A rapid, nongenomic pathway facilitates the synaptic transmission induced by retinoic acid at the developing synapse

All-trans retinoic acid (RA), a member of retinoid family, plays important roles during embryogenesis in modulating the growth and differentiation of a variety of cell types, particularly in the developing nervous system. The biological effects of RA are elicited by changes in gene expression brought about by the interactions of RA with its cellular binding proteins and retinoid nuclear receptors. Two types of receptors, the retinoic acid receptors (RAR) and the retinoid X receptors (RXR), mediate RA signaling (Chambon, 1994; Kastner et al., 1997). Both receptor types comprise three subtypes (RARα, β, γ and RXRα, β, γ) encoded by different genes and displaying distinct patterns of expression (De Luca, 1991). These receptors belong to the superfamily of the steroid-thyroid-retinoid hormone nuclear receptors with RAR forming heterodimers with RXR, whereas RXR, either as a homodimer or by forming heterodimers with orphan receptors, acts by binding to retinoid response elements in the promoters of target genes and functions in modifying the transcriptional activity of specific genes (the so called `genomic' effect) in the presence of ligands (Chambon, 1994; Mangelsdorf and Evans, 1995; Kastner et al., 1997). Besides this well-known classical mode of action for steroid-thyroid-retinoid hormone, which triggers genomic events that are finally responsible for delayed and long-term cellular responses, the number of reports on rapid effects of steroids, which are considered to be of `nongenomic' origin, has grown tremendously (Losel et al., 2003). Although substantial evidence reveals that steroid, thyroid hormones and the metabolite of vitamin D₃, 1α, 25-dihydroxyvitamin D₃, can act with a rapid onset rather than their classical mode of genomic steroid action, very little is known of the effects of RA on targets other than those classically described.

Successful synaptic transmission at the neuromuscular junction is a complex process regulated by interplay among intrinsic cellular programs, cell-cell and cell-substrate interactions, and plenty of soluble extracellular signaling molecules (Lu and Je, 2003). RA, which is highly expressed in the spinal cord of developing embryos, has been suggested to be capable of increasing the survival of developing motoneurons, stimulating neurite outgrowth and directing axons extending from embryonic spinal cord explants in vitro (Maden et al., 1998; Prince and Carlone, 2003). Apart from its known effect in patterning both the anteroposterior and dorsoventral axes, RA also has considerable significance as a neural differentiation factor. Recently, it has been suggested that RA, as an inductive signal, can direct embryonic stem cells to differentiate into motoneurons (Wichterle et al., 2002). We have previously provided the first evidence that activation of `RARβ' receptor is responsible for RA-induced rapid effect (∼8-15 minutes) on spontaneous synaptic current (SSC) frequency facilitation at the developing neuromuscular synapse. Moreover, the insensitivity of RA-induced SSC frequency facilitation toward inhibitors of transcription and/or translation and the short time frame of their response further suggest the involvement of an unusual nongenomic mechanism (Liao et al., 2004). Although results from our previous studies suggest that RA might act through a classical receptor to facilitate spontaneous neurotransmitter release by a non-classical mechanism, efforts are still needed to further explore the underlying molecular mechanisms through which RA induces facilitation of SSC frequency. How does RA enhance presynaptic efficacy? What are the intracellular signaling mechanisms that mediate such rapid synaptic effects of RA? It is well known that the intracellular Ca²⁺ concentration ([Ca²⁺]_i) level in the nerve terminal exerts a dominant effect on the rate of spontaneous transmitter release (Augustine et al., 1987). Many experimental approaches have suggested a regulatory role for steroid hormones in the control of [Ca²⁺]_i. Aldosterone induces rapid increase in intracellular protein kinase C (PKC) activity and a rise of Ca²⁺ in human distal colon cells (Harvey et al., 2002). In cultured skeletal muscle cells, 1,25-dihydroxyvitamin D₃ produced a rise in [Ca²⁺]_i by promoting a non-genomic release of Ca²⁺ from internal stores via activation of phospholipase C (PLC) and D and inositol 1,4,5-trisphosphate [Ins(1,4,5)P₃] and by Ca²⁺ influx through L-type and store-operated Ca²⁺ channels (Capiati et al., 2001). However, no information is available so far concerning the acute effects of RA on intracellular Ca²⁺ and on the turnover of membrane phosphoinositides. In the present study, we are seeking to fill these gaps in our understanding of the molecular machineries that are responsible for this facilitation. Results from our study show for the first time that RA rapidly triggers the liberation of Ca²⁺ from internal store, which is the result of pleiotropic convergent signaling pathways involving PLCγ, phosphatidylinositide 3-kinase (PI 3-kinase) and activation of Src tyrosine kinase.

