Abstract
Autism is characterized by two key diagnostic criteria including social deficits and repetitive behaviors. Although recent studies implicated ventral striatum in social deficits and dorsal striatum in repetitive behaviors, here we revealed coexisting and opposite morphologic and functional alterations in the dorsostriatal direct and indirect pathways, and such alterations in these two pathways were found to be responsible, respectively, for the two abovementioned different autism-like behaviors exhibited by male mice prenatally exposed to valproate. The alteration in direct pathway was characterized by a potentiated state of basal activity, with impairment in transient responsiveness of D1-MSNs during social exploration. Concurrent alteration in indirect pathway was a depressed state of basal activity, with enhancement in transient responsiveness of D2-MSNs during repetitive behaviors. A causal relationship linking such differential alterations in these two pathways to the coexistence of these two autism-like behaviors was demonstrated by the cell type-specific correction of abnormal basal activity in the D1-MSNs and D2-MSNs of valproate-exposed mice. The findings support those differential alterations in two striatal pathways mediate the two coexisting autism-like behavioral abnormalities, respectively. This result will help in developing therapeutic options targeting these circuit alterations.
SIGNIFICANCE STATEMENT Autism is characterized by two key diagnostic criteria including social deficits and repetitive behaviors. Although a number of recent studies have implicated ventral striatum in social deficits and dorsal striatum in repetitive behaviors, but social behaviors need to be processed by a series of actions, and repetitive behaviors, especially the high-order repetitive behaviors such as restrictive interests, have its scope to cognitive and emotional domains. The current study, for the first time, revealed that prenatal valproate exposure induced coexisting and differential alterations in the dorsomedial striatal direct and indirect pathways, and that these alterations mediate the two coexisting autism-like behavioral abnormalities, respectively. This result will help in developing therapeutic options targeting these circuit alterations to address the behavioral abnormalities.
Introduction
The interaction of multiple genetic and environmental factors contributes to the etiology of autism spectrum disorders (ASDs), which are characterized by two key diagnostic criteria including impaired social interactions, restrictive interests, and repetitive behaviors (American Psychiatric Association, 2013; Lai et al., 2014). Unfortunately, the pathophysiology in brain circuits underlying these behavioral symptoms remains unclear. Given the coexistence of social communication and motor control abnormalities in ASD and the surprising degree of overlap in brain regions involved in the control of social behavior and repetitive actions (Bariselli et al., 2016; Fuccillo, 2016), whether two separate pathophysiological alterations in the neural circuit level correspondingly produce these two abnormalities is unknown.
The striatum receives excitatory glutamate inputs from the cortex and thalamus and sends outputs that ultimately relay information back to the cortex via the thalamus. Through the dichotomous function of the direct and indirect pathways, the striatum involves in regulation of attention, behavioral flexibility, motivational state, reward processing and goal-directed behaviors (Kravitz et al., 2010; Calabresi et al., 2014; Nelson and Kreitzer, 2014; Cox and Witten, 2019; Smith et al., 2021). Evidence from clinical studies has also suggested that there is a pattern of core striatal dysfunction in ASD patients (Turner et al., 2006; Langen et al., 2007; Takarae et al., 2007). Recently, the findings from animal studies have suggested the striatum to be a common node for autism pathophysiology (Graybiel, 2008; Gunaydin et al., 2014; Portmann et al., 2014; Rothwell et al., 2014; Fuccillo, 2016; Wang et al., 2017; Schiavi et al., 2019; Folkes et al., 2020; Solié et al., 2022), but the exact striatal mechanisms mediating the two autism-like core symptoms remains unclear. The subregional hypothesis suggested that social behaviors are mostly linked to ventral striatum since the nucleus accumbens (NAc) mediate reward processing and goal-directed behaviors, while stereotyped and repetitive behaviors are likely linked to dorsal striatum which involves in motor control (Gunaydin et al., 2014; Fuccillo et al., 2016; Wang et al., 2017; Solié et al., 2022). Although the two core symptoms are apparently different and seemingly unrelated, but social behaviors need to be processed by a series of actions, and repetitive behaviors, especially the high-order repetitive behaviors such as restrictive interests, have its scope to cognitive and emotional domains (Langen et al., 2011). There is thereby a probability to rise another hypothesis that both ventral and dorsal striatum may involve in both of the two symptoms and imbalance of striatal direct and indirect pathways may play a critical role in expression of the two behavioral abnormalities.
The vast majority of striatal neurons are medium-sized spiny GABAergic projection neurons (MSNs), which can be divided into two types: direct pathway MSNs expressing D1 dopamine receptors (D1-MSNs) and indirect pathway MSNs expressing D2 dopamine receptors (D2-MSNs). Recent studies using genetic ASD animal models have reported that disrupted homeostatic alterations in glutamatergic transmission in direct or indirect striatal pathways are linked to the expression of repetitive behaviors (Rothwell et al., 2014; Wang et al., 2017). Considering that the expression of social behaviors is also mediated by direct pathway neurons (Gunaydin et al., 2014; Folkes et al., 2020), the abnormal activity of D1-MSNs may be involved in the social deficits exhibited by ASD individuals. We thus hypothesize that both direct and indirect pathway alterations may be associated with the coexistence of autism-like behaviors, including social deficits and repetitive behaviors.
Prenatal valproate (valproic acid; VPA) exposure affects global gene expression during key developmental stages and can replicate a relatively full spectrum of symptomatology of ASD (Mabunga et al., 2015; Nicolini and Fahnestock, 2018; R. Zhang et al., 2018). In the current study, we found that prenatal VPA exposure induced differential morphologic and functional alterations in D1-MSNs and D2-MSNs in the dorsomedial striatum (DMS), and that these alterations mediate the two coexisting autism-like behavioral abnormalities, respectively.
Materials and Methods
Animals
All procedures were approved by the Institutional Animal Care and Use Committee of Shaanxi Normal University and conformed to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals. Animals were kept with a standard light/dark circle (lights from 8 A.M. to 8 P.M.), constant temperature (22 ± 2°C), relative humidity (55 ± 5%), food and water were provided ad libitum. The D1-Cre (Drd1-Cre, 262, Gensat), D2-Cre (Drd2-Cre, ER44, Gensat), and Drd2-eGFP transgenic mice were described previously (Z. Zhang et al., 2017), Drd1-eGFP knock-in mice (C57BL/6J) were from Biocytogen (LC-060). All mouse lines had been crossed with the C57BL/6J background for more than ten generations. Given that severe autism-like behavior deficits were mainly observed in male offspring after VPA exposure (Schneider et al., 2008), and the male inclination to autism in humans, only males were used in the current study. All experiments were performed blindly to the experimenter and within each litter the mice were randomly assigned to different experiments.
VPA-induced models of autism
VPA (Sigma) was dissolved in 0.9% saline at a concentration of 250 mg/ml. Pregnant mice were treated with 500 mg/kg VPA or vehicle saline as vehicle control by a single intraperitoneal injection on embryonic day (E)12.5. Only male offspring were used in the current study.
Viral vectors
AAV-hSyn-Cre (PT-0136), AAV-hSyn-DIO-mCherry (PT-0115), AAV-hSyn-DIO-hM3D(Gq)-mCherry (PT-0019), AAV-hSyn-DIO-hM4D(Gi)-mCherry (PT-0020), AAV-Ef1α-DIO-GCaMP6m (PT-0283), AAV-Ef1α-DIO-EYFP (PT-0012) were purchased from BrainVTA. AAV-Ef1α-DIO-eNpHR3.0-EYFP (AG26966), AAV-Ef1α-DIO-hChR2(H134R)-EYFP (AG20298), and AAV-Ef1α-DIO-EYFP (AG20296) were purchased from Obio Technology. All vectors were serotyped with AAV2/9, and the viral titer was (2–7) × 1012 particles per ml. All viral vectors were aliquoted and stored at −80°C until used.
