Platelet-derived growth factor induces a long-term inhibition of N-methyl-D-aspartate receptor function.

Platelet-derived growth factor (PDGF) is a multifunctional protein that plays important roles in many tissues, including the mammalian central nervous system. PDGF and PDGF receptors (PDGFRs) are expressed in virtually every region of the central nervous system where they are involved in the development, survival, growth, and differentiation of both neuronal and glial cells. We now report that a brief activation of PDGFRs produced a long-lasting inhibition of N-methyl-D-aspartate (NMDA)-dependent excitatory postsynaptic currents in CA1 pyramidal neurons in rat hippocampal slices. PDGF also inhibited NMDA receptors (NMDA-Rs) in cultured hippocampal neurons by a mechanism that involves a decrease in single channel open probability. Non-NMDA receptor function was not affected by PDGF in hippocampal neurons. Experiments with mutant PDGFRs and chelation of intracellular Ca2+ in Xenopus oocytes indicate that this inhibition depends on a phospholipase C-γ-induced elevation of intracellular Ca2+ levels. The PDGF-induced inhibition of NMDA-Rs is produced by a mechanism different than the well characterized phenomenon of Ca2+-dependent NMDA-R run down because the effect of PDGF was blocked by the phosphatase inhibitor, calyculin A, and was not affected by the microtubule polymerizing agent, phalloidin. Because elevations of PDGF levels are associated with neurological trauma or disease, we propose that PDGF can exert neuroprotective effects by inhibiting NMDA-R-dependent excitotoxicity.

PDGFs are multifunctional proteins that regulate a number of physiological and pathophysiological processes, including embryonic and placental development, wound healing, atherosclerosis, cancer, renal diseases, and arthritis (1). In addition to its role in these processes, PDGF is particularly important for the regulation of both the developing and mature central nervous system. In contrast to the restricted localization of most neurotrophic factors, PDGFs and PDGFRs are expressed in virtually every region of the mammalian central nervous system (5)(6)(7)(8)(9). In the developing central nervous system, PDGF is important for the normal formation of neural plates and neural tubes (10), for the differentiation of progenitor oligodendrocyte cells (11), and for the chemotaxis and proliferation of glial cells (8,12). In the mature central nervous system, PDGF is important in the pathophysiology of several disease states. Cell lines from malignant glioma and other central nervous system tumors express PDGFs and PDGFRs, and it has been suggested that growth of some of these tumors could be mediated by an autocrine PDGF/PDGFR loop (13,14). Moreover, PDGF levels are elevated in non-neoplastic diseases of the central nervous system such as trauma, stroke, meningitis, cerebral abscesses, and glial and meningeal cysts (13,15,16). It is likely that this elevation in PDGF levels is involved not only in the pathogenesis of these conditions but also in the tissue repair processes associated with these diseases. In this regard, it has been shown that PDGF exerts neurotrophic effects on GABAergic and dopaminergic neurons (17,18) and that it protects hip-pocampal neurons against energy deprivation and oxidative injury in vitro (19).
In spite of the importance of PDGF for the development and maintenance of the mammalian central nervous system, little is known about its actions on synaptic transmission. To contribute to this issue, we examined the effects of this growth factor on the function of the N-methyl-D-aspartate subtype of glutamate receptors (NMDA-Rs). NMDA-Rs mediate excitatory synaptic transmission in the central nervous system and play important roles in many physiological and pathophysiological processes such as neuronal development and survival, synaptic plasticity, and neurotoxicity (20). We examined the effects of PDGF on NMDA-Rs in hippocampal slices, cultured hippocampal neurons, and Xenopus oocytes. Electrophysiological experiments with these preparations indicate that PDGF exerts long-lasting modulatory effects on the function of NMDA-Rs and that the mechanism of action of PDGF involves a complex intracellular signal transduction cascade that is triggered by PDGFR activation.