Xenopus laevis nerve-muscle cultures were prepared as previously reported (Liou and Fu, 1997). Briefly, the neural tube and the associated myotomal tissue of 1-day-old (stage 20-22) Xenopus embryos were dissected and dissociated in the Ca²⁺- and Mg²⁺-free Ringer's solution supplemented with 0.15 mM EDTA. The dissociated cells were plated and used for experiments after incubation at room temperature (20-25°C) for 1 day. The culture medium consisted of 50% (vol/vol) Ringer's solution (115 mM NaCl, 2 mM CaCl₂, 2.5 mM KCl, 10 mM Hepes, pH 7.6), 49% L-15 Leibovitz medium (Sigma), and 1% fetal bovine serum (Life Technologies), with antibiotics (100 U/ml penicillin and 100 μg/ml streptomycin sulfate). RA and various inhibitors were applied directly to the culture medium at the time of recording.

One day following cell plating, functional synapses are rapidly established between cultured spinal neurons and embryonic muscle cells. The present study utilized synapses of myocytes innervated by single co-cultured spinal neurons. The frequency of spontaneous synaptic events during the first day of synaptogenesis was found to vary greatly from cell to cell, over two orders of magnitude, and the frequency of SSC events increased with time of synapse development (Evers et al., 1989). To test the facilitation effect induced by RA treatment in more simple conditions, our analyses were performed mostly in low-activity synapses (<1.0 Hz) to mimic the early contact between motoneurons and myocytes (Evers et al., 1989). RA also facilitates, although to a lesser extent, the spontaneous transmitter release in neuromuscular synapse with higher activity (>1.5 Hz; data not shown).

Gigaohm-seal whole-cell recording methods followed those described previously (Hamill et al., 1981) were used. Patch pipettes (Hilgenberg) were pulled with a two-stage electrode puller (PP-830, Narishige, Tokyo, Japan), and the tips were polished immediately before the experiment using a microforge (MF-830, Narishige). SSCs were detected from innervated myocytes by whole-cell recording in the voltage-clamp mode. Recordings were made at room temperature in Ringer's solution, and the solution inside the recording pipette contained 150 mM KCl, 1 mM NaCl, 1 mM MgCl₂ and 10 mM Hepes (pH 7.2). Data were collected using a patch-clamp amplifier WPC-100 (ESF electronic, Göttingen, Germany) and Axoscope 8.0 (Axon). Signals were filtered at 10 KHz (Digidata 1322; Axon, Union City, CA). SSCs were detected and analyzed using the Mini Analysis Program 5.0 (Synaptosoft, Decatur, GA). To quantitatively measure the changes in neurotransmitter release, a time course of SSC frequency was first constructed on a minute-to-minute basis. The SSC frequencies for a 6-minute period right before drug application was averaged as control. The changes in SSC frequency were measured by averaging a 6-minute recording starting from the highest number after drug application (Liao et al., 2004), and the results were expressed as mean ± s.e.m. The statistical significance was evaluated using Student's paired t-test.

Presynaptic spinal neurons were dialyzed using a patch pipette containing 150 mM KCl, 1 mM NaCl, 1 mM MgCl₂, 10 mM Hepes, 5 mM GDP_βS and 1 mg/ml of the fluorescent dye Lucifer Yellow while the whole-cell recordings were being made at the cell body. The dialysis of the drug and Lucifer Yellow could be visualized directly in the inverted microscope under fluorescence mode. After a 10-minute dialysis, the patch pipette was pulled off after damaging the seal by injection of a large hyperpolarizing current.