Stereotaxic surgery
Surgeries were performed under anesthesia using isoflurane (4% for induction and 1% for maintenance). Each animal was mounted in a stereotaxic frame with nonpuncturing ear bars (RWD Life Science Inc). All skull measurements were made relative to bregma according to the mouse atlas. The stereotaxic coordinates for DMS injection were anteroposterior (AP) 0.7, mediolateral (ML) ±1.5 and dorsoventral (DV) 2.8 mm. Viral injection into the DMS was performed at a rate of 40 nl/min by using borosilicate glass pipettes connected with a 10 µl microsyringe (Hamilton), and the needle was held at least for 10 min at the site to allow diffusion of the virus. Only mice with the correct locations of viral expression and optical fibers were used for further analysis.
Sparse labeling and morphologic analyses
Sparse labeling was achieved by a two-virus system developed by BrainVTA, which expresses Cre recombinase with a highly diluted AAVs while simultaneously delivers a Cre-dependent reporter virus with high-titer. Thus, only few neurons infected with Cre-expressing virus could be brightly-labeled with many copies of the reporter gene which delivered by Cre-dependent virus. To sparsely label D1-MSNs or D2-MSNs, the Cre-expressing virus AAV-hSyn-Cre was diluted 1:30,000 in PBS and combined at a 1:1 ratio with the Cre-dependent virus AAV-hSyn-DIO-mCherry for injections into the DMS of D2-eGFP transgenic mice. A total volume of 200 nl mixed AAV virus was bilaterally injected into DMS. After three weeks, mice were perfused with 4% paraformaldehyde (PFA) and 200-µm-thick brain slices were prepared by using Vibratome (VT 1200S, Leica). The type of D1-MSNs or D2-MSNs was identified according to fluorescence color of soma. D1-MSNs were identified by expressing red fluorescent protein mCherry, while D2-MSNs were co-expressed red fluorescent protein mCherry and green fluorescent protein eGFP.
Confocal microscope (CR-DFLY-202–2540, Oxford Instruments) was used to image MSN cells whose neuronal dendritic trees and branching were intact and not obscured by dendrites of other neurons. To acquire dendritic trees, neurons near the center of the slice were imaged at 1-µm intervals under a 40 × microscope with the total thickness around 100 µm. Then, neurons were traced by Imaris software (version 9.4.2) for 3D reconstruction and measurements. Stack images (2048 × 2048 pixels) of spines were acquired using a 60 × oil microscope. Sholl analyses were performed by ImageJ software (NIH) with the plugin Sholl analysis.
Designer receptor that would be exclusively activated by designer drugs (DREADD) manipulation
For DREADD rescue experiment, AAV-hSyn-DIO-hM4D(Gi)-mCherry or AAV-hSyn-DIO-hM3D(Gq)-mCherry were injected into the DMS of VPA D1-Cre/D2-Cre mice, respectively. To induce autism-like behaviors by DREADD, AAV-hSyn-DIO-hM3D(Gq)-mCherry or AAV-hSyn-DIO-hM4D(Gi)-mCherry were bilaterally injected into the DMS of control D1-Cre/D2-Cre mice, respectively. AAV-hSyn-DIO-mCherry worked as control virus in all DREADD experiments. After two to three weeks of recovery, behavioral tests were performed at 30 min after intraperitoneal injection of clozapine-N-oxide (CNO) in viral-infected mice.
In vivo optogenetic manipulation
For the optogenetic rescue experiment, AAV-Ef1α-DIO-eNpHR3.0-EYFP, or AAV-Ef1α-DIO-hChR2(H134R)-EYFP was bilaterally delivered into DMS of VPA D1-Cre/D2-Cre mice, respectively. To induce autism-like behaviors by the optogenetic modulation, AAV-Ef1α-DIO-hChR2(H134R)-EYFP or AAV-Ef1α-DIO-eNpHR3.0-EYFP were bilaterally delivered into DMS of control D1-Cre/D2-Cre mice, respectively. AAV-Ef1α-DIO-EYFP worked as control virus. Optical fibers [1.25-mm diameter, 230-µm optical density (OD), 0.37 numerical aperture (NA), ThinkerTech] were implanted into the DMS (AP: 0.7; ML ±1.5; DV: −2.7).
For optical activation of ChR2, the optical fiber was connected to a 473-nm laser diode (Newdoon) via a FC/PC adaptor and a fiber optic rotary joint, blue light pulses were delivered to the DMS using 10% pulse width, 10 Hz, and 2 mW at the tip of the fiber. For activation of eNpHR3.0, the optical fiber was connected to a 596-nm laser diode (Newdoon), yellow light was delivered continuously at 8 mW. Mice received cycles of 8-s light on and 2-s light off. Optogenetic behavioral protocol was 15-min test containing three 5-min epochs (laser OFF-ON-OFF epochs) and mice were allowed to recover 1–3 min following connected to the patch cable.
Fiber photometry
Fiber photometry was used to record the population calcium signals of neurons in awake mice using fiber photometry recording system (ThinkerTech). To express genetically encoded calcium indicators, 200-nl AAV-Ef1α-DIO-GCaMP6m or AAV-Ef1α-DIO-EYFP was infused unilaterally into the DMS. Following virus injection, an optical fiber (200-µm core diameter, 0.37 NA; Shanghai Fiblaser) was implanted into the DMS (AP: 0.7; ML: −1.5; DV: −2.7). After two to three weeks of recovery, behavioral experiments were performed, and the placements of implantable optical fibers were confirmed after behavioral tests in all animals. The laser intensity was measured at the tip of optical fiber and adjusted to 30–40 µW to decrease laser bleaching. The signal was collected and digitalized at 50 Hz by CamFiberPhotometry software (ThinkerTech). The fluorescence change (ΔF/F) values were calculated as (F–F0)/F0, where F0 is the baseline fluorescence signals. To estimate the calcium response magnitudes, we calculated the area under the peri-event plot curve (AUC) during the 2-s time window following the beginning of behavioral bouts. Baseline recordings and analysis were performed according to Calipari's methods with minor modification (Calipari et al., 2016). All baseline MSN recordings were performed in D1-Cre/D2-Cre mice at animal's home cage for 2.5 min. Peak analysis were used to determine frequency and amplitude of the signals by the median average deviation (MAD) of the data set.
Behavioral tests
Open field test (OFT)
Mice were gently placed in the center of a square box (50 × 50 × 50, L × W × H in cm). Moving trajectory was automatically recorded for 10 min with a video camera positioned above the box and analyzed with the EthoVision XT software (version 1.9, Noldus Information Technology). The locomotor activity was evaluated by using the total traveled distance.
Three-chamber sociability test
The social interaction ability was evaluated by three-chamber test which was conducted in an acrylic plastic box with three communicating compartments (60 × 40 × 40 cm3). At the beginning of the sociability test, the tested mouse was introduced in the center of the middle chamber for 5-min habituation. Then, an unknown normal male mouse was putting in one of the side cages, the sociability test was performed for 10 min by putting the subject mouse into the middle chamber.
Home-cage interaction
Mice were allowed to habituate the home-cage for 10 min, then an unfamiliar juvenile male mouse (three to four weeks old) was introduced into the cage to explore freely for 10 min. We measured social interaction behaviors such as body sniffing, anogenital sniffing, following and direct contact initiated by the test mouse. All behaviors were video-recorded and analyzed using EthoVision software (version 1.9, Noldus Information Technology).
Self-grooming test
Mice were performed self-grooming test to estimate the repetitive behaviors. Each mouse was placed in a clean empty plastic cage (30 × 20 × 15 cm3) and habituated for 10 min. Bedding was not used to prevent digging behavior. Then, the subject mouse was recorded for cumulative time spent on self-grooming during the 10-min testing period. Grooming behavior was defined as rubbing the face, body, or head with the two forelimbs.
Marble burying task
Marble burying task was performed to assess the repetitive stereotyped behaviors. Plastic cages were filled with sawdust in depth of 5 cm, and 12 marbles were evenly spaced on the sawdust in a 3 × 4 grid. The time of burying marbles was recorded for 20 min. Numbers of marbles buried (approximately >75% of the size of the marble covered by bedding material) was assessed after every testing session.
