Early hyperactivity and precocious maturation of corticostriatal circuits in Shank3B^−/− mice

Abstract

Some autistic individuals exhibit abnormal development of the caudate nucleus and associative cortical areas, suggesting potential dysfunction of cortico-basal ganglia (BG) circuits. Using optogenetic and electrophysiological approaches in mice we identified a narrow postnatal period characterized by extensive glutamatergic synaptogenesis in striatal spiny projection neurons (SPNs) and a concomitant increase in corticostriatal circuit activity. SPNs during early development have high intrinsic excitability and respond strongly to cortical afferents despite sparse excitatory inputs. As a result, striatum and corticostriatal connectivity are highly sensitive to acute and chronic changes in cortical activity, suggesting that early imbalances in cortical function alter BG development. Indeed, a mouse model of autism with deletions in SHANK3 (Shank3B^−/−) has early cortical hyperactivity, which triggers increased SPN excitatory synapse and corticostriatal hyper-connectivity. These results show a tight functional coupling between cortex and striatum during early postnatal development and suggest a potential common circuit dysfunction caused by cortical hyperactivity.

Introduction

The basal ganglia (BG) are a group of phylogenetically conserved subcortical nuclei involved in the generation of purposeful movements and experience-dependent acquisition of complex motor skills and behavioral strategies¹. The striatum is the largest nucleus of the BG and receives diverse inputs from multiple brain regions, including glutamatergic projections from cortex that convey sensorimotor as well as cognitive information. Corticostriatal afferents synapse onto spiny projection neurons (SPNs), which project to output BG nuclei that modulate cortex and several motor centers, forming a series of recurrent circuit loops^1,2. Despite the central role of BG in controlling human behavior and their implication in numerous neurodevelopmental disorders, little is known about the early development of corticostriatal circuits. Studies across species show protracted maturation of SPN morphology and function for several weeks to months after birth^3–5. However, perturbing striatal activity during early postnatal periods in mice alters corticostriatal connectivity as early as postnatal day (P) 14, indicating that cortico-BG circuits are functional early in development and that recurrent network activity regulates BG circuit maturation⁶.

A subset of autistic individuals exhibit dysmorphic striatal regions suggesting that BG circuit dysfunction may contribute to the pathophysiology of autism spectrum disorders (ASD)^7–10. In addition, autism-associated behaviors such as limited and obsessive interests, and motor stereotypies are hallmarks of other BG-associated disorders such as Tourette’s and obsessive compulsive disorders (OCD)^1,11,12. Increasing evidence suggest a strong genetic component in the etiology of ASD, with many associated genes encoding proteins involved in glutamatergic synapse development and function^13,14. In particular, SHANK3 is highly prevalent in ASD with full deletions found in a large fraction of Phelan-McDermid syndrome cases and rare mutations associated with idiopathic ASDs^14–16. Shank family proteins are postsynaptic scaffolds at glutamatergic synapses where they organize an extensive protein complex¹⁴. Overexpression of Shank3 in dissociated neuronal cultures increases synapse formation and strength^17,18, whereas transgenic overexpression in vivo increases dendritic spines in hippocampus¹⁹. Conversely, knock-down of Shank3 reduces spine density in dissociated hippocampal neurons suggesting a correlation between Shank3 levels and excitatory synapse number²⁰. Notably, Shank3 is the only Shank expressed in the mouse striatum and mice with Shank3 deletions exhibit repetitive behaviors and reduced social interactions, two behavioral hallmarks of ASD^21–24. In particular, deletion of exons encoding the PDZ domain of Shank3 in the mouse (Shank3B^−/−) increases striatal volume and reduces corticostriatal transmission in adults²³. Given the developmental trajectory of ASD symptoms and the implication of Shank3 in glutamatergic synapse maturation, the corticostriatal phenotype exhibited by adult Shank3B^−/− mice could have a developmental origin.

Here we examined the development of corticostriatal circuits in mice by combining optogenetic approaches with in vitro and in vivo electrophysiological analyses. Our findings show that SPNs are primed to respond to cortical activity from very early developmental stages and undergo a phase of rapid maturation from P10-18. During this period, corticostriatal connectivity is highly sensitive to acute and chronic changes in cortical activity suggesting that early imbalances in cortical function can impair BG circuit development. Surprisingly, we found that Shank3B^−/− mice exhibit precocious maturation of SPN excitatory inputs due to increased corticostriatal network activity. These results reveal a developmental circuit defect induced by loss of Shank3 and suggest that abnormal corticostriatal maturation may be a common aspect of disorders with early imbalances in cortical activity.

Results

Rapid SPN excitatory synapse development after ~P10

To characterize the development of excitatory afferents onto SPNs we measured optically-evoked excitatory post-synaptic currents (oEPSC) in dorsomedial striatum of P6-P30 mice. Channelrhodopsin (ChR2) was expressed in a subset of corticostriatal projection neurons using Rbp4-Cre;Ai32 (Rbp4-Cre;ChR2^f/f) transgenic mice²⁵. These mice express Cre in layer 5 pyramidal neurons and activate ChR2(H134R)-EYFP expression from a Cre-conditional (floxed-stop) allele (Fig. 1a). Optical stimulation of ChR2⁺ Rbp4 fibers was induced by 5 ms pulses of 473 nm light delivered through the objective while performing voltage-clamp whole-cell recordings from SPNs. AMPA-type glutamate receptor (AMPAR)-mediated currents were isolated by pharmacologically inhibiting NMDA-type glutamate (NMDAR) and GABA_A receptors and recording at a holding potential (V_h) of −70 mV. All striatal neurons recorded at P6-7 exhibited oEPSCs (Figs. 1b–c; oEPSC amplitude; P6-7, 122±13 pA, n=16 cells/2 mice) revealing the presence of functional excitatory synapses at these early stages. Evoked current amplitude increased gradually until ~P10 (Fig. 1c, oEPSC amplitude; P8-9, 181±22 pA, n=19 cells/2 mice; P10-11, 595±80 pA, n=13 cells/2 mice) and exhibited a marked increase thereafter (Fig. 1c, oEPSC amplitude; P12-13, 1848±160 pA, n=31 cells/4 mice; P14-15, 3195±146 pA, n=57 cells/5 mice, one-way ANOVA, p<0.0001). To overcome space clamp limitations caused by large amplitude EPSCs, recordings at later developmental time points were performed at V_h=−20 mV to reduce the electrochemical driving force of AMPAR mediated ionic currents (Fig. 1d). Control recordings at P14 showed a linear relationship between oEPSCs recorded at V_h=−70 and −20 mV across a wide range of amplitudes (Supplementary Fig. 1). Evoked oEPSC amplitude increased ~2-fold from P14-15 to P17-18 and stabilized thereafter (Figs.1d–e; P14-15, 731±57 pA, n=20 cells from 3 mice; P17-18, 1269±59 pA, n=20 cells/3 mice; P29-30, 1337±97 pA, n=19 cells/3 mice, one-way ANOVA, p<0.0001). Normalization across data sets revealed that ~90% of Rbp4⁺ oEPSC amplitude develops from P8-9 to P17-18 (Fig. 1e).

