Scholarly article on topic 'Input- and Cell-Type-Specific Endocannabinoid-Dependent LTD in the Striatum'

Input- and Cell-Type-Specific Endocannabinoid-Dependent LTD in the Striatum Academic research paper on "Biological sciences"

Share paper
Academic journal
Cell Reports
OECD Field of science

Abstract of research paper on Biological sciences, author of scientific article — Yu-Wei Wu, Jae-Ick Kim, Vivianne L. Tawfik, Rupa R. Lalchandani, Grégory Scherrer, et al.

Summary Changes in basal ganglia plasticity at the corticostriatal and thalamostriatal levels are required for motor learning. Endocannabinoid-dependent long-term depression (eCB-LTD) is known to be a dominant form of synaptic plasticity expressed at these glutamatergic inputs; however, whether eCB-LTD can be induced at all inputs on all striatal neurons is still debatable. Using region-specific Cre mouse lines combined with optogenetic techniques, we directly investigated and distinguished between corticostriatal and thalamostriatal projections. We found that eCB-LTD was successfully induced at corticostriatal synapses, independent of postsynaptic striatal spiny projection neuron (SPN) subtype. Conversely, eCB-LTD was only nominally present at thalamostriatal synapses. This dichotomy was attributable to the minimal expression of cannabinoid type 1 (CB1) receptors on thalamostriatal terminals. Furthermore, coactivation of dopamine receptors on SPNs during LTD induction re-established SPN-subtype-dependent eCB-LTD. Altogether, our findings lay the groundwork for understanding corticostriatal and thalamostriatal synaptic plasticity and for striatal eCB-LTD in motor learning.

Academic research paper on topic "Input- and Cell-Type-Specific Endocannabinoid-Dependent LTD in the Striatum"

Cell Reports


Input- and Cell-Type-Specific Endocannabinoid-Dependent LTD in the Striatum

Graphical Abstract


Yu-Wei Wu, Jae-Ick Kim.....

Gregory Scherrer, Jun B. Ding


In Brief

Basal ganglia plasticity at glutamatergic synapses is required for motor learning. Wu et al. report that expression of endocannabinoid-dependent long-term depression (eCB-LTD) in the striatum is dependent on presynaptic input but independent of postsynaptic cell type. Furthermore, activation of dopamine receptors in the striatum bidirectionally modulates eCB-LTD expression.


• eCB-LTD is induced specifically at corticostriatal, but not thalamostriatal, synapses

• CB1 receptor expression levels are responsible for input specificity

• eCB-LTD in the striatum is present independent of postsynaptic SPN subtype

• Coactivation of dopamine receptors modulates eCB-LTD expression in the striatum

Wu et al., 2015, Cell Reports 10, 75-87 ciossMark January 6, 2015 ©2015 The Authors

http://dx.d0i.0rg/l 0.1016/j.celrep.2014.12.005


Cell Reports


Input- and Cell-Type-Specific Endocannabinoid-Dependent LTD in the Striatum

Yu-Wei Wu,1 Jae-Ick Kim,1 Vivianne L. Tawfik,2 Rupa R. Lalchandani,1 Gregory Scherrer,1'2 3 and Jun B. Ding14 *

1 Department of Neurosurgery, Stanford University School of Medicine, Palo Alto, CA 94304, USA

2Department of Anesthesiology, Perioperative and Pain Medicine, Stanford University School of Medicine, Palo Alto, CA 94304, USA 3Department of Molecular and Cellular Physiology, Stanford University School of Medicine, Palo Alto, CA 94304, USA 4Department of Neurology and Neurological Sciences, Stanford University School of Medicine, Palo Alto, CA 94304, USA 'Correspondence:

This is an open access article under the CC BY-NC-ND license (


Changes in basal ganglia plasticity at the cortico-striatal and thalamostriatal levels are required for motor learning. Endocannabinoid-dependent long-term depression (eCB-LTD) is known to be a dominant form of synaptic plasticity expressed at these glutamatergic inputs; however, whether eCB-LTD can be induced at all inputs on all striatal neurons is still debatable. Using region-specific Cre mouse lines combined with optogenetic techniques, we directly investigated and distinguished between corticostria-tal and thalamostriatal projections. We found that eCB-LTD was successfully induced at corticostriatal synapses, independent of postsynaptic striatal spiny projection neuron (SPN) subtype. Conversely, eCB-LTD was only nominally present at thalamostriatal synapses. This dichotomy was attributable to the minimal expression of cannabinoid type 1 (CB1) receptors on thalamostriatal terminals. Furthermore, coactivation of dopamine receptors on SPNs during LTD induction re-established SPN-subtype-depen-dent eCB-LTD. Altogether, our findings lay the groundwork for understanding corticostriatal and thalamostriatal synaptic plasticity and for striatal eCB-LTD in motor learning.


The basal ganglia are a group of subcortical nuclei that fulfill critical roles in motor control and action selection (Graybiel et al., 1994). The input nucleus of the basal ganglia, the striatum, is composed primarily of two distinct groups of GABAergic spiny projection neurons (SPNs): direct pathway SPNs (dSPNs), which project to substantia nigra pars reticulata (SNr) and express D1 dopamine receptors (D1R), and indirect pathway SPNs (iSPNs), which project to the globus pallidus and express D2 dopamine receptors (D2R) (Gerfen, 1989; Surmeieretal., 1996,2007). Stria-tal SPN dendrites receive intermingled excitatory glutamatergic inputs from both the cerebral cortex and the thalamus (Ding

et al., 2008; Smith et al., 2004). The function and plasticity of these synapses are modulated by endocannabinoids (eCBs) (Kano et al., 2009; Kreitzer and Malenka, 2008; Surmeier et al., 2014), and eCB-dependent long-term depression (eCB-LTD) is one of the most dominant forms of long-term plasticity expressed at these glutamatergic synapses (Gerdeman et al., 2002; Kreitzer and Malenka, 2005, 2007; Shen et al., 2008). eCBs are released by postsynaptic neurons and act as retrograde messengers to activate presynaptic CB1Rs, depressing neurotransmission (Kano et al., 2009). eCB-LTD induction requires the activation of postsynaptic calcium signaling and activation of G protein-coupled receptors (Kreitzer and Malenka, 2005). It has been suggested that this form of LTD is dependent on activation of postsynaptic D2Rs (Kreitzer and Malenka, 2007; Nazzaroetal., 2012; Shen etal., 2008). However, studies demonstrating that eCB-LTD can be induced in both SPN subtypes challenge this view (Bagetta et al., 2011; Wang et al., 2006).

Pharmacological tools have been used to probe the role of individual neuromodulatory systems in eCB-LTD induction, including dopaminergic, cholinergic, opioid, and serotoninergic inputs (Atwood et al., 2014a; Bagetta et al., 2011; Kreitzer and Malenka, 2005; Mathur et al., 2011; Shen et al., 2007; Wang et al., 2006). Nevertheless, it is still difficult to isolate the individual contributions of corticostriatal and thalamostriatal synapses, given that both are glutamatergic and are intermingled on SPNs dendrites (Doig et al., 2010). Moreover, these two groups of synapses exhibit very distinct properties: there are stark differences in release probability, short-term plasticity, and postsynaptic receptor composition (Ding etal., 2008), suggesting the properties of their synaptic plasticity might be very different. However, most previous eCB-LTD studies use conventional electrical stimulation paradigms in which the stimulation electrodes are placed either intrastriatally or in the white matter. These configurations inevitably coactivate cortico- and thalamostriatal synapses, as well as dopaminergic inputs, making it difficult to distinguish between the individual contributions of these inputs to striatal synaptic plasticity. We speculate that the discrepancies of past studies may be the result of nonspecifically exciting heterogeneous presynaptic striatal inputs. In order to achieve selective activation of presynaptic cortico- and thalamostriatal inputs, we combined region-specific Cre mouse lines with optogenetic tools to express channelrhodopsin-2 (ChR2) in either cortico-striatal or thalamostriatal projection neurons. We find that

Layer IV . Layer V/VI"


\ ■ %

;cpu ; ; i

Figure 1. Optogenetic Activation of Corticostriatal and Thalamos-triatal Axons in Dorsolateral Striatum

(A) Aconfocal image of a coronal section from aThy1-ChR2-YFP mouse. Inset image shows extensive ChR2-YFP-expressing axons in the CPu from a different section. Scale bar represents 10 mm. Pf, parafascicular nucleus of the thalamus; Fr, fasciculus retroflexus.

(B) A coronal section from a Vglut2-Cre;Ai32 mouse imaged as in (A).

(C) Left: a sample image of an oblique horizontal brain slice. SPNs were sampled in the circled area. Middle: the same slice visualized for green (Thy1-ChR2-YFP) and red (Drd1-tdTomato) fluorescence under an Arc lamp. Right: schematic of the recording configuration and optogenetic stimulation of axon terminals. Labeled as follows: blue, 450 nm blue light; Ctx, cortex; ic, internal capsule; CPu, caudate putamen; LGP, lateral globus pallidus; Tha, thalamus.

