Activation of Cortical Somatostatin Interneurons Rescues Synapse Loss and Motor Deficits after Acute MPTP Infusion.

Adult dendritic spines present structural and functional plasticity, which forms the basis of learning and memory. To provide in vivo evidence of spine plasticity under neurotoxicity, we generated an acute motor deficit model by single injection of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) into adult mice. Acute MPTP infusion impairs motor learnings across test paradigms. In vivo two-photon imaging further revealed MPTP-induced prominent dendritic spine loss and substantially increased calcium spikes in apical tufts of layer 5 pyramidal neurons in the motor cortex. MPTP infusion also decreased the activity of somatostatin (SST)-expressing inhibitory interneurons. Further chemogenetic re-activation of SST interneurons reversed MPTP-induced hyperactivation of dendrites, rescued spine loss, and enhanced motor learning. Taken together, our study reports MPTP-induced structural and functional deficits of dendritic spines and suggests the potency of modulating local inhibitory transmission to relieve neurological disorders.

Introduction

The learning of new motor skills leads to the genesis of postsynaptic dendritic spines in the motor cortex, and the retention of motor memory is associated with the maintenance of these learning-related new spines (Cichon and Gan, 2015, Xu et al., 2009, Xu et al., 2012, Yang et al., 2009, Yang et al., 2014). Under pathological conditions such as those in Parkinson disease (PD), dendritic spines in the motor cortex present instability and high levels of turnover (Guo et al., 2015). Agreeing with these findings, functional magnetic resonance imaging studies reveal pathological hyperactivity in the motor cortex of patients with PD (Pelled et al., 2002, Sabatini et al., 2000, Yu et al., 2007). Such evidence suggests dysfunctions of the motor cortex in motor learning impairments (Guo et al., 2015, Sabatini et al., 2000). Therefore, further studies for the relationship among synaptic activity, spine activity, and motor function may provide implications for the intervention of motor disorders.

Many lines of evidence indicate that the generation and propagation of Ca2+ spikes in pyramidal neuron dendrites coordinate synaptic activity and facilitate synaptic potentiation and/or depotentiation (Cichon and Gan, 2015, Weber et al., 2016). Dendritic Ca2+ spikes participate in the strengthening and/or remodeling of dendritic spines during both development and motor learning (Li et al., 2017), whereas aberrant dendritic activity is associated with synaptic depotentiation in the diseased brain (Bai et al., 2017). The generation of Ca2+ spikes is heavily regulated by local inhibitory interneurons, which also play an important role in modulating structural synaptic changes in pyramidal cells (Hattori et al., 2017). In Alzheimer disease (AD), dysfunction of somatostatin-expressing inhibitory interneurons (SST-INs) results in progressive hippocampal synapse loss (Schmid et al., 2016). Following repeated stress in mice, decreased activity of parvalbumin-expressing inhibitory interneurons causes a high rate of dendritic spine elimination in the somatosensory cortex (Chen et al., 2018). In the motor cortex, a recent study showed that SST-INs are tightly correlated with learning-induced activities of pyramidal neurons (Adler et al., 2019). However, no study has investigated the involvement of SST-INs under pathological conditions.

In this study, we generated an acute motor deficit model by introducing a single dose of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), which led to impaired motor memory retention. In vivo imaging of the motor cortex revealed enhanced elimination of learning-related dendritic spines, increased pyramidal neuron dendritic activity, as well as decreased activities of SST-INs. Activation of SST-INs in the motor cortex restored dendritic Ca2+ activity and synaptic stability and improved motor skill retention. Our results suggest that targeting SST-INs may help to relieve MPTP-induced cortical dysfunction and motor deficits.

