Transactivation of the Epidermal Growth Factor Receptor Mediates Parathyroid Hormone and Prostaglandin F2α-Stimulated Mitogen-Activated Protein Kinase Activation in Cultured Transgenic Murine Osteoblasts

Abstract

Recent data suggest that G protein-coupled receptors (GPCRs), including those for PTH and prostaglandins (PGs), contribute to the proliferation and differentiation of osteoblasts in vivo. To understand how these signals are transduced, we studied activation of the ERK1/2 MAPK cascade in cultures of differentiating TMOb murine osteoblasts. In TMOb cells, stimulation of endogenous Gs/Gq-coupled PTH receptors, Gq-coupled PGF2α receptors, and Gi/Gq-coupled lysophosphatidic acid receptors, but not Gs-coupled PGE2 receptors, caused a rapid 5- to 10-fold increase in ERK1/2 phosphorylation. GPCR-stimulated ERK1/2 activation coincided with increased tyrosine phosphorylation of epidermal growth factor (EGF) receptors and was blocked by the EGF receptor inhibitor, tyrphostin AG1478, and the metalloprotease inhibitor, batimastat, suggesting that the response involved transactivation of EGF receptors through the proteolytic release of an EGF receptor ligand. To further examine the mechanism of PTH-stimulated EGF receptor transactivation, we employed COS-7 cells expressing the rat PTH receptor. Here, stimulation with PTH(1–34) caused proteolysis of hemagglutinin epitope-tagged heparin binding-EGF, increased tyrosine autophosphorylation of EGF receptors, and AG1478-sensitive ERK1/2 activation. When PTH receptor-expressing COS-7 cells were placed in a mixed culture with cells lacking the PTH receptor but expressing a green fluorescent protein-tagged ERK2, stimulation with PTH(1–34) induced phosphorylation of green fluorescent protein-ERK2 that was abolished by either batimastat or tyrphostin AG1478. These data suggest that autocrine/paracrine cross-talk between EGF receptors and Gi- or Gq/11-coupled GPCRs represents the predominant mechanism of GPCR-mediated activation of ERK1/2 in cultured TMOb osteoblasts.

OVER THE PAST decade, it has become increasingly evident that G protein-coupled receptors (GPCRs) (1) function not only as short-term regulators of intermediary metabolism, but also as critical modulators of cellular growth and differentiation (1, 2). In bone, several hormones have been shown to increase bone turnover and remodeling through the activation of GPCRs. The PTH/PTH-related peptide receptor, in addition to acting as the principal regulator of calcium homeostasis, controls bone remodeling through direct actions on osteoblasts and their precursors, and indirect effects on osteoclasts (3). In vivo, intermittent, but not continuous, administration of N-terminal PTH fragments increases the rate of bone formation, bone density, and strength in rats, monkeys, and humans (4–8). This anabolic effect is reflected in increased numbers of mature osteoblasts (9, 10) that may result from PTH-induced differentiation of preexisting bone lining cells (11). Treatment of osteopenic ovariectomized rats with prostaglandin E2 (PGE2) or prostaglandin F2α (PGF2α) causes osteoblast recruitment and activation, and increases trabecular bone volume (12), while in vitro, PGs stimulate the proliferation and differentiation of murine osteoblastic MC3T3-E1 cells (13, 14). Activation of the extracellular calcium-sensing receptor likewise stimulates proliferation of MC3T3-E1 cells (15). The proliferative effect of fragments of osteogenic growth peptide on MC3T3-E1 cells requires the activation of pertussis toxin-sensitive G proteins (16). In addition, the mitogenic effect of fluoride, which is known to induce bone formation in vivo, probably results from direct activation of pertussis toxin-sensitive G proteins by a fluoroalumino complex (17, 18).

The ubiquitous MAPKs are a family of serine/threonine kinases involved in the transduction of externally derived signals regulating cell growth, division, differentiation, and apoptosis. Mammalian cells contain at least three major classes of MAPK, ERK1/2, c-Jun N-terminal kinase/stress-activated protein kinases, and p38/HOG1 MAPKs. MAPK activity is regulated by a series of parallel protein phosphorylation cascades, each comprised of three kinases that successively phosphorylate and activate the downstream component. In the ERK1/2 cascade, the proximal kinases, cRaf-1 and B-Raf, phosphorylate and activate MAPK kinase 1 and 2 (MEK1 and MEK2). MEK1 and -2 are dual-function threonine/tyrosine kinases that, in turn, phosphorylate and activate ERK1/2. Once activated, MAPKs phosphorylate a variety of membrane, cytoplasmic, and cytoskeletal substrates. Activated MAPKs also translocate to the nucleus, where they phosphorylate and activate nuclear transcription factors involved in DNA synthesis and cell division (19, 20).

Many of the anabolic effects of growth factors on bone, including those acting through GPCRs, depend on the activation of MAPKs. In human osteoblastic cells, expression of a dominant negative mutant of ERK1 inhibits basal and growth factor-stimulated proliferation, differentiation, and matrix mineralization (21). PTH exerts complex effects on ERK activity in osteoblasts. In human osteoblastic, bone marrow stromal, and UMR106 osteosarcoma cells, PTH inhibits the ERK2 activation caused by basic fibroblast growth factor, platelet-derived growth factor (PDGF) BB, epidermal growth factor (EGF), and lysophosphatidic acid (LPA), through a cAMP-dependent mechanism (22, 23). In contrast, treatment of UMR106 cells with PTH peptides that selectively activate the Gq/11-protein kinase C (PKC) pathway stimulates both ERK activity and cell proliferation (24). PGF2α and phorbol esters stimulate DNA synthesis in MC3T3-E1 cells through a PKC- and tyrosine kinase-dependent mechanism that involves activation of ERK1/2 (13, 25). Extracellular calcium, acting via the extracellular calcium-sensing receptor, activates the ERK1/2 and p38 MAPKs and stimulates ERK-dependent proliferation of human osteoblasts (15, 26).

