR products were separated on 1.5% agarose gels supplemented with 0.1% (v/v) ethidium bromide and visualised using a UV transilluminator (Herolab; Heidelberg, Germany). To ensure equal RNA loading, RT-PCR for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was performed for each experiment [29,30].

2.6. Electrophoretic mobility shift assay (EMSA)

Confluent cells were treated as indicated above. Reactions were terminated by aspiration of the medium. Cells were washed twice with chilled PBS (pH 7.4), harvested by scraping with a rubber scraper and pelleted by centrifugation at 500 g (4 °C; 3 min). Non-denatured, active nuclear proteins were isolated using NE-PER^® extraction reagents including Complete Mini protease inhibitors according to the manufacturer's suggestions. Protein concentrations were determined using the BCA™ protein assay kit. Nuclear extracts were aliquoted and stored at −70 °C until use. Nucleotide sequences of the oligonucleotides containing an Egr-1 consensus binding sequence as reported by Santa Cruz Biotechnology (SC-2529) are 5′-GGATCCAGCGGGGGCGAGCGGGGGCGA-3′ and 3′-CCTAGGTCGCCCCCGCTCGCCCCCGCT-5′. Complementary oligonucleotides were annealed to form double-stranded DNA followed by radioactive labelling with [γ-³²P]ATP using T4 Polynucleotide kinase and removal of non-incorporated nucleotides via Micro Bio-Spin-6 Chromatography Columns [31]. To perform EMSA binding reaction, 5 μg of nuclear extracts were incubated (10 min, 25 °C) with binding buffer (10 mM Tris, 50 mM NaCl, 1 mM dithiothreitol, 0.1 mM EDTA, 5% (v/v) glycerol, pH 7.5) containing 0.25 μg poly [d(I–C)] and 5 μg bovine serum albumin. Labelled oligonucleotides were added to each reaction mixture and incubated for 20 min at 25 °C. After addition of glycerol to a final concentration of 20% (v/v), samples were separated by electrophoresis on a 3.9% PAGE (acrylamide:bisacrylamide, 29:1, v/v) in 22.5 mM Tris boric acid buffer (0.5 mM EDTA, pH 8.3) at 25 °C (3.5 h, 120 V). The gel was dried under vacuum (80 °C, 1.5 h) and exposed to Hyperfilm™ MP Film for 12–24 h. For competition experiments, one μl of non-radiolabelled Egr-1 oligonucleotides (20-fold molar excess) was added to the binding mixture before addition of the radiolabelled probe.

2.7. siRNA experiments

MG-63 cells were transfected with siRNAs specific for Egr-1 (catalogue number: SI00030709, Qiagen) or LPA₁ (Edg-2, Santa Cruz: catalogue number: sc-43746) or with a scrambled negative control siRNA (catalogue number: 1022076, Qiagen). The siRNA transfection was performed with oligofectamine^® (Invitrogen) according to the manufacturer's suggestions. Briefly, cells were plated at a density of 5 × 10⁴ in 6-well plates and grown overnight to 40–60% confluence. Cells were transfected in 1 ml MEM without FCS containing 4 μl of oligofectamine and 50 nM of the respective siRNA. As a mock control, cells were treated with the transfection reagent alone [32]. After an incubation period of 6 h at 37 °C (5% CO₂), the transfection mix was removed and replaced by the standard growth medium (MEM, 5% FCS). Then, cells were cultivated for further 66 h followed by stimulation with indicated LPA concentrations. After indicated time periods total cell protein was isolated to determine the expression level of Egr-1, periostin and β-actin using specific antibodies.

All experiments listed above were performed at least three times.

3. Results

3.1. LPA receptor subtypes in human MG-63 cells

To identify LPA receptor subtypes, specific primer pairs were used (Supplement, Table I). After RT-PCR analysis, separating and sequencing of the corresponding PCR products, we show that transcripts for LPA_1–5 are expressed in MG-63 cells (Fig. 1). No transcripts for the recently identified LPA₆ (P2Y₅ [33,34]) were found under our experimental conditions (data not shown).

Fig. 1.

RT-PCR of LPA receptors in human MG-63 cells: cells were cultured in α-MEM until confluence, RNA was isolated and LPA_1–5 receptor transcripts were amplified by RT-PCR using specific forward and reverse oligonucleotide primer pairs (Supplement, Table I). The corresponding RT-PCR products were separated on 1.5% agarose gels.

