Introduction

Deficits in social interaction are a hallmark symptom of autism spectrum disorder (ASD) and other neuropsychiatric disorders. ASD genomic studies have identified a large number of genes that encode proteins involved in synaptic development and functions, suggesting a potential converging mechanism1,2,3,4,5,6,7,8,9,10,11,12. SHANK3 encodes a postsynaptic density scaffold protein and is one of the most common causal genes for autism, found in ~2% of patients, and has been extensively modeled in animals13,14,15.

To study the role of SHANK3 protein on synaptic function and behavior, we previously developed the Shank3 complete deletion model with an exon 4 to 22 deletion (Shank3Δe4-22)16. We demonstrated that Shank3Δe4-22 mice exhibit increased unidirectional social behavior in a naturalistic setting, are profoundly impaired in reward-seeking behavior in a lever-press task for food pellets, and display anxiety-like behavior compared to wild-type (WT) controls16. These data corroborate extensive literature showing that removing Shank3 isoforms impacts social behaviors, indicating SHANK3 protein expression is a fundamental component underlying the neural mechanisms of social interaction17,18,19,20,21,22,23. Yet, how Shank3 in specific brain regions coordinates social behaviors, motivation, and anxiety-like behavior is not well understood.

SHANK3 protein is highly expressed in the nucleus accumbens (NAc)23,24,25,26, a well-established regulator of reward, social interaction, and social motivation27,28,29,30,31,32,33. Imaging studies in individuals with ASD reveal altered NAc responses to social cues34,35,36, and many autism mouse models show critical NAc activity changes37,38,39,40,41,42,43, indicating it is an essential region for regulating social behaviors relevant to ASD. We previously demonstrated that the NAc and NAc neuronal networks have a blunted response to social stimuli in Shank3Δe4-22 mice16, and others have shown that removing Shank3 only from the NAc is sufficient to drive lower NAc firing rates26. This highlights the importance of the NAc as a central hub for the loss of social and reward-seeking behaviors in Shank3Δe4-22 mice; however, investigation into the role of SHANK3 in the NAc on social and reward-seeking behavior is limited.

Here, we tested the hypothesis that SHANK3 protein in the NAc is critical for driving reward behavior and neural deficits described in Shank3Δe4-22 mice. Our data indicate that removing Shank3 from the NAc selectively impairs social motivation but does not alter food motivation. This starkly contrasts with conventional Shank3Δe4-22 mice, which prefer a socially-paired chamber after conditioning and have a complete deficit of food reward-seeking in an instrumental conditioning assay. Our data suggest that Shank3 in the NAc underlies social motivation independent from other reward-seeking behaviors.

Materials and methods

Animals

Male and female mice were group housed by sex on a 12-hour light/ 12-hour dark cycle with lights on at 0700 hours with one nestlet in Tecniplast GM500 cages. All mice used were on a C57/Bl/6 J mouse line and were between 6 and 26 weeks old. Specific ages can be found in Supplementary Data 1. All cages were assigned to the same rack location in the animal facility, which remained consistent through all cohorts tested. Food and water were available ad libitum, except during rewarded lever press experiments (see experimental details). All animals were monitored daily in their home cages by experimenters and Yale Animal Resources Center veterinary staff for activity, signs of stress and pain, weight, and skin lesions. If any adverse health or activity events were observed, mice were treated by veterinary staff and were not used for this study. Pertinent to these studies, animals were monitored during testing for aggression and weight during food deprivation used in the lever press task. If mice received skin lesions from aggression or fell below our weight loss parameters, they were removed from this study; however, this endpoint was not met. All experiments were conducted during the light phase. We have complied with all relevant ethical regulations for animal use. Animal husbandry and all animal protocols received ethical approved by the Yale Animal Resources Center and the Institutional Animal Care and Use Committee at Yale.

The Shank3fl/fl mouse line was engineered via CRISPR/Cas genome editing. In our previous Shank3 e4-22flox/flox mouse line44, three loxP sites were located between intron 3, 9, and downstream of exon 22 of Shank3 genes (GenBank:NM_033517). A sgRNA was designed to target the second loxP site in Shank3 e4-22flox/flox mouse line. The validation of sgRNA was done in vitro. After pronuclear injection of the Cas9 and sgRNA vector, PCR and genomic sequencing were done to confirm the deletion of the second loxP sequence and the presence of the first and third loxP sequences in the founder mouse (Supplementary Fig. 1). The founder mouse was then bred with a C57/BL6 mouse to generate heterozygous offspring. The litters were genotyped using PCR, and the pups containing two loxP sites were backcrossed with C57/BL6 WT mice for 5 generations to eliminate any potential off-target events. Shank3fl- x Shank3fl- mating pairs were then used to establish Shank3fl/fl line. No unusual phenotypes were observed in the animals. After ~10 generations of breeding, we used homozygote breeding (Shank3fl/fl x Shank3fl/fl) pairs to maintain the mouse line and obtain the experimental mice.

Shank3Δe4-22 heterozygous mice were bred in the Jiang lab to obtain global Shank3Δe4-22 deletion and WT littermate controls for all experiments. WT targets for social behavior testing were C57BL/6 J mice obtained from Jackson Laboratories.

Genotyping and PCR confirmation of loss of Shank3

Shank3fl/fl mice were genotyped using the following protocol: Primers FLP-Neo-F: (5’-gggaggattgggaagacaat-3’), SH-AFLP-R (5’-ggctatgttcatgggatcttgt-3’), and SH-BFLP-F: (5’-ttgccgaggtaatcaagacc-3’) were used to amplify WT (390 bp) and Shank3fl/fl alleles (211 bp). To identify recombination of the loxP sites (Δe4-22), primers SH3-MTF (5’-ttgcatctgggacctactcc-3’), SH3-MTR (5’-aaagcactgactcctctcttgg-3’), SH3-WTF (5’ gtgccacgatcttcctctaaac-3’), SH3-WTR (5’ agctggagcgagataagtatgc-3’) were used to amplify a 200 bp product when the floxed allele was intact and a 600 bp product when recombination had occurred with a 30 s denaturing step at 94 °C, 30 s annealing step at 59 °C, and a 45 s extension step at 72 °C for 40 cycles. This protocol was also used to identify Shank3Δe4-22 mice16.