Results

Dendritic Spine Instability and Motor Memory Deficits Induced by Single MPTP Infusion

We repeatedly imaged fluorescently labeled dendritic spines of layer 5 pyramidal neurons (L5PRNs) in the primary motor cortex using transcranial two-photon microscopy (Figure 1A). Results showed that a single intraperitoneal (i.p.) injection of MPTP (30 mg/kg) caused a 2-fold increase in dendritic spine turnover within 4–24 h in 1-month-old mice (MPTP versus vehicle, all in %, 4 h, 9.0 ± 0.54 versus 4.4 ± 0.58; 24 h, 18.2 ± 0.95 versus 10.4 ± 0.74; p < 0.05 by t test in both cases; Figures 1B and 1C). The rates of spine formation and elimination were both higher in MPTP-treated mice than those in vehicle-treated control mice (MPTP versus vehicle, all in %, elimination: 4 h, 4.2 ± 0.36 versus 2.5 ± 0.40; 24 h, 8.0 ± 0.63 versus 4.2 ± 0.36; formation: 4 h, 5.3 ± 0.40 versus 1.9 ± 0.27; 24 h, 10.2 ± 0.57 versus 4.8 ± 0.39; p < 0.05 by t test in all comparison pairs). These results indicate that MPTP has potent effects on dendritic spine dynamics in the cortex.

Figure 1

Systemic Administration of MPTP Destabilizes Dendritic Spines and Impairs Motor Memory

(A) Upper, schematic diagram of in vivo two-photon imaging in apical dendrites of layer 5 pyramidal neurons (L5PRNs) in the motor cortex of Thy1-YFP mice. Lower, timeline of imaging and drug administration.

(B) Representative two-photon images of the same dendritic segment in time series from vehicle or MPTP-treated groups. Filled arrowheads, newly formed spines; open arrowheads, eliminated spines.

(C) From left to right, overall spine turnover, elimination, and formation rates during 0–4 h and 0–24 h. MPTP group showed significantly higher spine plasticity (two-sample student t test; 0–4 h and 0–24 h; turnover, t8 = 5.82, t8 = 6.51; elimination, t8 = 6.02, t8 = 6.99; formation, t8 = 3.15, t8 = 4.07; p < 0.05 in all cases. n = 5 per group).

(D) Experimental timeline of motor learning and two-photon microscopy.

(E) Representative images of dendritic segments before (0) and after motor learning (8 h).

(F) Spine formation rates over 0–8 h was significantly higher under learning paradigms (non-learning, n = 4; beam learning, n = 9; rotarod learning, n = 9; F (2, 19) = 7.231, p = 0.0046).

(G–I) (G) Representative dendritic segments in time series before and after motor learning under various conditions. Filled arrowheads, newly formed spine at 8 h; asterisks, filopodia. (H and I) The survival rates of newly formed spines 12 and 24 h after (H) rotarod learning (8–12 h, t7 = 5.496, p = 0.0009; 8–24 h, t7 =5.941, p = 0.0006; n = 4 in rotarod + vehicle and n = 5 in rotarod + MPTP group) or (I) or beam walking learning (8–12 h, t7 = 5.130, p = 0.0014; 8–24 h, t7 = 4.050, p = 0.0049; n = 4 in beam + vehicle and n = 5 in beam + MPTP group).

(J and K) Motor performance improvement 24 h and 4 days after (J) rotarod (24 h, t7 = 4.867, p = 0.0018; 4 days, t7 = 2.916, p = 0.0225) or (K) beam walking learning (24 h, t7 = 4.463, p = 0.0024; 4 days, t6 = 2.699, p = 0.0356).

(L and M) Pearson correlation analysis showed that new spine survival rates (0–8–24 h) were positively correlated with performance improvement at 24 h in rotarod (r = 0.78, p = 0.0016; L) and beam walking (r = 0.69, p = 0.0058; M) assays. *p < 0.05. Data were presented as mean ± SEM. Scale bar, 5 μm.