Figure 1. Rapid development of striatal SPN excitatory input in mice after ~P10.

(a) Whole cell voltage clamp recordings in SPNs of dorsomedial striatum in acute brain slices of Rbp4-Cre; ChR2-YFP^f/f mice and optogenetic fiber stimulation using whole field illumination. Scale bar, 1mm.

(b) AMPAR oEPSCs recorded in SPNs under voltage clamp (V_h=−70 mV) at different postnatal days in response to brief pulses of 473 nm laser light (blue rectangle).

(c) Mean oEPSC peak amplitude ± SEM recorded in SPNs at V_h=−70 mV from P6-P15 and (d) at V_h=−20 mV from P14-30.

(e) Developmental progression of oEPSC amplitude values normalized to P30.

(f) Representative traces of AMPAR (gray) and NMDAR (black) oEPSCs recorded in the same SPN of Rbp4-Cre; ChR2-YFP^f/wt at V_h=−70 mV and V_h=+40 mV, respectively. Red circle indicates time of NMDAR current amplitude analysis at 50ms post light stimulus (blue rectangle).

(g) Mean AMPAR and (h) NMDAR oEPSC peak amplitude ± SEM in P10-11 and P14-15 SPNs.

(i) Mean AMPAR to NMDAR ratio ± SEM for each SPN represented in (g–h). Scale bar, 1mm.

(j) Average (solid line) ± SEM (dashed line) NMDAR oEPSCs from cells represented in (g–h).

(k) Mean NMDAR oEPSC decay time constant (Tau) ± SEM of P10-11 and P14-15 SPNs.

(l) Coronal brain slice of P12 mouse infected with AAV8-CAG-EGFP. Ctx- cortex; Str- striatum. Scale bar, 1mm.

(m) Representative images of EGFP expressing SPN dendrites at different postnatal days. Scale bar, 10 μm.

(n) Average dendritic spine density ± SEM from infected SPNs at P8-P24.

In many brain areas synapse development is characterized by a gradual recruitment and stabilization of AMPARs resulting in a progressive increase of the ratio of currents carried by AMPARs in relation to NMDARs²⁶. To determine AMPAR/NMDAR current ratio (R_A/N) in developing SPN synapses we isolated the AMPAR and NMDAR components of oEPSCs by recording at V_h=−70 and +40 mV, respectively (Fig. 1f). Rbp4-Cre;ChR2^f/wt mice were used to reduce overall current amplitude and maximize voltage control in the absence of NMDAR inhibitors. There was a significant increase in both the AMPAR (P10-11, 155±24 pA, n=18 cells/3 mice; P14-15, 647±65 pA, n=20 cells/3 mice; unpaired t-test, p<0.0001) and NMDAR (P10-11, 158±19 pA, n=18; P14-15, 441±45 pA; unpaired t-test, p<0.0001) components of oEPSC from P10-11 to P14-15, and R_A/N increased significantly during this time period (P10-11, 0.85±0.07, n=18; P14-15, 1.56±0.14, n=20, unpaired t-test p=0.0006) consistent with ongoing synapse maturation (Fig. 1g–j)^26,27. NMDAR EPSC decay kinetics were not significantly different between the two age groups (Figs. 1j–1k; P10-13, τ=409±40 ms, n=18; P14-15, τ=415±23 ms, n=20) suggesting no change in the subunit composition of NMDARs across this developmental period.

Corticostriatal synapses are mainly localized in dendritic spines of SPNs¹. To address if spinogenesis is associated with oEPSC amplitude increase we analyzed dendritic spine density in developing SPNs in dorsomedial striatum using adeno-associated virus encoding GFP (AAV8-CAG-EGFP) and confocal microscopy (Fig. 1l–n). Consistent with the developmental increase of oEPSC amplitude, the density of spines increased markedly during the second postnatal week with the highest growth rate between P10 and P12 (P8, 0.33±0.03 μm⁻¹, n=16 dendrites/2 mice; P10, 0.43±0.02 μm⁻¹, n=17 dendrites/2 mice; P12, 0.65±0.03 μm⁻¹, n=35 dendrites/2 mice; P14, 0.71±0.02 μm⁻¹, n=23 dendrites/2 mice; P24, 0.87±0.03 μm⁻¹, n=25 dendrites/2 mice). Together these results indicate that a large fraction of SPN excitatory synapses develops rapidly during the end of the second postnatal week.

Increase in corticostriatal activity in vivo from P10-16

To characterize how corticostriatal circuit activity evolves during this period we recorded multi-unit activity in cortex and striatum of awake head-fixed mice from P10-16 (Fig. 2a) following 1 hour recovery from surgical head post implantation. Neuronal activity was recorded using multi-electrode arrays for 20 min after a 10 min stabilization period. Recordings were performed in somatosensory regions of cortex by positioning the electrode array 1250 μm deep from the surface to target cortical layer 5. The firing rate (FR) of cortical units increased ~6-fold from P10-16 (Median P10-11, 0.25 Hz, n=77 units/5 mice; P12-13, 0.48 Hz, n=125 units/5 mice; P14-16, 1.64 Hz, n=207 units/5 mice; Kruskal-Wallis test, p<0.0001), with a significant increase in the number of bursts fired (Median burst frequency P10-11, 0.2 min⁻¹, n=77; P12-13, 0.4 min⁻¹, n=125; P14-16, 1.0 min⁻¹, n=207; Kruskal-Wallis test, p<0.0001) and in the frequency of APs within a burst (Median intra burst average frequency P10-11, 4.9 Hz, n=53; P12-13, 9.1 Hz, n=89; P14-16, 16.4 Hz, n=198; Kruskal-Wallis test, *p=0.035, ***p=0.0008, ****p<0.0001) (Fig. 2b–e).