Viruses

Shank3fl/fl mice were injected with 300nL AAV5-CMV-eGFP-Cre (AAV.CMV.HI.eGFP-Cre.WPRE.SV40 was a gift from James M. Wilson; Addgene viral prep # 105545-AAV5; http://n2t.net/addgene:105545; RRID:Addgene_105545) or 300nL AAV5-CMV-eGFP (AAV.CMV.PI.EGFP.WPRE.bGH was a gift from James M. Wilson; Addgene viral prep # 105530-AAV5; http://n2t.net/addgene:105530; RRID:Addgene_105530)

Stereotaxic surgery

Male and female Shank3fl/fl mice underwent bilateral stereotaxic surgery at 6-12 weeks of age. Animals were initially anaesthetized at 5% isoflurane and administered Ethiqa XR 1.3 mg/mL at 0.6 mg/kg (S.Q., Covetrus) for preemptive analgesia. Animals were then moved to the stereotaxic frame (RWD) and anesthesia was maintained on 1% isoflurane, 1% oxygen. Following a craniotomy, viruses were loaded into a NanoFil 10uL syringe equipped with a 33-gauge blunt needle tip (World Precision Instruments). The needle was moved into the NAc (anterior/posterior [AP]: −1.50 mm, medial/ lateral [ML]: +/− 0.75 mm, dorsal/ ventral [DV]: 4.50 mm), paused for 1 minute, then moved to DV 4.45 mm and the virus was infused into the NAc at a rate of 100nL/ minute with an automated infuser (Harvard Instruments). Following the infusion of the virus, the syringe remained in place for 10 minutes and was then slowly removed from the brain. Following bilateral injection, the incision site was sutured, and mice were given 1 mL of saline S.Q. and 5 mg/kg Carprofen I.P. Mice received 5 mg/kg injections of carprofen every 24 hours for 72-hour recovery monitoring. Mice recovered for 3-4 weeks before any experiments. Viral expression targeting is represented in Supplementary Fig. 2A and B.

Immunoblot

Tissue

4 weeks after viral infusion, NAc brain tissue was collected. Mice were anesthetized using isoflurane in a bell jar, then rapidly decapitated, and the brain was extracted on ice. The brain was placed in a cold brain block (Harvard Instruments) and sections containing the NAc were isolated between razor blades. NAc sections were frozen on dry ice and the NAc was isolated using a 2 mm biopsy punch (Integra Miltex) and stored in an Epindorf tube in −80 °C.

Preparation of crude synaptic proteins

The NAc tissue was homogenized in Syn-PER Synaptic Protein Extraction Reagent solution (Thermo Scientific) with protease and phosphatase inhibitors. Homogenate was centrifuged at 1200 x g for 10 minutes at 4 °C. The supernatant was separated and centrifuged at 15000 x g at 4 °C for 20 minutes. This synaptosome pellet (P2) was suspended in Syn-PER Reagent, and the cytosolic fraction (S2) was saved.

Quantitative immunoblot analysis

Immunoblot analysis was adapted from previously reported methods16,44,45. Here, 20ug of protein from the synaptosome and cytosolic fraction were separated by SDS-PAGE on a precast 4-20% 15-well gel (BioRad), then transferred to PVDF membranes. The membranes were blocked for 1 hour in 0.02 M Tris-Buffered Saline (TBS) with 0.1% Tween-20 (TBST) and 5% non-fat milk solution then incubated overnight at 4 °C in primary antibody [Shank3, rabbit (Courtesy of Yong Q. Zhang lab at Hubei University, 1:1000) or β actin HRP conjugated (Cell Signaling Technology #5125S 1:10000)]. The blots were then washed with TBST 3 times for 10 minutes each and incubated with HRP-conjugated secondary antibody [Anti-rabbit IgG, HRP-linked Antibody (Cell Signaling #7074, 1:3000)] for 1 hour at room temperature. The membrane was then washed 3 times for 10 minutes in TBST. The membrane was incubated with ECL reagent for 1 minute and imaged using a Chemidoc (BioRad). Band sizes were analyzed using ImageJ and normalized to β-actin loading control.

Histology and imaging

Tissue

After completion of the behavior experiments, brain tissue was collected. Mice were anesthetized using isoflurane in a bell jar and transcardially perfused with phosphate buffered saline (PBS, 10 mL) followed by 4% paraformaldehyde in 0.1 phosphate buffer (PFA, 15-20 mL). Brains were harvested and stored overnight in 4% PFA and transferred the following day to a 30% sucrose solution for at least 48 hours. Brains were cut at 80uM using a Leica CM3050 S cryostat or a Leica SM2000R microtome and slices were stored in an antifreeze, ethylene-glycol solution and stored at −20° until analysis.

Viral validation experiments

Tissue was transferred to cell strainers in a bath of fresh PBS for 10 minutes 4 times. Slices were then mounted onto charged slides and mounted in Dapi-Fluromount-G (SouthernBiotech) and sealed with a coverslip and clear nail polish. Images were collected at 10x objective on a Zeiss 980 and manually reviewed for viral expression by a blind experimenter, which the lead author later unblinded.

Immunohistochemistry

Tissue was transferred to cell strainers and washed in a fresh bath of PBS for 10 minutes 4 times. Next, slices were transferred to a PBS + 0.3% Triton-X 100 solution and washed for 60 minutes. Slices were then washed in a blocking buffer (0.3% Triton-X 100 and 10% Blocking One (Nacalai USA, Inc.) in PBS filtered with 0.22uM filtering unit) for 60 minutes. Slices were then transferred to primary antibody solution overnight at room temperature. SHANK3 (Cell Signaling Technology #64555) solution was made at 1:1000 concentration in blocking buffer. The next day, the slices were washed 4 times in fresh PBS for 10 minutes each and then were incubated in Alexa 647 anti-rabbit (Fisher Scientific A-21245) at 1:1000 for 3 hours at room temperature. Slices were again washed 4 times for 10 minutes in fresh PBS then incubated in DAPI (1:1000, VWR) for 5 minutes and washed another 4 times for 10 minutes each in fresh PBS. Finally, tissue was mounted onto charged slides using Fluoromount-G (EMS Acquisition) and cover slips were sealed with clear nail polish.