Prior work on the mechanisms used by GPCRs to control MAPK activity in a variety of cell types has demonstrated remarkable heterogeneity. A number of distinct signaling mechanisms contribute to these responses, including signals mediated by second messenger-dependent protein kinases, signals mediated through cross-talk between GPCRs and receptor tyrosine kinases, and signals mediated by the interaction of GPCR-bound β-arrestins with components of the MAPK cascade (27–29). Indeed, the predominant mechanism of GPCR-mediated ERK activation varies significantly between different receptors and cell types.

Although the control of MAPK activity by GPCRs has clear importance for the regulation of anabolic bone metabolism, the mechanism of MAPK regulation in osteoblasts is not well understood. In this paper, we have examined the mechanisms of regulation of the ERK1/2 MAPK cascade by endogenous GPCRs in cultures of differentiating transgenic mouse osteoblasts (TMObs). We find that ERK1/2 activation in response to PTH(1–34), PGF2α, or LPA is mediated predominantly through an autocrine mechanism involving matrix metalloprotease-dependent release of EGF receptor ligands, leading to the transactivation of EGF receptors. These data suggest that EGF receptors represent an important point of convergence for growth-regulatory signals arising from several distinct GPCRs in osteoblasts.

RESULTS

PTH, PGF2α, and LPA Mediate Rapid and Transient Activation of ERK1/2 in Differentiating TMOb Osteoblasts

TMOb osteoblasts are clonal cell lines isolated from the calvariae of C57BL6J mice heterozygous for the SV40 large T antigen (30). Like MC3T3-E1 cells, TMOb osteoblasts exhibit a developmental sequence in vitro that resembles that of osteoblasts in bone tissue (31). During the initial phase of development, between d 1 and 9 in culture, the cells actively replicate and do not express bone-specific markers. After attaining confluence at approximately d 9, the cells undergo growth arrest, adopt a cuboidal morphology, and develop osteoblastic functions, including expression of PTH receptors and alkaline phosphatase, processing of procollagens to collagens, and deposition of a extracellular collagenous matrix. Mineralization of the extracellular matrix begins after d 16 in culture.

Figure 1 depicts the time course of agonist-induced phosphorylation of ERK1/2 in 12- to 14-d-old cultures of differentiating TMOb68 osteoblasts in response to maximally efficacious doses of three agonists for endogenous GPCRs: PTH(1–34), PGF2α, and LPA. The response to EGF, which stimulates ERK activation via a classical receptor tyrosine kinase, is shown for comparison. Each agonist provoked a similar 6- to 10-fold increase in ERK1/2 phosphorylation that was maximal within 5 min of stimulation and declined to about 20–30% of the peak value within 30 min. Qualitatively indistinguishable results were obtained using a second TMOb line, TMOb1223 (data not shown).

Fig. 1.

Time Course of Agonist-Stimulated ERK1/2 Phosphorylation in Differentiating Murine TMOb Osteoblasts TMOb68 osteoblasts (12- to 14-d cultures) were incubated overnight in serum-free medium before stimulation for the indicated times with human PTH(1–34) (100 nm), PGF2α (100 nm), LPA (1 μm), or EGF (10 ng/ml). Whole-cell phospho-ERK1/2 was determined as described in Materials and Methods. Data are presented as fold increase in ERK1/2 phosphorylation over the basal level in unstimulated cells and represent the mean ± sem values from three separate experiments. For each graph, the upper panel depicts a representative phospho-ERK1/2 immunoblot.

Protein Kinase A (PKA) and PKC-Dependent Signals Are Not Sufficient to Account for GPCR-Stimulated ERK Activation in TMOb Osteoblasts

The PTH/PTH-related peptide receptor couples to both Gs and Gq/11 family heterotrimeric G proteins, which mediate the activation of the adenylyl cyclase-protein kinase A (PKA) and phospholipase Cβ-PKC pathways, respectively. PGF2α receptors signal predominantly through Gq/11 proteins, while PGE2 receptors are predominantly Gs coupled. LPA receptors couple to both Gq/11 and pertussis toxin-sensitive Gi/o proteins. In many cases, activation of ERK1/2 in response to LPA has been shown to involve Gβγ subunits derived from pertussis toxin-sensitive G proteins, acting via as yet poorly defined effectors (32–34). Figure 2 depicts the effect of pertussis toxin, which catalyzes the ADP ribosylation and inactivation of Gi/o family G proteins, on ERK1/2 activation by PTH, PGF2α, and LPA receptors in TMOb68 cells. Consistent with the reported G protein-coupling specificity of each receptor, the response to PTH(1–34) and PGF2α were unaffected by pertussis toxin, whereas the LPA response was approximately 60% pertussis toxin sensitive.

Fig. 2.

Effect of Pertussis Toxin on GPCR-Stimulated ERK1/2 Phosphorylation in TMOb Osteoblasts TMOb68 osteoblasts (12- to 14-d cultures) were incubated overnight in serum-free medium in the presence or absence of pertussis toxin (PTX, 100 ng/ml) before stimulation for 5 min with human PTH(1–34) (100 nm), PGF2α (100 nm), or LPA (1 μm). Whole-cell phospho-ERK1/2 was determined as described. Data are presented as fold increase in ERK1/2 phosphorylation over the basal level in nonstimulated (NS) cells in the absence of pertussis toxin and represent the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

Depending on cell type, activation of the Gs-adenylyl cyclase-PKA pathway can lead to activation (35, 36), or inhibition (37–39), of ERK1/2. To determine the effects of adenylyl cyclase activation on ERK1/2 activity in TMOb osteoblasts, cells were stimulated with PGE2, cholera toxin, or forskolin, and ERK1/2 phosphorylation was assayed. As shown in Fig. 3A, stimulation with PGE2, or treatment with cholera toxin, which directly activates Gs, or with forskolin, which directly activates adenylyl cyclase, each produced a substantial rise in intracellular cAMP. Figure 3B depicts the level of ERK1/2 phosphorylation under these conditions. Cholera toxin, which causes a sustained increase in cAMP level, had no effect on ERK1/2 phosphorylation, whereas PGE2 and forskolin, which cause rapid cAMP increases, produced a 2-fold or less ERK1/2 response. These data suggest that PKA-dependent signaling pathways represent, at most, a minor component of GPCR-mediated ERK1/2 activation in TMOb cells. Consistent with this, preincubation of TMOb68 cells with the PKA inhibitor, H-89 (10 μm), had no significant effect on ERK1/2 phosphorylation in response to PTH(1–34), PGF2α, or LPA (data not shown).