Cultured mouse hippocampal neurons were grown according to previously described procedures (23) and were used for patch clamp recordings 12-20 days after plating. Recording electrodes with resistances of 3-5 megaohms were constructed from thin-walled borosilicate glass (1.5 mm diameter, WPI Inc., Sarasota, FL). Patch clamp recordings were made in the perforated patch and cell-attached configurations using Axopatch-1B amplifiers (Axon Instruments, Foster City, CA). For perforated patch recordings, data were digitized, filtered (2 kHz), and acquired on-line using the program pClamp 5.5.1 (Axon Instruments). For cell-attached recordings, single channel events were first recorded on videotape using a digital data recorder (VR-10, Instrutech Corp., Mineola, NY) and later played back and acquired using the pClamp 6 program (Axon Instruments). Single channel currents were filtered at 2 kHz and sampled at 5 kHz. Only patches with stable basal activities were used to ensure that the changes in activity were not due to run down or random fluctuations. To study the effects of PDGF on open probability of NMDA channels, a control period of 5 min to record the basal activity was allowed before the introduction of PDGF. The single channel open probability was determined from the ratio of the time spent in the open state to the duration of recording, P o ϭ (t 1 ϩ t 2 ϩ ⅐⅐⅐ ϩ t n )/Nt tot , where t is the amount of time that n channels are open, and N is the maximum number of levels observed in the patch. The extracellular solution contained (in mM) NaCl (140), CaCl 2 (1.3), KCl (5.4), HEPES (25), glucose (33), tetrodotoxin (0.0005-0.001) (pH 7.4, using NaOH, 320 -335 mOsm). For perforated patch recordings, 3 M glycine and 100 M NMDA were added to the extracellular solution to evoke NMDA currents. For non-NMDA currents, 200 M kainate was added to the extracellular solution. Perforated patch electrodes were filled with (in mM) KCl (55), K 2 SO 4 (70), MgCl 2 (7), HEPES (10), glucose (5), and nystatin 0.3 mg/ml. For cell-attached recording, patch electrodes contained (in mM) NaCl (70), Na 2 SO 4 (70), HEPES (10), CaCl 2 (1.3), CsSO 4 (5), and glucose (33). Glycine 1-3 and NMDA 10 M were also added to the electrode solution to induce NMDA channel current. All experiments were performed at room temperature (20 -22°C). A multi-barrel perfusion system was employed to achieve a rapid exchange of solutions.
Microinjection and Electrophysiological Recording of Xenopus Oo-cytes-Human NMDA receptor subunits cDNAs (NR1a, NR2A, NR2B) were cloned on the eukaryotic expression vector pCDNA-I-Amp (Invitrogen Corp., San Diego, CA); the cloning of these subunits is described elsewhere (24). Human wild-type PDGFR-␤ were cloned on pBS-as described by Kazlauskas et al. (25). The construction of the F5 and the Tyr 1021 add-back PDGFR-␤ mutants have been described elsewhere (3). Subcloning of PDGFR-␤ mutants and in vitro cRNA synthesis of PDGFRs was performed as described by Valenzuela et al. (22). The methods used for oocyte preparation/culture, cRNA/cDNA microinjection, and drug application/microinjection are the same as those described by Valenzuela et al. (22), except that Mg 2ϩ -free modified Barth's solution was used instead of complete modified Barth's solution.
Statistical Analysis-In most cases, currents are expressed as percentages of control responses. In the case of Xenopus oocytes, control responses correspond to, at least, 2-3 NMDA currents determined before PDGF-BB application. In all cases, the maximal percent PDGFRinduced inhibition was calculated relative to the average of these control responses. All values in this article are given as mean Ϯ S.E., and n values refer to the number of different oocytes, cells, or slices used in the statistical analysis. Statistical analyses were performed by using either Student's t tests or two-way ANOVA by using the Solo computer program (BMPD Statistical Software, Los Angeles, CA). Nonlinear regression curve fitting was performed with GraphPad computer program (San Diego, CA).

Effects of PDGFR Activation on Glutamate Receptor Function in Hippocampal Slices and Cultured
Hippocampal Neurons-We examined the effects of PDGFR activation on pharmacologically isolated NMDA-R-mediated EPSCs in the hippocampus, a brain region where these two types of receptors are known to be coexpressed (7,19,26). Application of PDGF-BB (6 nM for 3 min) significantly decreased the amplitude of NMDA-R-mediated EPSCs (by 44 Ϯ 7%, p Ͻ 0.001 by two-way ANOVA, n ϭ 7, Fig. 1). The inhibition began soon after the onset of PDGF-BB application and was not reversed even after a 20-min PDGF washout period. No apparent recovery was observed even in cells held for more than 1 h following PDGF treatment. Hippocampal slices treated with vehicle only did not display any inhibition of NMDA-R-mediated EPSCs under the same recording conditions (Fig. 1). Non-NMDA-dependent EPSCs were not affected by PDGF treatment; the value for the non-NMDA-dependent EPSC in the PDGFtreated slices was 110 Ϯ 6% of control (p Ͼ 0.05 by t test, n ϭ 5).