Imaging

Images of slices were taken using a Zeiss 980 using 10x, 20x and 63x objective. Slices were compiled into a tiled image on one plane. For SHANK3 expression analysis, images were first deidentified so the experimenter could remain unbiased, and the NAc brain region was identified using a Mouse Brain Atlas (Paxinos and Franklin). We then calculated corrected total cell fluorescence using ImageJ (corrected total cell fluorescence = Integrated Density – (Area * Mean Fluorescence of background)). All measurements were standardized as a percent change from total average eGFP readings. For each mouse, we used each hemisphere as an individual point (See Supplementary Data 1). Data was later unblinded by the lead investigator for analysis.

Behavior testing

Behavioral experiments were conducted across 11 cohorts of mice. The details of each cohort’s testing order are reported in Supplementary Data 1. Behavior videos were recorded and analyzed using Noldus EthoVision XT (Leesburg, VA USA). All test and target mice were handled and tail marked for at least 3 days prior to testing. On the day of testing, animals were acclimated to the testing room for at least 1 hour prior to experiments. All equipment was cleaned prior to and in between trials using 70% etoh spray.

3-chamber social investigation test

We used a 3-chamber arena comprised of a 420 mm x 565 mm x 358 mm box with 185 mm x 420 mm equal chambers made of 5 mm thick, clear Plexiglas with two doors (120 mm wide, 5 mm thick, 358 mm tall) that connected chambers to the middle (Yale Machine Shop). The arena was outfitted with two empty wire pencil cups (Organize-it) that were flipped upside-down in the top corners of the arena with a 5 cm path around each cup. Full water bottles were placed on top of the cups to secure them in place. Test mice were placed in the center chamber for 5 minutes with the doors closed to habituate to the arena. Next, an age and sex-matched target mouse was placed underneath the cup in one of the external chambers. The doors were opened, and the test mouse was given 5 minutes to explore the arena. The location of the target mouse alternated for each subsequent trial. Target mice were acclimated to the pencil cups for 3 days prior to study. Time spent in each chamber, time spent within 5 cm of each cup (called close interaction) and distance traveled was recorded. Experiment was conducted in 80-130 lux.

Juvenile social dyadic task

Test mice and sex-matched juvenile ( < or =5 week-old mice) were placed in the Noldus Phenotyper (Noldus) (30 cm x 30 cm) externally lined with white paper to obfuscate clear plexiglass walls. Mice were permitted to freely interact for 10 minutes. In addition to analyses in Ethovision, videos were analyzed post-hoc by a blinded coder using Behavioral Observation Research Interactive Software (BORIS)46 for self-grooming. The lead author later unblinded the data for analysis.

Social conditioned place preference

The social conditioned place preference (sCPP) assay was modified from previous studies47. All studies were completed in red light (Lux <5). Animals were placed into a white plexiglass box (30 cm x 30 cm) fitted with two clear plexiglass inserts that created a 2-chamber arena. One chamber was lined with black and white striped paper between the white plexiglass and clear liner, and a metal wire cloth floor (McMaster, 85385T937) was placed on the floor. The second chamber was lined with black polka dot paper and a perforated steel floor with a staggered pattern (McMaster, 9358T221). The two chambers were connected by a clear Plexiglass door. On day 1, the door was removed, and mice were permitted to freely explore both chambers for 20 minutes. After each test the floors were washed with Alconox soap and water and dried with paper towels. The second day, in the first session (between 0900 and 1400 mice were placed into a randomly assigned side of the chamber with a novel, sex-matched juvenile target for 15 minutes (social-paired side) and kept on that side for the duration of the session with the clear door inserted. In the second daily session, from 1400 to 1900, mice were placed in the opposite side of the same box for 15 minutes alone (empty-paired side). On the third day mice were trained on empty-paired side for 15 minutes in the first session and social paired side in the second session. This alternating pattern of social and empty-paired training first continued for days 4 and 5. A new target mouse was used daily for training. Finally, on day 6, mice were placed in the chamber alone with the door removed and permitted to freely explore the 2-chamber arena for 20 minutes. For analysis, only the first 5 minutes of testing was used. Training videos with 2 freely interacting mice were analyzed post-hoc by a blinded coder using BORIS46 for investigation time, which the lead author later unblinded for analysis. When a cup was used to contain the target mouse, Noldus Ethovision was used to automatically detect the time the test mouse spent around the cup investigating the social target.

Rewarded lever press

This paradigm was adapted from previous reports16. Mice were weighed 3 days prior to testing and then the food hopper was removed from cages for the duration of the task. Mice were given approximately 1 g of food a day per mouse in the cage and weighed daily. Mice continued a restricted diet for the duration of the task and adjusted food availability to maintain 85-90% of starting body weight. On the day before testing, 5-10 20 mg Dustless Precision Pellets (VWR) per mouse were placed into the home cage of animals to reduce neophobia. For testing, we used Noldus Phenotypers (Noldus) (30 cm x 30 cm) fitted with 1 wall that contained a lever, and 1 wall that contained a pellet receptacle attached to an automated dispenser filled with reward pellets. The lever was programmed to release 1 pellet for 1 lever press for the first 7 days of testing. On day 1, the lever was baited with 2 pellets and the pellet receptacle was baited with ~3 pellets. Mice were given a daily session of 60 minutes for 7 days to freely explore and interact with the lever. On day 8, we ran a break point task in which the lever was programmed on a progressive ratio 5 schedule (e.g. 1 press per pellet then 5, then 10, etc). Number of lever presses were recorded daily.