Fig. 3.

Effect of Increased cAMP on ERK1/2 Phosphorylation in TMOb Osteoblasts A, TMOb68 osteoblasts (12- to 14-d cultures) were labeled overnight in serum-free medium containing [³H]adenine, in the presence or absence of cholera toxin (CTX, 100 ng/ml), before stimulation for 15 min with PGE2 (100 nm) or forskolin (50 μm). Intracellular cAMP content was determined as described. B, TMOb68 osteoblasts were incubated overnight in serum free medium in the presence or absence of cholera toxin, before stimulation for 5 min with PGE2 or forskolin. Whole-cell phospho-ERK1/2 was determined as described. In each panel, data are presented as fold increase over the basal level in nonstimulated (NS) cells and represent the mean ± sem values from three separate experiments.

Acute exposure to agents that increase intracellular calcium, such as ionomycin and thapsigargin, and potent stimulators of PKC, such as phorbol esters, rapidly activate ERK1/2 in many cell types, including osteoblasts (24, 40, 41). As shown in Fig. 4A, the calcium ionophore, ionomycin, which permits entry of extracellular calcium, and thapsigargin, which releases intracellular calcium stores by blocking the endoplasmic reticulum Ca²⁺-ATPase, both stimulated ERK1/2 phosphorylation in TMOb68 osteoblasts. The inhibition of these signals by preincubation with the cell permeant calcium chelator, BAPTA-AM, supports their calcium dependence. Exposure to the phorbol ester, phorbol myristate acetate (PMA), as shown in Fig. 4B, also produced rapid activation of ERK1/2. The response to PMA was sensitive to GF109203X, an inhibitor of the conventional isoforms of PKC.

Fig. 4.

Effect of Increased Intracellular Calcium or PKC Activity on Basal ERK1/2 Phosphorylation and of Intracellular Calcium Chelation or PKC Inhibition on GPCR-Stimulated ERK1/2 Phosphorylation in TMOb Osteoblasts A, Serum-starved 12- to 14-d cultures of TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of BAPTA-AM (50 μm), before stimulation for 5 min with ionomycin (2 μm) or thapsigargin (2 μm). Whole-cell phospho-ERK1/2 was determined as described. B, Serum-starved TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of GF109203X (2 μm), before stimulation for 5 min with PMA (1 μm). C, Serum-starved TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of BAPTA-AM or GF109203X before stimulation for 5 min with human PTH(1–34) (100 nm), PGF2α (100 nm), or LPA (10 μm). In each panel, data are presented as fold increase in phospho-ERK1/2 over the basal level in nonstimulated (NS) cells and represent the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

Despite their ability to activate the phospholipase Cβ-PKC pathway, activation of ERK1/2 by many Gi- and Gq/11-coupled receptors does not require increased intracellular calcium or PKC activation. Rather, these signals often follow Ras-dependent pathways that involve GPCR-dependent activation of receptor or nonreceptor tyrosine kinases (27, 28). To determine whether intracellular calcium or PKC contributed to GPCR-mediated ERK1/2 activation in TMOb osteoblasts, we assayed the sensitivity of PTH, PGF2α, or LPA receptor-stimulated ERK1/2 phosphorylation to pretreatment with BAPTA-AM and GF109203X. As shown in Fig. 4C, neither agent had a significant effect on GPCR-mediated ERK1/2 phosphorylation. These data suggest that the GPCR-mediated ERK1/2 activation in TMOB osteoblasts cannot be adequately explained on the basis of classical G protein-dependent activation of either the adenylyl cyclase-PKA or the phospholipase Cβ-PKC pathways.

GPCR-Stimulated ERK1/2 Activation in TMOb Osteoblasts Is Mediated Predominantly through Transactivation of EGF Receptors

Recent data have suggested that mitogenic signals arising from sources as diverse as GPCRs, cytokine receptors, bacterial toxins, and ionizing radiation utilize transactivated EGF receptors as part of a final common pathway regulating ERK activity (42–44). At least two receptor tyrosine kinases, those for PDGF (45, 46) and EGF (42, 43, 47), can be transactivated by GPCRs. The best understood mechanism of EGFR transactivation involves the autocrine/paracrine release of EGFR ligands from the cell surface via a process termed “ectodomain shedding.” Each of the known ligands for the EGF receptor [EGF, transforming growth factor α, heparin-binding (HB)-EGF, amphiregulin, betacellulin, and epiregulin (48)], is synthesized as a transmembrane precursor. Regulated proteolysis of these precursors, thought to be mediated by members of the ADAM family of matrix metalloproteases (49), results in the production of a soluble growth factor that acts locally to stimulate the EGF receptor.

As shown in Fig. 5A, both EGF and PDGF stimulated ERK1/2 phosphorylation in TMOb osteoblasts. Pretreatment with tyrphostin AG1478, an inhibitor of the EGF receptor tyrosine kinase, selectively blocked the response to EGF, whereas treatment with tyrphostin AG1295, a related inhibitor of the PDGF receptor, selectively blocked the PDGF response. Figure 5B depicts the effects of tyrphostins AG1478 and AG1295 on ERK1/2 activation in TMOb68 osteoblasts in response to PTH(1–34), PGF2α, and LPA. As shown, the PTH and PGF2α receptor-mediated responses were almost completely AG1478 sensitive, whereas the LPA receptor-mediated response was inhibited by about 60%. The PDGF receptor inhibitor AG1295 had no effect on GPCR-mediated ERK1/2 phosphorylation, suggesting that cross-talk between GPCRs and the EGF receptor, but not the PDGF receptor, accounted for the majority of the acute stimulation of ERK1/2. These findings are consistent with a previous report that in myocytes expressing both EGF and PDGF receptors, the EGF receptor served as the preferred target for GPCR-mediated transactivation (46). To directly determine whether GPCR stimulation affected EGF receptor activity in TMOb cells, we assayed EGF receptor tyrosine phosphorylation after stimulation with PTH(1–34), PGF2α, EGF, and HB-EGF. As shown in Fig. 5C, stimulation of either the PTH or PGF2α receptors significantly increased tyrosine phosphorylation of the EGF receptor, consistent with transactivation of the EGF receptor after GPCR stimulation.

