INTRODUCTION

The function of the nervous system depends on the highly specific pattern of connections formed between neurons during development. The specificity of these connections requires neurite extension toward the correct targets guided by the growth cone and refinement and remodeling of the initial pattern of connections, referred to as synaptic plasticity, that are dependent on patterns of synaptic activity (1). The growth cone receives several kinds of environmental signals, such as diffusible chemoattractants and chemorepellants, extracellular matrix components, and cell adhesion molecules (2), and then it interprets them and changes shape and motility that results in the advance, turning, or collapse of the growth cone. The Rho family of small GTPases, Rac, CDC42, and Rho, has been shown to be involved in morphological changes of a variety of cells (3). In neuronal cells, Rac or CDC42 appears to be required for the outgrowth of neurites while Rho is required for their retraction (4).

Prostaglandin (PG)¹ E₂ is one of the major PGs synthesized in the nervous system (5). PGE₂ has several important functions in the nervous system, such as generation of fever (6, 7), regulation of luteinizing hormone-releasing hormone secretion (8), pain modulation (9), and regulation of neurotransmitter release (10, 11). Although cyclooxygenase products including PGE₂ have been suggested to be involved in regulation of memory consolidation (12), the biological significance of PGE₂ in synaptic plasticity is not yet understood. PGE₂ acts on cell surface receptors to exert its actions (13). PGE receptors are pharmacologically divided into four subtypes, EP1, EP2, EP3, and EP4, on the basis of their responses to various agonists and antagonists (14, 15). We have recently cloned the four subtypes of mouse PGE receptors and demonstrated that they are heterotrimeric GTP-binding protein (G protein)-coupled rhodopsin-type receptors (16, 17, 18, 19). Among these subtypes, the EP3 receptor was most abundant in the brain and was specifically localized to the neurons (20).

PC12 cells were derived from rat pheochromocytomas, expressing a chromaffin-like phenotype, and serve as a useful model system for the study of neuronal differentiation. When PC12 cells are exposed to NGF for several days, they acquire many features of sympathetic neurons, such as an outgrowth of neurites. We recently isolated four EP3 receptor isoforms (EP3A, EP3B, EP3C, and EP3D) from bovine adrenal chromaffin cells that are produced through alternative splicing (21). To assess the role of the EP3 receptors in the nervous system, we introduced the cDNA for the EP3B receptor isoform, the most abundant isoform, into PC12 cells. In this report, we show that the activation of the EP3B receptor causes neurite retraction via small GTPase Rho.

EXPERIMENTAL PROCEDURES

M&B28767 was a generous gift from Dr. M. P. L. Caton of Rhone-Poulenc Ltd. NGF 2.5S was purchased from Promega Corporation. Pertussis toxin (PT) and Clostridium botulinum C3 exoenzyme were obtained from Seikagaku Kogyo (Tokyo, Japan), dibutyryl cyclic AMP Bt₂cAMP) was from Sigma, and 12-O-tetradecanoylphorbol-13-acetate (TPA) and 4-phorbol 12,13-didecanoate were obtained from Funakoshi Pharmaceuticals (Tokyo, Japan). The sources of the other materials are shown in the text.

PC12 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 5% horse serum, 4 mM glutamine, 100 units/ml penicillin, and 0.2 mg/ml streptomycin under humidified conditions in 95% air and 5% CO₂ at 37 °C. cDNA encoding the EP3B receptor was inserted into the expression vector pcDNA3 (Invitrogen), and the plasmid constructed was transfected into PC12 cells by lipofection (22). Stable transformants were cloned by selection with G418 (Life Technologies, Inc.).

After the cells had been seeded on poly-D-lysine (Sigma)-coated 24-well plates at a density of 10⁴ cells/well in serum-containing Dulbecco's modified Eagle's medium, and cultured for 24 h, they were differentiated in serum-free Dulbecco's modified Eagle's medium containing 50 ng/ml NGF and 20 µM indomethacin for 5 days. In the neurite retraction experiments, the differentiated cells were maintained in Hepes-buffered Dulbecco's modified Eagle's medium (pH 7.4) at 37 °C. Cells were photographed at × 100 or 200 magnification under a phase contrast microscope. A neurite was identified as a process greater than 1 cell body diameter in length. To establish the time course, the average neurite length was measured from at least 30 cells in the same field, excluding the cells that did not respond to the agonist. For the dose-response relationships and quantitative examinations shown in Tables I and II, neurite-retracted cells were defined as the cells that retracted by more than 10% of their original length within 60 min of the addition of the test agents. The percentages of neurite-retracted cells were calculated by counting at least 50 cells in the same field. All data were obtained from triplicate experiments.