Elevated plus maze

Mice were placed in the center of the elevated plus maze (San Diego Instruments) and permitted to freely explore the open arms (lux~200) and closed arms (lux~50) for 300 seconds. Time spent in open arms, closed arms, center, and total distance traveled was quantified.

Open field

Mice were placed in the center of the Noldus Phenotyper (30 cm x 30 cm) fitted with 4 clear plexiglass walls and permitted to explore the chamber for 10 or 30 minutes freely. Total distance traveled, time in center (central 5 cm square), and grooming (automatically detected by Ethovision XT) was quantified.

Exclusion criteria

For all viral and implant studies, animals were excluded based on a priori standards. The injection site of all viral injections was identified by the presence of eGFP fluorescent marker. Animals were excluded from all data sets if the viral expression was not bilaterally expressed in the targeted region. 16 Shank3fl/fl mice were excluded for lack of bilateral, NAc viral expression across all cohorts. Genotypes were confirmed post hoc, and 8 Shank3Δe4-22 or WT controls were removed due to incorrect initial genotype results across all cohorts. Additionally, data were analyzed using Grubbs’ outlier test (alpha = 0.05) and any identified outliers were removed from the experiment; all raw values excluded by Grubbs’ outlier test are reported and highlighted in Supplementary Data 1.

Statistics and reproducibility

Statistical analysis was conducted using Prism 10 (GraphPad), where significance was set at α = 0.05, and a P value less than 0.05 was considered significant. Generally, unpaired t-tests were used to compare 1 variable between two groups, One-way ANOVA was used to compare 1 variable with 3 groups with Tukey’s multiple comparisons test to correct for multiple comparisons. Mixed-Effects Analysis was used to analyze multiple variables between multiple groups with Šídák’s multiple comparisons test to correct for multiple comparisons. We used a Brown-Forsythe test to compare the standard deviations between groups; when the standard deviations were significantly different, a Welch’s correction or Welch’s ANOVA was used, with a Dunnet’s multiple comparisons test. Mann-Whitney nonparametric test was used to compare discrete data (number of entries). Full statistical details and raw data values are in Supplementary Data 1 and sample size, statistical tests, and parameters are indicated in the figure legends. Data was not analyzed by sex, as we previously reported no sex differences in the behavioral phenotypes of Shank3Δe4-22 mice16. Sample size was determined by referencing previous published literature from our group and others that use the same methods16,22,44,47. Error bars represent mean ± standard error of the mean (SEM), with individual plot points overlaid. An annotation of *p < 0.05, **p < 0.01, and *** p < 0.001 is used in the figures. Detailed statistical results, including the tests and multiple comparisons used for all results, can be found in Supplementary Data 1.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Results

Conditional Deletion of Shank3 Δe4-22 from the Nucleus Accumbens Considerably Reduces SHANK3 Protein Levels in the Synaptosome

We conditionally deleted Shank3 exclusively from the NAc using a viral-mediated approach. AAV5-CMV-Cre-GFP or AAV5-CMV-eGFP control was injected into the NAc of adult male and female Shank3fl/fl mice (Supplementary Fig. 2 A-B). To assess the recombination efficacy between loxP sites, we first took NAc brain punches and conducted PCR to identify the presence of joint fragments between loxP sites (Fig. 1B). We did not see recombination in Shank3fl/fl mice injected with eGFP. We next confirmed that our approach of genetically removing Shank3 was sufficient to reduce protein expression. We found that the expression of SHANK3 protein in Shank3fl/fl mice injected with CRE was reduced by approximately 50 percent of GFP controls by immunoblot (Fig. 1C), which recapitulates the haploinsufficiency of SHANK3 commonly described in patients14,48. The ~50 percent reduction of SHANK3 expression was also confirmed by immunohistochemistry with SHANK3 antibodies (Fig. 1D-F, Supplementary Fig. 2 C-H). We found a considerable reduction in SHANK3 expression in Shank3fl/fl+CRE mice compared to Shank3fl/fl+eGFP and in Shank3∆e4-22 mice compared to Shank3fl/fl+eGFP (Fig. 1F). These results demonstrate that our viral-mediated approach to selectively reduce SHANK3 in the NAc of Shank3fl/fl mice is effective.

Fig. 1: Brain region-specific knockdown of Shank3 Reduces SHANK3 protein expression in the NAc.
figure 1

A Schematic design of Shank3 deletion strategy using CRE-loxP. Arrows indicate loxP sites. Exons removed are shown in purple. AAV5-CMV-Cre-eGFP was injected directly into the NAc of Shank3fl/fl animals to delete ∆e4-22 of Shank3. B PCR revealed the detection of Shank3 deletion e4-22 in the NAc but not the tail in Shank3fl/fl mice injected with CRE and Shank3∆e4-22 controls. C SHANK3 protein levels were quantified using western blot analysis in Shank3fl/fl+eGFP or Shank3fl/fl+CRE mice and Shank3∆e4-22 mice. SHANK3 is significantly decreased in Shank3fl/fl+CRE mice and Shank3∆e4-22 mice compared to Shank3fl/fl+eGFP controls. (Ordinary One-way ANOVA, ***p = 0.0007, F(2, 12) = 14.25, Tukey’s multiple comparison test eGFP vs CRE *p = 0.0106, eGFP vs Shank3∆e4-22 ***p = 0.0007; n1 = 6, n2 = 6, n3 = 3). D–F Immunostaining of SHANK3 (red) and DAPI (blue) with eGFP (green) from viral injection. Representative 20x (left) and 63x images (right) of the NAc. D SHANK3 was absent in Shank3∆e4-22 mice and E reduced in Shank3fl/fl+CRE mice compared to Shank3fl/fl+eGFP mice. F Fluorescent signal as a proxy for protein expression was used to measure SHANK3 levels in all groups. Fluorescence was significantly decreased in Shank3fl/fl+CRE and Shank3∆e4-22 mice compared to Shank3fl/fl+eGFP controls (Ordinary One-way ANOVA, P < 0.0001, F(2, 18) = 44.94, Tukey’s multiple comparison test eGFP vs CRE *** p = 0.0006, eGFP vs Shank3∆e4-22 ****p < 0.0001, CRE vs Shank3∆e4-22 **p = 0.0044; n1 = 8, n2 = 5, n3 = 8). Complete statistical analyses are provided in Supplementary Data 1. Triangles represent female samples and circles represent male samples in Figures C and F.