Electrophysiological recording
Acute brain slices for electrophysiological recording were performed as previously described (Huang and Uusisaari, 2013; Ting et al., 2014). Briefly, the mice were anesthetized with isoflurane and decapitated. The brains were immersed in cold artificial CSF (ACSF) containing (in mm): 125 NaCl, 2.5 KCl, 25 glucose, 25 NaHCO3, 1.25 NaH2PO4, 2 CaCl2, and 1 MgCl2, gassed with 5% CO2/95% O2. Sagittal slices (300 µm) of the striatum were cut with a microslicer (VT 1200S, Leica). Slices were then transferred to a holding chamber filled with oxygenated ACSF at 34°C and allowed to recover for at least 0.5 h before use.
Individual slices were transferred to a recording chamber and were continuously perfused with oxygenated ACSF for the duration of the experiment. Whole-cell recordings of MSNs were obtained using a Multiclamp 700B amplifier and a Digidata 1550 (Molecular Devices), and the data were collected and analyzed using pClamp 10.5 software (Molecular Devices). The recording pipettes had 3- to 5-MΩ resistance when filled with intracellular solution containing (in mm): 140 CsCH3SO3, 10 HEPES, 2 QX-314, 2 MgCl2, 0.2 EGTA, 4 MgATP, 0.3 Na2GTP, and 10 Na2-phosphocreatine (pH 7.2–7.4 with CsOH). Studies were performed 5 min after membrane breaking for balancing the ingredients of internal solution and intracellular fluid. All experiments were conducted in the presence of 100 µm picrotoxin (PTX; Sigma). Series and input resistances were monitored for stability throughout each experiment and data were abandoned if the resistance changed by >20%. Data were filtered at 2 kHz, digitized at 10 kHz. After recording in wild-type mice, the cytoplasmic contents of the recorded cells were harvested into the patch pipette for identifying cell types (D1-MSNs and D2-MSNs) by single-cell PCR technique as we previously described (Diao et al., 2021). The ACSF, sucked by patch pipette positioned close to the tissue in the recording chamber, was employed as negative control.
Evoked EPSCs (eEPSCs) were induced with a stainless-steel bipolar microelectrode located on the white matter between the cortex and the striatum, which was closing to the recording electrode and stimulated at a baseline frequency of 0.1 Hz. The neurons were voltage-clamped at −80 mV. The paired-pulse ratio (PPR) was calculated by determining the ratio of the peak amplitude values of eEPSC2/eEPSC1 induced by the pairs of stimuli (at 50-ms interpulse interval). To measure the input–output response of NMDAR-mediated EPSCs, the slices were perfused with Mg2+-free ACSF, and the cells were held at −80 mV containing 20 µm CNQX and 100 µm PTX. The stimulus intensity was set as a step from 0.1 to 1 mA. To detect the effects of quinpirole on eEPSCs, LTD was induced by high-frequency stimulation (HFS; 1-s duration, 100-Hz frequency, repeated four times at 20-s intervals). D2R agonist (-)-quinpirole hydrochloride (10 µm, Sigma) was applied in the bath for at least 20 min starting from 10 min before HFS, and then washed out. The effect of quinpirole on synaptic transmission was evaluated by comparing the averaged EPSC magnitudes of 10-min stable recording before HFS with that after HFS (averaged EPSC values from 30 to 40 min). Miniature EPSCs (mEPSCs) were collected in the presence of 1 µm tetrodotoxin (TTX; Fishery Science and Technology Development Co). Cumulative probability and mean mEPSC amplitude/frequency were analyzed. The AMPAR-mediated and NMDAR-mediated currents were recorded at holding membrane potentials of −80 and +40 mV, respectively. The AMPAR/NMDAR current ratio was calculated as the ratio of the EPSC peak amplitude recorded at −80 mV to the amplitude of EPSCs recorded at +40 mV (40 ms after the current peak). To further confirm the AMPAR/NMDAR ratio (A/N ratio) by pharmacologically isolation methods, cells were held at +40 mV, AMPAR-mediated currents were isolated with the selective NMDAR antagonist DL-AP5 (50 µm, Sigma). The NMDAR-mediated current was then digitally obtained by taking the difference current before and after DL-AP5 application (Ren et al., 2016). All EPSCs used for analysis were averaged from 10 consecutive traces.
The current-clamp recordings were performed to assess neuronal action potential (AP) and passive membrane properties in response to the injection of depolarizing current pulses (Dorris et al., 2015). The internal solution contained the following (mm): 120 potassium gluconate, 20 KCl, 2 MgCl2, 0.2 EGTA, 10 HEPES, 4 Na2ATP, and 0.3 GTP (pH 7.2–7.4 with KOH). Under the current-clamping mode, MSNs were injected with 1000 ms step-current pulses (ranging from −100 to 250 pA, at 25-pA increments, with an intertrial interval of 10 s) in the presence of 20 µm CNQX and 50 µm AP-V. To confirm the efficiency of DREADD expression, the effects of DREADD agonist CNO on APs under similar step-current pulse injections were observed. Neurons were allowed to recover from the first step-pulse procedure for 5 min before perfused with ACSF containing 10 µm CNO. Then the second step-pulse procedure was performed 15 min after CNO perfusion. For validation of ChR2-mediated stimulation and NpHR3.0-mediated inhibition experiments, APs were evoked by a 473-nm blue light pulses with frequency at 10 Hz in ChR2-expressing DMS neurons. Light-evoked hyperpolarization was recorded in the NpHR3.0-expressing DMS neurons by a 596-nm yellow light.
Experimental design and statistical analyses
All the data are presented as mean ± SEM. The normality and the homogeneity of variance were examined by the Shapiro–Wilk and Levene's test, respectively. Data that met these two conditions were analyzed using a two-tailed unpaired t test or two-way repeated-measures ANOVA analysis, followed by Bonferroni multiple comparisons test. Data that were not normally distributed were analyzed with a nonparametric test. A p value of <0.05 was considered significant and all data used the confidence level of 95%. Details of particular statistical analyses can be found in Extended Data Table 1-1. All data analyses were performed using Prism 8.0 (GraphPad Software) and the SPSS program (SPSS).
Results
Prenatal VPA exposure generates opposing effects on glutamatergic synaptic transmission and intrinsic excitability in DMS D1-MSNs and D2-MSNs, respectively
Prenatal exposure to VPA triggered autism-like behaviors, including social deficits (Fig. 1A–D) and repetitive and stereotypical behaviors (Fig. 1E–G), as well as anxiety (Fig. 1H–L) in mice, as previously reported (Nicolini and Fahnestock, 2018; R. Zhang et al., 2018). To evaluate the effects of VPA exposure on excitatory synaptic transmission to D1-MSNs and D2-MSNs in the DMS, we performed whole-cell voltage clamp recordings of mEPSCs on MSNs in D1-eGFP and D2-eGFP mice in the presence of TTX and PTX. We found that the frequency of mEPSCs in D1-MSNs in VPA offspring was increased and that the cumulative interevent interval distribution curve exhibited a significant left shift (Fig. 2A,B). In contrast, the opposite alterations were observed (decreased or right shifted) in D2-MSNs (Fig. 2D,E), suggesting that the probability of presynaptic glutamate release was increased in D1-MSNs but reduced in D2-MSNs of VPA mice. The amplitude of mEPSCs was increased only in D1-MSNs in VPA offspring (Fig. 2A,C) but not in D2-MSNs (Fig. 2D,F), which indicates that prenatal VPA exposure also enhanced the postsynaptic currents in D1-MSNs. Similar mEPSC results were also confirmed in C57BL/6J mice, with the two types of MSNs identified by single-cell PCR (Extended Data Table 1-1). These results indicate that prenatal VPA exposure leads to bidirectional glutamatergic synaptic transmission alterations in D1-MSNs and D2-MSNs.
