Glutamatergic Tuning of Hyperactive Striatal Projection Neurons Controls the Motor Response to Dopamine Replacement in Parkinsonian Primates

Abstract

Dopamine (DA) loss in Parkinson's disease (PD) alters the function of striatal projection neurons (SPNs) and causes motor deficits, but DA replacement can induce further abnormalities. A key pathological change in animal models and patients is SPN hyperactivity; however, the role of glutamate in altered DA responses remains elusive. We tested the effect of locally applied AMPAR or NMDAR antagonists on glutamatergic signaling in SPNs of parkinsonian primates. Following a reduction in basal hyperactivity by antagonists at either receptor, DA inputs induced SPN firing changes that were stable during the entire motor response, in clear contrast with the typically unstable effects. The SPN activity reduction over an extended putamenal area controlled the release of involuntary movements in the "on" state and therefore improved motor responses to DA replacement. These results demonstrate the pathophysiological role of upregulated SPN activity and support strategies to reduce striatal glutamate signaling for PD therapy.

Keywords: AMPAR; NMDAR; Parkinson’s disease; SPN; dopamine; dyskinesia; glutamate; medium spiny neuron; striatal projection neuron; striatum.

Figure 1. In Vitro Assessment of Drug Concentration-Effect Relationship, Prediction of Brain Dilution, and Recording Timeline

(A) The effects of increasing concentrations of LY235959 on responses to rapid application of 1 mM glutamate from GluA1/stargazin receptors expressed in HEK cells measured using patch-clamp recordings (black symbols; five HEK cells per concentration). The effects of LY235959 on GluN1/GluN2A receptor responses activated by 1 mM glutamate and 10 μM glycine expressed in oocytes held under two-electrode voltage clamp (red symbols; five oocytes). Error bars indicate SEM. (B) Concentration-response data for NBQX inhibition of GluN1/GluN2A responses to 1 mM glutamate and 10 μM glycine for receptors expressed in Xenopus oocytes measured using two-electrode voltage clamp (black symbols; four oocytes). NBQX concentration response data for GluA1/stargazin receptors activated by 1 mM glutamate obtained using patch-clamp recordings (red symbols; four HEK cells). Error bars indicate SEM. (C) Schematic representation of the injectrode used to apply aSCF, LY235959, or NBQX to the putamen of primates. (D) Peak LY235959 concentration (y axis) reached at various distances from the injection site (x axis) 1, 10, and 30 min following injections. The color-coded traces refer to data obtained when injecting 1, 3, or 9 mM LY235959. The gray shaded area between 100 and 200 μm represents the location from which the SPN activity was recorded. (E) As in (D) for 0.5, 1, or 3 mM NBQX. (F) Schematic drawing of the injection site depicting the injectrode with recording electrode in the putamen (left) and a coronal brain section with a small scar at the site of guide cannula penetration. The dashed blue line represents the injectrode trajectory (right). (G) Timeline of the continuous single-cell recording showing data storage before (“pre,” black box) and after the local injection (L.I.) of antagonist or vehicle (“post,” red box), and again after transitioning to the “on” state and the dyskinesia state (blue boxes) following the s.c. L-DOPA injection.

Figure 2. NMDAR Antagonist Injection at the Site of SPN Recording Reduced the Basal Activity and Stabilized DA Responses in the Parkinsonian Primate

(A–D) Control tests. Firing frequency changes of each SPN at baseline (pre), after local injection of aCSF (post), and after transitioning to “on” (on) and dyskinesias states (dys) following s.c. injection of L-DOPA. In each panel, each colored curve (top graph) represents an individual SPN grouped according to the type of DA response (increase, A and B, or decrease, C and D, in the “on” state followed by stable, A and C, or unstable, B and D, response in the dyskinesia state), and the averaged firing frequency change for the group is shown as percentage (bottom graph). In each SPN, differences between post and on are significant at p < 0.01 (A–D). Responses were classified as unstable by significant changes in the dyskinesia state (p < 0.01). Total SPNs, 29; units with activity increase, 14; units with activity decrease, 15. (E–H) NMDAR antagonist tests. The firing frequency changes of each SPN as described above for control tests are shown in (E) and (G) after local injection of LY235959. In each SPN, differences between pre and post and between post and on are significant at p < 0.01. Differences between on and dys were non-significant (see also Figure S3). SPN stable responses after LY235959 are compared with control tests in (F) and (H). Total SPNs, 39; units with activity increase in the “on” state, 24; units with activity decrease, 15. In each group analysis, ^p < 0.01 versus baseline, *p < 0.01 versus post aCSF or LY235959 injection, and ⁺p < 0.01 versus the “on” state (ANOVAs for repeated-measure followed by Bonferroni correction). In (F) and (H), *p < 0.05 between LY235959 and aCSF unstable response (one way ANOVAs). Error bars indicate SEM (n = 5 primates; see Table S1). (I–L) Examples of SPN unstable responses in aCSF tests (I and J) or stable responses in LY235959 tests (K and L). Short (5 s) spike trains and spectrograms for the segment duration (180 s) are shown for each segment (pre, post, on, and dys). The unit activity is colored after spike sorting as the corresponding curve in the frequency graphs (B), (D), (E), and (G).