Figure 2. Correlated increase in cortical and striatal activity in vivo from P10 to P16.

(a) Experimental diagram of in vivo recordings in a sagittal view of a mouse brain showing cortex (CTX) and striatum (STR).

(b) Representative recordings of multi-unit activity in cortex (left) and striatum (right) at P10 and P14.

(c) Median ± interquartile range of average FR of cortical units from P10-11 to P14-16.

(d) Median AP burst frequency and (e) Intra-burst frequency ± interquartile range of cortical units shown in (c).

(f) Median ± interquartile range of average FR of striatal units at different developmental time points.

(g) Median AP burst frequency and (h) Intra-burst frequency ± interquartile range of striatal units shown in (f).

The developmental progression of activity in striatum was similar to cortex with a ~5-fold increase in average firing rate from P10 to P16 (Median FR P10-11, 0.14 Hz, n=92 units/5 mice; P12-13, 0.36 Hz, n=167 units/5 mice; P14-16, 0.71 Hz, n=255 units/5 mice; Kruskal-Wallis test, ***p=0.0007, ****p<0.0001) and a significant increase in bursting (Median burst frequency, P10-11, 0.2 min⁻¹, n=92; P12-13, 0.6 min⁻¹, n=167; P14-16, 1.2 min⁻¹, n=255; Kruskal-Wallis test, ****p<0.0001) and intra-burst FR (Median intra burst frequency P10-11, 10.0 Hz, n=65; P12-13, 18.7 Hz, n=141; P14-16, 22.8 Hz, n=236, Kruskal-Wallis test, **p=0.0013, ***p=0.0002, ****p<0.0001) (Fig. 2f–h). These results indicate that P10-16 is a period characterized not only by rapid maturation of SPN synapses but also by a significant increase in corticostriatal activity.

Corticostriatal coupling during early development

To address if the observed rise in striatal neuronal activity is caused by the increased number of SPN synapses, we compared responses of striatal neurons to optogenetic stimulation of corticostriatal projection neurons at P10-11 and P14-16. Stimulation of cortical neurons was achieved using Rbp4-Cre; ChR2^f/f mice and extracranial optical stimulation using 473nm light (Fig. 3a). The stimulation protocol consisted of trains of 10 pulses at 10 Hz repeated 10 times every 30 seconds, similar to the median AP burst frequency of cortical units at P10-11²⁸ (Fig. 2e). Optogenetic stimulation reliably increased FR of cortical units at all ages (Peak FR; P10-11, 11.7±0.5 Hz, n=76 units/5 mice; P14-16, 14.2±1.3 Hz, n=139 units/6 mice) with maximum peak FR in response to the first pulse of the stimulation train (1^st pulse peak FR; P10-11, 24.6±1.2 Hz; P14-16, 30.9±3.3 Hz) (Fig. 3d–e). Response latencies were similar in the two age groups with peak FR 5-10 ms after stimulus onset (Supplementary Fig. 2). Thus, stimulation of Rbp4⁺ cortical cells at these two developmental time points increases cortical activity to a similar extent, allowing comparisons of downstream striatal activity across development.

Figure 3. Corticostriatal coupling during early development.

(a) Experimental diagram of in vivo recordings and extracranial optogenetic stimulation using 473nm laser (blue) in Rbp4-Cre;ChR2-YFP^f/f mice showing cortex (CTX) and striatum (STR).

(b) Example raster plot (top) and 20 ms bin peri-stimulus time histogram (PSTH, bottom) of action potentials of a P11 striatal unit during optogenetic stimulation of cortex with a 10 Hz light pulse train (blue). Note the robust response to the first pulse of the train.

(c) Raster plot (top) and 5 ms bin PSTH (bottom) of the unit shown in (b) in response to individual optogenetic pulses (blue).

(d) PSTH (20 ms bin) representing firing rate of cortical neurons during ChR2-stimulation (blue) at P10-11 (black) and P14-16 (red). Shaded regions represent ± SEM

(e) PSTH (5 ms bin) of units shown in (d) in response to the first pulse of the stimulation train. Shaded regions represent ± SEM

(f) PSTH (20 ms bin) of firing rate of striatal neurons during cortical stimulation (blue) at P10-11 (black) and P14-16 (red). Shaded regions represent ± SEM

(g) PSTH (5 ms bin) of striatal neurons in response to the first pulse of the stimulation train. Shaded regions represent ± SEM. Note the presence of secondary peaks indicative of burst firing in response to single light pulses.

At P14-16, cortical stimulation increased the activity of striatal units (Peak FR 12.7±1.3 Hz, n=161 units/6 mice) with peak responses 15–20 ms after stimulus onset (Supplementary Fig. 2). Striatal responses were maximal in response to the first pulse of the train (Peak FR in response to 1^st stimulus; P14-16, 29.4±2.6 Hz) and exhibited moderate depression in response to subsequent light flashes (Figs. 3f–g). However, striatal units also responded strongly to cortical stimulation at P10-11 (Post stimulus peak FR; P10-11, 7.9±2.1 Hz, n=64 units/5 mice) and responded to the first pulse of the stimulation train even more effectively than P14-16 cells (Peak FR in response to 1^st stimulus; P10-11, 37.6±5.4 Hz) (Fig. 3b–g). Compared to P14-16, P10-11 striatal units exhibited longer 20–25 ms response latencies (Supplementary Fig. 2) and more pronounced depression during the stimulation train (Fig. 3f). There was no significant difference in pair-pulse ratios (PPR) of electrically-evoked EPSC (eEPSC) in SPNs at P9-10 and P15-16 (Supplementary Fig. 3, WT PPR P9, 1.05±0.14, n=9 neurons/2 mice, P14, 0.97±0.05, n=13 neurons/3 mice) suggesting that the increased depression observed at P10-11 is not due to changes in presynaptic properties of striatal afferents but may be instead due to changes in AMPAR desensitization^29,30. Together, these results reveal that striatal units are already tightly coupled to cortical activity at P10 and can effectively respond to trains of APs.