(D) Left: infrared differential interference contrast image of an SPN under whole-cell patch-clamp recording. Middle: a sample two-photon image of the same SPN showing expression of tdTomato (red) and surrounding ChR2-YFP-expressing axons (green). Scale bar represents 5 mm. Right, a sample two-photon image showing a dendritic branch of Alexa Fluor 594 (10 mM)-filled SPN with ChR2-YFP-expressing axons. Scale bar represents 1 mm

(E) Examples of blue-light-evoked EPSCs recorded from SPNs. The AMPA receptor component (black) was recorded at —70 mVand NMDAR component (red) at +40 mV. Dashed line indicates 50 ms after light stimulation at which the NMDAR component was measured to calculate the ratio.

when other neuromodulatory systems are not activated, eCB-LTD is reliably induced by (S)-3,5-dihydroxyphenylglycine (DHPG) at corticostriatal synapses but minimally at thalamostria-tal synapses, regardless of postsynaptic SPN subtype. We show that this differential eCB-LTD expression is attributable to CB1R expression patterns at corticostriatal and thalamostriatal presynaptic terminals. Understanding how striatal neurons integrate information from different synaptic inputs is essential for deciphering basal ganglia function. Our findings suggest that information carried by different glutamatergic inputs may undergo different forms of pathway-specific long-term plasticity that are critical for their unique roles in motor learning and action selection.


Optogenetic Targeting of Corticostriatal and Thalamostriatal Neurons

To achieve selective activation of corticostriatal terminals, we used Thy1-ChR2-YFP mice (Arenkiel et al., 2007) in which ChR2 is highly expressed in cortical layer V neurons, but not thalamic neurons (Figure 1A). We achieved selective expression of ChR2 in thalamic neurons by crossing Vglut2-Cre mice, in which Cre recombinase expression is under the control of vesicular glutamate transporter 2 (Vglut2) promoter (Vglut2-Cre line), with the Ai32 transgenic mouse, in which ChR2-eGFP expression is Cre dependent (Ai32 line) (Madisen et al., 2012). As Vglut2 is predominantly expressed in glutamatergic neurons in the thalamus and in layer IV cortical neurons (Fremeau et al., 2004), we observed robust ChR2-eGFP expression in thalamus and layer IV of the cortex in 8- to 10-week-old mice resulting from this cross (Figure 1B). Axonal fibers expressing ChR2/eGFP densely innervated the striatum (Figures 1A and 1B, inset). Because cortical layer IV neurons do not project afferents to the striatum (Wall et al., 2013), ChR2/eGFP-expressing axons in the striatum of Vglut2-Cre;Ai32 mice (Figure 1B, inset) arise exclusively from thalamostriatal inputs. Furthermore, in order to restrict ChR2 expression to a more confined area of the thalamus, we also injected adeno-associated virus (AAV) expressing a Cre-inducible ChR2-mCherry (AAV-DIO-ChR2-mCherry) into the parafascicular nucleus (Pf) of the thalamus in Vglut2-Cre mice (Figure S1). We did not observe any significant difference in NMDA/AMPA ratio between Vglut2-Cre;Ai32 mice and Vglut2-Cre mice combined with AAV injection (Ai32: 0.26 ± 0.06, n = 5; AAV: 0.23 ± 0.03, n = 17; p = 0.70, Mann-Whitney U test [U test]), which suggests that the majority of the axons activated in Vglut2-Cre;Ai32 mice shared similar properties with those arising from Pf-in-jected neurons. Therefore, we grouped the data from these two approaches for simplicity with the abbreviation Vglut2-Cre;ChR2 mice.

We performed whole-cell patch-clamp recordings in oblique horizontal brain slices and recorded from SPNs in the dorsolateral striatum (Figures 1C and 1D). In most cases, we also documented the cell type as dSPN or iSPN by determining the

(F) Summary of NMDA/AMPA ratios of corticostriatal (Thy-ChR2-YFP, n = 9) and thalamostriatal (Vglut2-Cre; ChR2, n = 22) synapses. *p < 0.05, U test.

expression of Drd1-tdTomato or Drd2-GFP (Figure 1D, middle). Dense ChR2-YFP-expressing axons were observed surrounding the SPN dendrites (Figure 1D, right). Next, we tested whether we could specifically and reliably activate either the corticostriatal or thalamostriatal pathway in this paradigm. We delivered short blue light pulses (0.1-0.15 ms at 450 nm, 0.5-4 mW under objective) to stimulate ChR2-expressing axons to evoke gluta-matergic excitatory postsynaptic currents (EPSCs) in SPNs. It has been shown that corticostriatal synapses have higher NMDA receptor (NMDAR) components as compared to thalamostriatal synapses (Ding et al., 2008). Consistent with this previous finding, we found that optogenetic stimulation-evoked EPSCs at corticostriatal synapses also had larger NMDAR components than at thalamostriatal synapses. The NMDAR and AMPAR current ratio (NMDA/AMPA ratio) was significantly larger in Thy1-ChR2-YFP mice than in Vglut2-Cre;ChR2 mice (Figures 1E and 1F; Thy1-ChR2-YFP: 0.38 ± 0.05, n = 9; Vglut2-Cre;ChR2: 0.23 ± 0.03, n = 22; p < 0.05, U test). These results further confirm the reliability of our optogenetic approach for the selective activation of either corticostriatal or thalamos-triatal axons in the striatum.

Differential Expression of CB1R-Dependent LTD in Corticostriatal and Thalamostriatal Synapses on SPNs

Next, we asked if eCB-LTD is expressed in both corticostriatal and thalamostriatal synapses formed on SPNs. Because this form of LTD can be modulated by activation of dopamine and acetylcholine receptors (Kreitzer and Malenka, 2005; Wang et al., 2006), high-frequency electrical stimulation (eHFS) might recruit dopaminergic and cholinergic signaling, complicating LTD induction. To bypass these alternative neuromodulator systems, we instead directly activated group 1 mGluR by applying 50 mM (S)-DHPG for 10 min to induce LTD. Together, with postsynaptic depolarization to -50 mV to activate L-type calcium channels (Kreitzer and Malenka, 2005, 2007), DHPG reliably induced LTD in almost all of the recorded SPNs in Thy1-ChR2-YFP mice (16 out of 18, 89% of neurons expressed LTD with EPSC peaks decreased by more than 20%) (Figures 2A and 2C; 62% ± 3% of baseline, n = 18; p < 0.001, Wilcoxon signed rank [Wilcoxon]). Surprisingly, in SPNs of Vglut2-Cre;ChR2 mice, DHPG caused only a small reduction in EPSCs (Figures 2B and 2D; 90% ± 4% of baseline, n = 13, p < 0.05, Wilcoxon), suggesting that this LTD is primarily restricted to corticostriatal afferents.

Next, we investigated whether DHPG-induced eCB-LTD is expressed in both dSPNs and iSPNs (Figure 1D, middle). eCB-LTD was reliably induced by DHPG in both dSPNs and iSPNs in Thy1-ChR2-YFP mice (Figure 2E; dSPN: 63% ± 5% of baseline, n = 8, p < 0.05, Wilcoxon; iSPNs: 63% ± 4% of baseline, n = 7; p < 0.05, Wilcoxon; p = 1 for comparison between dSPNs and iSPNs, U test). In stark contrast, DHPG application produced minimal or no LTD in either dSPNs and iSPNs in Vglut2-Cre;ChR2 mice (Figure 2F; dSPN: 94% ± 5% of baseline, n = 5, p = 0.795, Wilcoxon; iSPNs: 87% ± 5% of baseline, n = 5; p = 0.11, Wilcoxon; p = 0.53 for comparison between dSPNs and iSPNs, U test). Consistent with previous findings that striatal eCB-LTD is expressed pre-synaptically, paired-pulse ratios (PPRs) were significantly increased by the LTD induction protocol in both dSPNs and iSPNs in Thy1-ChR2-YFP mice (Figure 2G; dSPNs: baseline:

1.07 ± 0.09, DHPG: 1.32 ± 0.11, n = 8, p < 0.05, Wilcoxon; iSPNs: baseline: 1.05 ± 0.05, DHPG: 1.21 ± 0.08, n = 7; p < 0.05, Wilcoxon), suggesting eCB-LTD in corticostriatal terminals was accompanied by a decrease in presynaptic release probability. PPRs were not significantly changed in the thalamostriatal synapses of Vglut2-Cre;ChR2 mice (Figure 2H; dSPNs: baseline: 0.57 ± 0.17, DHPG: 0.60 ± 0.15, n = 5, p = 0.63, Wilcoxon; iSPNs: baseline: 0.79 ± 0.12, DHPG: 0.84 ± 0.14, n = 5; p = 0.31, Wilcoxon), suggesting the small reduction of EPSCs at thalamos-triatal synapses was not caused by a presynaptic mechanism. LTD induced in both dSPNs and iSPNs of Thy1-ChR2-YFP mice was sensitive to a CB1R antagonist (AM251 5-10 mM), suggesting that LTD is eCB dependent (Figures 3A and 3B; dSPN: 91% ± 5% of baseline, n = 6; p = 0.09, Wilcoxon; iSPN: 95% ± 10% of baseline, n = 8; p = 0.25, Wilcoxon). Interestingly, the slight LTD observed at thalamostriatal synapses was not sensitive to AM251 (Figure 3C; 84% ± 4% of baseline, n = 6, p = 0.46 comparing to control without AM251, U test), indicating this is not eCB-dependent LTD. Taken together, these results suggest that DHPG-induced eCB-LTD in striatal SPNs is input specific and not postsynaptic cell-type dependent.