Fig. 5.

Effect of EGF and PDGF Receptor Inhibition on Agonist-Stimulated ERK1/2 Phosphorylation and EGF Receptor Tyrosine Phosphorylation in TMOb Osteoblasts A, Serum-starved 12- to 14-d cultures of TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of tyrphostin AG1478 (100 nm) or tyrphostin AG1295 (25 μm), before stimulation for 5 min with EGF (10 ng/ml) or PDGF (10 ng/ml). Whole-cell phospho-ERK1/2 was determined as described. B, Serum-starved TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of tyrphostin AG1478 or tyrphostin AG1295, before stimulation for 5 min with human PTH(1–34) (100 nm), PGF2α (100 nm), or LPA (1 μm). C, Serum-starved TMOb68 osteoblasts in 100-mm dishes were stimulated for 5 min with vehicle (NS), PTH(1–34) (100 nm), PGF2α (100 nm), EGF (10 ng/ml), or HB-EGF (1 ng/ml), before immunoprecipitation of EGF receptor and immunoblotting for phosphotyrosine, as described. The upper panel depicts a representative antiphosphotyrosine immunoblot, whereas the bar graph depicts mean ± sem values from three separate experiments. The right lane of the immunoblot shows a control immunoprecipitation from nonstimulated (NS) cells performed in the absence of primary antibody (Sham). D, Serum-starved TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of tyrphostin AG1478 before stimulation for 5 min with ionomycin (2 μm), thapsigargin (2 μm), or PMA (1 μm). In each panel depicting ERK1/2 phosphorylation, data are presented as fold increase in phospho-ERK1/2 over the basal level in unstimulated (NS) cells and represent the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

Figure 5D depicts the effect of AG1478 on ERK1/2 activation induced by ionomycin, thapsigargin, and PMA. Interestingly, the calcium-dependent ERK1/2 activation induced by ionomycin and thapsigargin was also predominantly AG1478 sensitive, whereas the PKC-mediated response to PMA was independent of the EGF receptor. Because phorbol esters can mediate Ras-independent ERK1/2 activation via direct PKC-mediated activation of Raf-1 (50), these data suggest that PMA-mediated ERK1/2 activation in TMOb osteoblasts bypasses the upstream requirement for tyrosine kinase activity. In contrast, PKC activation by the GPCRs, even those coupled to Gq/11, was apparently not sufficient to activate ERK1/2 in the presence of the EGF receptor inhibitor.

To determine whether EGF receptor transactivation in TMOb osteoblasts involves the proteolytic generation of EGF receptor ligands, we employed the broad-spectrum matrix metalloprotease inhibitor, batimastat (47). As shown in Fig. 6A, pretreatment with maximally efficacious concentrations of batimastat significantly reduced, but did not abolish, PTH(1–34)-, PGF2α-, and LPA-stimulated ERK1/2 phosphorylation. The response to EGF was predictably unaffected, because direct application of EGF bypasses the requirement for proteolytic release of endogenous EGF receptor ligands. As shown in Fig. 6B, the responses to ionomycin and thapsigargin, but not to PMA, were also inhibited by batimastat. These data, which mirror the effects of tyrphostin AG1478, strongly suggest that proteolytic shedding of EGF receptor ligands, leading to EGF receptor transactivation, is the predominant mechanism of ERK1/2 activation in response to GPCR stimulation and intracellular calcium release in TMOb osteoblasts.

Fig. 6.

Effect of the Broad-Spectrum Matrix Metalloprotease Inhibitor, Batimistat, on Agonist-Stimulated ERK1/2 Phosphorylation in TMOb Osteoblasts A, Serum-starved 12- to 14-d cultures of TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of batimastat (5 μm), before stimulation for 5 min with PTH(1–34) (100 nm), PGF2α (100 mm), or LPA (10 μm), or EGF (10 ng/ml). Whole-cell phospho-ERK1/2 was determined as described. B, Serum-starved TMOb68 osteoblasts were preincubated for 15 min in the presence or absence of batimastat, before stimulation for 5 min with ionomycin (2 μm), thapsigargin (2 μm), or PMA (1 μm). In each panel, data are presented as fold increase in phospho-ERK1/2 over the basal level in nonstimulated (NS) cells and represent the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

PTH Receptors Expressed in COS-7 Cells Stimulate HB-EGF Proteolysis and EGF Receptor-Dependent Autocrine/Paracrine ERK1/2 Activation

To further characterize the molecular mechanisms underlying PTH receptor-mediated ERK activation, we employed a COS-7 cell model system in which the rat PTH receptor was transiently overexpressed. As shown in Fig. 7A, stimulation of PTH receptor-expressing COS-7 cells with either PTH(1–34) or EGF induced a rapid increase in tyrosine autophosphorylation of endogenous EGF receptors. PTH receptor-mediated EGF receptor phosphorylation and ERK1/2 activation were both sensitive to tyrphostin AG1478, duplicating the pattern observed in TMOb cells. To directly determine whether PTH receptor activation could stimulate the proteolysis of an EGF receptor ligand precursor, COS-7 cells were simultaneously transfected with plasmids encoding the rat PTH receptor and a modified HB-EGF containing both influenza virus hemagglutinin (HA) and myc epitopes. This epitope-tagged HB-EGF, which appears on immunoblots as a heterogeneous series of bands between approximately 18 and 30 kDa, carries the HA epitope within the extracellular HB-EGF moiety that is cleaved from the transmembrane domain upon proteolysis (51). As shown in Fig. 7B, stimulation with PTH(1–34) resulted in a time-dependent loss of HA epitope from the cell surface, consistent with the proteolytic degradation of HB-EGF in response to PTH receptor activation.

Fig. 7.