[Ca²⁺]_i was measured at room temperature with the fura-2 fluorescence ratio method on a single cell fluorimeter as previously described (Strotmann et al., 2000). In brief, cells attached on a coverslip were loaded with 2 μM fura-2/acetoxymethyl ester (fura-2/AM; Molecular Probes, Eugene, OR, USA) in DMEM culture medium at room temperature for 60 minutes. After loading, cells were washed three times with Ringer or Ca²⁺-free Ringer. After washing, coverslips were placed on the stage of an Olympus IX71 inverted microscope equipped with a xenon illumination system and an IMAGO CCD camera (Till Photonics, Grafelfing, Germany). The excitation wavelength was alternated between 340 nm and 380 nm using the Polychrome IV monochromator (Till Photonics, Grafelfing, Germany). The fluorescence intensity was monitored at 510 nm, stored digitally and analyzed using TILLvisION 4.0 (Till Photonics, Graefelfing, Germany) software. [Ca²⁺]_i was represented as the ratio of F340/F380. F340 and F380 are the emissive fluorescence intensity when cell were excited at 340 nm and 380 nm.

The following chemicals were used: all-trans retinoic acid (RA), 8-(dethylamino) octyl 3,4,5-trimethoxybenzoate (TMB-8), Lucifer Yellow, 1,2-bis-(2-aminophenoxy)-ethane-N,N,N′,N′-tetraacetic acid tetrakis (acetoxy-methyl) ester (BAPTA-AM), ruthenium red and thapsigargin were all obtained from Sigma (St Louis, MO, USA); PD98059, wortmannin, U73122, LY294002, bisindolmaleimide was from Tocris Cookson (Bristol, UK); Xestospongin C (XeC), 2-APB from Calbiochem (San Diego, CA, USA). All drugs were applied directly to the culture media at the times indicated.

In Xenopus nerve-muscle cultures, functional synaptic transmission can be detected within minutes after nerve-muscle contact, although morphological maturation of the synapse requires many days to complete (Evers et al., 1989). SSCs are readily detectable from the innervated muscle cell with the whole-cell voltage-clamp recordings. These currents were shown to be caused by spontaneous acetylcholine (ACh) secretion from the neuron because they were abolished by bath application of D-tubocurarine and unaffected by tetrodotoxin, which blocks action potentials in neurons (Xie and Poo, 1986). Bath application of RA at 30 μM dramatically enhanced spontaneous transmitter release, as evidenced by a marked increase in the frequency of spontaneous synaptic events. The increase in SSC frequency produced by RA was rapid and reached a plateau within ∼8-15 minutes after bath application of RA, and the effect persisted for more than 30 minutes (Fig. 1A,C). On average, the frequency increased by 29.0±7.1 (n=17)-fold over the control SSC frequency.

How does RA enhance presynaptic efficacy? Given the suggestion that the [Ca²⁺]_i level in the nerve terminal exerts a dominant effect on the rate of spontaneous transmitter release, we set out to examine whether a rise in [Ca²⁺]_i is required for the RA-induced SSC frequency facilitation (Augustine et al., 1987). The membrane-permeable Ca²⁺ chelator BAPTA-AM has been widely used as a probe to test the role of Ca²⁺ in a large variety of cell functions (Yang et al., 2003). The SSC frequency facilitation by RA was completely blocked by buffering the [Ca²⁺]_i rise with a 60- to 80-minute pretreatment with 30 μM BAPTA-AM (1.4±0.5-fold of control, n=6; Fig. 1B,C), suggesting that the effect of RA on synaptic transmission requires an increase in [Ca²⁺]_i in the presynaptic neurons.

This increase in [Ca²⁺]_i may be due to influx of Ca²⁺ from the extracellular fluid or release of Ca²⁺ from intracellular stores. We first examined the role of Ca²⁺ influx in the action of RA. Ca²⁺ was eliminated from the culture medium after several washes with the Ca²⁺-free Ringer's solution. Treating the cells with RA still elicited an increase in SSC frequency under the zero external Ca²⁺ condition (Fig. 2A,C). The SSC frequency was increased by 29.6±6.4-fold (n=8) under the Ca²⁺-free condition. To further examine the role of membrane Ca²⁺ channels, we blocked Ca²⁺ influx by bath application of Cd²⁺ (0.5 mM), which competes with Ca²⁺ and blocks Ca²⁺ influx through Ca²⁺ channels. RA was still capable of facilitating SSC frequency in the presence of Cd²⁺ (20.1±7.3-fold of control, n=8, P<0.05; Fig. 2B,C). Thus, RA-induced facilitation of neurotransmitter release does not require Ca²⁺ influx from extracellular fluid.