Full size image

Shank3 deficiency in the NAc diminishes social preference

We first used the 3-chamber social interaction task to assess how conditional, NAc Shank3 deletion and conventional Shank3Δe4-22 deletion alters social preference (Fig. 2A)49. We found that WT mice showed a preference for the mouse chamber, but Shank3Δe4-22 mice did not show a preference to either chamber (Fig. 2B). Similarly, WT mice preferred to be in close proximity to the mouse cup compared to the empty cup (Fig. 2C), while Shank3Δe4-22 mice did not prefer either close proximity zone. We found no changes in distance traveled during the test between WT and Shank3Δe4-22 mice (Fig. 2D). In the conditional deletion, Shank3fl/fl +CRE mice showed a preference for the empty chamber, while Shank3fl/fl+eGFP mice preferred the mouse chamber (Fig. 2E). Additionally, Shank3fl/fl+CRE mice spent substantially less time in the mouse chamber than Shank3fl/fl+eGFP controls (Fig. 2E). Shank3fl/fl+eGFP mice also have a preference to be in close proximity with the mouse cup and spent more time in close proximity of the mouse cup compared to Shank3fl/fl+CRE mice (Fig. 2F). There were no differences in distance traveled between Shank3fl/fl+CRE and Shank3fl/fl+eGFP mice during the 3-chamber task (Fig. 2G).

Fig. 2: Conventional Shank3∆e4-22 deletion and conditional Shank3-NAc deletion eliminates social preference.
figure 2

A Schematic of 3-chamber arena. B WT mice prefer the mouse chamber but not Shank3∆e4-22 mice (Mixed-Effects Analysis, Chamber x Genotype p = 0.9249 F(1, 84) = 0.008935, Šídák’s multiple comparison test, E-M WT *p = 0.0297, E-M KO ns p = 0.1073); n1 = 26, n2 = 18). C WT mice also spend significantly more time in close proximity ( ~ 5 cm perimeter) to the mouse-containing cup compared to empty, but not Shank3∆e4-22 mice (Mixed-Effects Analysis, Genotype x Chamber p = 0.5987, F(1, 84) = 0.2790, Šídák’s multiple comparison test, E-M WT **p = 0.0100, E-M KO ns p = 0.1732); n1 = 26, n2 = 18). D There was no effect on the distance traveled between groups in the 3-chamber task (Unpaired t-test, Two-tailed, p = 0.3505, t = 1.212 df=42; n1 = 26, n2 = 18). E Shank3fl/fl-eGFP mice showed a preference for the mouse chamber, and Shank3fl/fl+-CRE mice spent significantly less time in the mouse chamber than eGFP-injected mice (Mixed-Effects Analysis, Chamber x Virus p < 0.0001, F(1, 46) = 30.48, Šídák’s multiple comparison test, E-M eGFP ****p < 0.0001, eGFP-CRE Mouse *** p = 0.0005; n1 = 12, n2 = 14). F Shank3fl/fl-eGFP mice, but not Shank3fl/fl+CRE mice, spend significantly more time closely interacting with the mouse cup than the empty cup and Shank3fl/fl+eGFP mice spend more time in the mouse close interaction zone than Shank3fl/fl+CRE mice. (Mixed-Effects Analysis, Chamber x Virus p < 0.0001, F(1, 46) = 23.02, Šídák’s multiple comparison test, E-M eGFP ****p < 0.0001, eGFP-CRE Mouse *** p = 0.0004; n1 = 12, n2 = 14). G There were no changes in distance traveled between Shank3fl/fl+eGFP and Shank3fl/fl+CRE mice. (Unpaired t-test, Two-tailed, p = 0.0863, t = 1.792, df=23; n1 = 12, n2 = 14). Complete statistical analyses are provided in Supplementary Data 1.

Full size image

Next, we extended our investigation to understand how Shank3 deficiency impacts a naturalistic social environment using the juvenile social dyadic task (Fig. 3A). Mice were paired with sex-matched juvenile mice. Shank3Δe4-22 mice showed increased body contact time with the target compared to WTs (Fig. 3B). However, there was no difference between groups in the average distance between the test mouse and the juvenile target (Fig. 3C) or time spent in side-by-side facing the same direction (Fig. 3D). Shank3Δe4-22 mice spent considerably more time grooming than WTs (Fig. 3E). Shank3fl/fl+CRE mice were not different from Shank3fl/fl+eGFP controls in body contact time (Fig. 3F), average distance from juvenile mouse (Fig. 3G), time in side-by-side interaction (Fig. 3H), or in grooming time (Fig. 3I). Taken together, these data indicate that Shank3 in the NAc does not regulate the time the animal spends in naturalistic social engagement.

Fig. 3: Shank3∆e4-22 Mice Show Increased Social Contact While Shank3-NAc Deletion Does Not Alter Dyadic Social Behavior.
figure 3