Previous studies have shown that motor learning induces the rapid formation of new spines in the motor cortex, and the persistence of these new spines associated with motor learning is critical for motor memory retention (Yang et al., 2009). We next investigated whether MPTP similarly altered the dynamics of motor learning-induced new spines (Figure 1D). We trained 1-month-old mice to run on an accelerating rotarod or perform a beam walking task, and immediately after motor skill learning, mice were treated with MPTP. Consistent with previous studies (Li et al., 2017, Xu et al., 2009, Yang et al., 2009, Yang et al., 2014), new spine formation increased significantly within 8 h after motor training when compared with untrained controls (all in %, non-learning, 4.4 ± 0.5; beam learning, 6.2 ± 0.3; rotarod learning, 6.8 ± 0.4; p < 0.01 by t test for both comparison pairs; Figures 1E and 1F). We found that among new spines formed within 8 h after motor training, fewer spines persisted over next 4 h (8–12) or 16 h (8–24) in the MPTP-treated mice when compared with vehicle-treated controls (p < 0.01 by t test for both rotarod and beam walking, Figures 1G–1I). Furthermore, motor performance improvement was lower in MPTP-treated mice 1 day or 4 days after motor training (p < 0.05 by t test for both 1 day and 4 days, Figures 1J and 1K). There was a significant correlation between the percentages of motor training-related persistent new spines and motor skill performance improvement (Pearson correlation, rotarod learning, r = 0.78; p = 0.0016; beam learning, r = 0.69; p = 0.0058; Figures 1L and 1M). Together, these results demonstrate dendritic spine instability and impaired motor memory retention by single MPTP administration.

Increased Ca2+ Spike Generation in Apical Tuft Dendrites of Pyramidal Neurons

Previous studies have correlated dendritic spine dynamics with the dendritic activity of pyramidal neurons (Cichon and Gan, 2015, Weber et al., 2016). To study cortical activity in MPTP-treated mice, we performed in vivo Ca2+ imaging in apical tuft branches of L5PRNs (Figure 2C). Wild-type (WT) mice were given intracranial injection of an adeno-associated virus (AAV) delivered vector encoding the Ca2+ indicator GCaMP6s to monitor pyramidal neuron activity (Figure S1). First, the levels of Ca2+ activity in dendritic spines exhibited a 3-fold increase in MPTP-treated mice when compared with vehicle treatment (p < 0.0001 by t test, Figures 2A and 2B). In addition to Ca2+ activity in spines, we also observed a substantial increase in dendritic Ca2+ spikes following systemic MPTP injection: the number of Ca2+ spikes in the field of view (FOV) rapidly increased within 1 h post-injection, reached the peak level at 2 h, and remained higher than the pre-injection baseline at 4 h (p < 0.0001 by two-way ANOVA; Figure 2D). Vehicle treatment had no effects on the generation of dendritic Ca2+ spikes (Figures S2A and S2B). Analyses of individual dendrites showed significantly higher Ca2+ frequency and integrated Ca2+ levels 2 h after MPTP injection relative to pre-injection (p < 0.0001 by two-way ANOVA; Figures 2E–2H, Video S1). The amplitudes of dendritic Ca2+ spikes were comparable before and after MPTP injection (Figure S2C). These MPTP-induced dendritic Ca2+ spikes were largely reduced in the presence of locally delivered N-methyl-d-aspartate (NMDA) receptor antagonist MK801 (p < 0.0001 by t test; Figures 2I and 2J). These results show that MPTP injection causes an increase of dendritic spine and branch activity in pyramidal neuron tufts.

Figure 2

Systemic Administration of MPTP Increases Ca2+ Activity in Apical Dendrites of L5PRNs

(A) Representative fluorescence images of dendritic spines in apical tufts of L5PRN expressing GCaMP6s over 50-s time series. Dotted circles, region of interest (ROI) showing Ca2+ transients in spines.

(B) MPTP injection significantly increased frequency (t98 = 9.30, p < 0.0001) and total integrated levels (t98 = 7.32, p < 0.0001) of Ca2+ transients in dendritic spines. Vehicle, n = 50 ROIs from four animals; MPTP, n = 50 ROIs from four animals.

(C) Experimental design for apical dendritic imaging after MPTP.

(D) The number of Ca2+ spikes in the same field of view (FOV) over hours after MPTP systemic administration (one-way ANOVA, F(3, 44) = 10.18, p < 0.0001; n = 12 FOVs from five mice). Gray, individual traces; blue, averaged levels.