We next explored the routes of Ca²⁺ released from intracellular stores that resulted in RA-induced SSC frequency facilitation. Because of the possibility that depletion of internal Ca²⁺ might induce a Ca²⁺ entry through store-operated channels in the plasma membrane, the following experiments were performed, mainly in Ca²⁺-free medium to avoid unnecessary interference (Tempia et al., 2001). There are two major pathways that result in the release of Ca²⁺ from intracellular stores: the Ins(1,4,5)P₃-sensitive and the ryanodine-sensitive Ca²⁺ stores (Berridge, 1998). Preincubation of the culture for 15-20 minutes with membrane-permeable inhibitors of Ins(1,4,5)P₃-induced Ca²⁺ release, XeC (1 μM) or 2-APB (50 μM), effectively blocked the increase of SSC frequency elicited by RA (Fig. 3A,B). The synaptic facilitation of RA under the presence of XeC and 2-APB was 5.1±1.5 (n=9) and 2.5±0.8 (n=6) times that of control, respectively. The release of Ca²⁺ from Ins(1,4,5)P₃ receptors could further trigger Ca²⁺-induced Ca²⁺ release from ryanodine receptors. Pretreatment of the cultures with ryanodine receptor antagonist TMB-8 (3 μM) or ruthenium red (10 μM) blocked the action of RA (3.1±0.9, n=15 and 1.5±0.3, n=5, times the control values, for TMB-8 and ruthenium red pretreatment, respectively; Fig. 3C,D). Thus, an intracellular liberation of Ca²⁺ from both Ins(1,4,5)P₃- and ryanodine-sensitive pools, rather than an influx of extracellular Ca²⁺, is responsible for the RA-induced synaptic facilitation.

How might RA come into play in mobilizing intracellular Ca²⁺ stores? To approach this problem, we examined the signaling pathway that is responsible for the action of RA in developing Xenopus neuromuscular synapses. Activation of PLC is an attractive candidate for the mediation of synaptic facilitation because its activation would result in intracellular Ca²⁺ release via the second messenger Ins(1,4,5)P₃. To evaluate whether PLC is part of the RA signaling mechanism facilitating neurotransmitter release, we set out to examine the effect of inhibition of PLC on the action of RA. Pretreatment of cells with the PLCγ inhibitor U73122 (5 μM) prior to RA treatment completely blunted the RA-induced increase in SSC frequency (5.8±1.9-fold of control, n=5; Fig. 4A), suggesting that PLCγ activity is required for RA-induced SSC frequency facilitation.

Two possible mechanisms are indicated for PLC activation: PLCβ by G protein-coupled receptor activation and PLCγ via kinase-mediated phosphorylation. PLCγ is phosphorylated by diverse receptor tyrosine kinases and nonreceptor protein tyrosine kinases through a high affinity interaction with the SH2 domain of PLCγ (Rhee, 2001). Also, it has been shown that the binding of the PH domain of PLCγ to phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P₃] present in the membrane as a result of PI 3-kinase activation leads to the activation of PLCγ (Bae et al., 1998; Falasca et al., 1998). To evaluate a possible participation of G protein in the RA-induced facilitating effects, GDP_βS (a non-hydrolyzable GDP analogue and inhibitor of G protein) were used. Loading presynaptic neurons with the G protein inhibitor GDP_βS (5 mM) prior to RA treatment did not affect the RA-induced SSC frequency facilitation (24.2±5.9-fold of control, n=5; Fig. 4B). These results suggest that the facilitating effects of RA on the spontaneous transmitter release are independent of a G protein serving as a signal transducer in the signaling pathway. Next, we addressed whether the inhibition of PI 3-kinase by its specific inhibitor wortmannin would impair RA-induced SSC frequency facilitation. The RA-induced SSC frequency facilitation was abolished in the presence of wortmannin (100 nM; 1.2±0.2-fold of control, n=8; Fig. 4C). Pretreatment with another PI 3-kinase inhibitor, LY294002 (5 μM), also prevented the RA-induced increase in SSC frequency (1.1±0.2-fold of control, n=7; Fig. 4D).