A Schema of experimental design of the social dyadic task. B Shank3∆e4-22 mice showed increased time in body contact with a same-sex juvenile mate than WT controls (*p = 0.0278, Unpaired t-test, Two-tailed, t = 2.291, df=37; n1 = 23, n2 = 16). C No difference between groups in the average distance between test mouse and juvenile target during testing (p = 0.1958, Unpaired t-test with Welch’s correction, Two-tailed, t = 1.334, df=21.97; n1 = 23, n2 = 16). D No difference for the time in side-by-side facing the same direction between subjects during testing (p = 0.0956, Unpaired t-test with Welch’s correction, Two-tailed, t = 1.750, df=19.84; n1 = 23, n2 = 16). E A random selection of Shank3∆e4-22 and WT mice were hand scored for grooming. Shank3∆e4-22 mice spend significantly more time grooming (Unpaired t-test with Welch’s correction, Two-tailed, *p = 0.0269, t = 2.416, df=17.54; n1 = 16, n2 = 14). F–I We found no differences between Shank3fl/fl+eGFP and Shank3fl/fl+CRE groups in time spent in (F) body contact (Unpaired t-test with Welch’s correction, Two-tailed, p = 0.9788, t = 0.02677, df=35.24; n1 = 22, n2 = 25), G distance between subjects (Unpaired t-test with Welch’s correction, Two-tailed, p = 0.3834, t = 0.8802, df=45; n1 = 22, n2 = 25), H or time side-by-side same direction (Unpaired t-test, Two-tailed, p = 0.317, t = 1.012, df=45; n1 = 22, n2 = 25). I) A random selection of grooming was hand scored for Shank3fl/fl-eGFP and Shank3fl/fl+CRE mice. No changes were detected in time grooming (Unpaired t-test, Two-tailed, p = 0.9524, t = 0.06040, df=23 n1 = 12, n2 = 13). Complete statistical analyses are provided in Supplementary Data 1.

Full size image

Shank3 deficiency in the NAc does not alter locomotion, grooming, or anxiety-like behavior

Previously, we showed that conventional Shank3Δe4-22 mice are hypoactive in the open field task, show hyper grooming, and increased anxiety-like behavior16. We sought to rule out the possibility that changes in basic locomotive behaviors, grooming, or anxiety-like behaviors contribute to the loss of social preference seen in Cre-injected Shank3fl/fl mice by running a 10-minute open field trial (Fig. 4A). We first recapitulated our previous findings and showed that Shank3Δe4-22 mice travel less distance (Fig. 4B), spend less time in the center (Fig. 4C), enter the center fewer times (Fig. 4D), and groom more (Fig. 4E). We found no change in the distance traveled of Shank3fl/fl+CRE compared to Shank3fl/fl+eGFP mice (Fig. 4F). We also did not find any differences between Shank3fl/fl+CRE and Shank3fl/fl+eGFP for the time spent in the center (Fig. 4G), entries to the center (Fig. 4H), or grooming (Fig. 4I). We also extended the open field to 30 minutes in these animals to look for changes in habituation over the longer trial (Supplementary Fig. 3A)50, as we have previously reported Shank3Δe4-22 mice are hypo-locomotive and spend less time in the center during an extended open field trial16. We did not see any differences between Shank3fl/fl+eGFP and Shank3fl/fl+CRE mice in the 30-minute open field trial (Supplementary Fig. 3B-F). We also ran the Elevated Plus Maze (EPM) task (Supplementary Fig. 3G) and found no differences between Shank3fl/fl +eGFP and Shank3fl/fl +CRE mice (Supplementary Fig. 3 H-K).

Fig. 4: Shank3∆e4-22 mice show reduced locomotion and increased anxiety-like behavior, but Shank3-NAc deletion does not alter anxiety-like behaviors.
figure 4

A Diagram of open field task. B–E SShank3∆e4-22 mice traveled significantly less distance than WT controls (Unpaired t-test, Two-tailed, **p = 0.0057, t = 3.033, df=24; n1 = 14, n2 = 12) C Shank3∆e4-22 mice show a trend toward less center time than WTs but it is not significant. (Unpaired t-test, Two-tailed, p = 0.0583, t = 1.989, df=24; n1 = 14, n2 = 12.) D Shank3∆e4-22 mice enter the center fewer times (Mann Whitney test, two-tailed, **p = 0.0051, U = 31; n1 = 14, n2 = 12) and (E) groom more (Unpaired t-test, Two-tailed, ****p < 0.0001 t = 4.934, df=24; n1 = 14, n2 = 12) compared to WTs. F–H There is no difference between Shank3fl/fl+CRE and Shank3fl/fl+eGFP mice in distance traveled (Unpaired t-test, Two-tailed, p = 0.1286, t = 1.597, df=17; n1 = 9, n2 = 10), G time spent in the center (Unpaired t-test, Two-tailed, P = 0.3634, t = 0.9339, df=17; n1 = 9, n2 = 10), H entries to the center (Mann Whitney test, two-tailed, p = 0.1387, U = 26.5; n1 = 9, n2 = 10), or I grooming (Unpaired t-test, Two-tailed, p = 0.8077, t = 0.2472, df=17; n1 = 9, n2 = 10) Complete statistical analyses are provided in Supplementary Data 1.

Full size image

Shank3 deficiency in the NAc decreases social reward-seeking but not food reward-seeking

The NAc is a well-established modulator of motivation32, and we have shown that Shank3Δe4-22 mice exhibit depleted motivation during rewarded lever-pressing tasks16. Therefore, we next tested Shank3fl/fl mice in a rewarded lever press task (Fig. 5A, B). Shank3fl/fl+CRE mice showed no differences in the number of lever presses across 6 training days (Fig. 5C) or the test day (Fig. 5D) compared to Shank3fl/fl+eGFP mice. To assess any differences in motivation, we extended the lever press task to include a breakpoint task on day 851. We found no differences between Shank3fl/fl+CRE and Shank3fl/fl+eGFP mice (Fig. 5E).

Fig. 5: Removing Shank3 from the NAc does not affect palatable food reward-seeking behavior.
figure 5

A Timeline of experiment. B Schema of lever press apparatus. C Lever pressing counts do not differ between Shank3fl/fl+CRE and Shank3fl/fl+eGFP mice across the 6 days of training or test day of the experiment (Mixed-Effects Analysis, Day x Virus p = 0.7999, F(6, 204) = 0.5106; n1 = 20, n2 = 16). D There are no differences in the number of lever presses of animals between Shank3fl/fl+CRE and Shank3fl/fl+eGFP mice on test day (Unpaired t-test, Two-tailed; p = 0.1248, t = 1.574, df=34; n1 = 20, n2 = 16) or E during breakpoint on day 8 (Unpaired t-test, Two-tailed; p = 0.7801 t = 0.2820, df=27; n1 = 16, n2 = 13). Complete statistical analyses are provided in Supplementary Data 1.