Increased excitability of SPNs during early development

The robust striatal firing evoked by corticostriatal stimulation at P10-11 appears at odds with the much reduced number of SPN excitatory inputs during this early developmental period. To test if the early SPN responsivity is due to increased intrinsic SPN excitability³¹, we performed whole-cell recordings in current-clamp from SPNs in dorsomedial striatum across development and measured membrane potential changes and AP frequency in response to somatic current injections (Fig. 4a–f). The resting membrane potential (V_rest) at P10-11 was more depolarized than at later stages (P10-11, −68.7±1.1 mV, n=26 cells/3 mice; P13-14, −80.1±0.9 mV, n=26 cells/3 mice; P16-17, −80.6±1.4 mV, n=15 cells/3 mice; one-way ANOVA, p<0.0001). In addition, AP firing threshold (P10-11, −29.1±1.3 mV, n=26; P13-14, −34.5±1.3 mV, n=26; P16-17, −35.4±0.8 mV, n=15; one-way ANOVA, **p<0.01) and input resistance (R_i, P10-11, 438±14 MΩ; P12-13, 245±5 MΩ; P16-17, 151±4 MΩ) decreased during this period, whereas rheobase significantly increased with age (P10-11, 93±8 pA; P12-13, 159±11 pA; P16-17, 183±12 pA, one-way ANOVA, ***p=0.0002, ****p<0.0001). Thus, SPNs undergo extensive maturation of their intrinsic properties and are most excitable during early development.

Figure 4. Hyperexcitability of SPNs during early development.

(a) Example membrane responses to discreet current injection steps in SPNs of dorsomedial striatum.

(b) Mean resting membrane potential ± SEM and (c) Mean spike threshold potential ± SEM of SPNs recorded at different postnatal periods.

(d) Mean ± SEM current-voltage (I–V) relationship in SPNs across development. Dashed lines represent linear fits to voltage steps to 10, 25 and 50 pA whose slopes were used to determine the input resistance.

(e) Current-firing rate (I–F) plot of SPNs across development. Error bars represent ± SEM

(f) Mean SPN rheobase current ± SEM from P10-11 to P16-17.

Precocious SPN maturation in Shank3B^−/− mice

Shank3B^−/− mice display reduced corticostriatal connectivity in adults²³, suggesting that loss of Shank3 might impair the early development of striatal afferents. We recorded AMPAR mEPSCs from SPNs in dorsomedial striatum of wild-type (WT) and Shank3B^−/− littermates across development (Figs. 5a–h). Because of the rapid rate of cellular and synaptic maturation at these ages all comparisons were made within litters from mice recorded on the same day. At P10, there was no difference in SPN mEPSC frequency between genotypes (mEPSC frequency; WT, 0.37±0.04 Hz, n=25 cells/3 mice; Shank3B^−/−, 0.43±0.05 Hz, n=27 cells/3 mice, one-way ANOVA p>0.999). However, at P14 Shank3B^−/− SPNs had elevated mEPSC frequency compared to WT controls (mEPSC frequency; WT, 0.93±0.09 Hz, n=21 cells/4 mice; Shank3B^−/−, 1.71±0.29 Hz, n=22 cells/4 mice, one-way ANOVA *p=0.015). Furthermore, whereas in control mice the rate of SPN mEPSC increased gradually until adulthood, in Shank3B^−/− mice it plateaued at P30 (P30 WT, 1.97±0.18 Hz, n=16 cells/3 mice; P30 Shank3B^−/−, 1.98±0.20, n=15 cells/3 mice; P60 WT, 2.84±0.26 Hz, n=15 cells/3 mice; P60 Shank3B^−/−, 2.00±0.19 Hz, n=12 cells/3 mice). Interestingly, Shank3B^−/− SPNs exhibited significantly increased mEPSC amplitude relative to WT controls at P10, but not at any later stages (P10 WT, 16±1, Shank3B^−/−, 19±1 pA, one-way ANOVA, p=0.0009, P14 WT 18±1 pA; Shank3B^−/−, 19±1 pA; P30 WT, 17±1 pA; Shank3B^−/−, 15±1 pA; P60 WT, 16±1 pA; Shank3B^−/−, 17±1 pA). Importantly, PPR of eEPSC was not significantly different between genotypes at P14 (Supplementary Fig. 3b; WT, 0.97±0.05, n=13 cells/2 mice; Shank3B^−/−, 1.04 ± 0.04, n=14 cells/2 mice), suggesting that the higher mEPSC frequency observed in Shank3B^−/− SPNs is not due altered release probability.

Figure 5. Precocious maturation of striatal glutamatergic inputs in Shank3B^−/− SPNs.

(a) Representative mEPSC recordings in SPNs of dorsomedial striatum of WT and Shank3B^−/− mice at P14.

(b) Cumulative distribution of amplitude and (c) inter-event interval of mEPSCs recorded from SPNs of WT and Shank3B^−/− littermates at P14.

(d) Representative mEPSC recordings of WT and Shank3B^−/− SPNs at P60.

(e) Cumulative distribution of amplitude and (f) inter-event interval of mEPSCs recorded from WT and Shank3B^−/− litter mates at P60.

(g) Mean mEPSC frequency and (h) amplitude ± SEM of SPNs from WT and Shank3B^−/− animals at different developmental time points. WT maturation is characterized by a continuous increase in mEPSC frequency throughout development whereas Shank3B^−/− show a precocious maturation followed by an arrest in later stages.

(i) Experimental diagram depicting whole cell voltage clamp recordings in SPNs of dorsomedial striatum in acute brain slices of Shank3B^−/−;Rbp4-Cre;ChR2-YFP^f/wt mice and optogenetic fiber stimulation using whole field illumination. Scale bar, 1mm.

(j) Representative average AMPAR and NMDAR oEPSCs from WT (black) and Shank3B^−/− (red) SPNs. Blue rectangle represents 5ms 473nm light stimulation.

(k) Mean AMPAR and (h) NMDAR oEPSC peak amplitude ± SEM in P14 WT and KO SPNs.

(l) Mean AMPAR to NMDAR ratio ± SEM for SPNs represented in (k).