Table I. Effects of PT and Bt2cAMP on M&B28767-induced neurite retraction The percentage of neurite-retracted cells was determined as described under "Experimental Procedures." Data are means ± S.E. of triplicate experiments. Treatment Neurite-retracted cells % of total cells None 93.9  ± 1.0 PT 92.2  ± 2.3 Bt2cAMP 13.4  ± 1.9

Table II. Effect of long term exposure to TPA on M&B28767- or TPA-induced neurite retraction The percentage of neurite-retracted cells was determined as described under "Experimental Procedures." Data are means ± S.E. of triplicate experiments. Treatment Neurite-retracted cells M&B28767 TPA % of total cells None 93.0  ± 1.5 81.4  ± 4.6 TPA 87.5  ± 2.5 0.0  ± 0.0

For microinjection, cells were seeded on poly-D-lysine-coated 35-mm dishes, which were marked with a cross to facilitate the localization of microinjected cells. Microinjection was performed using an IMM-188 microinjection apparatus (Narishige, Tokyo, Japan). C3 exoenzyme was microinjected at 100 µg/ml in reversed phosphate-buffered saline (3.7 mM NaCl, 137 mM KCl, 1 mM KH₂PO₄, 8 mM Na₂HPO₄). We used 100 µg/ml C3 exoenzyme according to the method of Ridley and Hall (23). At this concentration, the inhibition of neurite retraction by C3 exoenzyme was complete.

RESULTS

To assess the role of EP3 receptors in neuronal morphology we established PC12 cells that stably expressed the EP3B receptor (151 fmol/mg protein). We initially examined the coupling of the EP3B receptor to the classical signal transduction pathways, the adenylate cyclase and Ca²⁺ mobilization pathways. In these cells, the EP3B receptor inhibited adenylate cyclase activity through a PT-sensitive heterotrimeric GTP-binding protein, but the receptor could not stimulate adenylate cyclase, phosphatidylinositol hydrolysis, or Ca²⁺ mobilization (data not shown), suggesting that the EP3B receptor is coupled to G_i, but not to G_s or G_q. We next examined the effect of M&B28767, a specific EP3 agonist, on the neuronal morphology of the EP3B receptor-expressing PC12 cells. As shown in Fig. 1, treatment with NGF induced neurite outgrowth. The addition of M&B28767 caused a dramatic morphological change in the NGF-differentiated PC12 cells. Within 10 min of the addition of the agonist neurites began to retract, and within 30 min most neurites had retracted completely. Fig. 2A shows the time course of the effect of the agonist on neurite length. The neurite length became half of the original length within 20 min. The rate of neurite retraction was not dependent on the concentration of the agonist (data not shown). M&B28767 concentration-dependently increased the population of neurite-retracted cells, and more than 90% of the total cells responded to 1 µM M&B28767 (Fig. 2B). However, a few cells resisted even high concentrations of the agonist. M&B28767-induced neurite retraction was not seen for more than 60 min in the untransfected cells (data not shown).

The EP3B receptor inhibited adenylate cyclase via a PT-sensitive heterotrimeric G protein. Therefore, we examined whether M&B28767-induced neurite retraction was mediated through a PT-sensitive G protein. As shown in Fig. 3 and Table I, pretreating the EP3B receptor-expressing cells with PT for 12 h did not affect agonist-induced neurite retraction.

We further examined the effect of protein kinase A activation on EP3B receptor-mediated neurite retraction. As shown in Fig. 4, whereas Bt₂cAMP did not affect the morphology of the differentiated cells, it strongly prevented M&B28767-induced neurite retraction. Table I shows the quantitative examination of the effect of Bt₂cAMP. It decreased the percentage of neurite-retracted cells in response to M&B28767 from 93.9 to 13.4%. These results indicate that the activation of protein kinase A suppresses the EP3B receptor-mediated neurite retraction.