VPA mice exhibit core autistic-like behaviors including social interaction deficits and repetitive behaviors. A, Representative traces during three-chamber test in Control and VPA mice. B, C, Control mice spent more time in stranger compartment (B), and displayed higher social index (C), but these were not detected in VPA mice. D, Behavioral paradigm (left), VPA mice displayed less social time in home-cage social interaction test. E, Behavioral paradigm (left). Summary data showing that VPA mice displayed more grooming time. F, Representative photograph after marble burying test. G, The numbers of marbles buried were increased in VPA mice compared with Controls. H, Representative traces during OFT. I, Summary data showing no significant changes in locomotor activities of VPA mice. J–L, Summary data showing that time duration in center area during OFT (J), time spent in open arm (K), and the percentage of open arm time to total time (L) in EPM were decreased in VPA mice compared with Control mice (Control: n = 11 mice; VPA: n = 9 mice, two-way ANOVA with Bonferroni post hoc for B, two-tailed unpaired t test for C–E, G, I–L). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; *p < 0.05, **p < 0.01, ***p < 0.001.
Prenatal VPA exposure generates opposing effects on glutamatergic synaptic transmission and intrinsic excitability in DMS D1-MSNs and D2-MSNs, respectively. A, D, Representative recordings of mEPSCs from D1-MSNs (A) and D2-MSNs (D) in the Control and VPA groups. B, E, Summary data for mEPSC frequency with cumulative probability plots of interevent intervals in D1-MSNs (B) and D2-MSNs (E). C, F, Summary data for mEPSC amplitude with cumulative probability plots in D1-MSNs (C) and D2-MSNs (F; D1-MSNs: Control, n = 15 cells from eight mice; VPA, n = 18 cells from 6 mice. D2-MSNs: Control, n = 15 cells from seven mice; VPA, n = 17 cells from 5 mice, Mann–Whitney U test for B, two-tailed unpaired t test for C, E, F). G, K, Representative recording traces of PPR of D1-MSNs (G) and D2-MSNs (K) from the Control and VPA groups. H, L, Summary data for PPR recording at 50-ms time interval in D1-MSNs (H) and D2-MSNs (L) from the Control and VPA groups (D1-MSNs: Control, n = 10 cells from 6 mice; VPA, n = 7 cells from 4 mice; D2-MSNs: Control, n = 10 cells from 6 mice; VPA, n = 9 cells from 5 mice, two-tailed unpaired t test for H, Mann–Whitney U test for L). I, M, Representative traces of NMDAR EPSCs at +40 mV (upper traces) and AMPAR EPSCs at –80 mV (lower traces) in D1-MSNs (I) and D2-MSNs (M) from the Control and VPA groups. J, N, Statistics of the ratio of AMPAR to NMDAR EPSCs in D1-MSNs (J) and D2-MSNs (N; D1-MSNs: Control, n = 18 cells from 5 mice; VPA, n = 18 cells from 5 mice; D2-MSNs: Control, n = 19 cells from 6 mice; VPA, n = 16 cells from eight mice, two-tailed unpaired t test). O, S, Representative AMPAR-mediated and NMDAR-mediated currents determined by pharmacological isolation at +40 mV in D1-MSNs (O) and D2-MSNs (S) from the Control and VPA groups. P, T, Statistics of the ratio of AMPAR to NMDAR EPSCs determined by pharmacological isolation in D1-MSNs (P) and D2-MSNs (T; D1-MSNs: Control, n = 13 cells from 4 mice; VPA, n = 12 cells from 3 mice; D2-MSNs: Control, n = 10 cells from 3 mice VPA, n = 12 cells from 3 mice, two-tailed unpaired t test). Q, U, Example mean amplitude traces of evoked NMDAR-mediated EPSCs from D1-MSNs (Q) and D2-MSNs (U) in Control and VPA groups. R, V, Summary data showing no significant changes in input–output curve between VPA and Control group in D1-MSNs (R) and D2-MSNs (V; D1-MSNs: Control, n = 9 from 3 mice; VPA, n = 11 from 4 mice, D2-MSNs: Control, n = 9 from 4 mice; VPA, n = 8 from 3 mice, Friedman's M test). W, Y, Summary data for AP Rheobase from D1-MSNs (W) and D2-MSNs (Y) in Control and VPA groups. X, Z, Average instant firing frequency-current (F/I) curves for D1-MSNs (X) and D2-MSNs (Z) in the Control and VPA groups (D1-MSNs: Control, n = 15 cells from 4 mice; VPA, n = 16 cells from 5 mice; D2-MSNs: Control, n = 16 cells from 4 mice; VPA, n = 17 cells from 6 mice, two-tailed unpaired t test for W, Y, Friedman's M test for X, Z). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; *p < 0.05, **p < 0.01, ***p < 0.001.
To further confirm the presynaptic and postsynaptic alterations that we observed in D1-MSNs and D2-MSNs, respectively, we first measured the PPR at a 50-ms interpulse interval to assess the presynaptic release probability. A significant reduction in PPR was found in D1-MSN recordings (Fig. 2G,H), but a remarkable enhancement was found in D2-MSN recordings (Fig. 2K,L) in VPA mice, indicating an increased presynaptic release probability of glutamatergic presynaptic terminals innervating D1-MSNs and a decreased presynaptic release probability in D2-MSNs in the VPA group. Then, to determine whether prenatal VPA exposure affects glutamatergic transmission at the postsynaptic receptor level, we recorded the A/N ratio in MSNs by two commonly used methods (Fig. 2I,J,M–P,S,T). NMDAR-mediated eEPSCs were not significantly changed in either MSN recording (Fig. 2Q,R,U,V). In D1-MSNs, the A/N ratio was significantly larger in the VPA group (Fig. 2I,J,O,P), implying an increase in AMPAR-mediated postsynaptic currents on D1-MSNs after prenatal VPA exposure. Consistent with the findings for the amplitude of mEPSCs in D2-MSNs, there were no significant differences in AMPAR-mediated postsynaptic currents on D2-MSNs in the VPA group compared with the controls (Fig. 2M,N,S,T).
In addition to synaptic input, neuronal intrinsic excitability is also essential for the determination of functional neuronal output (Crabtree and Gogos, 2014), we thus examined MSN membrane properties to evaluate the possible impact of VPA exposure on intrinsic excitability. There were no significant differences in resting membrane potential (RMP) in D1-MSNs and D2-MSNs between VPA and wild-type mice, respectively (Extended Data Table 1-1). The intrinsic excitability of D1-MSNs in the VPA group was significantly increased, illustrated by a decreased rheobase current and a left-shifted average instantaneous frequency (Fig. 2W,X). In contrast, the intrinsic excitability of D2-MSNs in the VPA group was significantly decreased, illustrated by an increased rheobase current and a right-shifted average instantaneous frequency (Fig. 2Y,Z). Altogether, the electrophysiological data demonstrate that prenatal VPA exposure generates opposing effects on glutamatergic synaptic input and intrinsic excitability of D1-MSNs and D2-MSNs, respectively.