Recently, it has been suggested that RA modulates PKC activity by competing with acidic phospholipids to bind to the C2 domain of PKC. Furthermore, results from amino acid alignments and crystal structure analysis suggest the existence of an RA binding site on the PKC molecule (Ochoa et al., 2003). Bath application of a protein kinase C inhibitor H-9 (100 μM) or bisindolmaleimide (10 μM) failed to reduce the RA facilitating effect (19.1±6.3, n=9 and 18.1±4.6, n=5, times the control, respectively; Fig. 5A,B and Fig. 6), suggesting that PKC is exclusive in the signaling pathway of RA-induced synaptic facilitation. To go further into the mechanisms that relay the signal from RA/receptor complex to PI 3-kinase, we examined the possible involvement of other protein kinases in RA-induced SSC frequency facilitation. Pretreatment of the cultures with genistein (100 μM), a broad-spectrum tyrosine kinase inhibitor, completely abolished RA-enhanced SSC frequency, while 100 μM daidzein, an inactive analog of genistein, did not block the facilitating effect of RA (1.9±0.3 and 29.4±6.2-fold of control, n=5∼13 for genistein and daidzein, respectively; Fig. 5C,D and Fig. 6). Furthermore, the RA-induced SSC frequency facilitation was also abolished when 4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo[3,4-d] pyrimidine (PP2, 10 μM), which predominantly inhibits the Src family of nonreceptor tyrosine kinase, was present in the culture medium (1.0±0.2-fold of control, n=9; Fig. 5E, Fig. 6). The enhancement of spontaneous neurotransmitter release induced by RA was unaffected while MAP kinase inhibitor PD98059 (10 μM) was tested (26.1±12.5-fold of control, n=8; Fig. 5F, Fig. 6).

The effect of RA on intracellular Ca²⁺ was further investigated by imaging of a fluorescent intracellular Ca²⁺ probe. Cultures were loaded with Fura-2 AM, and ratio fluorometric measurement of intracellular Ca²⁺ was made on a single motoneuron. As shown in Fig. 7, perfusion of the cultures with RA either in Ringer's or Ca²⁺-free medium caused a significant rise in the [Ca²⁺]_i. The increase in the fluorescence ratio (F₃₄₀/F₃₈₀) under RA perfusion in Ringer or Ca²⁺-free medium was 129.0±19.6% and 113.5±21.0% of control, respectively. The effect of RA in mobilizing the intracellular Ca²⁺ store was next examined. The endoplasmic reticulum Ca²⁺ store was depleted by pretreatment of the culture with thapsigargin (TG; 2 μM) for 1 hour and the cells were then washed with Ca²⁺-free medium. Perfusion of these TG-treated cultures with RA in Ca²⁺-free medium failed to significantly change the fluorescence ratio (14.8±14.1%, Fig. 7D). The involvement of the PLCγ/PI 3-kinase/Src kinase signaling pathway was further tested. The PLCγ inhibitor U73122, the PI 3-kinase inhibitor LY294002, the tyrosine kinase inhibitor genistein and the Src kinase-specific inhibitor PP2 all effectively inhibited the RA-induced rise of intracellular Ca²⁺. The increase in fluorescence ratio under RA perfusion in the presence of the various inhibitors was 13.4±2.2% (U73122), 9.5±4.9% (LY294002), 6.7±2.1% (genistein) and 12.7±1.4% (PP2) of control (Fig. 7D).