Protein kinase C activation has been reported to induce membrane ruffling (24). Next, we examined the effect of protein kinase C activation on the morphology of PC12 cells. TPA, a protein kinase C activator, also induced neurite retraction in differentiated PC12 cells, as did M&B28767 (Fig. 5, A-D). However, 4-phorbol 12,13-didecanoate, an inactive phorbol ester analogue, did not induce any change in cell shape (data not shown). Since the activation of protein kinase C triggered neurite retraction, we examined whether the morphological action of the EP3B receptor was mediated by protein kinase C. Exposing PC12 cells to TPA for more than 24 h has been reported (25) to induce down-regulation of protein kinase C. When the cells had been exposed to 1 µM TPA during differentiation, NGF normally induced morphological differentiation of PC12 cells. Long term exposure to TPA completely suppressed TPA-induced neurite retraction, but M&B28767-induced neurite retraction occurred normally in protein kinase C-down-regulated cells (Fig. 5, E-H, and Table II). These results indicate that EP3B receptor-mediated morphological change does not involve the activation of protein kinase C. Fig. 6 shows the time course of the effects of M&B28767 and TPA on neurite length in differentiated cells. M&B28767-induced neurite retraction occurred within 5 min in most of the cells that responded, and the time taken for 50% retraction was estimated to be 20 min. In contrast, TPA-induced neurite retraction in most of the cells that responded was preceded by a lag period ranging from 5 to 10 min.

A small G protein, Rho, appears to be required for the retraction of neurites (4). To evaluate the participation of Rho in the EP3B receptor-mediated neurite retraction, the cells were microinjected with a C. botulinum C3 exoenzyme that specifically ADP-ribosylates Rho and suppresses the actions of Rho (26, 27). As shown in Fig. 7, A and B, M&B28767-induced neurite retraction was completely inhibited in a cell microinjected with C3 exoenzyme. TPA-induced neurite retraction was also inhibited in a C3 exoenzyme-microinjected cell (Fig. 7, C and D). No inhibition was observed in a cell microinjected with the buffer alone, indicating that microinjection itself did not inhibit neurite retraction (data not shown).

DISCUSSION

Information on the function of the EP3 receptors in the nervous system is scarce, although several findings have been reported, for example, they inhibit neurotransmitter release and have hyperalgesic effects (28, 29). In this study, we described a new possible function of the EP3 receptors, modulation of a neuronal morphological change through a novel pathway distinct from adenylate cyclase inhibition or protein kinase C activation. In the EP3B receptor-expressing PC12 cells differentiated with NGF, M&B28767 induced neurite retraction. However, in the untransfected cells, no morphological change was caused by the agonist, indicating that M&B28767-induced neurite retraction is mediated by EP3B receptor.

Growing evidence suggests that Rho plays a key role in the regulation of the actin cytoskeleton. The functions of Rho have been explored using the C3 exoenzyme, which ADP-ribosylates and inactivates the Rho protein (26, 27). Many studies using C3 exoenzyme demonstrated that Rho is involved in the control of cell shape, adhesion, and motility by regulating the actin cytoskeleton in fibroblasts (3), and it is also involved in neurite retraction in neuronal cells (4). Since microinjection into the EP3B receptor-expressing cells with the C3 exoenzyme completely inhibited M&B28767-induced neurite retraction (Fig. 7, A and B), Rho was shown to be an essential element in the EP3B receptor-mediated neurite retraction.

Among classical second messenger pathways, the EP3B receptor is coupled to adenylate cyclase inhibition through G_i in PC12 cells, but the receptor stimulates neither adenylate cyclase nor phospholipase C, indicating that the EP3B receptor is exclusively coupled to G_i. However, neurite retraction mediated by this receptor was not suppressed by PT treatment (Fig. 3 and Table I), indicating that this action is not mediated by G_i. These observations suggest that the EP3B receptor is coupled to G_i, which inhibits adenylate cyclase, and to a PT-insensitive heterotrimeric G protein other than G_s or G_q that induces neurite retraction dependent on Rho activity. It has recently been reported (30) that GTPase-deficient constitutively activated G₁₂ and G₁₃ stimulate Rho-dependent stress fiber formation and focal adhesion assembly in Swiss 3T3 cells. As G₁₂ and G₁₃ are PT-insensitive G proteins, we suggest that the EP3B receptor may regulate neuronal morphology through G₁₂ or G₁₃.