Abnormal dopamine receptor activation was associated with the expression of autism-like behaviors (Lee et al., 2018), and dopamine modulators could improve these behaviors abnormalities (LeClerc and Easley, 2015; Hara et al., 2017). One of the distinguishing features of MSNs is their differential dopamine receptor expression (Gerfen and Surmeier, 2011; Calabresi et al., 2014). In addition, dopamine D1 and D2 receptors dichotomously modulate the intrinsic excitability and glutamatergic synaptic transmission of striatal neurons (Kreitzer and Malenka, 2005, 2007; Surmeier et al., 2007, 2014; Shen et al., 2008; Nakanishi et al., 2014; Zhai et al., 2018). Thus, we next wanted to explore whether the changes in synaptic transmission and intrinsic excitability in MSNs observed above are linked to the impacts of prenatal VPA exposure on dopamine receptors. We found that incubation with the specific D1R agonist SKF81297 (20 µm) significantly enhanced glutamatergic synaptic transmission presynaptically and postsynaptically in the D1-MSNs of control mice, but this was not the case in the VPA group (Fig. 3A–C). Similar results were observed at 5 and 30 µm SKF81297 (Extended Data Table 1-1). In parallel, the selective D2R agonist quinpirole (10 µm) significantly decreased the magnitude of eEPSCs after HFS induction in control mice but not in VPA mice (Fig. 3D–F). As expected, in control mice, intrinsic excitability was significantly increased by the D1R agonist SKF81297 in D1-MSNs (Fig. 3G,H) and decreased by the D2R agonist quinpirole in D2-MSNs (Fig. 3K,L), but there were no significantly changes in intrinsic excitability in the D1-MSNs (Fig. 3I,J) and D2-MSNs (Fig. 3M,N) of VPA mice. Taken together, these data indicate that prenatal VPA exposure results in the activation of D1R and D2R in vivo, thereby increasing synaptic transmission and intrinsic excitability in D1-MSNs but decreasing those in D2-MSNs, which blunts exogenous agonists of dopamine receptors to further induce changes in synaptic transmission and intrinsic excitability in vitro.
Prenatal VPA exposure blunts dopamine receptor agonist-mediated effects on synaptic transmission and intrinsic excitability of MSNs in the DMS. A, Representative traces of mEPSCs in D1-MSNs. B, C, Summary data showing that SKF81297 incubation (20 µm) enhanced the mEPSC frequency (B) and amplitude (C) of D1-MSNs in Control mice, but not in VPA mice (Control, n = 12 cells from 5 mice; VPA, n = 14 cells from 6 mice, two-way ANOVA with Bonferroni post hoc). D, Representative averaged traces of eEPSC in D2-MSNs. Light traces represent the baseline EPSC average from 0 to 10 min (labeled with light color number 1), dark traces represent the average EPSCs from the last 10 min after LTD induction (dark color number 2). E, Normalized EPSC amplitude of D2-MSNs showing a less reduction of eEPSCs in VPA mice than those in Control mice by quinpirole incubation. F, Summary of the eEPSC magnitude in D2-MSNs (Control, n = 9 cells from 4 mice; VPA, n = 12 cells from 6 mice, Friedman's M test for E, two-tailed unpaired t test for F). G, I, K, M, Representative current-response curves of D1-MSNs (G, I) and D2-MSNs (K, M) in brain slices in Control and VPA mice. H, J, L, N, Average instant firing frequency-current (F/I) was increased by SKF81297 perfusion (5 µm) in D1-MSNs (H) and reduced by quinpirole perfusion (10 µm) in D2-MSNs (L) from Control mice, but not in VPA mice (J, N; D1-MSNs: Control, n =12 cells from 3 mice; VPA: n = 13 cells from 4 mice; D2-MSNs: Control, n =18 cells from 6 mice; VPA: n = 13 cells from 5 mice, Friedman's M test). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; **p < 0.01, ***p < 0.001.
Prenatal VPA exposure differentially alters the morphology of D1-MSNs and D2-MSNs in the DMS
The functional alteration of striatal MSNs is attributed to structural changes (Gertler et al., 2008). Morphologic changes in striatal MSNs have been extensively reported in different models of ASD (Bringas et al., 2013; Fuccillo, 2016; Wang et al., 2017). To verify whether prenatal VPA exposure induces structural remodeling in DMS MSNs, we performed spare labeling with a two-virus system by injecting hSyn-Cre (highly diluted) together with hSyn-DIO-mCherry virus into the DMS of VPA and control D2-eGFP transgenic mice (Fig. 4A). Mice were analyzed three weeks after viral injection to allow for the virus to be fully expressed. D1-MSNs were identified by mCherry fluorescence, while D2-MSNs showed coexpression of mCherry and eGFP (Fig. 4B). In D1-MSNs, we observed that the dendritic branches were longer and more complex as measured by Sholl analysis, illustrated by an increased number of intersections as a function of distance from the soma, total dendrite length, branch points and tips (Fig. 4C–G). The spines, especially mushroom and stubby spines, were denser in the D1-MSNs of VPA mice (Fig. 4H–J). In contrast, the dendritic branches were shorter and less complex in the D2-MSNs of VPA mice, illustrated by a decreased number of intersections as a function of distance from the soma, total dendrite length, branch points and tips (Fig. 4K–O). However, the spine density, including the total and the three subtypes, showed no difference in D2-MSNs (Fig. 4P–R). Thus, prenatal VPA exposure disrupted the structural homeostasis of D1-MSNs and D2-MSNs. D1-MSNs might receive more glutamatergic input, while D2-MSNs might receive less glutamatergic input in VPA mice than in controls. These different changes in dendritic anatomy could partly explain the opposite functional alterations observed in the two types of MSNs after VPA exposure.
Prenatal VPA exposure differentially alters the morphology of D1-MSNs and D2-MSNs in the DMS. A, Schematic representation and injecting site of AAV virus in DMS to sparse labeling MSNs. B, C, K, Representative confocal stacks and three-dimensional reconstruction images (right in C, K) of D1-MSNs (B, C) and D2-MSNs (B, K) in the DMS in the Control and VPA D2-eGFP mice. Arrowheads indicate cell body in B. Scale bars, 50 µm. D–G, L–O, Statistical results showing that dendritic branching complexity (D, L), the total dendrite length (E, M), branch points (F, N) and tips (G, O) were increased in D1-MSNs and decreased in D2-MSNs from VPA D2-eGFP mice (D1-MSNs: Control, n = 21 neurons from 3 mice; VPA, n = 21 neurons from 4 mice; D2-MSNs: Control, n = 14 neurons from 4 mice; VPA, n = 15 neurons from 5 mice, Friedman's M test for D, L, two-tailed unpaired t test for E–G, M–O). H, P, Representative confocal stacks and three-dimensional reconstruction images of dendrites in D1-MSNs (H) and D2-MSNs (P) from the Control and VPA D2-eGFP mice. Scale bar, 3 µm. I, J, Q, R, Statistical results showing that total spine density (I, Q) and density of different subtypes of spines (J, R) were increased in D1-MSNs, but not in D2-MSNs in VPA D2-eGFP mice (D1-MSNs: Control, n = 34 dendrites from 3 mice; VPA, n = 38 dendrites from 4 mice; D2-MSNs: Control, n = 28 dendrites from 4 mice; VPA, n = 35 dendrites from 5 mice, two-tailed unpaired t test for I, J, Q, R). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; *p < 0.05, ***p < 0.001.