Application of PDGF-BB (7 nM for 15 min) to rat cultured hippocampal CA1 pyramidal neurons in the perforated patch configuration significantly decreased the amplitude of NMDAgated currents (by 23 Ϯ 3%, p Ͻ 0.001 by two-way ANOVA, n ϭ 8) (Fig. 2). The inhibitory effects of PDGF-BB appeared gradually over the 15-min application and were not reversible after a 10-min washout period. Control cells treated with vehicle only did not display any run down of NMDA-R-dependent currents under the same recording conditions (Fig. 2). The ratio of peak/steady state NMDA-gated currents was significantly reduced (p Ͻ 0.05 by t test) by PDGFR activation; the ratios in control and PDGF-treated neurons were 1.7 Ϯ 0.07 and 1.5 Ϯ 0.1 (n ϭ 9), respectively. Kainate-evoked currents were not affected by PDGF treatment in cultured hippocampal neurons under the same recording conditions (Fig. 2); the values for kainate-gated currents after 5, 10, and 15 min of PDGF-BB application were 99.5 Ϯ 2, 101 Ϯ 3, and 102 Ϯ 4% of control (n ϭ 7).
The effects of PDGF on NMDA-R function were also studied at the single channel level (Table I, (Table I). Closed time (2) was significantly increased (p Ͻ 0.05 by t test) by PDGF treatment from 12.3 Ϯ 2 to 19.5 Ϯ 3 ms, which is a reflection of the lower frequency of channel openings in the treated group.
We next determined the effects of PDGFR activation on the NMDA and glycine dose-response curves (Fig. 4, B and C). In this batch of oocytes, PDGFR activation produced a 45 Ϯ 8% (n ϭ 9) decrease in the NMDA E max (p Ͻ 0.001 by two-way ANOVA) with no significant change in the EC 50 (p Ͼ 0.8 by t test); the NMDA EC 50 values before and during PDGFR inhibition were 77 Ϯ 16 and 84 Ϯ 25 M, respectively. The NMDA Hill coefficients were not affected by PDGFR activation; the values before and after PDGFR activation were 1.7 Ϯ 0.2 and 1.9 Ϯ 1.3, respectively. PDGFR activation produced a significant (p Ͻ 0.001 by two-way ANOVA) decrease of 65 Ϯ 7% (n ϭ 10) in the glycine E max with no significant change in the EC 50 (p Ͼ 0.7 by t test); the glycine EC 50 values before and during PDGFR inhibition were 4 Ϯ 0.6 and 4 Ϯ 0.7 M, respectively. The glycine Hill coefficients were not affected by PDGFR activation; the values before and after PDGFR activation were 2 Ϯ 0.3 and 1.9 Ϯ 0.4, respectively.
Finally, we measured the effect of PDGFR activation on the NMDA-R current/voltage relationships. PDGFR activation inhibited NMDA-R currents independently of the membrane holding potential (Fig. 4D). The reversal potentials for the NMDA-R-mediated currents were not significantly (p Ͼ 0.5 by t test, n ϭ 9) affected by PDGFR activation; the values before and during PDGFR-induced inhibition were Ϫ26 Ϯ 3 and Ϫ25 Ϯ 3 mV, respectively.
Effect of Calyculin A and Phalloidin-We next tested the effects of the phosphatase inhibitor, calyculin A, and of the microtubule polymerizing agent, phalloidin, to determine whether the PDGF-induced inhibition of NMDA-Rs was produced by a similar mechanism to that of the phenomenon of Ca 2ϩ -dependent run down described by Rosenmund and Westbrook (27,28). These investigators showed that Ca 2ϩ -dependent NMDA-R run down was unaffected by phosphatase inhibitors (27) and blocked by the microtubule polymerizing agent, phalloidin (28). However, microinjection of the potent inhibitor of protein phosphatases 1 (PP1) and 2A (PP2A), calyculin A (20 -30 nM), into Xenopus oocytes significantly blocked (p Ͻ 0.05 by two-way ANOVA, n ϭ 10) the PDGFR-induced inhibition of NMDA-R function (Fig. 6A). Treatment of cultured hippocampal neurons with calyculin A (40 nM) also significantly blocked the PDGF inhibitory effects on both NMDA-R whole cell currents (p Ͻ 0.001 by two-way ANOVA, n ϭ 7) (Fig.   6B) and single channel open probability (Fig. 6C, Table I). Moreover, phalloidin did not significantly block the effects of PDGF on NMDA-R function in Xenopus oocytes (Fig. 6D, p Ͼ 0.15 by two-way ANOVA). It should be noted that we also tested the effects of deltamethrin, a potent inhibitor of the Ca 2ϩ /calmodulin-dependent protein phosphatase 2B, calcineurin. Microinjection of deltamethrin into Xenopus oocytes did not block the inhibitory actions of PDGFR; the PDGFRinduced maximal inhibition of NMDA-Rs in control and deltamethrin-treated oocytes (200 nM for 6 -9 h) was 66 Ϯ 12 and 72 Ϯ 5%, respectively (n ϭ 7-10).