(E) Time-series and z-stacked images showing dendritic Ca2+ spikes within 50 s from the same FOV before (pre-MPTP) or 2 h after drug infusion (post-MPTP).

(F) Representative fluorescence traces of L5PRN apical tufts expressing GCaMP6s.

(G) Distribution of Ca2+ spike frequency from individual dendritic segments during 3-min sampling windows. MPTP injection significantly increased Ca2+ frequency (t305 = 3.15, p < 0.0001; n = 61 ROIs before MPTP and n = 246 ROIs post-MPTP from five animals).

(H) Total integrated dendritic Ca2+ levels were enhanced by MPTP application (t305 = 4.74, p < 0.0001).

(I) Schematic showing MK801 local injection and dendritic Ca2+ recording in MPTP-treated mice.

(J) Local application of MK801 reduced dendritic Ca2+ spike frequency and averaged dendritic Ca2+ activity after MPTP systemic injection (n = 88, t332 = 5.54, p < 0.0001).

*p < 0.05. Data were presented as mean ± SEM. Scale bar, 10 μm. See also Figures S1 and S2.

To further understand the effects of MPTP on the cortex, we topically applied MPTP (30 μM in artifical cerebrospinal fluid, aCSF) to the superficial layer of the motor cortex and performed in vivo Ca2+ imaging on apical dendrites of L5PRNs (Figure 3A). We found that the number of Ca2+ spikes in the FOV increased markedly over 30 min after MPTP local application and returned to the baseline 2 h post-washout (150 min post-MPTP) (p < 0.05 by t test, Figures 3B–3D, Video S2). Both frequency and integrated activity of dendritic Ca2+ spikes peaked at 30 min after MPTP application and recovered to pre-injection levels 2 h post-washout (p < 0.0001 by two-way ANOVA, Figures 3E and 3F). The amplitudes of dendritic Ca2+ spikes induced by local MPTP are similar to those that occurred after systemic MPTP injection (Figure 2). These results suggest that MPTP-induced dendritic Ca2+ spikes are partly mediated by its local action on the cerebral cortex.

Figure 3

Local Application of MPTP Increases Dendritic Ca2+ Activity

(A) Experimental diagram and timeline of MPTP local application and two-photon imaging in L5PRN apical dendrites.

(B) The frequency of dendritic Ca2+ spikes rapidly increased after local application of MPTP and returned to baseline levels 150 min later (F(2, 33) = 160.4, p < 0.0001; n = 12 fields of view [FOVs] from four mice). Gray, individual traces; blue, averaged levels.

(C) Time-series and z-stacked images showing dendritic Ca2+ spikes within 50 s from the same FOV before (pre-MPTP), 30 min after (post-MPTP), and 2 h after removal of drugs (MPTP-recovery). Scale bar, 10 μm.

(D) Representative fluorescence traces of L5PRN apical dendrites expressing GCaMP6s.

(E) Distribution of dendritic Ca2+ spike frequency in individual dendrites over 3-min recording showed significantly elevated Ca2+ activity after local MPTP infusion (F(2, 247) = 23.51; p < 0.0001; n = 35, 165, and 50 regions of interest from pre-MPTP, post-MPTP, and MPTP-recovery groups, respectively, from four animals).

(F) MPTP application reversibly increased dendritic Ca2+ activity averaged over 3-min recording (F(2, 247) = 47.38, p < 0.0001) *p < 0.05. Data were presented as mean ± SEM.