Results from current studies demonstrate for the first time that through mobilization of intracellular Ca²⁺, RA induced a nongenomic and rapid effect on facilitation of spontaneous transmitter release in the developing neuromuscular synapses. Furthermore, our systematic investigation provided groundbreaking evidence that the RA-induced liberation of Ca²⁺ from intracellular store appears to be the result of pleiotropic signaling molecules involving PLCγ, PI 3-kinase and activation of Src tyrosine kinase. We examined the acute effect of RA on synaptic transmission in cultured Xenopus nerve-muscle cultures – a simple and easily accessible model – providing insight into related mechanisms. Cultures derived from Xenopus embryos offer several advantages in studying the early events of synaptogenesis. First, previous studies of neuromuscular synapses in Xenopus cell cultures provide detailed descriptions of the morphological and physiological events associated with various stages of development. Second, Xenopus myoblasts do not fuse to form polynucleated myotubes in culture but remain mono-nucleated as long as they survive, providing good conditions for whole cell patch-clamp recording and easy access for drug application. Third, the cells remain viable for many hours in the open air at room temperature on the microscope stage, which is ideal for electrophysiological recordings. We found that a rise in intracellular Ca²⁺ through mobilization of intracellular stores via Ins(1,4,5)P₃ receptors and ryanodine receptor activation, but not by Ca²⁺ influx from extracellular sources, were responsible for the synaptic facilitation induced by RA. Our experimental supports for this claim include the following observations. First, buffering the intracellular Ca²⁺ rise, using the membrane-permeable Ca²⁺ chelator BAPTA-AM, effectively prevented SSC frequency facilitation induced by RA. This suggests that RA-induced SSC frequency facilitation results from elevated intracellular Ca²⁺ concentration. Second, the effect of RA in facilitating transmitter release is not hampered when Ca²⁺ is omitted from the extracellular fluid or by pharmacological blockade of Ca²⁺ channels with Cd²⁺. In recent years, many studies have shown that RA induces L and/or N type Ca²⁺ channel expression and plays a major role in regulating intracellular Ca²⁺ homeostasis and cell excitability in human NT2N neurons and H9C2 cardiac cells (Gao et a., 1998; Menard et al., 1999). Furthermore, it has been suggested that long-term treatment of the neuroblastoma C1300 cell line with RA and activin A resulted in cooperative enhancement of the functional activity of L-type Ca²⁺ channels (Fukuhara et al., 1997). However, our experiments show that RA rapidly (within 8-15 minutes) facilitates spontaneous transmitter release either in Ca²⁺-free or Cd²⁺-containing medium, implying that RA modulates the machinery of transmitter release in a different way. Third, pretreatment of the culture with TG abolished the capacity of RA to raise [Ca²⁺]_i. Transmitter release modulated by the release of Ca²⁺ from intracellular stores has been shown in a number of systems, such as the cholinergic synapse in Aplysia and the reticulospinal synapse in lamprey, and sympathetic nerve terminals (Smith and Cunnane, 1996; Cochilla and Alford, 1998; Mothet et al., 1998). In the present study, RA induced Ca²⁺ release through Ins(1,4,5)P₃ receptors and subsequently triggered Ca²⁺-induced Ca²⁺ release from ryanodine receptors, leading to an increase in spontaneous transmitter release at the terminals of developing spinal neurons. The profile of the RA-induced rise in presynaptic [Ca²⁺]_i sheds the first light on this biphasic nature. Also, the finding that either Ins(1,4,5)P₃ or ryanodine receptor antagonist can effectively prevent synaptic facilitation induced by RA suggests that both pathways are necessary for the synaptic facilitation and the primary effect of RA is probably on the Ins(1,4,5)P₃ receptor.