Figure 3. AMPAR Antagonist Injection at the Site of SPN Recording Reduced the Basal Activity and Stabilized DA Responses in the Parkinsonian Primate

(A–D) The same control data as presented in Figures 2A–2D, respectively, for comparison with results obtained with the AMPAR antagonist. See Figure 2 for details. (E–H) AMPAR antagonist tests. Firing frequency changes of each SPN as described for control tests are shown in (E) and (G) after local injection of NBQX. In each SPN, differences between Pre and Post, and between Post and On are significant at p < 0.01. Differences between On and Dys were non-significant (see also Figure S3). SPN stable responses after NBQX are compared to control tests in (F) and (H). Total SPNs, 43; units with activity increase in the “on” state, 26; units with activity decrease, 17. In each group analysis, ^p < 0.01 versus baseline, *p < 0.01 versus Post aCSF or NBQX injection, and +p < 0.01 versus the “on” state (ANOVAs for repeated measure followed by Bonferroni’s correction). In (F) and (H), *p < 0.05 between NBQX and aCSF unstable response (one way ANOVAs). Error bars indicate SEM (n = 5 primates; see Table S1). (I–L) Examples of SPN unstable responses in aCSF tests ([I and J], same examples as in Figures 2I and 2J for comparison) or stable responses in NBQX tests (K and L). Short (5 s) spike trains and spectrograms for the segment duration (180 s) are shown for each segment, Pre, Post, On, and Dys. The unit activity is colored after spike sorting as the corresponding curve in the frequency graphs B, D, E and G.

Figure 4. DA Responses after Local NMDAR or AMPAR Blockade Are Stable across SPNs

The proportion of SPNs with stable DA responses is compared after local injection of aCSF, LY235959, or NBQX. More than 90% of SPNs with LY235959 or NBQX injection exhibited stable DA responses (activity increase or decrease), but fewer than 50% of SPNs with aCSF injection. Total SPNs, 117; stable responses, 39 of 43 in LY235959 tests, 43 of 45 in NBQX tests, and 12 of 29 in aCSF tests (see complementary data on unstable DA responses in Figure S3).

Figure 5. Relationship between the Effect of NMDAR or AMPAR Antagonist and the Magnitude of the SPN Response to DA

(A–D) Significant (blue) and non-significant (red) correlations between the activity reduction induced by LY235959 (A and C) or NBQX (B and D) and the DA response analyzed in the “on” state or dyskinesia state (top and bottom graphs, respectively in each panel). Firing frequency reduction post LY235959 or NBQX: ratio of post-antagonist injection to baseline frequencies. Firing frequency increase or decrease in “on” or dyskinesia state: ratio of motor state to post-antagonist injection frequencies. SPNs included in all regression analyses had stable responses to DA (total SPNs, 82; in LY235959 tests, 24 SPNs with activity increase in response to DA and 15 with activity decrease; in NBQX tests, 26 SPNs with activity increase in response to DA and 17 with activity decrease).

Figure 6. Infusion of NMDAR Antagonist over the Putamen Reduced Contralateral Dyskinesias Induced by L-DOPA in Parkinsonian Primates

(A and B) Time course of global (A) and contralateral (CL) (B) dyskinesias induced by s.c. injection of a suboptimal dose of L-DOPA after unilateral infusion of LY235959 into the posterolateral putamen. Reduced global scores by LY235959 infusion (red) (9 mM) compared with the control vehicle infusion (black) reflect differences in scores on the contralateral side of the body (see also Movies S1 and S2). Time 0, before L-DOPA injection (“off” state before infusion). Scores after L-DOPA injection: 30 min post-injection and thereafter every 20 min interval until dyskinesias disappear and the mobility was returning to the “off” state. *p < 0.01, two-way ANOVAs for repeated-measures followed by Fisher’s PLSD test. (C and D) Total and peak scores of global (C) and contralateral (D) L-DOPA-induced dyskinesias after infusion of LY235959 (red) and aCSF (black). Scales are adjusted for the contralateral side (D). AUC, area under the curve. Peak values, 50 min interval scores. *p < 0.01, paired t tests. (E) Motor disability scores (MDS) showing no changes in the antiparkinsonian effects of L-DOPA after LY235959 infusion in comparison with aCSF infusion. Each animal (n = 3; see details in Table S1) received one infusion of each dose (LY235959 0 and 9 mM). Error bars indicate SEM.