Decreased Activity in Somatostatin-Expressing Inhibitory Interneurons

To understand the mechanisms underlying MPTP-induced dendritic activity in pyramidal neurons, we examined the activity of local inhibitory interneurons, focusing on dendritic-targeting SST-INs. These cells project their axons to the superficial layers of the cortex, where they primarily target pyramidal neuron dendrites and form inhibitory synapses (Urban-Ciecko and Barth, 2016). To determine the effects of MPTP on SST-IN activity, we used in vivo Ca2+ imaging to examine the activity of SST-INs expressing GCaMP6s. Here, SST-IRES-Cre mice were injected with a Cre-dependent AAV-dio-GCaMP6s into the motor cortex to induce the expression of GCaMP6s specifically in SST-INs (Figures 4A and 4B). In contrast to the increased activity of dendritic spines and branches in L5PRNs (Figure 2), we found that the activity of SST-IN somas was significantly reduced within 1 h after MPTP injection (p < 0.01 by t test, Figures 4C–4E). This reduction in SST-IN activity was also observed in their axon fibers projecting to layer 1 (p < 0.01 by t test, Figures 4F–4H), indicating that MPTP induces a marked reduction of SST-IN activities in the motor cortex.

Figure 4

MPTP Decreases SST-IN Activity in the Motor Cortex

(A) Left, experimental timeline; middle, expression of GCaMP6s in the motor cortex; right, in vivo Ca2+ imaging of SST-INs expressing GCaMP6s. Yellow arrowhead indicates activated SST-IN soma. Scale bar, 200 μm in middle panel and 50 μm in right panel.

(B) Schematic diagram showing two-photon imaging planes in the superficial layer of the motor cortex.

(C–E) (C) Representative fluorescence traces of GCaMP6s-labeled SST-IN somas before and after MPTP injection. Recording duration: 150 s. (D and E) Distribution (D) and quantitative analysis (E) of total integrated Ca2+ in SST-IN somas (t129 = 3.15, p = 0.002; n = 65 somas from pre-MPTP and n = 66 somas from post-MPTP in four mice).

(F) Representative Ca2+ traces of SST-IN axons before and after MPTP injection.

(G and H) Distribution (G) and quantitative analysis (H) of total integrated Ca2+ in SST-IN axons (t334 = 6.011, p < 0.0001; n = 182 axons from pre-MPTP and n = 154 axons from post-MPTP in four mice).

*p < 0.05. Data were presented as mean ± SEM.

Activation of SST-INs Reduces Dendritic Ca2+ Spikes in MPTP-Treated Mice

Previous studies have shown that SST-INs regulate dendritic Ca2+ spike generation in apical tuft dendrites of pyramidal neurons (Cichon et al., 2017). To test whether the decreased SST-IN activity contributes to MPTP-induced dendritic Ca2+ spikes in L5PRNs, we activated SST-INs in vivo by using the designer receptors exclusively activated by designer drugs (DREADD). In this experiment, we infected SST-INs in the motor cortex with AAV encoding Cre-dependent hM3Dq DREADD receptors in SST-IRES-Cre mice (Figure 5A). An i.p. injection of DREADD receptor ligand, clozapine-N-oxide (CNO), selectively activated hM3Dq-expressing SST-INs as reported by cFos immunostaining (Figure 5B). To test the effect of SST-IN activation on MPTP-induced dendritic activity in L5PRNs, we imaged GCaMP6s-expressing apical dendrites in mice expressing hM3Dq in SST-INs (Figures 5A and S3). Dendritic Ca2+ activities were measured before and after CNO injection in MPTP-treated mice. We found that activation of SST-INs substantially reduced the number of dendritic Ca2+ spikes within 1 h and such effects persisted for at least 4 h (p < 0.0001 by two-way ANOVA, Figures 5C–5E, Video S3). Analyses of individual dendrites revealed a reduction of both Ca2+ spikes' frequency and total integrated Ca2+ levels after the activation of SST-INs (p < 0.001 by t test, Figures 5F and 5G). Comparing among all the treatment groups, activation of SST-INs reduced the number of dendritic Ca2+ spikes in MPTP-treated mice to levels that are comparable with the pre-MPTP group (Figure S4). Taken together, these findings indicate that activation of SST-INs in the motor cortex is sufficient to suppress pyramidal neuron dendritic hyperexcitability induced by MPTP.

Figure 5

Chemogenetic Activation of SST-INs Reduces MPTP-Induced Pyramidal Neuron Dendritic Activity

(A) Experimental timeline. In brief, AAV-hsyn-GCaMP6s and AAV-dio-hM3Dq-mCherry viruses were injected into layer 5 and layer 2 or 3 of the motor cortex of SST-IRES-Cre mice, respectively. MPTP and CNO were sequentially applied 18 days after viral infection, along with two-photon imaging.