PTH(1–34)-Stimulated EGF Receptor Phosphorylation, ERK1/2 Activation, and HB-EGF Proteolysis in COS-7 Cells Transiently Expressing the Rat PTH Receptor A, Serum starved COS-7 cells transiently expressing HA epitope-tagged rat PTH receptors were preincubated for 15 min in the presence or absence of tyrphostin AG1478 (100 nm), before stimulation for 5 min with PTH(1–34) (100 nm) or EGF (10 ng/ml). The upper panel depicts an antiphosphotyrosine immunoblot of immunoprecipitated EGF receptor, whereas the lower panel depicts an anti-phospho-ERK1/2 immunoblot performed on parallel whole-cell detergent lysates. Data shown are representative of three separate experiments. B, Serum-starved COS-7 cells transiently expressing HA epitope-tagged rat PTH receptor and HA/myc epitope-tagged HB-EGF were stimulated with PTH(1–34) (100 nm) for the indicated times before immunoprecipitation of intact HA/myc-HB-EGF as described. The upper immunoblot depicts intact HA/myc-HB-EGF present in the cell lysate after PTH(1–34) treatment, detected by immmunoblotting with anti-HA IgG. The bar graph represents the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

To determine whether PTH receptors can stimulate EGF receptor-dependent activation of ERK1/2 through the metalloprotease-dependent production of soluble endogenous EGF receptor ligands, we employed a mixed cell culture system designed to detect paracrine signals generated in response to PTH stimulation. This system, which is depicted schematically in Fig. 8A, consisted of two COS-7 cell populations cocultured on the same surface. COS-7 cells expressing the rat PTH receptor were used as ligand donors, while COS-7 cells lacking the PTH receptor, but expressing an epitope-tagged ERK2 construct, were used as ligand acceptors. In these assays, green fluorescent protein (GFP)-tagged ERK2, which migrates with an apparent molecular mass of approximately 70 kDa, was expressed in the acceptor cells to permit simultaneous visualization of both endogenous ERK1/2 and GFP-ERK2 on immunoblots of whole-cell lysates. This GFP-ERK2 chimera has previously been shown to undergo GPCR-stimulated phosphorylation and nuclear translocation in a manner analogous to endogenous ERK1/2 (52).

Fig. 8.

Paracrine Activation of GFP-ERK2 by PTH(1–34) in a in a Mixed COS-7 Cell Culture System A, Schematic depicting the mixed culture model for detecting paracrine ERK1/2 in response to stimulation of PTH receptors. Cultures are comprised of donor COS-7 cells expressing HA epitope-tagged rat PTH receptor and acceptor cells expressing GFP-tagged ERK2, which permit monitoring of ERK2 phosphorylation in a mixed culture. B, Time course of PTH(1–34)-mediated ERK1/2 and GFP-ERK2 phosphorylation. Donor cells alone (upper immunoblot), acceptor cells alone (center immunoblot), or a 50:50 mixed culture of donor-acceptor cells (lower immunoblot) were stimulated with PTH(1–34) (100 nm) for the indicated times before immunoblotting for phospho-ERK1/2. The graph depicts the time course of PTH(1–34)-stimulated phosphorylation of GFP-ERK2 in acceptor cells when cultured alone (open symbols) or in the presence of PTH receptor-expressing donor cells (closed symbols). Data shown represent the mean ± sem values from three separate experiments. C, Effect of batimastat and tyrphostin AG1478 on PTH(1–34)-mediated ERK1/2 and GFP-ERK2 phosphorylation. Donor and acceptor cells were preincubated with batimastat (BB-94, 5 μm) or tyrphostin AG 1478 (100 nm) for 15 min before stimulation for 5 min with PTH(1–34) or EGF (10 ng/ml) as indicated. The immunoblot depicts ERK1/2 and GFP-ERK2 phosphorylation in a 50:50 mixed culture of donor-acceptor cells. The upper bar graph depicts the effect of the inhibitors on phosphorylation of endogenous ERK1/2 in donor cells cultured alone. The lower bar graph depicts the effect of the inhibitors on phosphorylation of GFP-ERK2 expressed in acceptor cells grown in mixed culture with donor cells. Data shown represent the mean ± sem values from three separate experiments. *, Less than control; P < 0.05.

As shown in Fig. 8B, stimulation of pure donor cell cultures with PTH(1–34) produced a rapid and transient activation of the endogenous ERK1/2 that mimicked the time course observed in TMOb cells. PTH (1–34) failed to stimulate either endogenous ERK1/2 or GFP-ERK2 in acceptor cells when cultured alone, indicating the absence of endogenous PTH receptor expression in COS-7 cells. However, when donor cells were stimulated in a mixed culture containing acceptor cells, a clear increase in GFP-ERK2 phosphorylation was observed, indicating that in the presence of cells expressing the PTH receptor, the cells lacking the PTH receptor were able to respond to PTH(1–34). The time course of the GFP-ERK2 response paralleled the time course of ERK1/2 activation seen in both the donor cells and in TMOb68 cells, with a detectable increase occurring within 2 min of stimulation and a maximal response by 5 min. To determine the mechanistic basic for this paracrine signal, we determined the effects of batimastat and tyrphostin AG1478 on the PTH(1–34)-induced activation of GFP-ERK2 in the acceptor cells in mixed culture. As shown in Fig. 8C, both the endogenous ERK1/2 response in donor cells, and the GFP-ERK2 response in acceptor cells were markedly inhibited by both agents. As in the TMOb68 cells, ERK phosphorylation in response to EGF, which can activate both donor and acceptor cells directly, was insensitive to batimastat, but inhibited by tyrphostin AG1478. These data strongly support the hypothesis that autocrine/paracrine EGF receptor transactivation, occurring as a consequence of metalloprotease-dependent release of endogenous EGF receptor ligands, represents the predominant mechanism of PTH receptor-mediated ERK activation in COS-7 cells.

DISCUSSION

Our data indicate that a complex set of inputs controls the activity of the ERK1/2 cascade in differentiating TMOb osteoblasts. As depicted schematically in Fig. 9, cross-talk between GPCRs and the EGF receptor accounts for most of the rapid activation of ERK1/2 in response to stimulation of endogenous PTH, PGF2α, and LPA receptors. Consistent with findings in other cell types (27), these signals are independent of the adenylyl cyclase-PKA and phospholipase Cβ-PKC pathways. Rather, ERK1/2 activation is mediated primarily by metalloprotease-dependent shedding of endogenous EGF receptor ligands. Agents that cause sustained increases in intracellular calcium, such as ionomycin and thapsigargin, act via a similar mechanism. Phorbol ester mediates PKC-dependent ERK1/2 activation, via a mechanism that, unlike the GPCR-mediated signals, is independent of the EGF receptor.