DISCUSSION
In spite of being linked by history and name to platelets, PDGF should be considered a "classical" neurotrophic factor from a functional perspective. PDGF, like the neurothrophins, is 1) produced locally and is important for the development, differentiation, proliferation, and survival of neuronal and glial cells (7,10,11,17); 2) coupled to tyrosine kinase receptors that activate complex intracellular signaling pathways (2); and 3) released as part of the compensatory response to central nervous system injury or disease (13,15,16). We now report that PDGF exerts another function that is characteristic of the neurotrophic factors, which is the modulation of neurotransmitter receptors in the central nervous system (29).
Regulation of NMDA-R Function by PDGF-Our experiments demonstrate that PDGF is a potent modulator of NMDA-Rs. PDGFR activation produced a long-lasting inhibition of NMDA-Rs in cultured hippocampal neurons and in Xenopus oocytes and also inhibited synaptically evoked NMDAdependent EPSCs in CA1 pyramidal neurons in hippocampal slices. The inhibition gradually appeared within minutes of PDGF application and lasted for at least 20 min in both cultured hippocampal neurons and hippocampal slices and for more than 3 h in the oocytes. At the single channel level, the inhibition was produced by a decrease in the open channel probability and not by a decrease in single channel conductance or open time(s). PDGFR activation decreased the efficacy of both NMDA and glycine but not their potency for NMDA-Rs. PDGFR activation inhibited NMDA-R function independently of the membrane holding potential and did not affect the reversal potential. The inhibitory effects of PDGF were specific for NMDA-Rs since this growth factor did not affect synaptically evoked non-NMDA-mediated EPSCs in CA1 pyramidal neurons in hippocampal slices or kainate-gated currents in cultured hippocampal neurons. Taken together, these findings suggest that PDGF receptor activation specifically affects NMDA-R function and that it does not affect glutamate release. An effect of PDGF on glutamate release would be expected to affect both NMDA-and non-NMDA-dependent currents. In Kainate currents were recorded in the presence and absence of PDGF at the same time intervals as NMDA currents; kainate currents 5, 10, and 15 min after PDGF application were 99 Ϯ 2, 101 Ϯ 3, 102 Ϯ 4% of control (n ϭ 7). PDGF-BB concentration was 7 nM. Currents were activated by a 2-s application of NMDA (100 M) and glycine (3 M) or kainate (200 M). Holding potential was Ϫ60 mV. Shown in the two-axis graph is a summary of the effects of PDGF on NMDA-activated currents in cultured hippocampal neurons. Perfusion of 7 nM PDGF (q) gradually decreased the peak amplitude of NMDA currents (p Ͻ 0.001 by two-way ANOVA, n ϭ 5 for untreated and 8 for treated neurons). The inhibition of NMDA currents was not reversed after washout of PDGF for more than 10 min. Currents recorded from untreated neurons (E) did not display any run down over the same period of time. All current amplitudes were normalized to that recorded 10 min after seal formation. addition, PDGF not only affected synaptically evoked NMDA currents but also currents produced by both application of NMDA and glycine to cultured hippocampal neurons and Xenopus oocytes.