To compare overall excitatory drive onto WT and Shank3B^−/− SPNs we measured AMPAR and NMDAR oEPSCs in dorsomedial striatum of mice expressing Rbp4-Cre and ChR2 (Shank3^−/−;Rbp-Cre;ChR2^f/wt) (Figs. 5i–l). Rbp4⁺ oEPSC amplitude was increased in Shank3B^−/− SPNs compared to WT at P13-14 (WT 399±68 pA, n=21 cells/3 mice; Shank3B^−/− 710±110 pA, n=22 cells/3 mice, unpaired t-test p=0.022). This relative difference was also observed between heterozygote and Shank3B^−/− mice (Supplementary Fig. 4, Shank3B^+/−, 386±48 pA, n=37 cells/3 mice; Shank3B^−/− 541±45 pA, n=40 cells/3 mice, unpaired t-test p=0.021). The increase in AMPAR oEPSC amplitude in Shank3B^−/− SPNs also resulted in increased R_A/N (oEPSC R_A/N WT, 0.86±0.01, n=14, Shank3B^−/− 1.21±0.12, n=16, unpaired t-test, p=0.036) consistent with a larger fraction of synapses exhibiting AMPAR currents in relation to controls. Taken together, these results show a premature increase and subsquent arrest in the development of SPN excitatory inputs in Shank3B^−/− mice when compared to WT.

Early corticostriatal hyperactivity in Shank3B^−/− mice

To characterize corticostriatal circuit activity in Shank3B^−/− mice during early development we performed multi-unit recordings in cortex and dorsomedial striatum of awake P13-14 WT and Shank3B^−/− littermates (Figs. 6a–f). Cortical activity was elevated in Shank3B^−/− animals ~2-fold compared to WT (Median FR WT 0.73 Hz, n=154 units/4 mice; Shank3B^−/−, 1.34 Hz, n=155 units/4 mice, Mann Whitney U test, p=0.0006). The frequency of cortical AP bursts was not statistically different between genotypes, but the intra-burst average FR was significantly increased in the cortex of Shank3B^−/− animals (Median burst frequency: WT 0.8 min⁻¹, n=154, Shank3B^−/−, 0.8 min⁻¹, n=155; Median intra-burst frequency: WT 13.8 Hz, n=102, Shank3B^−/−, 27.9 Hz, n=121; Mann Whitney U test, p< 0.0001).

Figure 6. Cortical hyperactivity in neonatal Shank3B^−/− mice.

(a) Experimental diagram of in vivo recordings in a sagittal view of a mouse brain showing cortex (CTX) and striatum (STR).

(b) Representative recordings of multi-unit activity in cortex and (c) striatum of WT and Shank3B^−/− animals at P13-14.

(d) Median ± interquartile range of average FR of cortical units from WT and Shank3B^−/− mice.

(e) Median frequency of AP bursts and (f) Intra-burst frequency ± interquartile range of cortical units shown in (d).

(g) Median ± interquartile range of average FR of striatal units from WT and Shank3B^−/− mice.

(h) Median frequency of AP bursts and (i) intra-burst firing rate ± interquartile range of cortical units shown in (e).

Interestingly, striatal activity was also increased in Shank3B^−/− mice compared to WT (Median FR: WT 0.27 Hz, n=155 units/4 mice; Shank3B^−/−, 0.48 Hz, n=144 units/4 mice; Mann Whitney U test, p=0.002). Moreover, in striatum of Shank3B^−/− mice the frequency of AP bursts (Median burst frequency: WT 0.6 min⁻¹, n=155; Shank3B^−/−, 1.2 min⁻¹, n=144; Mann Whitney U test, p<0.0001) and the intra-burst AP frequency (Median intra burst frequency: WT 16.85 Hz, n=125; Shank3B^−/−, 20.53 Hz, n=138; Mann Whitney U p=0.0209) were also increased (Figs. 6g–h). Importantly, we detected no difference in SPN R_i, V_rest, rheobase or spike threshold potential between genotypes (Supplementary Fig. 5, V_rest, WT −79.3±2.4 mV, Shank3B^−/− −81.5±1.8 mV; Spike Threshold, WT −38.0±0.8 mV, Shank3B^−/− −37.4±1.0 mV; Rheobase, WT −155±19 pA, Shank3B^−/− 175±17 pA) indicating that increased striatal activity observed in KO animals is not due to increased SPN excitability.

Cortical activity drives corticostriatal connectivity

To directly test whether elevated cortical activity can increase corticostriatal connectivity, we silenced cortical GABAergic interneuron output by conditional deletion of the vesicular GABA transporter (vGAT) in Slc32a1^f/f transgenic mice. Recombination was achieved by unilateral injection of AAVs expressing Cre recombinase under control of the neuron specific synapsin promoter (AAV-Syn-Cre-GFP) in frontal, motor and somatosensory cortex at P3-4 (Fig. 7a). Injected animals did not present signs of seizures but exhibited epileptiform spike and wave activity patterns in local field potential recordings of injected cortical regions (Fig. 7b–c) and overall movement deficits in an open chamber test (Supplementary Fig. 6; Distance moved: control 1.70±0.40 m, Cre injected 0.84±0.21 m, t-test p=0.044; Velocity: control 1.96±0.37 cm/s, Cre injected 0.99±0.20 cm/s, t-test p=0.037; Time moving: control 244.5±66.3 s, Cre injected 109.7±32.8 s, t-test p=0.040, n=4 mouse pairs).

Figure 7. Elevated cortical activity during early development increases corticostriatal connectivity.

(a) Silencing of cortical interneuron output was achieved by injecting Cre expressing adenovirus in the cortex of Slc32a1^f/f animals at P4.

(b) Local field potential (LFP) recordings from cortex of control and AAV injected animals at P14 show epileptiform patterns of activity after VGAT deletion.

(c) Spectrogram of LFPs shown in (b). Scale bar, 1 min. Color scale represents normalized power.

(d) Example mEPSC recordings in SPNs of dorsomedial striatum of control and AAV-Cre injected animals.

(e) Cumulative distribution of mEPSC amplitude and (f) mEPSC inter-event interval values for the total pool of mEPSCs recorded from control (black) and Cre injected (red) litter mates at P12-14.

(g) Cell average mEPSC frequency and (h) amplitude ± SEM of SPNs from control and Cre injected animals.

(i) Schematic showing optogenetic cortical stimulation using extracranial implant of a low mass LED (blue) in Rbp4-Cre; ChR2^f/f animals (top panel) and subsequent oEPSC measurements in SPNs in the ipsilateral (stimulated) and contralateral (control) hemispheres (bottom panel).