Different responsiveness of DMS D1-MSNs and D2-MSNs are specifically linked to social deficits and repetitive behaviors in prenatally VPA-exposed offspring, respectively
The alteration of synaptic input and intrinsic excitability putatively affects MSN activity, which is essential for locomotion and emotional behavior (Kravitz and Kreitzer, 2012; Gunaydin et al., 2014). After having identified the differential structural and functional alterations in D1-MSNs and D2-MSNs reported above, we next wanted to investigate whether the core behavioral phenotypes, including social deficits and repetitive behaviors, were associated with alterations in basal activity in DMS D1-MSNs and/or D2-MSNs in VPA mice. To address this question, we sought to evaluate the population Ca2+ transients of the two main groups of projection neurons in the DMS in freely moving mice during these behavioral periods. We expressed the calcium indicator GCaMP6m in each population using a Cre-dependent viral strategy. Cre-dependent AAV GCaMP6m were stereotactically injected into the DMS of D1-Cre/D2-Cre mice, and an optical fiber was implanted above the injection site. After waiting two to three weeks for expression, we conducted fiber photometry to record GCaMP signal fluctuations in population fluorescence during social exploration and marble burying tasks (Fig. 5A–C). In both D1-MSNs and D2-MSNs of the control and VPA mice, we observed pronounced increases in transient Ca2+ fluorescence signaling during social interaction and spontaneous movements such as digging, which is considered a repetitive behavior, in the marble burying task (Fig. 5), suggesting the concurrent activation of striatal direct and indirect pathways during action initiation, consistent with prior research (Cui et al., 2013; London et al., 2018). To our surprise, we observed decreased transient Ca2+ signaling in D1-MSNs during social interaction (Fig. 5D–G), whereas no significant changes in signaling were observed in the D2-MSNs of VPA mice compared with controls (Fig. 5H–K). In contrast, an increase in Ca2+ response was detected in D2-MSNs during repetitive behaviors (digging in the marble burying task; Fig. 5P–S) but not in the D1-MSNs of VPA mice (Fig. 5L–O). GFP-expressing control animals did not exhibit transience or modulation during any behavior. The reduced transient Ca2+ responses in D1-MSNs during social interaction could be because of the potentiated state of excitatory input and intrinsic excitability, and such basal potentiation could occlude the transient increases in activity of D1-MSNs during social exploration. Conversely, the elevation in transient Ca2+ responses in D2-MSNs during repetitive behaviors could be because of a relatively larger increment in calcium fluorescence ratio from a depressed state of basal synaptic transmission and intrinsic excitability. To test this hypothesis, we performed baseline MSN recordings in D1-Cre/D2-Cre mice at animal's home cage, baseline D1-MSN activity was higher in VPA mice, illustrated by an increased Ca2+ transient frequency (Fig. 5T,U), whereas baseline D2-MSN activity was lower in VPA mice, illustrated by an decreased Ca2+ transient frequency (Fig. 5V,W). Together, the concurrent activation of two types of DMS MSNs during social interaction and repetitive behavior assays in VPA and control mice suggest that both pathways are involved in these two behaviors. Prenatal VPA exposure decreases D1-MSN Ca2+ responses, which are associated with social interactions, while increases in D2-MSN transient Ca2+ responses are linked to repetitive behaviors in prenatally VPA-exposed offspring.
Prenatal VPA exposure induces different responsiveness of DMS D1-MSNs and D2-MSNs during social interaction and marble burying tasks. A, Photometry setup used to record calcium signaling in GCaMP-expressing neurons. Light path for fluorescence excitation and emission is through a fiber optic implanted in DMS. B, Viral expression of GCaMP6m and placement of the fiber-optic probe in DMS. C, Schematic of viral injection and following behavioral tests. D, H, Heatmap illustrating the calcium responses (ΔF/F, %) for D1-MSNs (D) and D2-MSNs (H) during social interaction. E, I, Peri-event plots of averaged calcium signals (ΔF/F, %) during social behavior in D1-MSNs (E) and D2-MSNs (I). Solid line and shaded regions are the mean ± SEM. F, J, The calcium responses on social interaction were weaker in D1-MSNs (F), but not in D2-MSNs (J) of VPA mice compared with Control mice. G, K, Summary data of social interaction in D1-Cre (G) and D2-Cre (K) mice (D1-Cre: Control, n = 13 mice; VPA, n = 13 mice, D2-Cre: Control, n = 14 mice; VPA, n = 11 mice, Mann–Whitney U test for F, G, J, K). L, P, Heatmap of the calcium response (ΔF/F, %) in D1-MSNs (L) and D2-MSNs (P) during marble burying behaviors. M, Q, Peri-event plots of averaged calcium signals (ΔF/F, %) during marble burying in D1-MSNs (M) and D2-MSNs (Q). N, R, The calcium responses during marble burying were higher in D2-MSNs (R), but not in D1-MSNs (N) in VPA mice compared with Control mice. O, S, Summary data of the numbers of marbles in D1-Cre (O) and D2-Cre (S) mice (D1-Cre: Control, n = 10 mice; VPA, n = 11 mice, D2-Cre: Control, n = 13 mice; VPA, n = 13 mice, two-tailed unpaired t test). T, U, V, W, Representative baseline calcium traces and peak analysis from a D1-Cre (T, U) and D2-Cre (V, W) mouse. Baseline D1-MSN activity was higher in VPA mice, illustrated by an increased calcium transient frequency, whereas baseline D2-MSN activity was lower in VPA mice, illustrated by a decreased calcium transient frequency (D1-Cre: Control, n = 10 mice; VPA, n = 11 mice, D2-Cre: Control, n = 11 mice; VPA, n = 12 mice, Mann–Whitney U test). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; *p < 0.05, **p < 0.01.
Cell type-specific inhibition of DMS D1-MSNs improves social deficits, whereas activation of D2-MSNs alleviates repetitive behaviors in VPA offspring
To further investigate the relationship between D1-MSNs/D2-MSNs alterations and autism-like behaviors in VPA offspring, we performed a DREADDs to regulate the basal activity of D1-MSNs and D2-MSNs. We bilaterally injected Cre-dependent AAV-hSyn-DIO-hM3D(Gq)-mCherry (hM3Dq-mCherry) or AAV-hSyn-DIO-hM4D(Gi)-mCherry (hM4Di-mCherry) into the DMS of control and VPA D1-Cre/D2-Cre mice to selectively express hM3Dq-mCherry or hM4Di-mCherry in D1-MSNs and D2-MSNs, respectively (Fig. 6A). In parallel, all control mice received bilateral injections of AAV-hSyn-DIO-mCherry. After waiting two to three weeks for expression, the spiking response to current stimulation in hM3Dq-mCherry-positive MSNs was significantly increased, and those in hM4Di-mCherry-positive MSNs were decreased by bath application of CNO (10 µmol/l); this effect was absent in the control mCherry-expressing D1-MSNs and D2-MSNs (Fig. 6B). Since these direct and indirect pathways have an important impact on motor activity, to avoid confounding subsequent behavioral experiments, we first tested the effects of different doses of CNO on motor activity in D1-Cre and D2-Cre mice injected with hM3Dq-mCherry and hM4Di-mCherry. Similar to previous reports (Wang et al., 2017), we found that 0.3 mg/kg CNO did not significantly affect motor activity (Extended Data Table 1-1), so a low dose of 0.3 mg/kg CNO was used for subsequent behavioral experiments. When the DMS D1-MSNs were inhibited by injection of CNO, the VPA mice spent significantly more time in the compartment of the stranger mouse in the three-chamber test (Fig. 6C–E) and performed more social interaction with the novel partner in the home-cage than mCherry control mice (Fig. 6F). However, the self-grooming time and number of buried marbles were not significantly affected (Fig. 6G,H). We also found that activation of D1-MSNs in control D1-Cre mice could mimic social deficits (Fig. 7A–F) but not repetitive behaviors (Fig. 7G,H). To further confirm the chemogenetic results, we used an optogenetic approach with more space accuracy to investigate the causal relationship between D1-MSNs and social interaction deficits in VPA mice, and similar results were observed (Fig. 8A–H). These results further indicate that enhanced basal D1-MSNs activity plays a causal role in the social interaction deficits exhibited by prenatally VPA-exposed offspring.
Cell-specific inhibition of DMS D1-MSNs improves social deficits, whereas activation of D2-MSNs alleviates repetitive behaviors in VPA offspring. A, Schematic representation and injecting site of virus in the DMS. B, Current-voltage relationship of representative D1-MSNs and D2-MSNs recorded before and after 10 µmol/l CNO perfusion in brain slices. Gray traces represent the rheobase of APs from D1-MSNs and D2-MSNs. C, I, Representative traces during three-chamber test in hM4Di-expressing VPA D1-Cre mice (C) and hM3Dq-expressing VPA D2-Cre mice (I). D, E, After CNO injection, hM4Di-expressing VPA D1-Cre mice spent more time in the stranger compartment (D), and showed higher social index during three-chamber test (E). F, Summary data showing that social time in home-cage social interaction test was increased in hM4Di-expressing VPA D1-Cre mice compared with mCherry-expressing VPA D1-Cre mice after CNO injection. G, H, Summary data showing no significant changes of self-grooming time (G) and numbers of marbles buried (H) both in hM4Di-expressing VPA D1-Cre mice and in mCherry-expressing VPA D1-Cre mice after CNO injection. J, K, Summary data showing no significant changes of time duration in stranger compartments (J) and social index (K) in hM3Dq-expressing VPA D2-Cre mice during three-chamber test after CNO injection. L, Summary data showing similar social time during home-cage social interaction test in hM3Dq-expressing VPA D2-Cre mice and mCherry-expressing VPA D2-Cre mice after CNO injection. M, N, Summary data showing that self-grooming time (M) and numbers of marbles buried (N) were significantly decreased in hM3Dq-expressing VPA D2-Cre mice compared with mCherry-expressing VPA D2-Cre mice after CNO injection (VPA: D1-mCherry, n = 9 mice; D1-hM4Di, n = 11 mice; D2-mCherry, n = 10 mice; D2-hM3Dq, n = 13 mice, two-way ANOVA with Bonferroni post hoc for D, J, two-tailed unpaired t test for E–H, K–N). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; *p < 0.05, **p < 0.01, ***p < 0.001.