Studies with PDGFR-␤ mutants expressed in Xenopus oocytes provided two important pieces of information about the mechanism of the inhibition of NMDA-Rs. First, these experiments indicate that the intrinsic tyrosine kinase activity of  Table I). PDGFR-␤ is not sufficient to produce inhibition of NMDA-Rs because the F5 PDGFR-␤ mutant, which does not activate a number of SH2 domain proteins but possesses intact intrinsic tyrosine kinase activity, did not inhibit NMDA-R function. Second, these experiments show that the SH2 domain protein that relays the inhibitory signal from PDGFR to NMDA-Rs is PLC-␥ because restoration of the activation site for PLC-␥ (Tyr 1021 PDGFR-␤ add-back mutant) rescues the inhibitory actions of PDGFR. In addition, the effects of PDGFR-␤ require an elevation of intracellular Ca 2ϩ levels because microinjection of EGTA into Xenopus oocytes before PDGFR activation blocked its inhibitory actions. Microinjection of EGTA after maximal inhibition was reached did not block the effects of PDGF, suggesting that a transient elevation of intracellular Ca 2ϩ levels produces activation of an NMDA-R intracellular modulator. This modulator appears to be PP1 and/or PP2A because the phosphatase inhibitor, calyculin A, blocked the inhibitory effects of PDGF. This signaling cascade is schematically shown in Fig. 7.
Our finding that the PDGFR-induced inhibition of NMDA-R function is blocked by calyculin A is consistent with a number of recent reports showing that protein phosphatases decrease NMDA-R function. Wang et al. (30) reported that NMDA receptor currents are enhanced by calyculin A in cultured hippocampal neurons studied with the perforated patch technique. The authors also found that, like PDGF, PP1 and PP2A decrease the open probability of NMDA-Rs in inside-out patches (30). Inhibition of calcineurin (PP2B) resulted in prolonged single channel openings recorded with the cell-attached patch technique in adult rat dentate gyrus neurons (31). Moreover, calcineurin (PP2B), but not of PP1 and PP2A, appear to be involved in the development of the glycine-insensitive form of NMDA-R desensitization (32,33). In addition, tyrosine kinase inhibitors decrease NMDA-R function in spinal dorsal horn neurons whereas tyrosine phosphatase inhibitors enhance its function, suggesting that tyrosine phosphatases inhibit NMDA-R function (34). Taken together, these studies indicate that protein phosphatases exert inhibitory actions on NMDA-R function and, consequently, are consistent with our finding that the PDGFR-induced inhibition of NMDA-Rs is mediated by PP1 and/or PP2A. It should be emphasized, however, that whether phosphatases produce NMDA-R inhibition by directly dephosphorylating the receptor or by acting indirectly on a NMDA-R regulatory protein remains to be determined biochemically.
The precise role that Ca 2ϩ plays on the PDGFR-induced inhibition of NMDA-Rs is unclear. PP1 and PP2A are not directly regulated by Ca 2ϩ , unlike calcineurin (PP2B) which is activated by Ca 2ϩ and calmodulin. A signal transduction cascade where calcineurin activates PP1 via dephosphorylation of the endogenous PP inhibitor-1 was recently described (35). However, our results are inconsistent with this mechanism because the calcineurin inhibitor deltamethrin did not block the inhibitory actions of PDGFR-␤ in Xenopus oocytes. Consequently, the elevation of intracellular Ca 2ϩ levels could produce activation of PP1 and/or PP2A by a mechanism different than the calcineurin-dependent dephosphorylation of PP inhibitor-1. The endogenous PP inhibitor-1 and the dopamine/ cAMP-regulated phosphoprotein-32 are activated by protein kinase A (36), and elevations in intracellular Ca 2ϩ levels mediated by L-type calcium channels have been shown to decrease both adenylyl cyclase activity and cAMP levels in cardiac myocytes (37). Therefore, it is possible that the PDGFR-induced elevation of intracellular Ca 2ϩ levels results in inhibition of protein kinase A activity which, in turn, could decrease the activities of PP inhibitor-1 and/or dopamine/cAMP-regulated phosphoprotein-32. Whether this is the mechanism by which the elevation in intracellular Ca 2ϩ levels results in activation of PP1 and/or PP2A remains to be tested directly; it should be kept in mind that others have shown that elevation of intracellular Ca 2ϩ levels stimulate protein kinase A activity (38). Another mechanism by which elevations in intracellular Ca 2ϩ levels could result in activation of PP1 and/or -2B could involve  0.001 by t test). B, shown is the 6 nM PDGF-induced maximal inhibition of NMDA receptor responses in oocytes expressing NR1a/2A subunits with wild-type PDGFR-␤. Effects of PDGF were measured in control oocytes (n ϭ 10), oocytes microinjected with 500 M EGTA 10 -15 min before PDGF application (n ϭ 11), and oocytes microinjected with 500 M EGTA ϳ20 min after PDGF application (n ϭ 6). Microinjection of EGTA into the oocytes before PDGF-BB application significantly reduced the PDGFR-induced inhibition of NMDA-Rs (p Ͻ 0.005 by t test). NMDA and glycine concentrations were 100 and 10 M, respectively. tyrosine kinases. Increases in intracellular Ca 2ϩ levels are known to activate tyrosine kinases (39,40), and the activity of PP inhibitor-2 is inhibited by tyrosine phosphorylation in vitro (41). Thus, the elevation in intracellular Ca 2ϩ levels could result in activation of tyrosine kinases and inhibition of PP inhibitor-2 activity. It would be interesting to determine whether the PDGFR-induced elevation of intracellular Ca 2ϩ levels results in activation of PP1 and/or PP2A via these mechanisms or via other, as of yet, unidentified intracellular signaling cascades.