Full size image

Given that conditional removal of Shank3 from the NAc and conventional Shank3Δe4-22 mutation decreases social investigation in the 3-chamber task, we predicted that Shank3Δe4-22 and Shank3fl/fl+ CRE mice are less rewarded by social interaction and, therefore, show less motivation to enter the mouse chamber. To test this hypothesis, we used a social condition place preference (sCPP) assay to measure social motivation (Fig. 6A)47. WT mice show an increase in time spent in the social chamber compared to empty in the post-test, but not the pre-test (Fig. 6B). Interestingly, we also found Shank3Δe4-22 mice spend more time in the social-paired chamber on post-test day (Fig. 6C). Shank3fl/fl+eGFP mice do not show differences in chamber time on post-test day (Fig. 6D), whereas Shank3fl/fl+CRE mice showed a preference for the empty-paired chamber on post-test day, revealing an opposite result from all other groups (Fig. 6E). When we compared the time in the social-paired chamber on post-test day, we found Shank3fl/fl+CRE mice spend considerably less time in the social-paired chamber compared to Shank3fl/fl+eGFP mice (Fig. 6F). Similarly, when we calculated the CPP score, Shank3fl/fl+CRE mice had markedly lower scores than WT mice and Shank3fl/fl +eGFP mice (Fig. 6G). Moreover, we report a significant difference between the groups in standard deviation (Fig. 6G) (Brown-Forsythe test p = 0.011), driven by the variability in the Shank3Δe4-22 mice. Shank3Δe4-22 mice are hypoactive on the pre-test day but also show a substantial decrease in the distance traveled on the pre-test day compared to the post-test day (Fig. 6H), which we did not find in any other group (Fig. 6I). We also measured body contact or social sniffing during conditioning and found no differences across groups (Supplementary Fig. 4 A-F).

Fig. 6: Removing Shank3 from the NAc causes social conditioned place aversion.
figure 6

A Social Conditioned Place Preference (SCPP) Experimental Design. B WT (Mixed-Effects Analysis, Time x Chamber p = 0.1849, F(1, 48)=1.809, Šídák’s multiple comparisons test, Post Empty vs Social *p = 0.021; n = 13) and C Shank3∆e4-22 mice (Mixed-Effects Analysis, Time x Chamber *p = 0.0227, F(1, 40) = 5.616, Šídák’s multiple comparisons test, Post Empty vs Social: **p = 0.0097; n = 11) spend significantly more time in the social-paired chamber in the post-test. D Shank3fl/fl+eGFP mice do not show a statistical preference for a chamber (Mixed-Effects Analysis, Time x Chamber p = 0.0914, F(1, 12) = 3.367; n = 13), while (E) Shank3fl/fl+CRE mice spend significantly more time in the empty-paired chamber than social-paired during the post-test day (Mixed-Effects Analysis, Time x Chamber **p = 0.0012, F(1, 13)=17.06, Šídák’s multiple comparisons test Post Empty vs Social, **p = 0.0012; n = 14). F Shank3fl/fl+CRE mice also spend significantly less time in the social-paired chamber on the post-test day than Shank3fl/fl+eGFP mice (Welch’s ANOVA test **p = 0.0088, W(3.000, 24.16) = 4.856, Dunnets Multiple Comparisons Test, WT vs CRE *p = 0.0152, eGFP vs CRE *p = 0.0141; Brown-Forsythe test of Standard Deviations *p = 0.011, F(3,47) = 4.138; n1 = 13, n2 = 11, n3 = 13, n4 = 14). G Shank3fl/fl+CRE mice have a significantly lower CPP score than WTs and eGFP-injected mice (Welch’s ANOVA test *p = 0.0109, W(3.000, 24.17) = 4.617, Dunnets Multiple Comparisons Test, WT vs CRE *p = 0.0427, eGFP vs CRE *p = 0.0236; Brown-Forsythe test of Standard Deviations **p = 0.0002, F(3,47) = 5.731; n1 = 13, n2 = 11, n3 = 13, n4 = 14). H Distance traveled is significantly lower in Shank3∆e4-22 mice compared to WTs during the pre-test and post-test and Shank3∆e4-22 mice travel less distance in the post-test compared to pretest. (Mixed-Effects Analysis, Time x Genotype *p = 0.0264, F(1,21) = 5.704, Šídák’s multiple comparisons test, Pre vs Post KO *p = 0.0307, WT vs KO PRE *p = 0.0111, WT vs KO POST ****p < 0.0001; n1 = 12, n2 = 11). I There are no changes in distance traveled in the pretest to posttest analysis or between Shank3fl/fl+eGFP and Shank3fl/fl+CRE mice (Mixed-Effects Analysis, Time x Genotype p = 0.8332, F(1, 25) = 0.04529; n1 = 13, n2 = 14). Complete statistical analyses are provided in Supplementary Data 1.

Full size image

Discussion

In the present study, we found that removing Shank3 from the NAc selectively eliminates social motivation but does not affect naturalistic social behaviors, motivation for palatable food, or anxiety-like behaviors, all of which are disrupted in the conventional Shank3Δe4-22 mouse model. We conclude that Shank3 in the NAc is an essential regulator specifically for social motivation and that social motivation is driven by neural mechanisms separate from other reward-seeking behaviors.