(B) Immunofluorescence images of the motor cortex (upper left and lower panels) showing cFos-positive neurons among mCherry-transfected cells. CNO injection remarkably increased the co-labeling ratio (t8 = 56.67, p < 0.0001; n = 5 animals per group; upper right panel). Filled arrowheads, cFos-positive neurons. Scale bar, 500 μm in top panel and 50 μm in lower panels.

(C) The frequency of L5PRN dendritic Ca2+ spikes in the same field of view (FOV) increased after MPTP administration and decreased after CNO injection (F (3, 48) = 17.98, p < 0.0001; n = 13 FOVs from five mice).

(D and E) (D) Time-series and z-stacked images showing dendritic Ca2+ spikes over 50 s in the same FOV before (MPTP + pre-CNO) and after CNO injection (MPTP + post-CNO). Scale bar, 10 μm. (E) Representative fluorescence traces of dendritic Ca2+ spikes in L5PRNs.

(F) Distribution of Ca2+ spike frequencies in individual dendrites over 3 min before and after MPTP and CNO injection. CNO injection reduced the frequency of dendritic Ca2+ spikes (t214 = 3.77, p = 0.0002; n = 164 regions of interest from MPTP + pre-CNO and n = 52 from MPTP + post-CNO in four mice).

(G) Ca2+ activity in apical dendrites averaged over 3-min recording before and after CNO injection (t214 = 5.84, p < 0.0001).

*p < 0.05. Data were presented as mean ± SEM. See also Figures S3 and S4.

Activation of SST-INs Alleviates Dendritic Spine Instability and Motor Memory Deficits

We next investigated whether activation of SST-INs could rescue dendritic spine loss and motor memory deficits induced by acute MPTP infection. We crossed SST-IRES-Cre mice with Thy1-YFP mice and bilaterally injected Cre-dependent AAV-dio-hM3Dq-mCherry into the primary motor cortex (Figure 6A). Mice were trained on the accelerating rotarod motor learning task and administered MPTP and CNO 8 h after training. The cFos immunofluorescent staining showed that a single injection of CNO activated more than 75% of hM3Dq-expressing cells (Figure 6B). When dendritic spine plasticity was examined, we found that CNO treatment significantly increased the survival rates of learning-induced new spines in the motor cortex (p < 0.01 by t test for both 8–12 h and 8–24 h, Figures 6C and 6D). Furthermore, CNO-treated mice showed higher performance improvement in the rotarod task 24 h or 4 days after initial training (p < 0.01 by t test, Figure 6E). Notably, the improvement of rotarod performance positively correlated with the survival rate of training-related persistent new spines (p = 0.0042; r = 0.77, Figure 6F), suggesting that activation of SST-INs can rescue spine instability and motor memory deficits induced by acute MPTP infusion.

Figure 6

Activation of SST-INs Alleviates MPTP-Induced Dendritic Spine Loss and Motor Memory Deficits

(A) Experimental timeline.

(B) Immunostaining of the motor cortex (upper left and lower panels) and quantification analysis showing prominent activation of mCherry-positive SST-INs after CNO injection (t6 = 55.98, p < 0.0001; n = 4 per group). Filled arrowheads, cFos-positive neurons. Scale bar, 500 μm in top panel and 50 μm in lower panels.

(C) Representative dendritic segments from L5PRN apical tufts before rotarod skill learning (0), after learning and MPTP and/or CNO injection (8 h), and afterward (12–24 h). Filled arrowheads, newly formed spines at 8 h; asterisks, filopodia. Scale bar, 5 μm.

(D) CNO injection enhanced the survival rates of newly formed spines within 8- to 12-h and 8- to 24-h time windows (8–12 h, t6 = 6.35, p = 0.0007; 8–24 h, t6 = 4.378, p = 0.0047; n = 4 per group).