With regard to a possible involvement of Ca²⁺ in the rapid regulatory role of RA in spontaneous synaptic transmission, the following findings of the nongenomic effects of steroid hormones and other lipophilic agents such as thyroid hormones should be mentioned. Aldosterone induces rapid increase in intracellular PKC activity and a rise of Ca²⁺ in human distal colon cells (Doolan et al., 1998). In cultured skeletal muscle cells, 1,25-dihydroxyvitamin D₃ produced a rise in [Ca²⁺]_i by promoting a non-genomic release of Ca²⁺ from internal stores via activation of PLC and D and Ins(1,4,5)P₃ receptor and by Ca²⁺ influx through L-type and store-operated Ca²⁺ channels (Capiati et al., 2001). Furthermore, Dufy-Barbe et al. (Dufy-Barbe et al., 1992) showed an increased Ca²⁺-dependent spiking activity in pituitary cells after estradiol application. Here we show the first evidence that events leading to the RA-induced synaptic facilitation involve PLCγ, PI 3-kinase and activation of Src tyrosine kinase. Activation of PLCγ is an attractive candidate for the mediation of synaptic facilitation because its activation would result in intracellular Ca²⁺ release via the second messenger Ins(1,4,5)P₃. There are two possible mechanisms that resulted in PLCγ activation. PLCγ is phosphorylated by diverse receptor tyrosine kinases and nonreceptor protein tyrosine kinase through a high affinity interaction with its SH2 domain (Rhee, 2001). It has also been shown that the binding of the PH domain of PLCγ to phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P₃] present in the membrane as a result of PI 3-kinase activation leads to the activation of PLCγ (Falasca et al., 1998; Bae et al., 2001). Results in our experiments show that either the PI 3-kinase inhibitor or the PLCγ inhibitor can effectively prevent the synaptic facilitation and [Ca²⁺]_i rise induced by RA, suggesting that PLCγ activated from the PI 3-kinase pathway appears to play an important role. How might RA come into play in activation of PI 3-kinase? PI 3-kinase, a heterodimeric complex composed of a regulatory 85-kDa and catalytic 110 kDa subunit, is typically activated by receptors with an intrinsic or associated protein tyrosine kinase activity, proteins that are tyrosine-phosphorylated in response to external stimuli, and by G protein-coupled receptors (Leevers et al., 1999). Although the usual action of RA is transcriptional regulation, our results, on the basis of the rapid RA-induced synaptic facilitation, suggest that RA may activate PI 3-kinase through a nongenomic action of its receptor. This is in line with an increasing number of other studies, which suggest the nongenomic activation of PI 3-kinase by steroid hormones and other lipophilic agents such as thyroid hormones. The involvement of PI 3-kinase activation in vitamin D₃-induced myeloid cell differentiation and progesterone-induced oocyte maturation in Xenopus has been reported (Hmama et al., 1999; Hehl et al., 2001). Furthermore, others have demonstrated that 17β-estradiol induces the nongenomic activities of its receptor ERα to evoke the PI 3-kinase/AKT signaling pathway committed to the regulation of cell proliferation (Acconcia et al., 2004). In further support of this hypothesis, it has recently been suggested that an extragenomic activation of PI 3-kinase by RA is required for neural differentiation of SH-SY5Y human neuroblastoma cells (Lopez-Carballo et al., 2002).

An important mechanistic question that arises from the results shown here is the nature of the molecular mechanism by which RA activates PI 3-kinase. Much to our surprise, while trying to go further into the mechanisms that relay the signal from the RA receptor to PI 3-kinase, we found the first evidence of the participation of Src tyrosine kinase in the synaptic facilitating effect of RA. Preincubation of the cultures with pharmacological inhibitors genistein, a broad-spectrum tyrosine kinase inhibitor, or PP2, which predominantly inhibits the Src family of nonreceptor tyrosine kinase, completely abolished RA-induced synaptic facilitation and rise in [Ca²⁺]_i. It is known that nonreceptor tyrosine kinases, such as JAKs, Syk, Src and ZAP70, which are recruited and activated by a variety of receptors, are involved in regulation of PI 3-kinase activity (Cantrell, 2001). For example, there are compelling results showing that Src kinase, acting through a PI 3-kinase-dependent pathway, is required for neuronal survival and neurite outgrowth, suggesting that Src kinase may be involved in the regulation of PI 3-kinase activity in the nervous system (Encinas et al., 2001). Although there is no information in the literature to date concerning the association of Src kinase and the RA receptor, it is worthwhile to note that there is increasing evidence on the role and biological significance of the Src family of nonreceptor tyrosine kinase within the rapid, nongenomic action of steroid and thyroid hormones (Shupnik, 2004). There is evidence indicating that in chick skeletal muscle cells, 1α, 25 dihydroxyvitamin D₃ promotes complex formation between vitamin D receptor and Src kinase and is thus responsible for the rapid modulation of Ca²⁺ influx by opening both L-type voltage-dependent and store-operated Ca²⁺ channels (Buitrago et al., 2001). In rat enterocytes, the Src kinase is involved in parathyroid hormone-dependent PI 3-kinase activation (Gentili et al., 2002). It has been suggested that Src kinase plays a central role in estrogen and progesterone-dependent PI 3-kinase and MAPK activation (Migliaccio et al., 1996). Moreover, recent results from binding assays revealed that the estrogen receptor interacted with the SH2 domain of Src, whereas the androgen receptor and progesterone receptor interacted with SH3 domain of Src kinase (Boonyaratanakornkit et al., 2001). Precisely how Src kinase couples the RA receptor to PI 3-kinase remains a subject for further study. A possible mechanism by which RA stimulates Src activity in Xenopus developing neuromuscular synapses may be to bind RA to its receptor, thus inducing a conformational change on this protein, which is then sensed by the Src kinase. We have previously provided evidence that the activation of RARβ receptor is responsible for RA-induced SSC frequency facilitation (Liao et al., 2004). It could be informative in the future to explore, by the use of amino acid alignments and crystal structure analysis, if there is a Src kinase-binding motif in RARβ.