3.2. Activation of the MAPK pathway in human MG-63 cells

As LPA receptor signalling may be coupled with MAPK activation in cancer cells, we investigated the involvement of endogenously expressed LPA receptor subtypes in MG-63 cells. Due to the lack of commercially available specific agonists for LPA_4–5, only members of the Edg family were tested. Concentration-dependent activation of MG-63 cells with different LPA-receptor agonists revealed pronounced phosphorylation of p42/44 MAPKs with l-NASPA (LPA₁) and OMPT (LPA₃) while the effect of DDP (an LPA₂ agonist) on p42/44 MAPK phosphorylation was less pronounced (Fig. 2A). From these experiments, we conclude that preferentially LPA₁ and LPA₃ are operative to promote phosphorylation of p42/44 MAPK.

Fig. 2.

Western blot of LPA- and synthetic LPA receptor agonist-induced concentration- or time-dependent phosphorylation of p42/44 MAPK in human MG-63 cells: (A) cells were stimulated for 5 min with l-NASPA (LPA₁), DDP (LPA₂) and OMPT (LPA₃) at indicated concentrations (b = blank [non-stimulated cells], c = control [cells were stimulated with 20 μM LPA]). (B–C) cells were incubated in medium containing (B) 20 μM LPA for indicated time periods or with (C) indicated concentrations of LPA for 5 min. Cells were lysed, aliquots of protein lysates were subjected to SDS-PAGE and transferred to nitrocellulose. Phospho-specific mouse monoclonal anti-p42/44 (dilution 1:1000) was used as a primary antibody. For normalization, membranes were stripped and reprobed with rabbit anti-p42/44 (dilution 1:1000). Immunoreactive bands were visualized with HRP-conjugated secondary antibodies (dilution 1:10,000) using the Super signal system.

Naturally occurring LPA is effective to promote phosphorylation of p42/44 MAPK at comparable concentrations as shown for the synthetic agonists l-NASPA and OMPT (Fig. 2A). Therefore, LPA was used as an agonist in all ongoing experiments. Time-dependent stimulation of MG-63 cells with LPA resulted in a rapid and transient phosphorylation of p42/44 MAPK, reaching a maximum activation at approx. 5 min after stimulation (Fig. 2B). Fig. 2C shows that phosphorylation of p42/44 MAPK in response to LPA is concentration-dependent.

3.3. Activation of the MAPK pathway in human MG-63 cells via G_αi-protein

Next, specific toxins were used to reveal the G-protein coupling involved in LPA-mediated p42/44 MAPK phosphorylation. PTX (known to directly inactivate the G_αi subunit of heterotrimeric type of G-proteins) completely inhibited LPA-induced p42/44 MAPK phosphorylation (Fig. 3A). To investigate a putative role for small monomeric G-proteins downstream of G_αi, cell permeable C3 exoenzyme (inhibiting the Rho family of small G proteins) and toxin B (inhibiting in addition small GTPases such as Rac and Cdc42) were used. No involvement of these small monomeric G-proteins in LPA-mediated p42/44 MAPK phosphorylation was found (Fig. 3A).

Fig. 3.

Western blot of LPA-induced phosphorylation of p42/44 MAPK in human MG-63 cells in the presence of bacterial toxins or pharmacological inhibitors: cells were incubated overnight with (A) PTX (200 ng/ml), C3 exoenzyme (5 μg/ml), toxin B (100 ng/ml) for 15 min with (B) Ki16425 (10 μM), SQ22536 (25 μM), rp-cAMPS (25 μM), PP2 (10 μM) and for (C) 15 min with PD98059 (25 μM), SB203580 (25 μM), JNK inhibitor II (1 μM) before stimulation with 20 μM LPA for 5 min. Cells were lysed, aliquots of protein lysates were subjected to SDS-PAGE and immunoreactive phosphorylated and total p42/44 MAPK bands were detected as described in Fig. 2; (blank = non stimulated cells; control = cells stimulated with 20 μM LPA).

