Report

Dichotomous Dopaminergic Control of Striatal Synaptic Plasticity

See allHide authors and affiliations

Science  08 Aug 2008:
Vol. 321, Issue 5890, pp. 848-851
DOI: 10.1126/science.1160575

Abstract

At synapses between cortical pyramidal neurons and principal striatal medium spiny neurons (MSNs), postsynaptic D1 and D2 dopamine (DA) receptors are postulated to be necessary for the induction of long-term potentiation and depression, respectively—forms of plasticity thought to underlie associative learning. Because these receptors are restricted to two distinct MSN populations, this postulate demands that synaptic plasticity be unidirectional in each cell type. Using brain slices from DA receptor transgenic mice, we show that this is not the case. Rather, DA plays complementary roles in these two types of MSN to ensure that synaptic plasticity is bidirectional and Hebbian. In models of Parkinson's disease, this system is thrown out of balance, leading to unidirectional changes in plasticity that could underlie network pathology and symptoms.

The striatal release of dopamine (DA) is intimately linked to associative learning and habit formation (1, 2). This role is thought to be mediated by controlling corticostriatal synaptic plasticity (35). However, efforts to characterize how this control is exerted have met with only modest success. Principal striatal medium spiny neurons (MSNs) are heterogeneous in their expression of DA receptors (6), falling into one of two equally sized, morphologically similar groups. One group expresses predominantly D1 DA receptors (D1 MSNs), whereas the other expresses D2 DA receptors (D2 MSNs). These two receptors appear to modulate long-term changes in glutamatergic synaptic plasticity in MSNs in different ways. D1 DA receptor signaling promotes long-term potentiation (LTP) (3, 7), whereas D2 DA receptor signaling promotes long-term depression (LTD) (8, 9). In animal models of Parkinson's disease (PD), where striatal DA levels are very low, both forms of synaptic plasticity in MSNs appear to be lost, which suggests that DA receptor signaling is necessary for their induction (7, 8).

This simple picture poses an obvious conceptual puzzle. If DA is necessary for the induction of synaptic plasticity and D1 and D2 receptors are expressed by different MSNs, then synaptic plasticity must be unidirectional in each population. However, a high percentage of MSNs display both forms of plasticity (3, 7, 9).

In an attempt to resolve this paradox, we reexamined glutamatergic synaptic plasticity in brain slices from transgenic mice in which the expression of D1 or D2 receptors was reported by coexpression of green fluorescent protein (GFP). GFP expression in these mice faithfully reports MSN phenotype, allowing D1 and D2 receptor–expressing MSNs to be reliably sampled (10, 11). We induced plasticity by pairing afferent stimulation with postsynaptic spikes in short bursts that were repeated at a theta frequency (5 Hz). At most synapses, a Hebbian form of spike-timing-dependent plasticity (STDP) is induced by this protocol (12). That is, when presynaptic activity precedes postsynaptic spiking, LTP is induced, whereas reversing the order induces LTD (1214). To induce STDP, glutamatergic afferent fibers were stimulated with a small pipette close (∼100 μm) to the soma of an identified MSN that was driven to spike by current injected through a somatic perforated membrane patch (Fig. 1, A to C) (12, 13, 15).

Fig. 1.

D2 MSNs displayed bidirectional Hebbian STDP dependent upon D2 and A2a receptors. (A) Schematic illustration of the recording/stimulation configuration. (B and C) The theta-burst pairing protocols for induction of (B) LTP and (C) LTD. Scale bars, 40 mV × 200 ms. (D) LTP induced by a positive timing pairing. Plots show EPSP amplitude and input resistance as a function of time. The dashed line shows the average EPSP amplitude before induction. The induction was performed at the vertical bar. Filled symbols show the averages of 12 trials (±SEM). The averaged EPSP traces before and after induction are shown at the top. Scale bars, 2 mV × 100 ms. (E) LTD induced by a negative timing pairing. Plots and EPSP traces as in (D). Scale bars, 2 mV × 100 ms. (F) In the presence of D2 receptor antagonist sulpiride (10 μM), negative timing pairing failed to alter EPSP amplitude, but coapplication of A2a adenosine receptor agonist CGS 21680 (100 nM) and sulpiride led to LTP (n = 6; P < 0.05, Wilcoxon). (G) LTP induction (n = 11; P < 0.01, Wilcoxon) was disrupted by the NMDA receptor antagonists APV (50 μM) and MK-801 (20 μM) or the A2a receptor antagonist SCH58261 (100 nM). (H) In the presence of D2 receptor agonist quinpirole (10 μM, n = 6), the application of the positive timing protocol leads to induction of LTD. Application of quinpirole and CGS21680 (100 nM, n = 6) together restored LTP with a positive timing pairing. (I) Schematic illustration shows that activation of A2a and NMDA receptors leads to LTP and that activation of D2 and mGluR5 receptors and Cav1.3 channels leads to LTD. Moreover, A2a and D2 receptor activation oppose each other in inducing plasticity. Glu, glutamate; EC, endocannabinoid.