It has been shown that dopamine release is critical for inducing eCB-LTD in the striatum. Specifically, D2R activation is required for eCB-LTD when induced by eHFS (Bagetta et al., 2011; Kreitzer and Malenka, 2005, 2007; Wang et al., 2006). Therefore, we tested whether DHPG-induced eCB-LTD at corticostriatal and thalamostriatal synapses requires activation of D1R and D2Rs. Consistent with previous findings, this form of eCB-LTD was not NMDAR dependent (Figure S2A; R-CPP 10 mM, 71% ± 4% of baseline, n = 6; p < 0.05, Wilcoxon). The LTD was also not sensitive to D1R antagonist (Figure S2B; SCH23390 3 mM, 68% ± 6% of baseline, n = 7; p < 0.05, Wilcoxon). Surprisingly, we found that D2R antagonist (sulpiride 5 mM) did not prevent eCB-LTD induction by DHPG in either dSPNs and iSPNs (Figures 3D-3F; dSPNs: 75% ± 6% of baseline, n = 6, p < 0.05, Wilcoxon; iSPNs: 68% ± 7% of baseline, n = 7; p < 0.05, Wilcoxon). In addition, in the presence of D2R antagonist, eCB-LTD was accompanied by increased PPRs (dSPNs: baseline: 1.02 ± 0.12, DHPG: 1.14 ± 0.13, n = 6, p < 0.05, Wilcoxon; iSPNs: baseline: 1.04 ± 0.07, DHPG: 1.28 ± 0.12, n = 7; p < 0.05, Wilcoxon). This finding, consistent with the data reported in Kreitzer and Malenka (2005), suggests that prolonged activation of mGluR1/5 and L-type calcium channels is sufficient to induce eCB-LTD at cor-ticostriatal synapses. Furthermore, this DHPG LTD-induction paradigm bypasses the requirement for D2R activation. Therefore, contrary to previous findings, eCB-LTD can be induced in both dSPNs and iSPNs in the presence of dopamine receptor antagonists with this model.

eCB-LTD in Corticostriatal Synapses in Cortex-Specific Cre Mouse Lines

Although it is generally believed that ChR2 is expressed primarily in the cortical layer V neurons of Thy1-ChR2-YFP mice (Wang et al., 2007) and, therefore, that light predominantly stimulates cortical inputs to the striatum, it is still possible that light may activate axons arising from other brain areas (Arenkiel et al., 2007). To ensure more specific expression of ChR2 in cortical neurons, we took advantage of the Emx1-Cre mouse line in which Cre

recombinase is expressed in cortical and hippocampal excitatory neurons (Madisen et al., 2012) and Rbp4-Cre mouse line that selectively expresses Cre recombinase in a dense population of layer V cortical neurons (Glickfeld et al., 2013) (Figures 4A and

Figure 2. Differential Expression of CB1R-Dependent LTD in Corticostriatal and Thalamostriatal Synapses on SPNs

(A) Upper plot: an individual experiment showing that activation of mGluR1/5 with 50 mM DHPG induced LTD in SPNs of Thy1-ChR2-YFP mice (postsynaptic neurons held at —50 mV). Light-induced paired pulses were evoked every 20 s. DHPG was applied for 10 min after 10 min of baseline recording, as indicated by the red bar. Lower left plot: averaged EPSC traces (1, black) and drug effect (2, red). Lower right plot: baseline and drug-effect traces were normalized to the first EPSC peak to show the changes in paired-pulse ratio (PPR).

(B) Same as (A), except the SPN was from a Vglut2-Cre mouse with AAV DIO-ChR2-mCherry injected in the Pf.

(C and D) Summary of DHPG induced LTD on SPNs from Thy1-ChR2-YFP mice and Vglut2-Cre;ChR2-expressing (Ai32- or AAV-injected) mice. See also Figure S1.

(E and F) Summary of EPSC amplitudes in dSPNs (filled circles) and iSPNs (open circles) in both Thy1-ChR2-YFP (E) and Vglut2-Cre;ChR2 (F) mice.

(C-F) Data are presented as mean ± SEM. (G and H) Summary PPRs of baseline and 20 min after DHPG treatment on dSPNs (left) and iSPNs (right) from Thy1-ChR2-YFP mice (G) and Vglut2-Cre;ChR2 mice (H).

4E). We then tested whether the corticostriatal synapses on the SPNs in these two mouse lines shared similar properties with corticostriatal synapses of Thy1-ChR2-YFP mice. Indeed, the NMDA/ AMPA ratios measured from optogeneti-cally evoked EPSCs recorded in Emx1-Cre;Ai32 and Rbp4-Cre;Ai32 mice were similar to those in the Thy1-ChR2-YFP mouse (Figures S3A and S3B; Emx1-Cre;Ai32: 0.40 ± 0.06, n = 11; Rbp4-Cre;Ai32: 0.37 ± 0.04, n = 10; p = 1.0 for comparison of Emx1-Cre;Ai32 and Rbp4-Cre;Ai32 with Thy1-ChR2-YFP mice, U test). In all three mouse lines, the NMDA/AMPA ratio was significantly larger than that recorded in Vglut2-Cre;ChR2 mice (Figure S3B; p < 0.05, U test). These results indicate that the synaptic properties of these two mouse lines are similar to those of corticostriatal synapses recorded from Thy1-ChR2-YFP mice.

Using the same DHPG induction protocol, eCB-LTD was successfully induced and Rbp4-Cre;Ai32 mice (Figures 4B and : 8 of baseline, n = 9; p < 0.05, Wil-6 of base-

in Emx1-Cre;Ai32 4C; Emx1-Cre;Ai32 mice 70 : coxon; Figures 4F and 4G; Rbp4-Cre;Ai32 mice 66 : line, n = 6; p < 0.05, Wilcoxon). The PPRs were also significantly

Figure 3. eCB-LTD Is Not Blocked by D2R Antagonist

(Aand B) The DHPG-induced LTD in corticostriatal synapses on both dSPNs (A) and iSPNs (B) is CB1R dependent. CB1R antagonist (AM251 510 mM) was applied throughout the recording (gray bar).

(C) AM251 did not affect the DHPG-induced LTD in thalamostriatal synapses. (D and E) The DHPG-induced LTD in corticostriatal synapses on both dSPNs (D) and iSPNs (E) is not blocked by D2R antagonist (sulpiride 5 mM). (A-E) Data are presented as mean ± SEM. (F) Summary PPRs of baseline and 20 min after DHPG treatment on dSPNs (left) and iSPNs (right) in the presence of sulpiride from Thy1-ChR2-YFP mice.

increased in both mouse lines along with the expression of LTD (Figures 4D and 4H; Emx1-Cre: baseline: 0.39 ± 0.04, DHPG: 0.48 ± 0.06, n = 9, p < 0.05, Wilcoxon; Rbp4-Cre: baseline: 0.41 ± 0.09, DHPG: 0.52 ± 0.08, n = 6; p < 0.05, Wilcoxon), suggesting a similar presynaptic mechanism of eCB-LTD to the Thy1-ChR2-YFP mice reported above. We observed that LTD was induced in all recorded neurons in Emx1-Cre;Ai32 and Rbp4-Cre;Ai32 mice, suggesting eCB-LTD is present in both dSPNs and iSPNs. Together, these results further strengthen our conclusion that eCB-LTD in glutamatergic synapses onto striatal SPNs is determined by cortical input.

Dopamine Modulation of eCB-LTD Induced by Spike-Timing-Dependent Protocol

Our results indicate that dopamine is not required in DHPG-induced eCB-LTD. However, eCB-LTD could also be induced by other induction protocols, such as eHFS or spike-timing-dependent plasticity (STDP) induction protocol (Bagetta et al., 2011; Gerdeman et al., 2002; Kreitzer and Malenka, 2005,

2007; Shen et al., 2008; Wang et al., 2006). We attempted to induce LTD with optogenetic high-frequency-stimulation (oHFS) in brain slices made from either Thy1-ChR2-YFP or Vglut2-Cre;ChR2 mice (Figure S4). However, because ChR2 and ChR2(H134B) are subject to strong inactivation and desensitization (Lin et al., 2009), the synaptic release could not reliably follow the optical stimulation when the stimulation frequency is above 20 Hz (Figure S4), preventing us from directly testing whether eCB-LTD induced by oHFS stimulation protocol is also input specific. Thus, we asked whether DHPG protocol and high-frequency-stimulation (HFS) share similar features by testing the occlusive effect of DHPG on HFS-induced LTD. Indeed, with local electrical stimulation, LTD was induced in only iSPNs and was occluded by DHPG (Figure S5), suggesting DHPG-induced eCB-LTDs share a similar mechanism.