To study signalling events downstream of G_αi, MG-63 cells were pretreated with various inhibitors of key molecules in signal transduction pathway(s). The LPA receptor antagonist Ki16425 [35] completely inhibited p42/44 phosphorylation indicating that stimulation of this MAPK pathway is specifically related to LPA receptor activation (Fig. 3B). Involvement of the PKA pathway could be excluded as neither SQ22536 (an inhibitor of adenylate cyclase) nor rp-cAMPS (a cAMP antagonist) impaired p42/44 MAPK phosphorylation (Fig. 3B). Next, a possible involvement of the Src family of proteins was tested. These non-receptor tyrosine kinases can be activated by G-protein-mediated activation of PKC as well as by growth factor receptor-mediated signalling. PP2, a specific Src inhibitor, completely blunted phosphorylation of p42/44 MAPK (Fig. 3B). In contrast to Src, inactivation of the likely downstream effector Ras by manumycin A did not impair LPA-induced p42/44 MAPK phosphorylation (Supplement Fig. IA) indicating that Ras is not involved.

To further investigate involvement of G_αq/11-mediated signalling, MG-63 cells were pre-treated with U73122 or Gö6983. As these inhibitors did not impair LPA-induced phosphorylation of p42/44 MAPK, activation via PLC and PKC can further be excluded (Supplement Fig. IB). Propranolol and LY294002 (Supplement Fig. IB) as well as the Akt inhibitor (Supplement Fig. IA) had also no effect on p42/44 MAPK phosphorylation, the PLD/PI3 kinase/Akt axis is obviously also not involved in LPA-mediated activation of p42/44 MAPK in MG-63 cells.

As PKC is not contributing to p42/44 MAPK phosphorylation (Supplement Fig. IB), a possible involvement of Src induction via transactivation of growth factor receptor tyrosine kinases was tested. Treatment of MG-63 cells with tyrphostins AG183 and AG1478 (two inhibitors of the EGFR) and AG1295 (an inhibitor of the PDGFR) had no effect (Supplement, Fig. II); this indicates that transactivation of EGFR and/or the PDGFR is not operative in LPA-induced p42/44 MAPK activation. Time-dependent incubation of MG-63 cells with LPA isolation of RNA and subsequent RT-PCR analysis for EGFR and PDGFRα/β revealed that LPA treatment did not alter expression of these receptors on mRNA level (Supplement Fig. IIIA). These findings were parallelled on the protein level when cells were incubated with LPA up to 6 h (Supplement Fig. IIIB).

To confirm that LPA specifically activates the p42/44 MAPK cascade, different inhibitors of the MAPK pathway were used. As expected SB203580 (inhibitor of p38 MAPK) and the JNK inhibitor II (inhibitor of SAPK/JNK) had no inhibitory effect on LPA-induced p42/44 MAPK phosphorylation. In contrast, PD98059 (an inhibitor of the MAPK kinase upstream of p42/44) completely abrogated immunoreactive pp42/44 signal (Fig. 3C).

3.4. G_αi and p42/44 MAPK-dependent functional expression of Egr-1 in human MG-63 cells

Stimulation of the p42/44 MAPK pathway may result in a multitude of events, including expression of early response genes e.g. proto-oncogenes c-Myc and Egr-1. We here show that c-Myc on mRNA level is constitutively expressed in MG-63 cells in response to LPA (Fig. 4A). In contrast, LPA induced a rapid and transient expression of Egr-1 on mRNA level reaching a maximum after 30 min; Egr-1 mRNA levels declined to basal levels after 120 min (Fig. 4A). Time-dependent expression of Egr-1 on the protein level revealed maximum expression in between 1 and 3 h after LPA stimulation (Fig. 4B).

Fig. 4.

LPA-induced time-dependent expression of early response genes in human osteosarcoma cells: time-dependent expression of (A) c-Myc and Egr-1 on mRNA level and (B) Egr-1 on protein level after stimulation of MG-63 cells with 20 μM LPA at indicated time periods. RNA was isolated and amplification was performed by RT-PCR using specific forward and reverse oligonucleotide primer pairs (Supplement, Table I). PCR products were separated on 1.5% agarose gels. To ensure equal gel loading, control GAPDH RT-PCR was performed. (B) For Western blot experiments, cells were lysed, protein aliquots were subjected to SDS-PAGE and transferred to nitrocellulose. Rabbit polyclonal anti-Egr-1 (dilution 1:1000) was used as a primary antibody. Normalization was performed with mouse monoclonal anti-β-actin (dilution 1:1000).