In D2 MSNs (fig. S1), repeated pairing of a synaptic stimulation with a postsynaptic spike 5 ms later resulted in LTP of the synaptic response (Fig. 1D). In contrast, preceding synaptic stimulation (–10 ms) with a short burst of postsynaptic spikes induced LTD (Fig. 1E). There were no lasting alterations in synaptic strength with unpaired presynaptic or postsynaptic activity (fig. S1).

Previous studies of striatal LTD induced by conventional plasticity protocols have underscored the importance of D2 receptors (7, 8, 16). In D2 MSNs, timing-dependent LTD was disrupted by antagonizing D2 receptors with sulpiride (control n = 5; sulpiride n = 5; P < 0.05, Mann-Whitney rank sum test), suggesting a similar involvement of D2 receptors (Fig. 1F). Moreover, LTD was disrupted by antagonizing CB1 endocannabinoid (fig. S2) or mGluR5 glutamate receptors (fig. S3). The combination of presynaptic activity and activation of terminal CB1 receptors leads to a lasting reduction in glutamate release probability underlying conventional LTD (17). Indicating a presynaptic expression mechanism, LTD was accompanied by increased trial-to-trial variation in excitatory postsynaptic potential (EPSP) amplitude (fig. S4).

The determinants of LTP at glutamatergic synapses of MSNs are less well characterized. D1 receptors are thought to be important, but they are not expressed by D2 MSNs. Moreover, blocking D1 receptors with SCH23390 had no effect on timing-dependent LTP in D2 MSNs (fig. S5). Adenosine A2a receptors, which couple to the same second messenger cascades as D1 receptors, are robustly and selectively expressed by D2 MSNs (18). Antagonizing these receptors—not D1 receptors—disrupted the induction of LTP in D2 MSNs (control n = 11; SCH58261 n = 7; P < 0.001, Mann-Whitney test) (Fig. 1G). Blocking N-methyl-d-aspartate (NMDA) receptors also prevented the induction of timing-dependent LTP, as with conventional LTP (7) [d,l-2-amino-5-phosphonovaleric acid (APV) and MK-801 n = 5; P < 0.01, Mann-Whitney test] (Fig. 1G). Furthermore, the variation in EPSP amplitude was unchanged after LTP induction, which suggests that the expression of plasticity was postsynaptic (fig. S4).

Bidirectional STDP depends upon opponent processes controlling LTD and LTP (13, 19, 20). The net change in synaptic strength is hypothesized to reflect the interaction between cellular processes sensitive to the timing of pre- and postsynaptic activity (e.g., NMDA receptors and L-type Ca2+ channels) and G-protein coupled receptor (GPCR)–regulated intracellular signaling cascades. By altering the balance between GPCR cascades controlling plasticity, the timing dependence of STDP can be diminished (19, 20). We therefore elevated D2 receptor stimulation by bath application of quinpirole during pairing of presynaptic stimulation with a trailing (+5 ms) postsynaptic spike, a protocol that normally induces LTP. This resulted in a robust LTD (n = 6; P < 0.05, Wilcoxon signed rank test) (Fig. 1, H and I). Boosting A2a receptor signaling by bath application of CGS21680 restored LTP (n = 6; P < 0.05, Wilcoxon test) (Fig. 1, H and I). Conversely, bath application of the A2a receptor agonist in the presence of a D2 receptor antagonist led to the induction of LTP, even when postsynaptic spiking preceded presynaptic stimulation (Fig. 1F).