We initially anticipated that disruption of Shank3 expression in the NAc would alter all motivation and social behaviors and recapitulate the blunted social and reward-seeking phenotypes of Shank3Δe4-22 mice. However, we show that removing Shank3 from the NAc only disrupts social motivation. We thus predict that Shank3 in NAc-connected regions regulates food motivation and naturalistic, reciprocal social interaction. The ventral tegmental area (VTA) is a prime site for further investigation as it shares reciprocal circuitry with the NAc and is a well-established modulator of motivation27,52,53. Knocking down SHANK3 in the VTA also disrupts social preference54. However, Shank3-VTA deletion mice also lack sucrose preference, indicating a non-specific loss of reward-seeking behavior54. Taken together with the present study, SHANK3 protein expression is likely necessary for social motivation at both ends of the NAc-VTA circuit, but removing SHANK3 has a different effect on neural activity and encoding of behavior on each side of the circuit. Beyond social motivation, removing Shank3 from the vCA1 of the hippocampus disrupts social memory in mice55, and removing Shank3Δ14-16 from the anterior cingulate cortex (ACC) disrupts social preference in the 3-chamber assay56. The vCA1 of the hippocampus and the ACC project to the NAc and regulate social behaviors57,58. There may be a convergent mechanism where the loss of Shank3 at one loci of social brain networks is sufficient to change social behavior due to its influence across the network. Further investigation of the effect of Shank3 NAc deletion on neural activity across brain-wide networks, particularly those involved in social motivation, will provide a comprehensive mechanistic understanding of social motivation and social behaviors.

It was unexpected that Shank3e4-22 mice would show decreases in social preference in the 3-chamber task but increases in time spent in the social-paired chamber in the sCPP task. Previous studies have shown that stress before test day of a CPP task causes mice to arbitrarily choose a side and remain immobile for the duration of the task, as reflected by decreased distance traveled and increased variability in chamber time59. We therefore hypothesize that heightened stress levels underlie our findings that Shank3Δe4-22 mice have increased chamber time variability and decreased distance traveled during CPP test day59. This hypothesis is further supported by the increased anxiety-like behaviors in Shank3Δe4-22 mice described here and in our previous study16, and by reports that a negative affective state influences social motivation in mice and patients with ASD60,61. Future studies should determine how Shank3Δe4-22 mice perform in social operant tasks, which eliminates the influence of stress on side-choice, particularly as Shank3B-/-(Shank3Δ14-16) mice also show increased anxiety-like behavior but decreased social reward in an operant task17,57,62,63. Importantly, we demonstrate that Shank3fl/fl+CRE mice develop a conditioned place aversion to the social paired chamber, without exhibiting any changes in baseline anxiety-like behaviors. We thus posit that Shank3 in the NAc directly underlies social motivation but that heightened basal stress levels in Shank3Δe4-22 mice cause the opposing sCPP results. Our data support future investigations that examine the influence of basal stress levels on social motivation in mouse models of ASD.

Our data next led us to question how neural circuitry underlies this communication between anxiety-like states and social motivation. Others have shown that removing Shank3 only from the Bed Nucleus of the Striata Terminalis (BNST) replicates the anxiety-like behaviors of Shank3Δe4-22 mice64. The BNST-NAc circuit regulates stress adaptations and social behaviors in a higher-stress state and, therefore, could regulate the influence of stress on social motivation65. Future experiments should investigate beyond one brain region to understand how the social brain integrates the affective state with social motivation.

Here, we establish the Shank3fl/fl mouse model with two loxP sites flanking exons 4 and 22 that significantly improves recombination efficiency with a predictable pattern related to diverse Shank3 isoforms compared to previous tools developed by our lab44. Phelan-McDermid Syndrome (PMS) or 22q13.3 deletion syndrome, is defined by a loss of the distal long arm of one chromosome 22, which includes SHANK3, and the majority of patients are diagnosed with autism66. Our findings show that we successfully knock down SHANK3 protein expression to approximately 50% of WTs, which recapitulates the haploinsufficiency of SHANK3 found in patients14,48. This model thus provides an avenue for exploring conditional Shank3 deficiency that replicates the genetic abnormalities found in patients. Findings from our recent study of WT, Shank3Δe4-9, and Shank3Δe4-22 mice using a targeted RNA capture and long-read sequencing method presented a more complex structure for WT and mutant Shank3 transcripts67. We unexpectedly found fusion transcripts of Shank3 coding exons and the downstream protein-coding gene Acr in WT mice. Shank3Δe4-22 deletion also results in highly expressed fusion transcripts between the remaining exons 1-4 of Shank3 and the coding exons of Acr that are not expressed in WTs. The function of these Shank3 and Acr fusion transcripts remains to be defined. Nevertheless, Shank3Δe4-22 deletion disrupts the largest portion of the coding region within the previously characterized Shank3 transcripts16,68. Shank3Δe4-22 deletion recapitulates the loss of function of the known SHANK3 protein, demonstrating molecular validity and recapitulating the SHANK3 mutations found in ASD patients13,14. Therefore, Shank3Δe4-22 and Shank3fl/fl mice are appropriate tools for investigating the pathomechanisms of SHANK3-associated ASD. Further investigation of the function of Shank3-Acr fusion transcripts is also warranted.

We used a CRE-loxP approach to conditionally remove exons 4-22 of the Shank3 gene, in the NAc of adult mice. This approach gave us the advantage of high-precision targeting in the NAc to establish our initial findings, but it does not address the developmental role of Shank3 in the NAc on behavior. Past reports have shown that viral delivery to the NAc of siRNA that targets exon 21 of Shank3 eliminates social preference, but only when done at P6 and not in adults26. However, it is noted that exon 21 is an alternatively spliced exon in the brains of WT mice67,68. The siRNA design targeting exon 21 may only disrupt a portion of SHANK3 isoforms, whereas our approach disrupts most Shank3 isoforms compared to other Shank3 deletion approaches, as described above67. Thus, the discrepancy may reflect the molecular validity of SHANK3 protein between 2 studies.

Overall, these data establish a specific role of Shank3 in the NAc on social motivation and support a growing body of evidence that social motivation mechanisms may underlie ASD-related phenotypes. The social motivation theory of autism is a classic theory for understanding social deficits in individuals with autism69, but recent publications argue that increased anxiety and overinterpretation of human behavior could be confounding our understanding of social motivation in autism61,70,71. Our data establish a prospective mechanism and model for how co-occurring anxiety-like phenotypes can shift social motivation and provide an avenue to investigate these hypotheses. We present a new Shank3fl/fl tool that can be used to model the haploinsufficiency found in PMS patients to explore these mechanisms further.