Next, MG-63 cells were preincubated with bacterial toxins or other inhibitors prior to LPA stimulation and both RT-PCR (Fig. 5, upper panel) and Western blot experiments were performed (Fig. 5, lower panel). The cell permeable C3 and toxin B had no effect while decreased Egr-1 expression was found in PTX-treated cells. These findings suggest involvement of G_αi but not of small GTPases (Rho, Rac and Cdc42) in LPA-induced Egr-1 expression. The LPA receptor antagonist Ki16425 and the Src kinase inhibitor PP2 impair Egr-1 expression on mRNA and protein level in response to LPA while SQ22536 and rp-cAMPS had no effect (Fig. 5B). All other inhibitors (as already mentioned in Supplement Fig. I) failed to alter Egr-1 expression (Supplement Fig. IVA–D). In general, expression of Egr-1 on mRNA level could be verified on the protein level (Fig. 5A, B) indicating that Egr-1 expression is regulated at the transcriptional as well as at the translational level similarly. Only those inhibitors (PTX, Ki16425, and PP2) being effective to suppress immunoreactive pp42/44 signal (Fig. 3) were also effective to impair Egr-1 expression on RNA and protein level (Fig. 5A, B).

Fig. 5.

RT-PCR and Western blot of LPA-induced expression of Egr-1 in human MG-63 cells: cells were incubated overnight with (A) PTX (200 ng/ml), C3 exoenzyme (5 μg/ml), toxin B (100 ng/ml) or (B) for 15 min with Ki16425 (10 μM), SQ22536 (25 μM), rp-cAMPS (25 μM), PP2 (10 μM) or (C) for 15 min with PD98059 (25 μM), SB203580 (25 μM), JNK inhibitor II (1 μM) followed by stimulation with 20 μM LPA for 30 min (mRNA) or 2 h (protein). Upper panel: RNA was isolated and amplification was performed by RT-PCR using specific forward and reverse oligonucleotide primer pairs. PCR products were separated on 1.5% agarose gels. To ensure equal gel loading, control GAPDH RT-PCR was performed. Lower panel: the cells were lysed and aliquots of protein lysates were subjected to SDS-PAGE and transferred to nitrocellulose as described in Materials and methods. Rabbit polyclonal anti-Egr-1 (dilution 1:1000) was used as a primary antibody. For normalization, the membranes were stripped and reprobed with mouse monoclonal anti-β-actin (dilution 1:1000); (blank = non stimulated cells; control = cells stimulated with 20 μM LPA).

To confirm that expression of Egr-1 is mediated via the p42/44 MAPK axis, inhibitors of p42/44 and p38 MAPK or SAPK/JNK were tested. Only PD98059 was effective to impair Egr-1 expression on mRNA and protein level in response to LPA; SB203580 and the JNK inhibitor II had no effect (Fig. 5C).

3.5. DNA binding activity of Egr-1

To test Egr-1 at the functional level, MG-63 cells were stimulated with LPA to induce Egr-1 DNA binding activity. Nuclear proteins were examined by EMSA using a specific probe for Egr-1. LPA stimulation of cells led to a time-dependent increase in Egr-1 binding after 30 min (Fig. 6). Competition experiments with an excess of non-radiolabelled probe confirmed the identity and specificity of the EMSA band. These experiments led us to conclude that LPA induced an increased binding of Egr-1 to the promotor region of target genes in MG-63 cells.

Fig. 6.

DNA-binding activity of the transcription factor Egr-1 in LPA-treated human MG-63 cells: confluent cells were stimulated with LPA (20 μM) at indicated time periods and nuclear protein extracts were prepared and analysed by EMSA as described under Materials and methods. Specificity of the binding pattern was determined by competition with excess unlabelled oligonucleotides (oligos). Nucleotide sequences of the oligonucleotides used are listed in methods.

3.6. Involvement of LPA₁ in LPA-induced suppression of periostin is mediated by Egr-1

Next, expression of periostin, a protein known to be involved in both cancer-related events and osteogenic processes [36,37] was investigated. Under non-stimulated conditions, periostin is abundantly expressed in MG-63 cells (Fig. 7A). However, time-dependent stimulation of MG-63 cells with LPA led to a dramatic decrease of periostin after 6 h. Only trace amounts of periostin were found in between 12 and 48 h after LPA treatment.

Fig. 7.