We next investigated whether eCB-LTD induced by a Hebbian form of STDP is also input specific. Previous studies have shown that eCB1-LTD could be induced by a "post-pre"-STDP protocol, i.e., pairing postsynaptic spiking preceding presynaptic release (Figure 5A) (Nazzaro et al., 2012; Shen et al., 2008). In addition, this post-pre-STDP protocol could trigger eCB-depen-dent LTD only in iSPN and not in dSPN (Shen et al., 2008). Wefirst tested whether LTD could be induced in dSPNs or iSPNs by pairing cortical afferent stimulation using optogenetic stimulation (Thy1-ChR2-YFP mice) with preceding postsynaptic spikes in short bursts that were repeated at a theta frequency (5 Hz, oSTDP pairing protocol; Figure 5A). Surprisingly, we found that LTD at corticostriatal synapses was only induced in dSPNs, but not in iSPNs (Figures 5B-5E; dSPN: 72% ± 4% of baseline, n = 6; p < 0.05, Wilcoxon; iSPN: 91% ± 12% of baseline, n = 9; p = 0.50, Wilcoxon). Previous studies of striatal LTD that used conventional local electrical stimulation to evoke glutamatergic

Figure 4. Optogenetic Activation of Corticostriatal Synapses Using Emx1-Cre and Rbp4-Cre Mice

(A and E) Confocal images of coronal sections across cortex and thalamus of Emx1-Cre;Ai32 (A) and Rbp4-Cre;Ai32 (E) mice. Inset image shows the ChR2-eGFP expressing axon terminals in the striatum (CPu) from different sections. Tha, thalamus. Scale bar represents 10 mm.

(B and F) Left: an individual experiment of LTD induced in SPNs of Emx1-Cre;Ai32 (B) and Rbp4-Cre;Ai32 (F) mice. Upper right: averaged EPSC traces of baseline (1, black) and drug effect (2, red). Lower right: normalized EPSC to show the changes in PPR.

(C and G) Summary of DHPG-induced LTD on SPNs from Emx1-Cre;Ai32 (C) and Rbp4-Cre;Ai32 (G) mice. Data are presented as mean ± SEM.

(D and H) Summary PPRs of baseline and 20 min after DHPG treatment on SPNs from Emx1-Cre;Ai32 (D) and Rbp4-Cre;Ai32 (H) mice. *p < 0.05, Wilcoxon.

Box-and-whisker plots indicate the minimum, 25th, 50th, 75th, and maximum percentiles.

synaptic transmission also inevitably activated en passant dopamine fibers (Shen et al., 2008). It is therefore possible that coac-tivation of D1Rs reduces LTD in dSPNs and, conversely, D2R activation is required for oSTDP-LTD in iSPNs. To directly test this hypothesis, we re-examined the role of D1R and D2Rs in oSTDP-LTD using D1R and D2R agonists. In dSPNs, we found that D1R agonist (SKF 81297 3 mM) prevented oSTDP-LTD induction (Figure 5F; 95% ± 4% of baseline, n = 5; p = 0.44, Wil-coxon), which resembles the LTD induced by electrical stimulation using the same pairing protocol (Shen et al., 2008). On the other hand, in iSPNs, oSTDP-LTD was successfully induced in the presence of D2R agonist (quinpirole 10 mM) (Figure 5G; 81% ± 3% of baseline, n = 7; p < 0.05, Wilcoxon). It is worth mentioning that our oSTDP-LTD results obtained from control conditions closely resemble what was shown in previous findings, where D1R and D2Rs activation was abolished by either dopamine receptor antagonists or dopamine depletion (Shen et al., 2008). Lastly, oSTDP-LTD induced in both dSPNs and iSPN (in quinpirole) was blocked by AM251 (Figures 5F and 5G; dSPN: 101% ± 3% of baseline, n = 5; p = 0.63, Wilcoxon; iSPN in quinpirole: 98% ± 6% of baseline, n = 5; p = 0.81, Wilcoxon), suggesting that oSTDP-LTD at corticostriatal synapses on both dSPNs and iSPNs are CB1R dependent. Together, these data showed that eCB-LTD could be induced at corticostriatal synapses using an oSTDP induction protocol and that corticostriatal synapses on both dSPNs and iSPNs are capable of eCB-LTD expression. In addition, bidirectional modulatory effects exerted by activation of different dopamine receptors in dSPNs and iSPNs underlie the postsynaptic cell-type dependence reported

in previous studies: using conventional local electrical stimulation, coactivation of D1R in dSPNs could mask eCB-LTD, whereas activation of D2R could facilitate eCB-LTD in iSPNs.

Next, we asked if LTD could be induced by the same oSTDP paradigm at thalamostriatal synapses. Interestingly, the same post-pre pairing protocol (Figure 5A) successfully induced LTD at thalamostriatal synapses in all the recorded SPNs (nine out of nine) (Figures 5H and 5I; 82% ± 3% of baseline, n = 9; p < 0.05, Wilcoxon). This LTD, however, was not blocked by AM251 (Figure 5J; 70% ± 8% of baseline, n = 7; p < 0.05, Wilcoxon), again suggesting that oSTDP-LTD at thalamostriatal synapses is not CB1R dependent. We further tested whether D2R agonist could enhance LTD by facilitating eCB release (Giuffrida et al., 1999; Wang et al., 2006). However, D2R agonist (quinpirole 10 mM) did not enhance the LTD at thalamostriatal synapses in iSPNs (Figure 5J; 70% ± 6% of baseline, n = 7; p < 0.05, Wilcoxon; p = 0.11, compared to control, U test). Finally, this LTD was blocked by NMDAR antagonist (R-CPP 10 mM) (Figure 5K; 95% ± 7% of baseline, n = 5; p = 0.44, Wilcoxon), indicating that LTD induced by oSTDP paradigm at thalamostriatal synapses is NMDAR dependent rather than eCB dependent. This finding, together with a recent report (Ellender et al., 2013), suggests that thalamostriatal synapses express a different form of LTD that does not involve CB1R activation.

Differential Expression of Cbr1 mRNA in Cortical and Thalamic Projection Neurons

What could cause such a difference in eCB-LTD induction in corticostriatal and thalamostriatal synapses? It is possible that eCB

A oSTDP LTD induction

^0St J_I_I_I_L

200 CD

13 150- 1 • J

-Q • • Thyl- •

EPSP (% c un c o o c ChR2; dSPN #

ro -Q 100

Q_ LL1

10 20 30 Time (min)


t Thy1 -ChR2; dSPNs (6)

"Ôj £ 120 <u

j§ lOOfi

10 20 30 Time (min)


Thy1-ChR2; dSPNs

• SKF81297(5)

• AM25K5)

10 20 30 Time (min)

^200 0J

^53150 ra -Q

"a — 120

£ 60 LU

1 Pairing 2 IM, 2mv|_ ! \\ 20 ms

•• • ' Î

• Vglut2 •Cre;Ai32

10 20 30 Time (min) Pairing


• AM25p(6)

♦ Quinpirole(7)


10 20 30 Time (min)

-70 mV ^



Q_ L±J

Pairing 2 \ 2 mv|_ V 20 ms

Cpo cPO ^ °J °o 3, O 2

ccow o o° «Vroc&cP S 0 e/

O Thy1- ChR2; iSPN

10 20 30 Time (min)

ra -Q 100

û. LO 60

Pairing IhR2; iSPNs (9)


0 1 20 30 40

c 120-

Time (min) Pairing

Thy1-CHR2; iSPNs ' O Quinpirole (7) O Quinpirole + AM251 (5) 0 10 20 30 Time (min)

"a? ~120 OJ

ê 60 uj

'ÔT — 120 a,

£ 60 uj


• Vglut -Cre;Ai32 (9)

0 1 0 20 30 40 Time (min) Pairing

■ Vglut2-( • R-CPP Jli.____in,,

re;Ai32 5)

10 20 30 Time (min)

Figure 5. Spike-Timing-Dependent LTD at Corticostriatal Synapses

(A) Left: thetheta-burst optogenetic spike-timing dependent plasticity (oSTDP) pairing protocol for induction of LTD in Thy1-ChR2-YFP mice. Right action potentials and EPSPs recorded from postsynaptic SPNs during induction. (B and C) An individual experiment showing the change of corticostriatal EPSP amplitude in adSPN (B)and an iSPN (C) before and after post-pre pairing STDP induction. Inset: averaged EPSP traces are collected from 10 to 15 traces of the first 5 min of baseline (1, black) and the last 5 min after induction (2, red).

production and release are different at these synapses. D2R activation can facilitate eCB production (Giuffrida et al., 1999; Wang et al., 2006) and induction of eCB-LTD (Kreitzer and Malenka, 2007). Indeed, when using an oSTDP induction protocol, D2R agonist is required for inducing eCB-LTD at corticostriatal synapses in iSPNs. However, this could not explain why LTD at thalamostriatal synapses is insensitive to CB1R antagonism. Furthermore, the DHPG protocol, which combined postsynaptic depolarization and mGluR1/5 activation, could bypass the activation of postsynaptic dopamine receptors. Striatal eCB-LTD expression requires activation of CB1Rs on the presynaptic terminals. Therefore, it is possible that the level of CB1R expression is different at corticostriatal and thalamostriatal terminals. To directly test this hypothesis, we first examined the expression levels of Cbr1, the gene that encodes CB1R, in cortical and thalamic projection neurons (Figures 6A and 6B). Using in situ hybridization, we found that Cbr1 mRNA was highly expressed in cortical neurons, especially in layer V neurons, whereas Cbr1 mRNA was barely detected in thalamus (Figure 6C). To confirm the specificity of Cbr1 expression in cortical projection neurons, we quantified the number of layer V/VI cells that contained Cbr1 mRNA. We found that nearly 30% of cortical layer V cells are Cbr1 positive, while less than 3% thalamic cells expressed Cbr1 mRNA (Figure 6D; cortical layer V: 26% ± 3%, thalamus 2% ± 1%, n = 8 slices from two animals, p < 0.01, U test). These distinctions in Cbr1 expression levels revealed that differential CB1R signaling might contribute to the dichotomous regulation of cortical and thalamic inputs to SPNs.