A rich history of research dating back to the time of Hans Selye (1942) supports the observations that apart from the generally accepted theory of action of the steroid-thyroid-retinoid hormone nuclear receptor superfamily, there also exists a nongenomic action mechanism in many steroids. Much of our knowledge about the nongenomic effect of the nuclear receptor superfamily comes from the studies of steroid hormones; however, no information is available for the nongenomic effects of RA. During development, the RA receptors are present in developing motoneurons, and high levels of RA have been detected in the spinal cord (Maden et al., 1998). In addition to its acknowledged role in patterning both the anteroposterior and dorsoventral axes, compelling evidence has suggested that RA also has considerable significance as a neural differentiation factor (Maden et al., 1998; Prince and Carlone, 2003).

Previously we have shown the first physiological evidence that RA potentially enhances the spontaneous transmitter release at the developing neuromuscular synapses. Here our results provide not only a deeper insight in understanding the activation of Src kinase/PI 3-kinase as a novel signaling pathway in response to RA, but also open up an entirely new area of investigation in the field of nongenomic regulation of synaptic function elicited by RA, since these rapid changes in synaptic activity may represent a mechanism whereby synapse formation is initiated. What is the functional significance of facilitating ACh secretion by RA during the early phase of synaptogenesis? Neuronal activity at developing synapses is crucial in synapse maturation and competition as well as in the differentiation of postsynaptic properties (Balice-Gordon and Lichtman, 1993). The potentiation of the spontaneous ACh release at developing neuromuscular synapses may have profound developmental significance. Several studies have indicated that the gene expression and secretion of neurotrophic factors NT-3 and NT-4 in the neuromuscular junction are regulated by synaptic activity (Liou and Fu, 1997; Xie et al., 1997). It has also been suggested that activity-dependent secretion of neurotrophic factors is important in synaptic activity regulation and may be involved in Hebbian-type homosynaptic potentiation (Poo, 2001). Furthermore, SSCs at developing neuromuscular junctions in Xenopus cultures are capable of eliciting action potentials and spontaneous contractions in muscle cells. This frequent supra-threshold excitation produces a global influence on the development of contractile properties of the postsynaptic muscle cell. In addition, spontaneous synaptic potentials are accompanied by a localized influx of ions, including Ca²⁺, at the subsynaptic site of the muscle (Decker and Dani, 1990). Local Ca²⁺ accumulation and the consequent Ca²⁺-dependent enzymatic reactions are likely to play an important role in regulating the development of postsynaptic structure. Overall, conclusions drawn from current studies suggest that RA rapidly elicits synaptic facilitation through mobilizing Ca²⁺ from Ins(1,4,5)P₃ and/or ryanodine-sensitive intracellular Ca²⁺ stores of the presynaptic nerve terminal. This involves consecutive activation of pleiotropic signaling molecules including PLCγ, PI 3-kinase and Src kinase, leading to an enhancement of spontaneous transmitter release. It thus may have significant roles in initiating the consecutive and complex cross-interaction between presynaptic motoneurons and postsynaptic muscle cells that then lead to the maturation of the neuromuscular synapse.

The authors are most grateful for the continuous help provided by W.-M. Fu and Y.-J. Chan. This work was supported by the University Integration Program from the Ministry of Education and by a grant from the National Science Council of Taiwan (NSC 92-2320-B-110-012).