(A) Western blot of LPA-stimulated periostin in human MG-63 cells in the absence or presence of (un)specific siRNA for Egr-1: (A) time dependent decrease of periostin levels after stimulation of MG-63 cells in medium with 20 μM LPA. (B) Cells were transfected with siRNA specific for LPA₁ (si-LPA1) and (C) with siRNA for Egr-1 (si-Egr-1) or with a scrambled negative control siRNA (si-scr) as described in Materials and methods. MG-63 cells were stimulated with 20 μM LPA for 2 h (C, Egr-1 [upper panel]) or 12 h (periostin, B and C [lower panel]), respectively; (blank = non stimulated cells; control = cells stimulated with 20 μM LPA). Cells were lysed, aliquots of protein lysates were subjected to SDS-PAGE and transferred to nitrocellulose as described in Materials and methods. Rabbit polyclonal anti-Egr-1, anti-β-actin or rabbit polyclonal anti-periostin (recognizing an immunoreactive 90 kD band) were used as primary antibodies (all used at dilutions of 1:1000).

To reveal whether LPA₁, the most widely expressed LPA receptor subtype [7], is involved in downregulation of periostin, cells were preincubated with specific or scrambled siRNA for LPA₁. Only the specific siRNA reversed the effect of LPA on periostin suppression (Fig. 7B).

The next set of experiments aimed to reveal a possible cross-link between LPA-induced upregulation of Egr-1 and suppression of periostin. Using a highly specific siRNA for Egr-1 which suppressed LPA-induced expression of Egr-1 but prevented LPA-stimulated downregulation of periostin, the involvement of the transcription factor Egr-1 in periostin metabolism is demonstrated (Fig. 7C).

4. Discussion

LPA, an important signalling molecule, exerts its effects by specific binding to G-protein coupled receptors [38]. In MG-63, a human osteogenic sarcoma cell line [39], LPA enhances fibronectin binding to adherent cells [40] and stimulates fibronectin matrix assembly through a Rho-dependent pathway [41]. Stimulation of MG-63 cells with LPA leads to G_αi-dependent stress fibre formation, activation of the MAPK cascade [42,43], increased proliferation [44] and osteoblast maturation in conjunction to the Vitamin D metabolite 1α, 25(OH)₂D₃ [45]. Expression of LPA₁ and LPA₂ has been reported in MG-63 cells [43,44,46] and G_αi-coupling of both receptors is apparently responsible for LPA-mediated phosphorylation of p42/44 MAPK. We now show expression of additional LPA receptor transcripts, namely another member of the Edg family (LPA₃) and two members of the non-Edg family of G-protein coupled receptors (LPA_4/5) (Fig. 1). As these LPA receptors may couple via G_αq and G_α12/13 [16,47,48] we tested whether LPA-induced signalling pathway(s) leading to p42/44 MAPK activation is strictly related to G_αi- or if G_αq- and/or G_α12/13-mediated signalling is involved.

LPA induces osteoblastic differentiation and a lack of LPA₁ or its inhibition by Ki16425 results in abnormal bone development [49] or completely abrogates osteogenesis [50]. Stimulation with specific agonists for the Edg family of G-protein coupled receptor (LPA₁–LPA₃) led to activation of p42/44 MAPK and demonstrates functional coupling of these receptor subtypes. Most importantly, p42/44 MAPK activation was inhibited by Ki16425 a subtype selective antagonist for members of the Edg family of G-protein coupled receptors in the following order: LPA₁ ≥ LPA₃ » LPA₂ [35]. As LPA-induced phosphorylation of p42/44 MAPK was completely suppressed by PTX (Fig. 3A), it is likely that LPA_1–3 are functionally coupled to the G_αi-subtype of G-proteins with no involvement of G_αq and G_α12/13 [43]. Due to the lack of specific agonists for members of the non-Edg family (LPA₄/LPA₅), activation of the p42/44 MAPK signalling via these receptors was not investigated. LPA₄, recently reported to inhibit osteogenesis [50], might also be involved in stimulation of the MAPK cascade due to G_αi coupling. By now LPA₅ is described to couple G_αq and G_α12/13 [16], so it is unlikely that this receptor subtype participates in LPA-induced p42/44 MAPK signalling in MG-63 cells.

We show that downstream of G_αi the Src family of protein tyrosine kinases, but not Ras, is required to phosphorylate p42/44 MAPK in response to LPA. A similar signalling route has been proposed for nontransformed murine osteoblast-like MC3T3-E1 cells [51]. In contrast to MC3T3-E1 cells, MG-63 osteosarcoma cells exhibited no involvement of G_αq and PKC-mediated induction of p42/44 MAPK. Also no requirement of the Rho family of small GTPases, as shown in ovarian theca cells [52] or cdc42 and Rac, could be observed in MG-63 cells.