Cell-specific activation of D1-MSNs by DREADD induces social deficits, whereas inhibition of D2-MSNs activity induces repetitive behaviors in control offspring. A, Schematic representation and injecting site of virus in DMS. B, Current-voltage relationship of representative D1-MSNs and D2-MSNs recorded before and after 10 µmol/l CNO perfusion in brain slices. C, I, Representative traces during three-chamber test in hM3Dq-expressing D1-Cre mice (C) and hM4Di-expressing D2-Cre mice (I). D, E, After 0.3 mg/kg CNO injection, hM3Dq-expressing D1-Cre mice spent more time in the stranger compartment (D), and showed lower social index during three-chamber test (E). F, Summary data showing that social time in home-cage social interaction test was decreased in hM3Dq-expressing D1-Cre mice compared with mCherry-expressing D1-Cre mice after CNO injection. G, H, Summary data showing no significant changes of self-grooming time (G) and numbers of marbles buried (H) both in hM3Dq-expressing D1-Cre mice and in mCherry-expressing D1-Cre mice after CNO injection. J, K, Summary data showing no significant changes of time duration in stranger compartments (J) and social index (K) in hM4Di-expressing D2-Cre mice during three-chamber test after 0.3 mg/kg CNO injection. L, Summary data showing similar social time during home-cage social interaction test in hM4Di-expressing D2-Cre mice and mCherry-expressing D2-Cre mice after CNO injection. M, N, Summary data showing that self-grooming time (M) and numbers of marbles buried (N) were significantly increased in hM4Di-expressing VPA D2-Cre mice compared with mCherry-expressing D2-Cre mice after 0.3 mg/kg CNO injection (Control: D1-mcherry: n = 10 mice; D1-hM3Dq: n = 12 mice; D2-mcherry: n = 10 mice; D2-hM4Di: n = 11 mice, two-way ANOVA with Bonferroni post hoc for D, J, two-tailed unpaired t test for E–H, K–N). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; **p < 0.01, ***p < 0.001.
Optogenetical inhibition of D1-MSNs improves social deficits, whereas activation of D2-MSNs alleviates repetitive behaviors in VPA offspring. A, Schematic representation and injecting site of virus in the DMS. B, C, Representative traces from in vitro slice recordings of DMS NpHR3.0-expressing D1-MSNs (B) illuminated by a 596-nm yellow light stimulation for 1 s, and ChR2-expressing D2-MSNs (C) illuminated by 10-Hz, 473-nm blue light stimulation. D, I, Representative traces of three-chamber test in the NpHR3.0-expressing VPA D1-Cre mice and EYFP-expressing VPA D2-Cre mice. E, F, Compared with the EYFP-expressing VPA mice, the NpHR3.0-expressing VPA D1-Cre mice spent significantly more time in the stranger compartment (E) and showed higher social index (F) during three-chamber test. G, Summary data showing that social time in home-cage social interaction test was increased in NpHR3.0-expressing VPA D1-Cre mice compared with EYFP-expressing VPA D1-Cre mice. H, Summary data showing no significant changes of self-grooming time both in NpHR3.0-expressing VPA D1-Cre mice and in EYFP-expressing VPA D1-Cre mice. J, K, Summary data showing no significant changes of time duration in stranger compartments (J) and social index (K) both in ChR2-expressing VPA D2-Cre mice and in EYFP-expressing VPA D2-Cre mice. L, Summary data showing similar social time during home-cage social interaction test in ChR2-expressing VPA D2-Cre mice and EYFP-expressing VPA D2-Cre mice. M, Summary data showing that self-grooming time was significantly decreased in ChR2-expressing VPA D2-Cre mice compared with EYFP-expressing VPA D2-Cre mice (VPA: D1-EYFP, n = 11 mice; D1-NpHR3.0, n = 10 mice; D2-EYFP, n = 10 mice; D2-ChR2, n = 12 mice, two-way RM measurement with Bonferroni post hoc for E–M). Data represent mean ± SEM. See Extended Data Table 1-1 for detailed statistical information; **p < 0.01.
In contrast, we found that activation of D2-MSNs in VPA D2-Cre mice expressing hM3Dq alleviated repetitive behaviors including prolonged self-grooming time (Fig. 6M) and the increase in the number of buried marbles (Fig. 6N) but did not improve the impaired social behaviors, as shown by the lack of a difference in the time spent in the compartment of the stranger mouse versus empty compartment and social index in the three-chamber test (Fig. 6I–K). In addition, no significant difference in home-cage social interactions with the novel partner was observed compared with the mCherry control mice (Fig. 6L). The inhibition of D2-MSNs expressing hM4Di in the control D2-Cre mice increased repetitive behavior, indicated by the significantly increased self-grooming time and number of marbles buried (Fig. 7M–N), but no effects on social interaction behaviors were observed in the three-chamber and home-cage tests (Fig. 7I–L). Optogenetic activation of D2-MSNs had no significant effects on social deficits (Fig. 8I–L) but decreased the self-grooming in VPA D2-Cre mice (Fig. 8M). These results further demonstrate that decreased basal D2-MSNs activity plays a causative role in the repetitive behaviors observed in offspring prenatally exposed to VPA.
Discussion
The hallmark of autism is the coexistence of social deficits and repetitive behaviors. Establishing the causality between circuit dysfunction and coexisting autism behaviors is helpful for understanding the pathophysiology of autism. The present study revealed that there were indeed coexisting and opposite morphologic and functional alterations in the direct and indirect pathways of the DMS, and such alterations in these two pathways were responsible for the expression of the two different autism-like behaviors observed in mice prenatally exposed to VPA. Alteration in D1-MSNs was characterized by a potentiated state of excitatory input and intrinsic excitability, and such basal potentiation could blunt the transient increases in activity of D1-MSNs during social exploration in VPA mice. Alteration in indirect pathway was a depressed state of excitatory input and intrinsic excitability in D2-MSNs, which relatively amplified the calcium transients of the D2-MSNs during repetitive behaviors. A causal relationship linking such differential alterations in the two pathways to the coexistence of these two autism-like behaviors was demonstrated by cell type-specific correction of abnormal basal activity in the D1-MSNs and D2-MSNs of VPA mice. Together, these results support the hypothesis that VPA exposure induces differential alterations concurrently in both direct and indirect pathways in the DMS, and such differential alterations are specifically associated with the two abovementioned autism-like behaviors. The results of the present work also suggest that correction of alterations in the DMS MSNs is a potential therapeutic target for developing interventions for autism-like behaviors.