Fig. 9.

Mechanisms of GPCR-Stimulated ERK1/2 Phosphorylation in TMOb Osteoblasts Activation of PTH, PGF2α, or LPA receptors results in the matrix metalloprotease (MMP)-dependent release of EGF-like peptides, such as HB-EGF (43, 47 ) and transactivation of EGF receptors. GPCR-stimulated ERK1/2 activation is mediated predominantly via the EGF receptor. Ionomycin and thapsigargin, which cause the sustained release of intracellular calcium, also induce EGF receptor-dependent ERK1/2 activation. Phorbol esters, which activate conventional and novel PKC isoforms, activate ERK1/2 predominantly via a direct, EGF receptor-independent mechanism.

The molecular mechanisms by which GPCRs transactivate the EGF receptor are incompletely understood. EGF receptor dimerization and tyrosine autophosphorylation occur within 1–2 min of stimulation of many GPCRs, among them the LPA, α-thrombin (53, 54), β2 adrenergic (55), 5-HT2a serotonin, and AT1a angiotensin receptors (36). Nonetheless, early studies looking at the mechanism of EGF receptor transactivation failed to detect the release of a ligand for the EGF receptor into the culture medium after GPCR stimulation (53, 56). However, it has recently been possible, using overexpressed epitope-tagged HB-EGF, or mixed cell culture systems, to demonstrate that rapid EGF receptor transactivation involves autocrine/paracrine signals initiated by GPCR-mediated proteolysis of EGF receptor ligand precursors (47, 56). Using these approaches, we have been able to demonstrate that PTH receptors expressed in COS-7 cells stimulate proteolysis of HB-EGF and induce autocrine/paracrine activation of ERK1/2 via the metalloprotease inhibitor-sensitive generation of EGF receptor ligands.

To date, only HB-EGF has been shown to undergo GPCR-stimulated release (47). Proteolysis of the HB-EGF precursor is thought to be mediated by one or more members of the ADAM family of matrix metalloproteases, which are membrane-anchored proteases with single transmembrane and short cytoplasmic domains (43). In fibroblasts, both Gi/o-coupled and Gq/11-coupled receptors stimulate metalloprotease inhibitor-sensitive HB-EGF release. For Gi/o-coupled receptors, HB-EGF shedding is mediated by Gβγ subunits (47, 56). HB-EGF shedding in response to stimulation of Gq/11-coupled receptors is apparently mediated by Gq/11α subunits. Indeed, Pasteurella multocida toxin, a potent mitogen and direct activator of Gq/11 proteins, transactivates EGF receptors through Gq/11-dependent HB-EGF shedding in COS-7 cells (44).

The proximal G protein effectors that regulate ectodomain shedding remain poorly defined. Both phosphatidylinositol-3′ kinases (54, 57) and Src family nonreceptor tyrosine kinases have been proposed as early intermediates in the EGF receptor transactivation pathway (56, 58), and several of the ADAMs, notably ADAM-9, -10, -12, -15, -17, and -19, possess consensus SH3 domain binding motifs within their short intracellular domains that might mediate interaction with Src kinases (49). Interestingly, classical G protein effector systems, such as the phospholipase Cβ-PKC pathway and the adenylyl cyclase-PKA pathway, do not appear to contribute significantly to GPCR-stimulated ectodomain shedding. In TMOb cells, we have found that GPCR-stimulated ERK1/2 activation apparently requires neither intracellular calcium nor activation of classical PKC isoforms. Nonetheless, sustained increases in intracellular calcium in response to ionomycin or thapsigargin, or activation of PKC by phorbol esters, each induce ERK1/2 activation, and in the former cases, do so via an EGF receptor-dependent mechanism. Further, phorbol ester treatment has been shown to induce PKC-dependent proteolysis of HB-EGF through activation of the matrix metalloprotease ADAM 9 (51, 59). The most likely explanation for this apparent paradox lies with the intensity and duration of the signals induced by calcium ionophores and phorbol esters in comparison to endogenously expressed GPCRs, which would be expected to undergo relatively rapid desensitization in the continued presence of agonist. In contrast to the sustained rise in intracellular calcium seen with ionomycin, GPCR-induced calcium transients are characteristically brief (52) and may not reach a threshold level required to support significant calcium-dependent metalloprotease activation. While phorbol ester treatment does stimulate HB-EGF shedding (51), the ability of classical PKC isoforms to directly activate c-Raf1 (50) probably bypasses the requirement for EGF receptor activation.

The identity of the matrix metalloproteases involved in GPCR-stimulated EGF receptor transactivation also remains poorly characterized. While phorbol esters have been shown to stimulate HB-EGF shedding by activating ADAM 9 (59), metalloprotease-dependent EGF receptor transactivation in Rat1 and COS-7 cells in response to stimulation of M1 muscarinic acetylcholine, endothelin-1, and thrombin receptors is insensitive both to PKC inhibitors and to expression of a dominant inhibitory mutant of ADAM 9, despite the ability of these receptors to stimulate phosphatidylinositol hydrolysis (43, 47). Another ADAM family metalloprotease, ADAM 12, has recently been implicated in GPCR-mediated HB-EGF shedding in the heart (60).

Although even less well characterized, it is apparent that alternative mechanisms for cross-talk between GPCRs and EGF receptors, that do not involve proteolytic release of EGF receptor ligands, also exist. For example, stimulation of endogenous serotonin 5-HT2A receptors in cultured rat renal mesangial cells causes PKC-dependent EGF receptor transactivation and ERK activation that is insensitive to matrix metalloprotease inhibitors (61). Our finding that maximally efficacious concentration of batimastat failed to inhibit GPCR-stimulated ERK1/2 activation in TMOb68 cells to the same extent as tyrphostin AG1478 could indicate the existence of an alternative pathway cross-talk that contributes to EGF receptor transactivation in these cells. Alternatively, the incomplete effect of batimastat might reflect incomplete penetration of the drug into the extensive extracellular matrix synthesized by the differentiating osteoblasts. The finding that both batimastat and tyrphostin AG1478 abolish the activation of GFP-ERK2 in the COS-7 cell system clearly supports the conclusion that the proteolytic generation of EGF receptor ligands accounts for the paracrine component of PTH receptor-mediated ERK activation.