CB1R Is Colocalized with Cortical, but Not Thalamic, Terminals

To further provide an estimate of CB1R protein expression at these presynaptic axonal terminals, we performed immunostain-ing in striatal brain slices. It is well established that presynaptic terminals originating from the cortex and thalamus contain different types of Vgluts: corticostriatal terminals express Vglut1 and thalamostriatal terminals express Vglut2 (Fremeau et al., 2004). Therefore, we performed double immunostaining with one antibody against either Vglut1 or Vglut2 and a second antibody against CB1R (Figures 6E-6G). We found that CB1R is highly expressed in the dorsal striatum, in agreement with previous findings (Kano et al., 2009; Uchigashima et al., 2007). Furthermore, we found that many CB1R puncta are highly

(D and E) Summary of EPSP amplitudes in dSPNs (D) and iSPNs (E) in Thy1-ChR2-YFP mice.

(F) D1R agonist (SKF81297 3 mM; blue) and CB1R antagonist (AM251 5-10 mM; gray) suppressed oSTDP-LTD in dSPNs.

(G) oSTDP-LTD was successfully induced in iSPNs in D2R agonist (quinpirole 10 mM; blue) and was blocked by CB1R antagonist (gray).

(H) An individual experiment showing the change of thalamostriatal EPSP amplitude.

(I) Summary of EPSP amplitudes in SPNs from Vglut2-ChR2 mice.

(J) The oSTDP-LTD in thalamostriatal synapses on dSPNs was not blocked by CB1R antagonist (orange) and not facilitated by D2R agonist (dark orange). (K) The oSTDP-LTD in thalamostriatal synapses on SPNs was blocked by NMDAR antagonist (R-CPP 10 mM).

Data are represented as mean ± SEM. Gray bars indicate the period of drug application, and blue bars indicate the period for oSTDP induction.

colocalized with Vglut1-positive terminals. In stark contrast, CB1R puncta were rarely colocalized with Vglut2-positive terminals. To estimate the degree of colocalization of CB1Rs with

Figure 6. Differential Expression of CB1R at Cortico- and Thalamostriatal Inputs

(A) A sagittal section of brain slice showing the in situ hybridization of Cbr1 mRNA. Signal presented in pseudocolor (red).

(B) Enlarged images of Cbr1 in situ hybridization in cortex (left) and in thalamus (coronal section, right). Labeled as follows: Ctx, cortex; II—VI, layer of cortex; cc, corpus callosum; Tha, thalamus; Pf, parafascicular nucleus of the thalamus; Fr, fasciculus retroflexus.

(C) Overlapped images of Cbr1 in situ hybridization (red) and DAPI (blue) in cortex (left) and in thalamus (right) in the boxed areas in (B).

(D) Summary of percentage of Cbr1-positive cells in cortical layer V and thalamus (Tha; n = 8,4 slices each from two mice; *p < 0.05, U test).

(E and F) Double IHC detected Vglut1(E) and Vglut2(F) (left, red) and CB1R (middle, green) and merged image (right) in dorsolateral stratum.

(G) High-magnification images of merged images from dotted box areas in (E) and (F) revealed CB1R was more highly expressed in corticostriatal (Vglutl positive, left) than in thalamostriatal (Vglut2 positive, right) axon terminals.

(H) Summary of CB1R expression levels on Vglutl-and Vglut2-positive immunoreactive puncta (n = 10; 2 slices each from five mice; *p < 0.05, U test). In (D) and (H), box-and-whisker plots indicate the minimum, 25th, 50th, 75th, and maximum percentiles.

these terminals, we quantified the CB1R density on Vglutl - and Vglut2-containing axonal boutons (see the Experimental Procedures and Figure S6). We found that CB1R density is significantly higher on Vglutl-positive terminals than on Vglut2-positive terminals (Figure 6H; Vglutl: 12.2 ± 1.7 and Vglut2: 2.9 ± 0.8 CB1R immunofluorescence/pixel; n = 10 slices of five mice; p < 0.05, U test). Our results thus demonstrate a dramatic difference in CB1R expression pattern in corticostriatal and thalamostriatal synapses. This lack of CB1R expression on tha-lamostriatal terminals explains why eCB-LTD is nearly absent at thalamostriatal synapses.

Lack of CB1 Receptor Modulation at Thalamostriatal Synapses

Because we observed a dramatic difference in the CB1R expression pattern at cortico- and thalamostriatal terminals, we speculated that direct activation of CB1Rs by a CB1R agonist would produce a neuromodulatory effect. It has been shown that prolonged activation of CB1Rs alone (without postsynaptic depolarization, holding potential = -70 mV) is sufficient to induce LTD in

Figure 7. Direct Activation of CB1R Reveals a Presynaptic Mechanism in Pathway-Specific LTD in Corticostriatal and Thalamostriatal Synapses

(A) Left: individual experiments showing that activation of CB1R with CB1R agonist (Win-2 2 mM) for 20 min followed by 20 min application of CB1R antagonist (AM251 5-10 mM) induced a larger LTD in SPNs of Thy1-ChR2-YFP (upper left) than in Vglut2;ChR2 (lower left) mice. Middle: averaged EPSC traces for baseline (1, black) and for drug effect (2, red). Right: normalized EPSPs to show the changes in PPR.

(B) Summary of Win-2 induced LTD on SPNs from Thy1-ChR2-YFP (blue) and Vglut2-Cre;ChR2 (orange) mice. Error bar indicates SEM.

(C) Summary time courses of PPR changes during Win-2 induced LTD on SPNs showing that PPRs were increased in Thy1-ChR2-YFP (blue) but not Vglut2-Cre; ChR2 (orange) mice. Error bar indicates SEM. *p < 0.05, Wilcoxon. (D and F) Left: Summary time courses of Win-2-induced LTD on SPNs from Emx1-Cre;Ai32 (D) and Rbp4-Cre;Ai32 (F) mice. Error bar indicates SEM. Upper right: averaged EPSC traces for baseline (1, black) and for drug effect (2, red). Lower right: normalized EPSCs.

(E and G) Summary PPRs of baseline and 20 min after Win-2 treatment on SPNs from Emx1-Cre; Ai32 (E) and Rbp4-Cre; Ai32 (G) mice. *p < 0.05, Wilcoxon. Box-and-whisker plots indicate the minimum, 25th, 50th, 75th, and maximum percentiles.

striatal glutamatergic synapses (Kreitzer and Malenka, 2005). If the difference in presynaptic CB1R density accounts for the input-specific eCB-LTD, direct activation of presynaptic CB1R should recapitulate the difference in LTD between cortical and thalamic inputs. To test this, we applied 2 mM Win55,212-2 (Win-2), a selective CB1R agonist, for 20 min, followed by a CB1R antagonist (AM251 5-10 mM), while washing out Win-2 to ensure the changes in EPSC amplitudes were long-lasting. We found that Win-2 strongly reduced EPSCs at corticostriatal synapses, and this reduction in EPSCs was indeed long-lasting. In agreement with a presynaptic mechanism, we observed an increase in PPRs in Thy1-ChR2-YFP mice treated with Win-2 (Figures 7A-7C; EPSCs: 39% ± 3% of baseline, n = 11, p < 0.001; PPRs: baseline: 1.11 ± 0.10, Win-2: 1.37 ± 0.09, n = 11, p < 0.01, Wilcoxon). Although we also observed a reduction in EPSC amplitudes by Win-2 in Vglut2-Cre;ChR2 mice (Figures 7A and 7B; EPSCs: 77% ± 2% of baseline, n = 6, p < 0.05, Wilcoxon), the reduction level was significant smaller than in Thy1-ChR2-YFP mice (p < 0.05, U test) and there was no change in PPR (Figure 7C; PPRs: baseline: 0.65 ± 0.09, Win-2: 0.67 ± 0.10, n = 6, p = 0.44, Wilcoxon), suggesting the Win-2 effect was not presynaptic. The reduction in EPSC amplitudes in Vglu2-Cre;ChR2 mice might be caused by nonspecific (Matyas et al., 2006) or postsynaptic effects (Kreitzer et al., 2002). These results suggest that the differential eCB-LTD between cortical and thalamic inputs is due to a difference in presynaptic CB1R density. Finally, we confirmed our findings by testing Win-2-induced LTD in cortex-specific mouse lines. In both Emx1-Cre;Ai32 and Rbp4-Cre;Ai32 mice, more than 45% reduction in EPSC amplitudes was induced by Win-2 (Figures 7D and 7F; Emx1-Cre: 52% ± 3% of baseline, n = 7, p < 0.05; Rbp4-Cre: 54% ± 5% of baseline, n = 7, p < 0.05; Wilcoxon). The reduction in EPSC amplitudes was significantly larger in cortex-specific Cre mice than in Vglut2-Cre;ChR2 mice (p < 0.05, U test). LTD was accompanied by increases in PPR (Figures 7E and 7G; Emx1-Cre: baseline: 0.62 ± 0.14, Win-2: 0.83 ± 0.18, n = 7, p < 0.05; Rbp4-Cre: baseline: 0.50 ± 0.08, Win-2: 0.65 ± 0.09, n = 7, p < 0.05, Wilcoxon), suggesting the effect of Win-2 resulted from selective action on presynaptic terminals of cortical inputs. Thus, our results demonstrate that the input-specific eCB-LTD on striatal SPNs is due to a differential activation of presynaptic CB1Rs at cortical and thalamic terminals.