(E) Performance improvement (fold increase normalized to that in the learning session at 0 h) was elevated at 24 h and 4 days after initial learnings (24 h: t6 = 4.099, p = 0.0064; 4 days: t6 = 5.493, p = 0.0015).

(F) Pearson correlation analysis showed positive correlation between new spine survival rates (0–8–24 h) and rotarod performance improvements at 24 h (r = 0.77; p = 0.0042).

*p < 0.05. Data were presented as mean ± SEM.

Discussion

In this study, we investigated the effects of acute MPTP treatment on synaptic plasticity and motor memory functions. Using two-photon microscopy, we found that a single injection of MPTP in adolescent mice destabilized dendritic spines on apical dendrites of L5PRNs in the motor cortex, which is associated with an increase of pyramidal neuron dendritic activity and a decrease of SST-IN activity. Notably, activation of SST-INs in the motor cortex following MPTP administration rescued pyramidal cell dendritic and synaptic deficits and preserved motor skill learning memory. Together, our results show that SST-IN dysfunction in the motor cortex plays an important role in synaptic and learning deficits induced by acute MPTP administration. As one parkinsonism-inducing agent, chronic MPTP exposure causes dopamine neuron degeneration, leading to learning and behavioral abnormalities (Guo et al., 2015, Halliday et al., 2014). The current study reported that a single dose of MPTP (30 mg/kg) led to impaired retention of motor skills that were acquired before the MPTP exposure, in association with the loss of learning-related new synapse structure, hyperactivity of pyramidal neuron dendrites, and suppression of SST-IN activity. These results indicate that at early stage of PD, motor cortex dysfunction may form one pathological feature and that targeting SST-INs has the potency to retard the progression of motor syndromes.

Structural and functional synaptic plasticity in the primary motor cortex has been shown to be critical for motor skill learning and retention (Cichon and Gan, 2015, Xu et al., 2009, Yang et al., 2009, Yang et al., 2014). Previous studies reported that rotarod learning over 8 h in 1-month-old mice causes a 2%–4% increase in new spine formation in the motor cortex (Li et al., 2017, Yang et al., 2014). Learning-induced new spines persist over days to weeks, and the survival of these new spines strongly correlates with the animals' performance on motor skill tasks. After acute MPTP administration, we observed a marked increase in dendritic spine dynamics and a substantial loss of learning-induced new spines. This rapid loss of dendritic spines after MPTP exposure may contribute to the impairments of motor skill retention in these mice. To capture the functional plasticity of cortical spines, we employed in vivo Ca2+ imaging and observed a 3-fold increase in Ca2+ activity in dendritic spines after MPTP injection. Previous studies in different neurological diseases, such as seizure (Segal et al., 2000), AD (Bai et al., 2017), and neuropathic pain (Cichon et al., 2017), have linked overload of Ca2+ to the shrinkage or elimination of dendritic spines. MPTP-induced loss of dendritic spines, therefore, could be attributed to the increased Ca2+ transients in pyramidal neuron dendrites and spines. Moreover, we found that MPTP-induced dendritic Ca2+ spikes are NMDA receptor dependent, as blockade of NMDA receptor with MK801 normalized dendritic activity in cortical pyramidal neurons. Together, these results suggest that cortical pyramidal neuron dendritic hyperactivity may contribute to MPTP-induced synapse loss and motor memory impairments.

The activities of pyramidal neurons in the cortex are regulated by a great variety of local inhibitory interneurons. Among them, SST-INs constitute ∼30% of all cortical interneurons (Urban-Ciecko and Barth, 2016). Although SST-INs themselves are a heterogeneous population and target different cells in different layers, they mainly innervate the distal portion of apical dendrites (Hattori et al., 2017, Muñoz et al., 2017, Urban-Ciecko and Barth, 2016). It has been shown that SST-INs regulate dendritic spine activity and dendritic Ca2+ spike generation (Cichon et al., 2017, Cichon and Gan, 2015). Consistent with the increase of dendritic activity in our acute MPTP administration model, we observed a rapid and substantial decrease of SST-IN activity within 60 min after MPTP infusion, which is in favor of dendritic hyperactivity. Interestingly, measures from postmortem brain of patients with PD-related dementia showed reduced SST immunoreactivity in the cortex (Allen et al., 1985, Beal et al., 1986, Epelbaum et al., 1983), and decreased expression of SST in pluripotent stem cells derived from patients with Parkin mutations (Iwasawa et al., 2019), supporting the involvement of SST-INs in PD pathogenesis.