VPA exposure induced coexisting and opposite striatal circuit functional alterations at two levels. One was basal activity alterations, illustrated by excitatory synaptic input and intrinsic excitability, and baseline Ca2+ signaling in D1-MSNs and D2-MSNs (Figs. 2, 3, 5). The other was the differential magnitude of transient increases in activity of MSNs during social interaction and repetitive behaviors, illustrated in vivo via Ca2+ signaling recording (Fig. 5). Unfortunately, the concrete underlying physiological mechanisms, from the altered transient responses in the direct and indirect pathway MSNs during the behavioral performance to the expression of impaired social communication and increased repetitive behavior, are still not clear. Recent advances have revealed the concurrent activation of both pathways during the initiation and completion of complex natural behaviors (Calabresi et al., 2014; Nelson and Kreitzer, 2014; Geddes et al., 2018). The coordinated activity of the direct and indirect pathways is critical for the appropriate timing/synchrony of basal ganglia circuits and thereby sustains normal function during reward processing, goal-directed learning, and motor control (Cui et al., 2013; London et al., 2018). In the present study, apart from the reduction in D1-MSN responsiveness during social exploration and the enhancement in D2-MSN responsiveness during repetitive behaviors, we also observed the concurrent activation of D1-MSNs and D2-MSNs during the initiation of these autism-like behaviors in control and VPA mice (Fig. 5). The mechanistic relationship between the changes in the responsiveness of MSNs and the expression of behavioral abnormalities remains to be elucidated by embedding their roles in the dynamic oscillation and continuous functioning of the whole cortex-basal ganglia-thalamus-cortex network.
Another interesting finding in this study is that D1-MSN dysfunction was more likely associated with social deficits while D2-MSN dysfunction was related to repetitive behaviors in the DMS of the VPA model of autism. However, whether the diverse nature of ASD symptomatology represents distributed dysfunction in brain networks or abnormalities within specific neural circuits is still unclear, and how dysfunction in specific brain circuits may contribute to specific behavioral deficits remains to be explored (Fuccillo, 2016). Genetic mouse models of autism provide an opportunity to probe brain function in a highly specific manner and explore alterations in the activation of striatal MSNs (Portmann et al., 2014; Fuccillo et al., 2016; Rothwell, 2016; Wang et al., 2017; Folkes et al., 2020). Our results in D2-MSNs are consistent with the finding that impaired excitatory synaptic transmission in dorsal striatal D2-MSNs is associated with an increase in repetitive behaviors in Shank3 mutant mice (Wang et al., 2017). Interestingly, recent studies have described that NAc D1-MSNs are linked to social behavior in wild-type mice (Gunaydin et al., 2014; Folkes et al., 2020) as well as repetitive behaviors in mice carrying ASD-associated mutations in Neuroligin3 (Rothwell et al., 2014; Rothwell, 2016). The present study did find an association between VPA-induced D1-MSN dysfunction in DMS and social deficits in VPA mice. The VPA-induced mouse model of autism has been motivated by clinical discoveries of the etiology of a set of ASD patients and accurately mimics the core symptoms of the disease (Mabunga et al., 2015; Nicolini and Fahnestock, 2018). As a histone deacetylase inhibitor, VPA exposure at the early developmental stage globally perturbs gene expression in the brain, including many well-studied autism-associated genes, such as Mecp2, Shank3, and Neuroligin 3 (Cohen et al., 2013; R. Zhang et al., 2018). Thus, the VPA model gives us a chance to study relatively more complete alterations in striatal circuits and to link them to the coexistence of two core symptoms of ASD.
Autism is characterized as the coexistence of social interaction deficits and repetitive behaviors (American Psychiatric Association, 2013). A deeper understanding of the pathophysiology of ASD can be derived from how striatal circuits normally function to regulate these behaviors and how alterations in such circuits mediate the major symptoms. In addition to the dichotomous function of the direct and indirect pathways mentioned above, subregions of the striatum are also relatively specialized for the regulation of various dimensions of behaviors. Previous clinical studies have reported structural and functional alterations in the caudate (roughly corresponding to DMS) of ASD patients (Turner et al., 2006; Langen et al., 2007; Takarae et al., 2007). Evidence from animal studies supports that through the coordinated activity of the direct and indirect pathway MSNs, DMS plays an important role in reward processing and motor control (Kravitz et al., 2010, 2012; Cox and Witten, 2019). Social interaction can be considered as a natural reward, aberrant reward processing seen in ASD may contribute to the reduced sensitivity to social stimuli, and result in social deficits (Insel, 2003; Dölen et al., 2013; Rothwell, 2016). DMS is involved in goal-directed behaviors that are flexible and sensitive to outcome, which is often the case early in “habitual” learning, and the process of habit formation could contribute to some of the repetitive routines and rituals observed in ASD patients (Yin and Knowlton, 2006; Balleine et al., 2007; Rothwell, 2016; Cox and Witten, 2019). Thus, it is not surprising that the dysfunction of DMS circuits results in the coexistence of autism-like behaviors in VPA mice.
Although our present work supports the hypothesis that coexisting and differential alterations in DMS direct and indirect pathways mediate the coexistence of two core symptoms of ASD, an alternative hypothesis for coexistence of the two core symptoms could be the coexistence of circuit alterations in different subregions. The involvement of other striatal subregions, including the DLS and NAc, and even the whole cortex-basal ganglia-thalamus-cortex network, should also be considered. DLS is involved in the control of self-grooming and other stereotypical behavioral patterns (Kalueff et al., 2016; Yu et al., 2018; Gandhi and Lee, 2020), while the NAc is commonly implicated in social behavior and reward processing (Gunaydin et al., 2014) and was recently shown to be involved in autism-like behaviors (Portmann et al., 2014; Rothwell et al., 2014; Schiavi et al., 2019; Folkes et al., 2020). We hope that competition between different hypotheses will motivate more enthusiastic research in the field and ultimately lead to a unified pathophysiological theory for ASD.
Finally, the reason for the opposite alterations observed in D1-MSNs and D2-MSNs in the present study is unknown. Studies employing genetic models of ASD have given prominence to a series of synaptic proteins in forming alterations in striatal circuits and have suggested a number of molecular and cellular mechanisms for the underlying pathophysiology (Fuccillo et al., 2016; Rothwell, 2016; Wang et al., 2017). Since VPA induces changes in the expression of a large number of genes, including many genes related to maintaining structural and functional homeostasis in synaptic strength, it is possible to determine the molecular mechanism by evaluating the dysfunction of such genes in the VPA model of ASD. Furthermore, the present work found that circuit alterations in DMS D1-MSNs and D2-MSNs were associated with the activation of the two types of dopamine receptors (Fig. 3). Previous studies also support that abnormal dopamine receptor activation could induce autism-like behaviors in mice (Lee et al., 2018). It has been well established that the activation of D1 and D2 dopamine receptors can induce long-term changes in the form and function of the striatal circuit (Kreitzer and Malenka, 2005; Gerfen and Surmeier, 2011; Kravitz and Kreitzer, 2012). Thus, it is also possible that an upstream mechanism (e.g., nigrostriatal pathway dysfunction) of the endogenous concurrent activation of excitatory D1 and inhibitory D2 dopamine receptors could induce the differential alterations observed in DMS D1-MSNs and D2-MSNs. Elucidation of the possible mechanisms could help in developing therapeutic options targeting these circuit alterations to address these behavioral abnormalities.
Footnotes
This work was supported by National Natural Science Foundation of China Grants 11727813, 81971285, and 82071516; the Innovation Capability Support Program of Shaanxi Province, China Grant 2020TD-037; and Fundamental Research Funds for the Central Universities Grants GK202005001 and GK202105001. We thank Dr. Fuqiang Xu for sharing D1-Cre and D2-Cre transgenic mice, Dr. Guoping Feng and Dr. Charles Gerfen for sharing the D2-eGFP transgenic mice, Dr. Yong Liu (Xi'an Jiaotong University) and Dr. Ruixi Li (Fudan University) for suggestions and discussions, Dr. Yihui Liu (Shaanxi Normal University) for suggestions on data statistics, and Miss Siying Ren (Beijing Institute of Graphic Communication) for figure color matching.
The authors declare no competing financial interests.
- Correspondence should be addressed to Wei Ren at renwei{at}snnu.edu.cn or Yingfang Tian at yingfang_tian{at}snnu.edu.cn