The complex effects of PTH on ERK1/2 activation in osteoblasts in vitro (22–24) probably reflects opposing roles of Gs- and Gq/11-dependent signals. In fibroblasts that express primarily the cRaf-1 isoform, PKA-mediated phosphorylation of Raf-1 attenuates ERK activation (37–39). In COS-7 cells, stimulation of Gs-coupled β2 adrenergic receptors provokes a net increase in ERK1/2 activity via a Gβγ subunit- and tyrosine kinase-dependent pathway. Nonetheless, overexpression of an activated mutant of Gαs, or treatment with the cell permeant cAMP analog, 8-Br-cAMP, not only fails to activate ERK in COS-7 cells, but markedly attenuates ERK activation in response to either isoproterenol or EGF (62). Our data, along with previous reports (22–24), suggest that Gs activation in cultured osteoblasts has similar effects. Increasing intracellular cAMP, using either PGE2, cholera toxin, or forskolin, has little effect on basal ERK1/2 phosphorylation in TMOb osteoblasts. Rather, the activation of ERK1/2 by PTH (1–34), like the response to PGF2α, is apparently mediated by pertussis toxin-insensitive G proteins, probably of the Gq/11 family.

Due to the tremendous heterogeneity of pathways influencing MAPK activity (27–29), a thorough understanding of the mechanisms underlying GPCR-mediated MAPK regulation requires that endogenously expressed receptors be studied in their native context. Our data, obtained in differentiating TMOb osteoblasts, clearly suggest a physiological role for cross-talk between GPCRs and classical receptor tyrosine kinases in control of the ERK1/2 pathway in osteoblasts. Similar data from primary vascular smooth muscle (60, 63, 64) and renal mesangial cells (61) support the concept that receptor tyrosine kinases of the EGF receptor family can serve as a point of convergence for growth-regulatory signals arising from GPCRs. Autocrine/paracrine activation of EGF receptors might therefore be accurately characterized a final common pathway of mitogenic signaling for diverse cell stimuli in multiple cell types.

The ultimate control of osteoblast recruitment, proliferation, and differentiation is a complex process involving the coordinated inputs of several classes of membrane and nuclear receptors and requiring activation of the adenylyl cyclase-PKA, phospholipase Cβ-PKC, and ERK1/2 MAPK pathways (12–14, 21, 22–24). For GPCRs, such as those for PTH and PGF2α, ERK1/2 activation represents a necessary, but not sufficient, signal for the control of osteoblast development. Understanding the contributions of each of these pathways to the ultimate expression of the mature osteoblast phenotype in vivo may prove a key element in the development of optimal strategies for increasing anabolic bone metabolism in the clinical setting.

MATERIALS AND METHODS

Materials

Human PTH(1–34), PGF2α, PGE2, LPA, PMA, forskolin, and isobutylmethylxanthine were purchased from Sigma Chemical Co. (St. Louis, MO). EGF, HB-EGF, PDGF, H-89, BAPTA-AM, GF109203X, tyrphostin AG1478, tyrphostin AG1295, ionomycin, and thapsigargin were from Calbiochem (La Jolla, CA). Pertussis toxin was from List Biological Laboratories (Campbell, CA). Cholera toxin, tissue culture media, and supplements were from Life Technologies, Inc. (Gaithersburg, MD). [³H]Adenine and [¹⁴C]cAMP were from Dupont-NEN (Boston, MA). Batimastat, BB-94, was generously provided by A. Ullrich. Rabbit polyclonal antirat EGF receptor IgG and rabbit polyclonal antiphosphotyrosine IgG were from Calbiochem. Mouse monoclonal antihuman EGF receptor R1 IgG and rabbit polyclonal anti-HA epitope IgG were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Rabbit polyclonal anti-ERK1/2 IgG and rabbit polyclonal anti-phospho-ERK1/2 IgG were from New England Biolabs (Beverly, MA). Monoclonal anti-HA11 affinity agarose was from CoVance (Princeton, NJ). Protein G plus/Protein A agarose was from Oncogene Research Products (San Diego, CA). Horseradish peroxidase-conjugated donkey antirabbit IgG was from Jackson Research Laboratories (Bar Harbor, ME).

The pBKCMV expression plasmid encoding HA epitope-tagged rat PTH receptor was from R. F. Spurney (Duke University Medical Center). The pCR3.1 expression plasmid encoding HA/myc epitope-tagged HB-EGF (51) was from M. Klagsbrun (Harvard Medical School). The pEGFP-N1 expression plasmid encoding GFP-ERK2 (52) was from K. A. DeFea and N. Bunnett (University of California at San Francisco).

Cell Culture and Transient Transfection

The TMOb68 and TMOb1223 osteoblast lines were derived from the calvariae of C57BL6J mice heterozygous for the SV40 large T antigen, as previously described (30). Stock cultures of TMObs were maintained in α-MEM supplemented with 10% fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 μg/ml) in a humidified atmosphere of 10% CO₂, 90% air at 37 C, and were split every 3–5 d to maintain subconfluence. In preparation for assay, cells were plated in six-well plates or 100-mm dishes depending on the assay to be performed. Cells were grown for 12–14 d in α-MEM supplemented with 10% fetal bovine serum, 5 mm β-glycerophosphate, and 25 μg/ml ascorbic acid, with the culture medium replaced every 3 d.

COS-7 cells were grown in DMEM supplemented with 10% fetal bovine serum and 100 μg/ml gentamicin. Transient transfection of COS-7 cells was performed using LipofectAMINE as previously described (56). For the assays involving mixed donor and acceptor cell populations, separate 100-mm dishes of COS-7 cells were transiently transfected using 6 μg/plate of either pBKCMV HA-PTH receptor or pEGFP-N1-ERK2. Twenty-four hours after transfection, cells were split into six-well plates containing either PTH receptor-transfected (donor) cells alone, GFP-ERK-transfected (acceptor) cells alone, or a 50:50 mixture of both cell populations. Before stimulation, both TMOb and COS-7 cultures were incubated for 18–24 h in serum-free growth medium supplemented with 10 mm HEPES (pH 7.4) and 0.1% BSA, in the presence or absence of inhibitors, as described in the figure legends.