(j) Example AMPAR oEPSCs recorded in SPNs located in dorsomedial striatum of the stimulated (ipsi, red) or opposite (contra, black) hemisphere in response to 5 ms pulses of 473 nm laser light (blue rectangle).

(k) Mean oEPSC amplitude ± SEM of control (contra) and stimulated (ipsi) SPNs.

(l) Pair wise comparison of average oEPSC amplitude in animals recorded in (k)

SPNs in dorsomedial striatum of injected P12-14 vGAT^f/f animals had 2-fold higher mEPSC frequency compared to controls (Fig. 7d–h; control 0.63±0.06 Hz, n=17 cells/3 mice; Cre injected 1.27±0.17, n=16 cells/3 mice; t-test p=0.001) with no significant change in mEPSC amplitude (control 19±1 pA, n=17; Cre injected 21±1, n=16). PPR of eEPSCs were not statistically different between experimental groups (Supplementary Fig. 7; eEPSC PPR; control, 0.90±0.09, n=10 cells/2 mice; Cre injected, 1.06±0.12, n=9 cells/2 mice) indicating that the increase in mEPSC frequency was likely not due to altered vesicular release probability.

To test whether increased cortical activity can acutely alter corticostriatal connectivity we performed extracranial optogenetic stimulation of corticostriatal projections from frontal regions of cortex by implanting a lightweight 470 nm LED unilaterally onto the skull of P10-11 of Rbp4-Cre;ChR2^f/f mice³² (Fig. 7i). Mice were stimulated with 5 pulses at 10 Hz every minute for one hour corresponding to a ~25% net increase over basal cortical AP firing and a ~5 fold increase in AP burst frequency (Figs. 2c–d). Acute brain slices were prepared immediately after stimulation and AMPAR-mediated oEPSCs were measured from SPNs located in dorsomedial striatum of the ipsilateral (stimulated) or contralateral (control) hemisphere (Figs. 7j–l). Ipsilateral SPNs exhibited elevated oEPSC amplitude compared to contralateral controls (Average oEPSC amplitude; ipsilateral 737±63 pA, n=26 cells/4 mice; contralateral 532±67 pA, n=25 cells/4 mice, t-test p=0.031) and this relationship was found in all mouse pairs studied (Mean difference of oEPSC amplitude: 200±39 pA, n=4 mice, paired t-test p=0.014). These results indicate that elevated cortical activity during early development rapidly enhances SPN excitatory input and suggest that the corticostriatal circuit abnormalities observed in developing Shank3B^−/− mice may be secondary to increased cortical activity.

Decreasing cortical activity rescues Shank3B^−/− phenotype

To directly test whether the abnormal development of striatal connectivity observed in Shank3B^−/− mice is caused by cortical hyperactivity, we reduced cortical activity during development and measured AMPAR mEPSCs in SPNs at P13-14. To gain spatio-temporal control over the activity of layer 5 corticostriatal projecting neurons, we bilaterally injected AAVs expressing Cre-dependent G_i-coupled inhibitory DREADD (hM4D_i-mCherry) into the cortex of Shank3B^−/−;Rbp4-Cre animals during the first postnatal days. This strategy resulted in expression of hM4D_i exclusively in Rbp4⁺ layer 5 cortical cells (Figs. 8a–b). Clozapine n-oxide (CNO, 1 mg/kg) was administered subcutaneously twice a day for 3 days to activate hM4D_i and decrease corticostriatal activity. SPNs of Shank3B^−/−;Rbp4-Cre animals injected with CNO had lower mEPSC frequency than saline injected controls (Figs. 8c–e, mEPSC frequency: Saline 1.23±0.1 Hz, n=25 neurons/3 mice, CNO 0.78±0.07 Hz, n=24 neurons/3 mice, unpaired t-test, p=0.0006; mEPSC amplitude: Saline 17±1 pA, CNO 17±1 pA).

Figure 8. Early increase in corticostriatal drive in Shank3B^−/− mice is due to cortical hyperactivity.

(a) Schematic representing bilateral injection of AAV8-DI-hM4D_i into cortex of Shank3B^−/−; Rbp4-Cre mice at P1-2 and bi-daily administration of CNO for 3 days before mEPSC recordings at P13-14.

(b) Coronal brain slice of P13 Shank3B^−/−;Rbp4-Cre mouse infected with AAV8-DI-hM4D_i-mCherry. Ctx- cortex, Str- striatum. Scale bar, 1 mm.

(c) Example mEPSC recordings in SPNs of dorsomedial striatum of saline or CNO injected animals.

(d) Cell average mEPSC frequency and (e) amplitude ± SEM of SPNs of saline or CNO injected animals.

Discussion

Despite evidence implicating BG dysfunction in neurodevelopmental disorders, the rules underlying the early development of subcortical circuits and the contribution of these processes to human disease remain unclear. In mice, glutamatergic afferents innervate striatum at P3-4³³ and, as revealed here, evoke synaptic responses in SPNs by P6 indicating that SPN synaptogenesis starts soon after axon innervation. The development of SPN excitatory inputs progresses gradually up to ~P10, and undergoes a period of rapid maturation from P10-18. This period of accelerated synaptogenesis is accompanied by a significant increase in cortical and striatal neuronal activity, possibly due to maturation of sensory systems and increases in thalamocortical input^34,35. Nevertheless, our results indicate that despite the much reduced corticostriatal drive and synapse density of immature SPNs, striatum can respond to cortical activity at P10. This early responsivity is due at least in part to elevated SPN V_rest and R_i that decrease progressively during development, similar to other neuronal cell types³⁶. These adaptations likely arise from changes in ion channel expression that optimally tune neuron responses to early patterns of activity^31,37 and contribute to activity-dependent maturation of sensorimotor circuits during early postnatal periods³⁸. Moreover, the tight functional coupling of cortex and striatum suggests that early imbalances in cortical activity can alter the trajectory of corticostriatal circuit maturation. Indeed, we found elevated cortical activity and both SPN mEPSC frequency and oEPSC amplitude in P13-14 Shank3B^−/− SPNs compared with WT. Importantly, this phenotype was rescued by chemogenetic reduction of cortical activity indicating that it is secondary to cortical hyper-activity. Moreover, this finding is unexpected as adult Shank3B^−/− mice have reduced corticostriatal drive²³ and Shank3 expression levels are correlated with glutamatergic synapse strength and number in several other in vivo and in vitro systems^{14,17,18,21–24,39}.