The PDGF-induced inhibition of NMDA-Rs appears to be different from both the Ca 2ϩ -dependent inactivation and run down of NMDA-Rs reported by several laboratories (27,28,(42)(43)(44)(45). Ca 2ϩ -dependent inactivation of NMDA-Rs is characterized by transient (10 -50 s) inhibition of ϳ50% that is not modulated by ATP and phosphatase or protease inhibitors and can be triggered by Ca 2ϩ entrance through NMDA-Rs or voltage-gated Ca 2ϩ channels (43,45). Ca 2ϩ -dependent NMDA-R run down occurs when intracellular Ca 2ϩ levels are elevated by repeated (every ϳ30 s) receptor activation (27,28). Run down is characterized by inhibition of ϳ50% that requires minutes to develop and by a reversibility rate that is dependent on the NMDA concentration used. NMDA-R run down does not occur with infrequent activation of NMDA-Rs, in Ca 2ϩ -free media, in the presence of an ATP-regenerating solution or when depolymerization of the actin cytoskeleton is prevented by application of phalloidin (27,28). Importantly, NMDA-R run down is not mimicked by intracellular dialysis of protein phosphatases (al-kaline phosphatase, PP1, and calcineurin) or blocked by phosphatase inhibitors (okadaic acid and microcystin) (27). Since the PDGF-induced inhibition of NMDA-Rs is blocked by the phosphatase inhibitor, calyculin A, and is not affected by phalloidin, present results suggest that PDGFR activation modulates NMDA-R function via a mechanism different from that of the Ca 2ϩ -dependent inactivation or the Ca 2ϩ -dependent run down of NMDA-Rs.

Modulation of Glutamate Receptors by Other Growth
Factors-Evidence in favor of the importance of growth factors as regulators of glutamate receptor function is beginning to emerge from several laboratories. Basic fibroblast growth factor enhances the elevation of intracellular Ca 2ϩ levels produced by activation of AMPA receptors but inhibits Ca 2ϩ responses produced by NMDA-R activation (46). Activation of tyrosine kinase-coupled insulin receptors produces a long-lasting potentiation of NMDA-mediated EPSCs in hippocampal slices (47). BDNF and NT-4/5 produce a transient augmentation of AMPAmediated synaptic currents and a transient increase in the frequency of miniature excitatory postsynaptic currents in cultured embryonic and postnatal rat hippocampal neurons (48). Moreover, Levine et al. (49) showed that BDNF rapidly enhances spontaneous firing rates and excitatory postsynaptic currents in cultured hippocampal neurons, and Kang and Schuman (50) demonstrated that BDNF and NT-3, but not NGF, produced a long-lasting enhancement of excitatory synaptic transmission in the Schaffer collateral-CA1 synapses. Kang and Schuman (50) also found that long term potentiation  Table I). D, oocytes coexpressing NR1a/2A human NMDA-R subunits and PDGFR-␤ were microinjected 15 min before PDGF-BB (6 nM) application with 100 M phalloidin (f). Untreated oocytes (Ⅺ) were from the same batch as the phalloidin-treated oocytes. Each point represents the mean Ϯ S.E. of 7 (control) and 10 (phalloidin-treated) oocytes. Phalloidin did not significantly affect the effect of PDGF (p Ͼ 0.15 by two-way ANOVA). NMDA and glycine concentrations were 100 and 10 M, respectively. could still be induced in slices where excitatory synaptic transmission had been enhanced by BDNF or NT-3. Interestingly, it was recently reported that hippocampal LTP appears to be impaired in mice deficient for the BDNF gene (51). In the case of PDGF, recent experiments in our laboratory indicate that it does not affect the generation, duration, or magnitude of NMDA-dependent LTP induced by tetanic stimulation (100 Hz/1 s) of the Schaffer collateral/commissural pathway to the CA1 region of the hippocampus. 2 The reason why PDGF did not block NMDA-dependent LTP under our recording conditions is unclear, but it might be due to PDGF-induced inhibition of GABAergic function in the hippocampus (22). Antagonism of GABAergic inhibition leads to an increase in LTP (52), and this may have counteracted the partial inhibition of NMDA-R function. Consequently, these results suggest that different classes of neurotrophic factors produce diverse effects on LTP and that they may be important modulators of this and other forms of synaptic plasticity in the central nervous system.