When the same STDP protocols were applied to D1 MSNs (fig. S1), a different picture emerged. Pairing presynaptic activity with a trailing postsynaptic spike (+5 ms) induced a robust LTP (Fig. 2A). LTP was dependent upon NMDA receptors (control n = 5; APV and MK-801 n = 4; P < 0.05, Mann-Whitney test) (Fig. 2B) and appeared to be postsynaptically expressed (fig. S4). However, when presynaptic activity followed postsynaptic spiking (–10 ms), EPSP amplitude did not change (Fig. 2C). The absence of LTD in D1 MSNs was consistent with recent work using a conventional plasticity protocol and has been attributed to the failure to generate endocannabinoids postsynaptically during induction, rather than to the absence of presynaptic CB1 receptors (8), an inference consistent with the inability of AM-251 to affect the response (control n = 6; AM-251 n = 5; P > 0.05, Mann-Whitney test) (Fig. 2D). We reasoned that this failure could be due to the activation of the GPCR signaling responsible for LTP induction. To test this hypothesis, D1 receptors—commonly believed to be necessary for LTP induction in MSNs (3, 7)—were blocked by bath application of SCH23390 and the protocol repeated. In the absence of D1 receptor activity, pairing postsynaptic spiking with a trailing presynaptic volley led to a robust LTD (n = 6; P < 0.05, Wilcoxon test) (Fig. 2D) that was blocked by an mGluR5 antagonist (fig. S3). However, antagonizing D2 receptors did not disrupt LTD induction in D1 MSNs (fig. S6), which suggests that D2 receptor sensitive interneuronal signaling was not engaged by local, minimal stimulation (16).

Fig. 2.

D1 MSNs displayed bidirectional Hebbian STDP dependent upon D1 receptors. (A) LTP induction by a positive timing pairing protocol. EPSP amplitude and input resistance of the recorded cell were plotted as a function of time. The averaged EPSP traces before and after induction are shown at the top. Scale bars, 2 mV × 100 ms. (B) LTP induction (n = 10; P < 0.01, Wilcoxon test) was blocked by APV (50 μM) and MK-801 (20 μM). (C) LTD was not induced in D1 neurons with a negative pairing. Plots and EPSP traces are from a single cell, as in (A). Scale bars, 2 mV × 100 ms. (D) In the presence of D1 receptor antagonist SCH23390 (3 μM), a negative timing pairing revealed LTD, but in the presence of CB1 receptor antagonist AM-251 (2 μM), negative pairing failed to alter EPSP amplitude. (E) LTP induced by a positive timing pairing was blocked by SCH23390, revealing LTD. LTD induced in the presence of SCH23390 was disrupted by AM-251. (F) Schematic drawing shows that activation of D1 and NMDA receptors evokes LTP and that activation of mGluR5 receptor and Cav1.3 channels evokes LTD. Moreover, D1 and mGluR5 receptor activation oppose each other in inducing plasticity.

To determine whether attenuating D1 receptor signaling altered the timing dependence of plasticity, the effect of the positive timing protocol that normally induced a robust LTP (Fig. 2E) was reexamined in the presence of a D1 receptor antagonist. This not only prevented LTP induction, it led to the induction of LTD (SCH23390 n = 6; P < 0.05, Wilcoxon test) that was dependent upon endocannabinoid CB1 receptors (SCH23390 n = 6; SCH23390 and AM-251 n = 5; P <0.01, Mann-Whitney test) (Fig. 2E, F), establishing a mechanistic parallel to LTD in D2 MSNs. Antagonizing downstream presynaptic CB1 receptors alone did not alter LTP induction (fig. S7).

These experiments show that while DA makes STDP in striatal MSNs bidirectional and Hebbian (12), it is not necessary for the induction of synaptic plasticity. This stands in contrast to previous work asserting that DA is essential for plasticity and that striatal DA depletion in animal models of PD eliminates both LTD and LTP at MSN glutamatergic synapses (7, 8). To revisit this issue, two well-established mouse models of PD were examined: In one, DA neurons were lesioned by injecting the toxin 6-hydroxydopamine (6-OHDA) into the medial forebrain bundle (Fig. 3A), and in the other, DAwas depleted by disrupting the monoamine vesicular transporter with reserpine (15). In both models, striatal DA levels were dramatically reduced and STDP altered in the same way. In D2 MSNs, LTP was induced not only by the usual pairing protocol (6-OHDA n = 6; P < 0.05, Wilcoxon test) (Fig. 3B) (reserpine n = 6; P < 0.05, Wilcoxon test) (Fig. 3C) but also when spiking preceded synaptic stimulation, a protocol that normally induced LTD (n = 10; P < 0.05, Wilcoxon test) (Fig. 3D). As in normal tissue, this LTP was sensitive to A2a receptor block (reserpine n = 6; SCH58261 n = 4; P < 0.05, Mann-Whitney test) (Fig. 3C). Reestablishing D2 receptor activity with exogenous quinpirole rescued LTD (n = 6; P < 0.05, Wilcoxon test) (Fig. 3D). In contrast, LTP was not induced in D1 MSNs by pairings in which synaptic activity preceded postsynaptic spiking; rather, this protocol induced a robust LTD (6-OHDA n = 6; P < 0.05, Wilcoxon test) (Fig. 3E) (reserpine n = 6; P < 0.05, Wilcoxon test) (Fig. 3F) that was sensitive to CB1 receptor block (reserpine n = 6; AM-251 n = 4; P < 0.01, Mann-Whitney test) (Fig. 3F). Reestablishing D1 receptor activity with exogenous SKF81297 rescued LTP with this protocol (n = 6; P < 0.05, Wilcoxon test) (Fig. 3F).