The transcription factor Egr-1 is a regulator for activation of several genes involved in cell proliferation and a role in growth control was proposed as Egr-1 expression and gene transcription was observed in many cell types stimulated with mitogens [53–56]. Egr-1 has also been connected with the development of cancer, as high Egr-1 levels have been found in a majority of prostate cancer [57]. Support for its role in cancer comes from animal models where prostate cancer was impaired in egr-1^−/− mice [58]. Moreover, Egr-1 is suggested to play a role in multistage carcinogenesis in skin [59]. Egr-1 also was described to be involved in growth factor synthesis, e.g. PDGFα/β, insulin-like growth factor II and transforming growth factor-β1; this interaction might represent an autocrine loop as PDGF, via its cognate receptor, can stimulate Egr-1 expression [60–62]. In human U2OS osteosarcoma cells, Egr-1 was shown to be involved in the upregulation of EGFR in response to hypoxia, a condition which might cause enhanced cell proliferation by sensitizing cells upon stimulation with EGF [26]. Here, expression of EGFR, PDGFRα/β on mRNA and protein levels was followed in response to LPA, however, no changes in expression of these growth factor receptors have been observed in MG-63 cells within 6 h of stimulation.

As Egr-1 is assumed to play a role in bone formation [22,63,64] MAPK-dependent expression of Egr-1 in response to LPA was investigated in MG-63 cells. Following LPA-induced expression of Egr-1 on mRNA and protein level, a rapid and transient synthesis of Egr-1, dependent on G_αi and phosphorylation of p42/44 MAPK, was observed. A similar pathway has recently been addressed in rat mesangial cells [21]. Involvement of the Src family of protein tyrosine kinases leading to MAPK activation and Egr-1 expression was also shown in rat osteosarcoma cells in a stimulated microgravity model [65]. We therefore assume that Egr-1 might play an important role in the proliferative action of LPA in MG-63 cells. In contrast to Egr-1, LPA was without effect on expression of c-Myc, a proto-oncogene. C-Myc may be abundantly expressed in human osteosarcoma [4] and is assumed to be upregulated in MG-63 cells after treatment with PDGF, forskolin and 12-O-tetradecanoylphorbol-13-acetate [66].

Egr-1 is suggested to be involved in an array of cellular events during cancer development. Therefore, the present study represents a basis for Egr-1-mediated regulation of relevant target genes in cancer initiation and development [67]. A protein assumed to play a key role in tumour growth and survival, angiogenesis and invasion of cancer cells is periostin [37]. This 90 kDa adhesion molecule, originally named osteoblast-specific factor-2, was first identified in bone, and implicated in regulating adhesion and differentiation of osteoblasts [36]. LPA-induced expression of periostin has been reported in stromal cells [68] and a recent report addressed the role of the LPA–LPA₁ axis in periostin formation in human adipose tissue-derived mesenchymal stem cells [69]. Different expression patterns of periostin on mRNA and protein level have been reported in various human osteosarcoma cell lines (e.g. MHM, KPDXM and Eggen) and expression of periostin mRNA was inversely related to the cell's abilities to differentiate and mineralize [70]. In human U2OS osteosarcoma cells knock down of periostin reduced entry into S-phase, migration, proliferation and invasion [71].

Previous studies have shown that in TNFα-stimulated chondrocytes inhibition of various extracellular matrix genes is directly related to binding of Egr-1 to specific regions in the promoter [67]. Our data demonstrate that periostin is abundantly expressed in human MG-63 osteosarcoma cells. Stimulation of cells with LPA resulted in a decrease of periostin that was strictly dependent on Egr-1 expression levels. Whether this effect is mediated by binding of this transcription factor to a specific Egr-1 response element in the promotor region (925–941 bps upstream of the transcription site) of the periostin gene is currently under investigation.

Taken together, our findings demonstrate expression of LPA receptor subtypes 1–5 in human MG-63 osteosarcoma cells and functional coupling of LPA_1–3 using receptor subtype specific agonists. LPA induced a rapid and transient synthesis of Egr-1 that is strictly mediated via G_αi/Src/p42/44 MAPK pathway with no involvement of the G_αq/11/PLC/PKC or the PLD/PI3 kinase/Akt pathways. Most importantly, we here show that the LPA-LPA₁-Egr-1 axis is tightly coupled with the expression of periostin.

Conflict of interest

None.