A recent study reported decreased number of Parvalbumin⁺ (PV) terminals in cortex of Shank3B^−/− mice⁴⁰. Given the large increase in thalamocortical afferent activity during early development^34,36,38 decreased inhibition could result in cortical hyper-activity, which would in turn enhance corticostriatal synaptogenesis. This hypothesis is consistent with the increase in SPN mESPC frequency resulting from either vGAT deletion in cortex or acute optogenetic stimulation of corticostriatal projection neurons. Moreover, these results indicate that both chronic and acute periods of elevated cortical activity can increase corticostriatal connectivity.

Although the ontogeny of BG circuits in humans remains unknown, the maturation of SPNs in Rhesus monkeys occurs gradually throughout the first year of life, which is analogous to the toddler/preschool stage in humans^3,41. Autistic individuals share a core set of behavioral symptoms suggesting the multiple genetic abnormalities underlying ASD may converge onto common neurological mechanisms. Notably, one third of autistic individuals develop seizures during infancy and an even larger fraction exhibit epileptiform EEG activity⁴², consistent with the cortical hyperactivity exhibited by several transgenic mouse lines carrying ASD-associated gene mutations^40,43–46.

Our results indicate that in mice, abnormal hyperactivity of corticostriatal afferents during a period of high SPN excitability alters the normal course of BG circuit development. If similar activity-dependent rules are conserved in humans, this anomalous developmental pattern could help explain why children with autism develop repetitive behaviors and often present precocious maturation and posterior regression of certain cognitive abilities. It is important to note that Shank3B^−/− mice only exhibit autistic-like behaviors such as over-grooming during adulthood, and the relationship between the early phenotype described here and the onset of these behaviors remains unknown. However, repetitive behaviors in autistic children emerge during the first 2 years of age^47,48 and often become persistent throughout life⁴⁹. Likewise, motor stereotypies associated with ADHD, OCD and Tourette syndrome exhibit a similar early developmental onset⁵⁰, suggesting that they are established during the early stages of BG circuit development. Thus, besides providing an experimental framework for characterizing circuit formation and cross-circuit interactions during mouse development, this study reveals mechanisms that may be implicated in the pathophysiology of multiple neurodevelopmental disorders.

Online Methods

Mice

All experimental manipulations on mice were performed in accordance with protocols approved by the Harvard Standing Committee on Animal Care and guidelines described in the US National Institutes of Health Guide for the Care and Use of Laboratory Animals. For electrophysiology studies, Rbp4-Cre transgenic mice (GENSAT #RP24-285K21) were bred to conditional channelrhodopsin-2 (ChR2) expressing mice expressing ChR2(H134R)-EYFP under control of an upstream loxP-flanked STOP cassette (Ai32; referred to as ChR2^f/f; The Jackson Laboratory #012569). For conditional deletion of the vesicular GABA transporter in cortical cells we used Slc32a1^f/f mice (The Jackson Laboratory #012897). Shank3B^−/− knock-out mutant mice were described previously²³ and obtained from The Jackson Laboratory (#017688). All experiments using Shank3B^−/− mice were performed in age matched littermates from breeding pairs between Shank3B^+/− hererozygous animals. For optogenetic studies triple transgenic Shank3^+/−; Rbp4-Cre; ChR2^f/wt animals were cross bred in order to obtain Cre and ChR2 expression in both heterozygous and Shank3B null backgrounds. In all experiments male and female mice were used.

Viruses and stereotaxic intracranial injections

For intracranial injections, P0-7 day old mice were anesthetized with cold or isofluorane and placed into a stereotaxic apparatus. Viruses were delivered by injecting 100–200nl at a maximum rate of 100 nl/min using a UMP3 microsyringe pump (WPI). For dendritic spine analysis of SPNs, AAVs expressing EGFP under the ubiquitous CAG promoter (AAV8-CAG-EGFP, Penn vector core) were injected in dorsomedial striatum using coordinates: 1 mm anterior to midpoint between ear and eye, 1 mm lateral from midline and 2.5 mm ventral to brain surface. For deletion of vGAT in cortical neurons, virus expressing Cre-EGFP under the neuron specific synapse promoter (AAV9-Syn-Cre-EGFP, Penn vector core) were injected in using coordinates: +1 mm and +2 mm anterior to midpoint between ear and eye, 1mm lateral from midline, 0.5 mm ventral to brain surface. For chemogenetic reduction of cortical activity, P1-2 Shank3B^−/−; Rbp4-Cre mouse pups were injected bilaterally with AAV8-DI-hM4D_i-mCherry using the following coordinates: hM4Following injections and wound closure, mice received ketoprofen for analgesia and were returned to home cages for 8+ days.

Brain tissue processing and imaging

Mice were deeply anesthetized with isoflurane and perfused transcardially with 4% paraformaldehyde in 0.1 M sodium phosphate buffer. Brains were fixed for 24 hours at 4°C, washed in phosphate buffer saline (PBS) and sectioned (50 μm) coronally using a vibratome (Leica VT1000s). Brain sections were mounted on glass slides, dried and mounted with ProLong antifade reagent containing DAPI (Molecular Probes). Whole brain sections were imaged with an Olympus VS110 slide-scanning microscope. For dendritic spine analysis high-resolution images of regions of interest were subsequently acquired using an Olympus FV1000 confocal microscope (Harvard Neurobiology Imaging Facility). Confocal stacks were acquired with a 63x objective and 0.75 μm spacing in Z. Confocal images were processed and analyzed using ImageJ software.