Fig. 3.

Bidirectional Hebbian STDP is disrupted in MSNs from parkinsonian mice. (A) Light microscopic image of a coronal section showing the loss of immunoreactivity for TH after unilateral 6-OHDA lesioning. Cx, cortex; CPu, caudate-putamen. (B and C) LTP was induced from (B) lesioned D2 mice and (C) reserpinetreated mice after a positive timing protocol. Plot of average EPSP amplitude as a function of time. In (C), timing-dependent LTP induced in reserpine treated animals was blocked by SCH58261 (100 nM). (D) LTP also was induced with a negative timing protocol that would normally induce LTD. In contrast, perfusion of quinpirole (10 μM) restored LTD. (E and F) Timing-dependent LTD was evoked in D1 MSNs from (E) lesioned D1 mice and (F) reserpine-treated mice. In (F), D1 receptor agonist SKF81297 (3 μM) restored LTP after a positive timing protocol. The LTD induced in reserpine-treated mice was disrupted by AM-251 (2 μM).

Our studies demonstrate that DA is critical for the induction of bidirectional, timing-dependent (Hebbian) plasticity at glutamatergic synapses formed on striatal MSNs. This finding is consistent with a recent study (14) but conflicts with another (21). The discrepancy between studies could be attributable to heterogeneity in glutamatergic synapses (22) or the engagement of striatal interneurons capable of modulating the induction of plasticity (7, 16). Although the role of these factors is important to sort out, our goal in using focal stimulation near synaptic sites was to isolate the direct actions of DA on the induction of plasticity at glutamatergic synapses of largely cortical origin. This strategy revealed that the type of DA receptor present at the postsynaptic site governed the actions of DA. Furthermore, the signaling determinants of STDP resolved in this way were very similar to those reported for conventional plasticity in MSNs (79), which suggests that they engaged the same cellular machinery. This insight and the recognition that DA does not act alone in regulating the induction of plasticity, but in concert with glutamate and adenosine, resolves the conflict posed by the apparent obligatory roles of D1 and D2 receptors in the induction of plasticity and their segregation in different MSN classes. It also shows how the activity of DA neurons might serve to reshape the striatal network during associative learning (4). In the absence of behaviorally important stimuli, DA neurons spike autonomously to maintain striatal DA concentrations at levels sufficient to keep high-affinity D2 DA receptors active, but not low affinity D1 DA receptors (23, 24)—in principle enabling bidirectional, Hebbian plasticity in D2 MSNs, but not in D1 MSNs, where the low level of D1 receptor activity should permit only LTD. However, when behaviorally important stimuli drive phasic spiking of mesencephalic DA neurons, striatal DA levels rise transiently and activate D1 DA receptors (23); this should enable the induction of Hebbian LTP in D1 MSNs. This stark dichotomy provides a cellular foundation for the view that the networks anchored by these two MSNs regulate distinct aspects of behavior and learning (25, 26).

The existence of opponent processes regulating the induction of plasticity also has implications for disease states where DA signaling is abnormal. In both hyperdopaminergic states, like drug abuse or schizophrenia (27), and hypodopaminergic states, like PD, the imposition of Hebbian rules on the sculpting of synaptic strength by experience will be disrupted, leading to inappropriate associations. This regulatory balance was absent in PD models. In this hypodopaminergic state, plasticity at glutamatergic synapses was still present, but it had lost its bidirectionality and dependence upon the temporal structure of pre- and postsynaptic activity. This might help to explain why striatal learning in PD patients is dysfunctional rather than simply absent (28). The cellular specificity of this imbalance also helps to explain the contrasting effects of DA depletion on the synaptic connectivity of D1 and D2 MSNs (10, 29) and the longer-term adaptations in network activity thought to underlie the motor symptoms of PD (30).

Supporting Online Material

www.sciencemag.org/cgi/content/full/321/5890/848/DC1

Materials and Methods

Figs. S1 to S7

References

References and Notes

View Abstract

Navigate This Article