Given the reduction of SST-IN activity after MPTP administration, we manipulated the activity of SST-INs using chemogenetic approach and showed that SST-IN activation in the motor cortex prevented MPTP-induced increase in dendritic Ca2+ spike generation (Figure S4), dendritic spine loss, and motor memory impairments. Activation of GABAergic inhibitory interneurons in cortical circuits has been shown to be beneficial in treating multiple neurological disorders with the characteristic of cortical hyperexcitability, including major depressive disorder (Fee et al., 2017, Lin and Sibille, 2015), neuropathic pain (Cichon et al., 2017), AD (Schmid et al., 2016, Verret et al., 2012), and PD (Lindenbach et al., 2016, Tyagi et al., 2015, Zhang et al., 2017). It is worth to note that besides SST-INs, there are other types of interneurons located in the superficial layer of the cerebral cortex (Murphy et al., 2016). These cells are known to target apical tuft dendrites, and their activation can suppress Ca2+ spike generation in L5PRNs.

To explore the molecular mechanisms underlying MPTP-induced memory deficits, we examined the amounts of various proteins in the cortex after MPTP treatment, focusing on those involved in synaptic plasticity and function (Lee et al., 2009, Yasuda, 2017). Previous studies have shown that activation of Ca2+/calmodulin-dependent kinase II (CaMKII) (Adler et al.; Lee et al., 2009, Lisman et al., 2012) and small GTPase proteins (Hedrick et al., 2016) facilitates the long-term potentiation of dendritic spines in hippocampal slices. Using western blot analysis, we found that MPTP injection had no significant effects on the total amounts of CaMKII in the brain but caused a marked reduction in the level of CaMKII phosphorylated at Thr-286 (p-CaMKII) when compared with vehicle-treated controls at 4–12 h (Figure S5). These results indicate that MPTP treatment acutely impairs CaMKII autophosphorylation in the cortex. Importantly, CNO treatment of SST-IRES-Cre mice infected with hM3Dq exhibited normal levels of p-CaMKII 4 h after MPTP injection, whereas the levels of cofilin, Ras1, RhoA, cdc42, and Rac1 remained unaltered (Figure S5). CaMKII autophosphorylation is an important regulator of synaptic plasticity (Chang et al., 2017, Lee et al., 2009, Okamoto et al., 2007), and its impairments have been observed in mouse hippocampus after acute or chronic MPTP infusion (Zhu et al., 2015). More importantly, a recent study suggested that pharmacological inhibition of CaMKII phosphorylation can disrupt motor learning (Adler et al., 2019). The fact that SST-IN activation also corrects CaMKII autophosphorylation further suggests that manipulating SST-IN activity could be an important approach for preventing cortical hyperexcitability and motor memory impairments.

In sum, our current work demonstrates rapidly hyperactivated dendritic spines in the motor cortex after acute MPTP infusion, in association with spine loss plus motor dysfunctions. The activation of SST-INs helps to relieve these pathological phenotypes to recover motor learning function. Future studies will be needed to examine whether other types of interneurons could also be targeted to protect synaptic loss and motor learning deficits in MPTP-treated mice.

Limitations of the Study

Our results demonstrate that potentiation of cortical SST-INs could relieve dendritic hyperactivity, spine loss, and motor learning deficits after acute MPTP injection. The current study, however, does not address if similar strategy can be used under chronic MPTP administration, which is one typical model of PD. Moreover, the long-term ameliorating effect from chemogenetics activation approach has not been evaluated. Further studies are thus required to address those issues to get a more complete picture of inhibitory transmission in motor cortical dysfunction.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.