Protein kinase C is also thought to be involved in the regulation of cell shape, contraction, and motility (31), and it has recently been reported (32) that protein kinase C activation induced membrane ruffling and translocation of activated Rho to the membrane ruffling area. We have shown here that protein kinase C activation also induced neurite retraction within 60 min in EP3B receptor-expressing PC12 cells (Fig. 5, C and D, and Table II). Long term exposure to TPA suppressed the TPA-induced neurite retraction by down-regulating protein kinase C (Fig. 5, G and H, and Table II). In contrast, long term exposure to TPA did not prevent M&B28767-induced neurite retraction (Fig. 5, E and F, and Table II), indicating that EP3 receptor induces neurite retraction through a mechanism not involving the protein kinase C pathway. Both M&B28767- and TPA-induced neurite retractions were sensitive to the C3 exoenzyme (Fig. 7), indicating that these effects are mediated by Rho activation. Thus, there are two different signaling pathways, protein kinase C-dependent and -independent pathways, involved in Rho-mediated neurite retraction. TPA-induced neurite retraction was preceded by a lag period (5-10 min), while the EP3B receptor activation induced neurite retraction without a lag period (Fig. 6). Thus, the EP3B receptor-mediated retraction is faster than that mediated by the protein kinase C pathway. This finding shows that the intermediate pathways from the EP3B receptor to Rho and from the protein kinase C to Rho differ, and the latter requires a longer time to activate Rho. It has recently been reported (4) that lysophosphatidic acid (LPA) caused neurite retraction dependent on Rho activity in N1E-115 cells as well as in PC12 cells and that the LPA-induced neurite retraction was mediated by G_q-coupled phospholipase C stimulation and subsequent protein kinase C activation (33). Furthermore, the time course of LPA-induced neurite retraction in PC12 cells appears to be slower than the EP3B receptor-mediated retraction but similar to that of TPA-induced retraction (34). These findings indicate that LPA induces neurite retraction through a protein kinase C-dependent pathway, while the EP3B receptor induces neurite retraction through a protein kinase C-independent pathway. Thus, there are two different types of heterotrimeric G protein-coupled receptors that induce Rho-dependent neuronal morphological change. One type is G_q-phospholipase C-coupled receptors, and the other is coupled to a heterotrimeric G protein other than G_q, leading to Rho-dependent action.

Our examination of the regulation of the EP3B receptor-mediated neurite retraction by protein kinase A revealed that the activation of protein kinase A inhibited the EP3B receptor-mediated neurite retraction and indicated that protein kinase A is a negative regulator. As the potential protein kinase A phosphorylation site is located in the first cytoplasmic loop of the EP3B receptor, it is possible that protein kinase A phosphorylates the EP3B receptor. The phosphorylation of this site by protein kinase A may stop the EP3B receptor-mediated signaling. The second possible site of the action of protein kinase A is on a downstream component. LPA-induced neurite retraction was recently reported (35) to be prevented by pretreatment with Bt₂cAMP in PC12 cells. Furthermore, it was recently reported (36) that protein kinase A directly phosphorylated Rho, and this phosphorylation resulted in termination of Rho signaling. As LPA and the EP3B receptors use different pathways to activate Rho, protein kinase A may block Rho-mediated morphological regulation by phosphorylating Rho in PC12 cells.

From this study, we propose new possible functions of PGE₂ acting through the EP3 receptors in the nervous system. We previously showed that EP3 receptors were expressed in a variety of neurons in the brain (20), and neurons in the dorsal root ganglion (37). PGE₂ is normally produced in the brain, and its production is dramatically stimulated by brain injuries such as concussion, trauma, and asphyxia (38, 39, 40, 41). When the brain is injured, newly synthesized PGE₂ may retract the neurites of the EP3 receptor-expressing neurons and reorganize damaged neuronal connections. In addition, the levels of PGE₂ are also increased in the brain by synaptic activity or during development (42). Furthermore, cyclooxygenase, a rate-limiting enzyme in PG synthesis, has been reported to be markedly induced by seizures or N-methyl-D-aspartate-dependent synaptic activity, especially in the cortex and hippocampus (43) where EP3 receptors are highly expressed (20). PGE₂ may be involved in refining and remodeling the initial neuronal connections through EP3 receptors.

In conclusion, in this report we have shown that the EP3B receptor elicits neurite retraction in a Rho-dependent fashion. This study will contribute not only to the understanding of neuronal PGE₂ function but will also help to elucidate the molecular mechanisms of signal transduction pathways between heterotrimeric G protein-coupled receptors and Rho.