[³H]cAMP Production

For the measurement of intracellular cAMP production, TMOb osteoblast cultures were preincubated for 18–24 h in α-MEM containing 3% fetal bovine serum and 2 μCi/ml [³H]adenine. Cholera toxin treatment was carried out overnight during the preincubation period. After labeling, cells were incubated for 15 min in Hanks’ balanced salt solution supplemented with 10 mm HEPES (pH 7.4) and 1 mm isobutylmethylxanthine, and stimulated for 20 min with PGE2 or forskolin, as described in the figure legends. After stimulation, cAMP was extracted and separated by sequential chromatography on AGW50-X4 and aluminum oxide columns, as previously described (65). Trace amounts of [¹⁴C]cAMP (10⁵ dpm/sample) were added to each sample to assess column recovery.

Proteolysis of HA/myc HB-EGF

Proteolytic cleavage of HB-EGF was measured as the PTH-stimulated loss of HA epitope from detergent lysates of COS-7 cells transiently expressing HA/myc-tagged HB-EGF. COS-7 cells were transiently transfected using 2 μg/plate pBKCMV HA-PTH receptor and 4 μg/plate pCR3.1 HA/myc HB-EGF. Serum-starved cells were stimulated with PTH(1–34) as described in the figure legends, washed with ice-cold PBS, solubilized in 1.0 ml of glycerol lysis buffer [50 mm HEPES, 50 mm NaCl, 10% (vol/vol) glycerol, 0.5% (vol/vol) Nonidet P-40, 2 mm EDTA, 100 μm Na₃VO₄, 1 mm phenylmethylsulfonyl fluoride, 4 μg/ml leupeptin, 2.5 μg/ml aprotinin], and clarified by centrifugation. Clarified lysates were agitated overnight at 4 C with 20 μl of 50% slurry of monoclonal anti-HA affinity agarose to immunoprecipitate intact HA epitope-tagged HB-EGF. Immune complexes were washed three times with glycerol lysis buffer, boiled in Laemmli sample buffer, resolved by SDS-PAGE, and transferred to polyvinylidine difluoride membrane for immunoblotting. Immunoblots were performed using a 1:1000 dilution of rabbit polyclonal anti-HA IgG, with horseradish peroxidase-conjugated polyclonal donkey antirabbit IgG as secondary antibody. Immunoprecipitated proteins on polyvinylidine difluoride filters were visualized by enzyme-linked chemiluminescence using SuperSignal chemiluminescence reagent, and immunoblots were quantified using a Fluor-S MultiImager.

EGF Receptor Tyrosine Phosphorylation

Tyrosine phosphorylation of endogenous EGF receptors in TMOb cells, and in COS-7 cells transiently expressing the rat PTH receptor, was determined after stimulation and immunoprecipitation of the EGF receptor. Serum-starved cultures in 100-mm dishes were stimulated as described in the figure legends and solubilized in 1.0 ml of glycerol lysis buffer. A 50 μl aliquot of each clarified whole-cell lysate was removed and mixed with an equal volume of 2× Laemmli sample buffer for simultaneous determination of total cellular phospho-ERK1/2. Murine EGF receptor was immunoprecipitated from the remaining lysate using 2 μg/sample of polyclonal antirat EGF receptor, while simian EGF receptor was immunoprecipitated using 2 μg/sample of monoclonal antihuman EGF receptor. Immune complexes were collected using 40 μl of a 50% slurry of Protein G plus/Protein A agarose beads tumbled overnight at 4 C. The phosphotyrosine content of immunoprecipitated EGF receptor was determined by immunoblotting using a 1:1000 dilution of rabbit polyclonal antiphosphotyrosine IgG, with horseradish peroxidase-conjugated polyclonal donkey antirabbit IgG as secondary antibody.

Phosphorylation of ERK1/2

Cultures (12–14 d) of differentiating TMOb osteoblasts were incubated for 18–24 h in serum-free growth medium in the presence or absence of inhibitors, as described in the figure legends. Agonist stimulations were carried out for 2–30 min, after which monolayers were washed once in 4 C PBS and lysed in 200 μl of Laemmli sample buffer. For the determination of total cellular ERK1/2 and phospho-ERK1/2, aliquots containing approximately 20 μg of cell protein were resolved by SDS-PAGE. ERK1/2 and phospho-ERK2 were detected by protein immunoblotting using polyclonal anti-ERK1/2 and anti-phospho-ERK1/2 antisera, respectively, with horseradish peroxidase-conjugated polyclonal donkey antirabbit IgG used as secondary antibody. Immune complexes were visualized by enzyme-linked chemiluminescence and quantified using a Fluor-S MultiImager (Bio-Rad, Hercules, CA). In each experiment, equal loading of ERK1/2 protein was confirmed by probing parallel immunoblots using anti-ERK1/2 antisera.

Acknowledgments

The authors thank Sabrina T. Exum and Francine L. Kelly for excellent technical assistance.

This work was supported by NIH Grants DK-55524 and DK-64353 (to L.M.L.) and AR-27032 (to M.K.D.).

Abbreviations:

 
• EGF,

  Epidermal growth factor;

 
• GFP,

  green fluorescent protein;

 
• GPCR,

  G protein-coupled receptor;

 
• HA,

  hemagglutinin;

 
• HB,

  heparin binding;

 
• LPA,

  lysophosphatidic acid;

 
• MEK 1 and 2,

  MAPK kinase 1 and 2;

 
• PDGF,

  platelet-derived growth factor;

 
• PGE2,

  prostaglandin E2;

 
• PGF2α,

  prostaglandin F2α;

 
• PKA,

  cAMP-dependent protein kinase;

 
• PKC,

  protein kinase C;

 
• PMA,

  phorbol myristate acetate;

 
• PTH(1–34),

  N-terminal 34-amino-acid fragment of human PTH;

 
• TMOb,

  transgenic mouse osteoblast.