Modulation of Other Ion Channels by PDGF-Work from our laboratory has recently shown that PDGF also modulates the function of GABA A receptors, which mediate the majority of fast inhibitory synaptic transmission in the central nervous system (22). PDGF induced a long-lasting inhibition of synaptically evoked GABA A -mediated inhibitory postsynaptic currents in CA1 pyramidal hippocampal neurons, mouse brain microsacs, and Xenopus oocytes. As for NMDA-Rs, the inhibition also depended on a PLC-␥-induced elevation of intracellular Ca 2ϩ levels. Voltage-gated ion channels are also modulated by PDGF. Timpe and Fantl (53) studied the effect of PDGF and fibroblast growth factor (FGF) on Xenopus oocytes coexpressing PDGFRs or FGF receptors and voltage-gated K ϩ channels. The authors found that PDGF and FGF inhibited the current amplitude of the K ϩ channels and that the inhibitory action of the growth factors depended on a PLC-␥-induced elevation of intracellular Ca 2ϩ levels. This inhibition was mimicked by intracellular application of inositol 1,4,5-triphosphate and phorbol 12-myristate 13-acetate which led the authors to conclude that protein kinase C was involved in this process. Although the effects of PDGF on K ϩ channels and other voltage-gated ion channels have not been studied in neurons, these findings suggest that PDGF is a modulator not only of the function of both excitatory and inhibitory neurotransmitter-gated ion channels but also of voltage-gated ion channels in the central nervous system. It should be emphasized that our finding that non-NMDA receptors are not affected by PDGF indicates that not all channels are targets for the acute modulatory actions of this growth factor.
Relevance of the Effects of PDGF on NMDA-R Function-Since PDGF and PDGFRs are widely expressed throughout the mammalian central nervous system, our findings suggest PDGF as an important modulator of synaptic neurotransmission, not only in the hippocampus but also in other brain regions. NMDA receptors are important for the development of the central nervous system, and, consequently, the long-lasting modulatory actions of PDGF may be important during embryogenesis. PDGF is also likely to be important for the normal functioning of the adult central nervous system. Indeed, evidence from behavioral studies indicate that PDGF exerts modulatory actions in the nervous system at the level of feeding regulation. Intracerebroventricular microinfusion of PDGF suppressed food intake in rats (54), and application of PDGF to Hydra depressed a component of the feeding response to glutathione (55). Interestingly, NMDA-Rs appear to be important mediators of the eating response in rats (56). Therefore, it will be important to determine whether the cross-communication between PDGF and NMDA-Rs is part of the system that controls food intake behavior in the brain.
Elevations of PDGF levels in the central nervous system have been detected during neurological diseases associated with excitotoxicity and neuronal death such as infections, trauma, and cerebrovascular ischemic disease (13,15,16). Therefore, the inhibitory actions of PDGF on NMDA-R function could also be important in the pathophysiology of these conditions because a decrease in Ca 2ϩ influx through NMDA-Rs should help to restore homeostasis and prevent cell death. The neuroprotective effects of PDGF were recently demonstrated in vitro by Cheng and Matsson (19). The authors showed that PDGF protects cultured hippocampal and cortical neurons against glucose deprivation and oxidative injury-dependent neurotoxicity, and they suggested that this is due to an increase in cellular antioxidant enzymes such as catalase, glutathione peroxidase, and superoxide dismutase. Our results indicate that, in addition to this mechanism, PDGF might exert its neuroprotective actions by inhibiting Ca 2ϩ influx via NMDA-Rs. A challenging task for future research will be to determine whether PDGF plays these neuroprotective roles in the injured brain in vivo.