Acute slice preparation and electrophysiology

Acute brain slices and whole-cell recordings from SPNs were performed using standard methods, as described previously ⁶. Briefly, mice (6–60 days old) were anesthetized by isoflurane inhalation and perfused transcardially with ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM): 125 NaCl, 2.5 KCl, 25 NaHCO₃, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄ and 25 glucose (310 mOsm per kg). Cerebral hemispheres were removed, placed in cold choline-based cutting solution consisting of (in mM): 110 choline chloride, 25 NaHCO₃, 2.5 KCl, 7 MgCl₂, 0.5 CaCl₂, 1.25 NaH₂PO₄, 25 glucose, 11.6 ascorbic acid, and 3.1 pyruvic acid), and transferred into a slicing chamber containing ice-cold choline-based solution. Coronal slices including striatum (275 μm thick) were cut with a Leica VT1000s vibratome, transferred for 10 min to a holding chamber containing ACSF at 34°C and subsequently maintained at room temperature (20–22°C). All recordings were obtained within 5 h of slicing. Both cutting solution and ACSF were constantly bubbled with 95% O₂/5% CO₂. Individual slices were transferred to a recording chamber mounted on an upright microscope (Olympus BX51WI) and continuously perfused (1–2 ml per minute) with ACSF at room temperature. Cells were visualized using a 40× water-immersion objective with infrared DIC optics. Whole-cell voltage- and current-clamp recordings were made from SPNs in dorsomedial regions of striatum. Patch pipettes (2–4 MΩ) pulled from borosilicate glass (BF150-86-7.5, Sutter Instruments) were filled either with a Cs⁺-based internal solution containing (in mM) 130 CsMeSO₄, 10 HEPES, 1.8 MgCl₂, 4 Na₂ATP, 0.3 NaGTP, and 8 Na₂-phosphocreatine,10 CsCl₂, 3.3 QX-314 (Cl⁻salt), (pH 7.3 adjusted with CsOH; 295 mOsm per kg) for voltage-clamp recordings, or with a K⁺-based low Cl⁻ internal solution composed of (in mM) 130 KMeSO₃, 3 KCl, 10 HEPES, 1 EGTA, 4 Mg-ATP, 0.3 Na-GTP, 8 Na₂-phosphocreatine (pH 7.3 adjusted with KOH; 295 mOsm) for current-clamp recordings. For all voltage-clamp experiments, errors due to voltage drop across the series resistance (<20 M ) were left uncompensated. For current clamp recordings all voltages reported are corrected for a junction potential of ~8 mV. In all experiments GABAR currents were blocked with 20μM SR95531 hydrobromide (Tocris) to eliminate inhibition from local interneuron or collateral SPNs. For light-evoked AMPAR EPSC recordings ACSF contained 10 μM (R)-CPP and recordings were performed at −70 mV holding potential except for ages >P15 in which −20 mV holding potential was used to maximize voltage control. For mEPSC recordings, ACSF contained 1 μM TTX besides CPP and TTX. For current clamp recordings membrane potentials were corrected for a ~8 mV liquid junction potential. Voltage clamp recordings were performed at room temperature (20–22°C) and current clamp recordings at 32°C. To activate ChR2-expressing fibers, light from a 473 nm laser (Opto engine LLC) was focused on the back aperture of the microscope objective to produce wide-field illumination of the recorded cell. TTL triggered pulses of light (5 ms duration; 6 mW/mm² under the objective) were delivered at the recording site at 30 s intervals.

In vivo recordings and optogenetic stimulation in mouse pups

Mouse pups (P10-16) were anesthetized with isoflurane and placed in a stereotaxic apparatus. After surgical removal of scalp and cleaning of the skull with saline and 70% ethanol, two craniotomies were made with a 0.25 mm burr micro drill at (AP- 0 mm; ML-2.0 mm and AP-0.3 mm; ML- +1.5 mm; from bregma) and sealed with Kwik-Cast silicone Elastomere. Animals were fitted with a custom-made Titanium head bar using transparent glue (Loctite 454) and allowed to recover from anesthesia for 1 h on a heat pad at 38°C. Following recovery from anesthesia, animals were head fixed and in vivo electrophysiological recordings were performed using 32 channel probes (A1x32-Poly2, 177 μm² site area, NeuroNexus Technologies) with recording sites spanning 750 μm. Cortical recordings were done by positioning the electrode tip 1250 μm deep from the brain surface, whereas striatal recordings were performed at a 2750 μm depth. Sessions of 20 min were recorded after a 10 min stabilization period, with ChR2 stimulation initiated at t=10 min. Extracranial optogenetic stimulation was achieved by coupling a 200 μm core optical fiber placed 7–8 mm from the craniotomy site at a 45° angle to a 473 nm laser (Ciel, Laser Quantum). Light pulses were controlled with an Acousto-Optic Modulator (AA opto-Electronic) for fast shuttering and intensity control. Final light power was 45–50 mW and pulse width was 5 ms. All recordings were validated by post hoc serial histological analysis of electrode placement. For in vivo unilateral optogenetic stimulation a bare surface mount emitter blue LED (470 nm) with flattened epoxy lens was attached directly to the surface of the skull using transparent glue following a similar surgical procedure. In these experiments brain slices were prepared immediately after completion of the stimulation protocol.

Data acquisition and analysis

For whole cell recordings, membrane currents and potentials were amplified and low-pass filtered at 3 kHz using Multiclamp 700B amplifier (Molecular Devices), digitized at 10 kHz and acquired using National Instruments acquisition boards and a custom version of ScanImage written in MATLAB (Mathworks). Off-line analysis was performed using custom routines written in MATLAB and Igor Pro (Wavemetrics). Statistical analyses were done using GraphPad PRIZM 5 software (GraphPad). For in vivo multi-unit recordings, data was filtered at 1–8000 Hz during acquisition, digitized at 40 kHz and sorted into single unit activity via principal component analysis using Plexon Inc.’s Omniplex and Offline Sorter systems, respectively. Data were further analyzed using custom routines in IGOR Pro and Neuroexplorer v5. Burst analysis was performed using an unbiased statistical method based on inter stimulus interval (ISI) “surprise” factor that takes into account the differential Gaussian distribution of intra-burst AP ISIs compared to ISIs of all APs in the recorded train⁵¹. AP clusters were considered bursts if the number of APs was >4 and surprise factor >5. All multi-unit activity data are represented as median ± interquartile range or mean ± standard error of the mean (SEM) and statistical tests utilized for each data set are specified in the main text. For two-group comparisons, statistical significance was determined by two-tailed Student’s t-tests (parametric) or Mann-Whitney U tests (non-parametric). Multi-groups were analyzed using one-way ANOVA with Tukey correction (parametric) or Kruskal-Wallis with Dunn’s multi comparison correction (non-parametric). P values less than 0.05 were considered statistically significant.