In this study, we take advantage of optogenetic approaches to examine presynaptic input- and postsynaptic cell-type specificity in striatal synaptic plasticity. Our tools allow us to selectively trigger glutamate release from corticostriatal or thalamostriatal synapses without coactivating other neuromodulatory systems. We report that eCB-LTD induced by DHPG and oSTDP at glutamatergic synapses on striatal SPNs is observed only at corticostriatal synapses, not at thalamostriatal synapses. This is due to differential expression of CB1Rs at presynaptic glutamatergic boutons; while CB1Rs are selectively expressed on axon terminals of cortical inputs, they are mostly absent on thalamic terminals (Figure S7). This dichotomy in CB1R expression patterns allows differential eCB-LTD expression at

corticostriatal synapses while preventing eCB-LTD of thalamos-triatal synapses.

eCB-LTD in Direct and Indirect Pathway SPNs

The question of selective eCB-LTD expression in iSPNs and not dSPNs continues to be a subject of debate in the field (Kreitzer and Malenka, 2007). Several observations support postsynaptic cell-type specificity of LTD induction: (1) selective blockade of D2Rs has been shown to abolish eCB-LTD in SPNs, (2) eCB-LTD has been shown to be absent in dSPNs (Kreitzer and Malenka, 2007; Nazzaro et al., 2012; Shen et al., 2008), and (3) D2R activation can facilitate eCB release (Giuffrida et al., 1999; Wang et al., 2006) and eCB-LTD in iSPNs (Kreitzer and Malenka, 2007). However, those views are challenged by studies demonstrating that eCB-LTD can, in fact, be induced in dSPNs and iSPN (Bagetta et al., 2011; Wang et al., 2006). These studies argue that the D2R dependency of eCB-LTD is mediated by disinhibition of cholinergic interneurons through M1 muscarinic receptors (Wang et al., 2006). A recent review article by Calabresi and colleagues (Calabresi et al., 2014) points out that crosstalk of different neuromodulatory systems, including dopamine, acetylcholine, eCB, and nitric oxide, might bridge the dichotomy between dSPNs and iSPNs. Depending on the experimental paradigms used, including stimulation electrode placement, stimulation intensity, slice preparation method (i.e., coronal versus parahorizontal), different neuromodulatory systems might be recruited and thus account for the conflicting observations between research groups. To avoid these complications, we used selective optogenetic activation of defined inputs to demonstrate that both dSPNs and iSPNs are capable of expressing eCB-LTD at corticostriatal glutamatergic synapses. We hypothesized that, because local electrical stimulation paradigms may activate heterogeneous presynaptic inputs, dopa-mine receptor signaling likely modulates, rather than determines, eCB-LTD induction in the striatum (Shen et al., 2008; Surmeier etal., 2014).

Our data directly support this conclusion. We show that eCB-LTD at corticostriatal synapses on dSPNs and iSPNs is indistinguishable when induced by DHPG, which bypasses dopamine modulation. Because DHPG-induced eCB-LTD is not sensitive to D1R or D2R antagonists, our data suggest that eCB production can simply be triggered by direct activation of intracellular signaling cascades that are downstream of dopamine receptors. This is consistent with previous studies that show that, when induction paradigms recruit sufficient downstream Ca2+ signaling, activation of dopaminergic receptors is not required for eCB-LTD (Adermark and Lovinger, 2007; Kreitzer and Mal-enka, 2005).

The dichotomous expression of eCB-LTD in dSPNs and iSPNs emerges when using an oSTDP protocol. We show that oSTDP-induced eCB-LTD is subject to dopamine modulation, suggesting that oSTDP activates intracellular signaling that converges with dopamine receptor modulation. In addition, the discrepancy between our data obtained with the oSTDP protocol and a previous STDP plasticity study can be explained by the general recruitment of dopamine fibers with local electrical stimulations (Shen et al., 2008). Our oSTDP protocol did not recruit dopamine fibers, and we observed eCB-LTD in dSPNs,

but not in iSPNs. This result is nearly identical compared to previous results obtained by using local stimulation and blockade of dopamine receptors. Shen and colleagues demonstrated that in the absence of D1R or D2R activity, LTD was successfully induced in dSPNs, but not iSPNs. We demonstrate that oSTDP combined with D2R activation could successfully induce eCB-LTD in iSPNs, which is consistent with previous findings that D2R activation can rescue LTD in iSPNs in Parkinson's disease animal models (Shen et al., 2008; Surmeier et al., 2014).

Previous studies have shown that LTD at corticostriatal synapses can be induced with yet another STDP protocol—a single spike followed by a single excitatory postsynaptic potential (EPSP)—and that this LTD is modulated by dopamine receptors (Pawlak and Kerr, 2008) and the local GABAergic circuit (Fino andVenance, 2010). However, most of these studies did notsys-tematically identify postsynaptic cell type (Fino and Venance, 2010), making it difficult to directly compare their results with ours. Nevertheless, these findings, together with ours, suggest SPN dendrites are a critical place for convergent glutamatergic, GABAergic, and dopaminergic signaling.

Dopamine Modulation of Cell-Type-Dependent eCB-LTD

There are several possible explanations for how the activation of mixed inputs via electrical stimulation may cause eCB-LTD expression in only D2R-expressing iSPNs. First, dopamine fibers might be activated by electrical stimulation and lead to dopa-mine receptor activation. D2R activation can enhance eCB production (Giuffrida et al., 1999; Wang et al., 2006) by activating Gbg and phospholipase C signaling, facilitating eCB-LTD in iSPNs (Hernandez-Lopez et al., 2000; Kreitzer and Malenka, 2007). Conversely, D1R activation in dSPNs can shift synaptic plasticity toward long-term potentiation (LTP) (Calabresi et al., 1992; Kerr and Wickens, 2001; Shen et al., 2008) by enhancing postsynaptic responsiveness or, presumably, by inhibiting Gq through a protein kinase A to RGS4 signaling pathway (Lerner and Kreitzer, 2012; Surmeier et al., 2014). Second, electrical stimulation may also activate cholinergic fibers, which would enhance the muscarinic tone in the striatum. M1 muscarinic receptors preferentially modulate Kir2 channels in iSPN dendrites (Shen et al., 2007), which in turn preferentially enhance iSPN dendritic excitability. Because iSPNs are more excitable than dSPNs (Day et al., 2008; Kreitzer and Malenka, 2007), they might favor eCB production in response to synaptic stimulation and postsynaptic depolarization. In our study, we utilized optoge-netic activation that avoids these complications. With this direct and selective methodology, we demonstrated that both dSPNs and iSPNs are capable of expressing eCB-LTD at corticostriatal glutamatergic synapses using DHPG combined with mild depolarization. Our findings suggest that striatal eCB-LTD is not cell-type dependent, per se. We postulate that the absence of eCB-LTD in dSPNs, as reported by Kreitzer and Malenka (2007), may in fact be the consequence of activating D1Rs, which suppress eCB release and thus prevent the eCB-LTD in dSPNs. Therefore, depending on the stimulation paradigm used, synaptic plasticity in striatum can be fine-tuned by different G protein-coupled receptors (Surmeieret al., 2014).

It is interesting that the oSTDP protocol induced eCB-LTD only in dSPNs and not in iSPNs. Why do iSPNs lack LTD when dopamine release is not triggered? It is possible that other modulatory systems are engaged. In addition to D2Rs, iSPNs also express adenosine 2A receptor (A2AR) (Schiffmann et al., 2007). Tonic activation of A2ARs by extracellular adenosine may suppress LTD induction via activating downstream Gs signaling pathways, which counteract the Ca2+ signaling required for eCB production (Higley and Sabatini, 2010; Lerneretal., 2010; Shen et al., 2008; Surmeier et al., 2014). dSPNs do not express A2ARs, but a similar role is thought to be carried out by D1Rs (Surmeier etal., 2014). It is also possible that D1R activation suppresses LTD induction in dSPNs by activating the Gs signaling pathway (Figure 5F). Our results further support the conclusion that dopamine, together with adenosine and acetylcholine, plays a modulatory role in eCB-LTD induction at corticostriatal synapses.

Input-Specific eCB-LTD on Striatal SPNs

Divergent long-term plasticity at corticostriatal and thalamos-triatal synapses has functional consequences: corticostriatal signaling is thought to play a role in cognitive and motivational goal-directed behavior and associative learning (Graybiel, 2000), while thalamostriatal projections convey salient sensory stimuli and are involved in attention shift (Ding etal., 2010; Matsu-moto et al., 2001; Minamimoto et al., 2009). The differences we observed in the physiological properties of corticostriatal and thalamostriatal synapses on SPNs are consistent with this functional divergence. The postsynaptic NMDAR component was significantly larger at corticostriatal synapses than at thalamostriatal synapses (Figures 1 and S3) (Ding et al., 2008). The relative abundance of NMDARs at corticostriatal synapses suggests preferential LTP induction in response to high-frequency afferent stimulation. The presence of eCB-LTD at corticostriatal synapses, as demonstrated in the current study, suggests that corticostriatal synapses exhibit a bidirectional plasticity that can be potentiated or depressed (Fino et al., 2005; Shen et al., 2008). These abilities are critical for motor learning and adaptive behavior (Costa, 2007; Koralek et al., 2012). Thalamostriatal signaling, in contrast, is thought to transmit salient sensory events and to play a role in attention and arousal, rather than learning-related plasticity (Bradfieldetal.,2013; Matsumotoetal.,2001).Therefore, lacking CB1Rs at presynaptic thalamostriatal terminals circumvents the effects of eCB spillover, ensuring that these synapses preserve their ability to redirect attention toward salient stimuli.

Although CB1Rs are only nominally expressed at thalamos-triatal axons, thalamostriatal synapses are still capable of undergoing long-term plasticity. It is interesting that the oSTDP protocol could induce an NMDAR-dependent LTD that is insensitive to eCB. It has been shown that thalamostriatal inputs from Pf exhibit NMDAR-dependent LTD, regardless of using pre-post or post-pre STDP induction protocols (Ellender et al., 2013). This could imply that when salient stimuli emerge with background stimulation, attentional shift mediated by thalamostriatal system wanes. Short-term synaptic depression and NMDAR-dependent LTD may be responsible for behavioral desensitization and attentional shift when salient events become background. Input-specific long-term plasticity, on the other hand, could be a general mechanism for motor learning and fine movement con-

trol. For example, serotonin-mediated LTD at striatal SPNs has been proposed to be restricted to cortical inputs (Mathur et al., 2011), whereas opioids induce LTD selectively at thalamostriatal synapses in the dorsal striatum (Atwood et al., 2014a). These findings, together with our present study, suggest a common input-specific plasticity mediated by presynaptic Gi/o coupled receptors at striatal excitatory synapses (Atwood et al., 2014b). It is possible that, during exercise or other repeated training paradigms for motor learning, different long-term plasticity mechanisms are engaged during different activity patterns of postsynaptic SPNs, ultimately fine tuning convergent inputs for precise movement control. The findings from our studies not only provide a neural substrate for corticothalamostriatal-rele-vant behaviors but also open avenues for the study of interactions of synaptic plasticity at highly convergent and intermingled synaptic inputs.


Adult (8-12 weeks) mice were used for this study. Stereotaxic injections were performed on postnatal day 17 (P17)-P18 Vglut2-Cre;Drd1-tdTomato or Vglut2- Cre;Drd2-eGFP mice and used at least 8 weeks postinjection. Injection sites were verified in slices fixed after each recording session (Figure S1A). All procedures were approved by Stanford University's Administrative Panel on Laboratory Animal Care.


Oblique horizontal brain slices (300 mm) containing the dorsal striatum (Figure 1C) were obtained using standard techniques (Ding et al., 2008). Whole-cell voltage- or current-clamp recordings were conducted using cesium methylsulfonate-based or potassium methylsulfonate-based internal solutions, respectively. Picrotoxin (50 mM) was included in the artificial cere-brospinal fluid to block GABAA receptors in experiments with voltage-clamp recording. Data were recorded with custom-made software written in MATLAB (MathWorks).

Optogenetic Stimulation

To stimulate ChR2-expressing axons, blue laser light (450 nm) wasfocused on the back focal plane of the objective. For the oSTDP induction paradigm, back-propagating APs (bAPs) were evoked by somatic current injection (2 ms, 2 nA). The LTD induction protocol consisted of 20 trains of five bursts (5 Hz), repeated at 1 Hz. The burst was composed of three bAPs followed by an EPSP evoked by blue laser stimulation with a 10 ms delay. Recorded neurons were depolarized to r—70 mV during the induction period (Figure 5A). The long-term changes in EPSCs/ EPSPs were calculated by averaging EPSC/EPSP amplitudes 25-30 min following the induction protocol and comparing the value to the average EPSC/ EPSP during the baseline. PPRs were measured by dividing the peak of the second EPSC by the first EPSC. For detailed methods, see Supplemental Experimental Procedures.


Supplemental Information includes Supplemental Experimental Procedures and seven figures and can be found with this article online at http://dx.doi. org/10.1016/j.celrep.2014.12.005.


Y.-W.W. and J.B.D. designed the experiments. Y.-W.W., J.I.K., and J.B.D. performed the electrophysiology and immunohistochemistry experiments. V.L.T. and G.S. performed the in situ hybridization experiments. Y.-W.W., R.R.L., and J.B.D. wrote the manuscript with contributions from J.I.K., V.L.T., and G.S.


The authors thank Drs. Lu Chen, John Huguenard, Zayd Khaliq, and members of J.B.D. laboratory for helpful discussions. Supported by grants from the NINDS/NIH NS075136 (J.B.D.) and the Klingenstein Foundation (J.B.D.), NIH/NIDA DA031777 (G.S.) and start-up funds from Stanford University Department of Anesthesiology, Perioperative and Pain Medicine and Stanford Institute for Neuro-Innovation and Translational Neurosciences (G.S). V.L.T. is supported by the Foundation for Anesthesia Education and Research RFG.

Received: February 2, 2014 Revised: November 11, 2014 Accepted: December 2, 2014 Published: December 24, 2014


Adermark, L., and Lovinger, D.M. (2007). Combined activation of L-type Ca2+ channels and synaptic transmission is sufficient to induce striatal long-term depression. J. Neurosci. 27, 6781-6787.

Arenkiel, B.R., Peca, J., Davison, I.G., Feliciano, C., Deisseroth, K., Augustine, G.J., Ehlers, M.D., and Feng, G. (2007). In vivo light-induced activation of neural circuitry in transgenic mice expressing channelrhodopsin-2. Neuron 54, 205-218.

Atwood, B.K., Kupferschmidt, D.A., and Lovinger, D.M. (2014a). Opioids induce dissociable forms of long-term depression of excitatory inputs to the dorsal striatum. Nat. Neurosci. 17, 540-548.

Atwood, B.K., Lovinger, D.M., and Mathur, B.N. (2014b). Presynaptic long-term depression mediated by Gi/o-coupled receptors. Trends Neurosci. 37, 663-673.

Bagetta, V., Picconi, B., Marinucci, S., Sgobio, C., Pendolino, V., Ghiglieri, V., Fusco, F.R., Giampa, C., and Calabresi, P. (2011). Dopamine-dependent long-term depression is expressed in striatal spiny neurons of both direct and indirect pathways: implications for Parkinson's disease. J. Neurosci. 31, 12513-12522.

Bradfield, L.A., Bertran-Gonzalez, J., Chieng, B., and Balleine, B.W. (2013). The thalamostriatal pathway and cholinergic control of goal-directed action: interlacing new with existing learning in the striatum. Neuron 79, 153-166. Calabresi, P., Maj, R., Pisani, A., Mercuri, N.B., and Bernardi, G. (1992). Long-term synaptic depression in the striatum: physiological and pharmacological characterization. J. Neurosci. 12, 4224-4233.

Calabresi, P., Picconi, B., Tozzi, A., Ghiglieri, V., and Di Filippo, M. (2014). Direct and indirect pathways of basal ganglia: a critical reappraisal. Nat. Neurosci. 17, 1022-1030.

Costa, R.M. (2007). Plastic corticostriatal circuits for action learning: what's dopamine got to do with it? Ann. N Y Acad. Sci. 1104,172-191. Day, M., Wokosin, D., Plotkin, J.L., Tian, X., and Surmeier, D.J. (2008). Differential excitability and modulation of striatal medium spiny neuron dendrites. J. Neurosci. 28, 11603-11614.

Ding, J., Peterson, J.D., and Surmeier, D.J. (2008). Corticostriatal and thala-mostriatal synapses have distinctive properties. J. Neurosci. 28, 6483-6492. Ding, J.B., Guzman, J.N., Peterson, J.D., Goldberg, J.A., and Surmeier, D.J. (2010). Thalamic gating of corticostriatal signaling by cholinergic interneurons. Neuron 67, 294-307.

Doig, N.M., Moss, J., and Bolam, J.P. (2010). Cortical and thalamic innervation of direct and indirect pathway medium-sized spiny neurons in mouse striatum. J. Neurosci. 30, 14610-14618.

Ellender, T.J., Harwood, J., Kosillo, P., Capogna, M., and Bolam, J.P. (2013). Heterogeneous properties of central lateral and parafascicular thalamic synapses in the striatum. J. Physiol. 591, 257-272.

Fino, E., and Venance, L. (2010). Spike-timing dependent plasticity in the striatum. Front Synaptic Neurosci 2, 6.

Fino, E., Glowinski, J., and Venance, L. (2005). Bidirectional activity-dependent plasticity at corticostriatal synapses. J. Neurosci. 25, 11279-11287.

Fremeau, R.T., Jr., Voglmaier, S., Seal, R.P., and Edwards, R.H. (2004). VGLUTs define subsets of excitatory neurons and suggest novel roles for glutamate. Trends Neurosci. 27, 98-103.

Gerdeman, G.L., Ronesi, J., and Lovinger, D.M. (2002). Postsynaptic endo-cannabinoid release is critical to long-term depression in the striatum. Nat. Neurosci. 5, 446-451.

Gerfen, C.R. (1989). The neostriatal mosaic: striatal patch-matrix organization is related to cortical lamination. Science 246, 385-388.

Giuffrida, A., Parsons, L.H., Kerr, T.M., Rodriguez de Fonseca, F., Navarro, M., and Piomelli, D. (1999). Dopamine activation of endogenous cannabinoid signaling in dorsal striatum. Nat. Neurosci. 2, 358-363.

Glickfeld, L.L., Andermann, M.L., Bonin, V., and Reid, R.C. (2013). Cortico-cortical projections in mouse visual cortex are functionally target specific. Nat. Neurosci. 16, 219-226.

Graybiel, A.M. (2000). The basal ganglia. Curr. Biol. 10, R509-R511.

Graybiel, A.M., Aosaki, T., Flaherty, A.W., and Kimura, M. (1994). The basal ganglia and adaptive motor control. Science 265, 1826-1831.

Hernandez-Lopez, S., Tkatch, T., Perez-Garci, E., Galarraga, E., Bargas, J., Hamm, H., and Surmeier, D.J. (2000). D2 dopamine receptors in striatal medium spiny neurons reduce L-type Ca2+ currents and excitability via a novel PLC[beta]1-IP3-calcineurin-signaling cascade. J. Neurosci. 20, 8987-8995.

Higley, M.J., and Sabatini, B.L. (2010). Competitive regulation of synaptic Ca2+ influx by D2 dopamine and A2A adenosine receptors. Nat. Neurosci. 13, 958-966.

Kano, M., Ohno-Shosaku, T., Hashimotodani, Y., Uchigashima, M., and Wata-nabe, M. (2009). Endocannabinoid-mediated control of synaptic transmission. Physiol. Rev. 89, 309-380.

Kerr, J.N., and Wickens, J.R. (2001). Dopamine D-1/D-5 receptor activation is required for long-term potentiation in the rat neostriatum in vitro. J. Neurophysiol. 85, 117-124.

Koralek, A.C., Jin, X., Long, J.D., 2nd, Costa, R.M., and Carmena, J.M. (2012). Corticostriatal plasticity is necessary for learning intentional neuroprosthetic skills. Nature 483, 331-335.

Kreitzer, A.C., and Malenka, R.C. (2005). Dopamine modulation of state-dependent endocannabinoid release and long-term depression in the stria-tum. J. Neurosci. 25, 10537-10545.

Kreitzer, A.C., and Malenka, R.C. (2007). Endocannabinoid-mediated rescue of striatal LTD and motor deficits in Parkinson's disease models. Nature 445, 643-647.

Kreitzer, A.C., and Malenka, R.C. (2008). Striatal plasticity and basal ganglia circuit function. Neuron 60, 543-554.

Kreitzer, A.C., Carter, A.G., and Regehr, W.G. (2002). Inhibition of interneuron firing extends the spread of endocannabinoid signaling in the cerebellum. Neuron 34, 787-796.

Lerner, T.N., and Kreitzer, A.C. (2012). RGS4 is required for dopaminergic control of striatal LTD and susceptibility to parkinsonian motor deficits. Neuron 73, 347-359.

Lerner, T.N., Horne, E.A., Stella, N., and Kreitzer, A.C. (2010). Endocannabi-noid signaling mediates psychomotor activation by adenosine A2A antagonists. J. Neurosci. 30, 2160-2164.

Lin, J.Y., Lin, M.Z., Steinbach, P., and Tsien, R.Y. (2009). Characterization of engineered channelrhodopsin variants with improved properties and kinetics. Biophys. J. 96, 1803-1814.

Madisen, L., Mao, T., Koch, H., Zhuo, J.M., Berenyi, A., Fujisawa, S., Hsu, Y.W., Garcia, A.J., 3rd, Gu, X., Zanella, S., et al. (2012). A toolbox of Cre-dependent optogenetic transgenic mice for light-induced activation and silencing. Nat. Neurosci. 15, 793-802.

Mathur, B.N., Capik, N.A., Alvarez, V.A., and Lovinger, D.M. (2011). Serotonin induces long-term depression at corticostriatal synapses. J. Neurosci. 31, 7402-7411.

Matsumoto, N., Minamimoto, T., Graybiel, A.M., and Kimura, M. (2001). Neurons in the thalamic CM-Pf complex supply striatal neurons with information about behaviorally significant sensory events. J. Neurophysiol. 85, 960-976. Matyas, F., Yanovsky, Y., Mackie, K., Kelsch, W., Misgeld, U., and Freund, T.F. (2006). Subcellular localization of type 1 cannabinoid receptors in the rat basal ganglia. Neuroscience 137, 337-361.

Minamimoto, T., Hori, Y., and Kimura, M. (2009). Roles of the thalamic CM-PF complex-Basal ganglia circuit in externally driven rebias of action. Brain Res. Bull. 78, 75-79.

Nazzaro, C., Greco, B., Cerovic, M., Baxter, P., Rubino, T., Trusel, M., Parolaro, D., Tkatch, T., Benfenati, F., Pedarzani, P., and Tonini, R. (2012). SK channel modulation rescues striatal plasticity and control over habit in cannabinoid tolerance. Nat. Neurosci. 15, 284-293.

Pawlak, V., and Kerr, J.N. (2008). Dopamine receptor activation is required for corticostriatal spike-timing-dependent plasticity. J. Neurosci. 28, 2435-2446. Schiffmann, S.N., Fisone, G., Moresco, R., Cunha, R.A., and Ferre, S. (2007). Adenosine A2A receptors and basal ganglia physiology. Prog. Neurobiol. 83, 277-292.

Shen, W., Tian, X., Day, M., Ulrich, S., Tkatch, T., Nathanson, N.M., and Surmeier, D.J. (2007). Cholinergic modulation of Kir2 channels selectively elevates dendritic excitability in striatopallidal neurons. Nat. Neurosci. 10, 1458-1466. Shen, W., Flajolet, M., Greengard, P., and Surmeier, D.J. (2008). Dichotomous dopaminergic control of striatal synaptic plasticity. Science 321, 848-851. Smith, Y., Raju, D.V., Pare, J.F., and Sidibe, M. (2004). The thalamostriatal system: a highly specific network of the basal ganglia circuitry. Trends Neurosci. 27, 520-527.

Surmeier, D.J., Song, W.J., and Yan, Z. (1996). Coordinated expression of dopamine receptors in neostriatal medium spiny neurons. J. Neurosci. 16, 6579-6591.

Surmeier, D.J., Ding, J., Day, M., Wang, Z., and Shen, W. (2007). D1 and D2 dopamine-receptor modulation of striatal glutamatergic signaling in striatal medium spiny neurons. Trends Neurosci. 30, 228-235.

Surmeier, D.J., Graves, S.M., and Shen, W. (2014). Dopaminergic modulation of striatal networks in health and Parkinson's disease. Curr. Opin. Neurobiol. 29C, 109-117.

Uchigashima, M., Narushima, M., Fukaya, M., Katona, I., Kano, M., and Watanabe, M. (2007). Subcellular arrangement of molecules for 2-arachido-noyl-glycerol-mediated retrograde signaling and its physiological contribution to synaptic modulation in the striatum. J. Neurosci. 27, 3663-3676.

Wall, N.R., De La Parra, M., Callaway, E.M., and Kreitzer, A.C. (2013). Differential innervation of direct- and indirect-pathway striatal projection neurons. Neuron 79, 347-360.

Wang, Z., Kai, L., Day, M., Ronesi, J., Yin, H.H., Ding, J., Tkatch, T., Lovinger, D.M., and Surmeier, D.J. (2006). Dopaminergic control of corticostriatal long-term synaptic depression in medium spiny neurons is mediated by cholinergic interneurons. Neuron 50, 443-452.

Wang, H., Peca, J., Matsuzaki, M., Matsuzaki, K., Noguchi, J., Qiu, L., Wang, D., Zhang, F., Boyden, E., Deisseroth, K., et al. (2007). High-speed mapping of synaptic connectivity using photostimulation in Channelrhodopsin-2 transgenic mice. Proc. Natl. Acad. Sci. USA 104, 8143-8148.