In light of the severe speech and cognitive deficits associated with SHANK3 mutations (25, 26), we leveraged variations in nasal flow during natural sniffing to conduct nonverbal assessments of odor valence discrimination in 21 children with SHANK3 mutations (14 with terminal deletions of chromosome 22q13.3 encompassing the SHANK3 gene, 3 with nonsense mutations, 3 with indels causing frameshift mutations, and 1 with a duplication; see table S1 for mutation details) and 42 age- and sex-matched typically developing (TD) controls. Pleasant (pineapple and chocolate) and unpleasant odors [trimethylamine (rotten fishlike odor) and isovaleric acid (sour stinky footlike odor)] were presented intermittently in random order while the children watched their favorite animation shows. Meanwhile, respirations were continuously measured using either a nasal cannula connected to a spirometer or a respiratory belt transducer (Fig. 1A). Each odor presentation lasted 7 s, typically encompassing two inhalations (odor sniffs).

Participants’ sniff traces during odor presentations were normalized by the average durations and peak flowrates of the three inhalations preceding the odor presentation, accounting for odor-irrelevant fluctuations in respiration caused by the storylines of the animation shows. TD children exhibited stronger sniffs for pleasant odors compared to unpleasant ones, consistent with their odor valence ratings (fig. S1), whereas no such differentiation was observed in children with SHANK3 (Fig. 1B). Instead, children with SHANK3 displayed a general reduction in inhalation duration when exposed to odors (fig. S2), reflecting both their ability to detect odors and an overall avoidance of olfactory stimulation.

Further comparisons of the normalized mean sniff flowrate, peak sniff flowrate, and sniff volume during the presentations of pleasant versus unpleasant odors revealed that odor valence significantly modulated sniff vigor in TD children (mean sniff flowrate: P = 5 × 10⁻⁵, peak sniff flowrate: P = 1.2 × 10⁻⁵, and sniff volume: P = 2.2 × 10⁻⁴) but not in children with SHANK3 (all Ps > 0.5), with significant between-group differences across all three parameters (Ps = 0.007, 0.002, and 0.006, respectively; Fig. 1, C to E). On the other hand, sniff duration and frequency remained unchanged between pleasant and unpleasant odors for both groups of children (TD: Ps = 0.79 and 0.24; SHANK3: Ps = 0.16 and 0.26; Fig. 1, F and G). These patterns persisted across both the first and second sniffs occurring during odor presentations (fig. S3). Moreover, in children with SHANK3 mutations, the absence of valence modulation of sniff vigor was evident both in those with additional gene deletions (SHANK3+, n = 13) and in those with SHANK3 mutations alone (SHANK3 only, n = 8) (fig. S4).

We also obtained usable scalp electroencephalography (EEG) recordings from an additional seven children with SHANK3 mutations and seven age- and sex-matched TD children. Using a modified testing procedure (fig. S5A), olfactory stimuli were presented intermittently and in random order via a computer-controlled olfactometer while the children watched their preferred animation shows. The stimuli included pineapple (pleasant) and trimethylamine (unpleasant) at clearly noticeable concentrations (1×), the same odors at 1/10 of those concentrations (0.1×), and purified air as a control. Each stimulus was delivered for 3 s with an interstimulus interval of ~18 s. During each olfactory stimulus presentation, the ongoing animation was replaced by a fixed frame (without audio) to facilitate EEG comparisons across olfactory conditions. EEG signals were aligned to sniff onsets during periods of olfactory stimulation.

Corroborating the results obtained with sniff parameters, time-frequency analysis of EEG signals in TD children revealed a significant differentiation of odor valence at Pz—the electrode strongly implicated in olfactory processing (27, 28)—in alpha-band power around 700 to 800 ms following inhalation onset (P = 0.007; fig. S5B). This observation aligns with earlier olfactory EEG findings in healthy adults (29–31). In contrast, no significant valence-related effect was observed in children with SHANK3 mutations (P = 0.13, BF₁₀ = 0.98, and error% = 0.0198; fig. S5B). The absence of valence differentiation was not attributable to general olfactory hyposensitivity. Compared with purified air, odors (pineapple and trimethylamine combined) at clearly perceptible concentrations elicited significant theta-alpha activity around 500 ms following inhalation onset in both TD children and children with SHANK3 (TD: P = 0.011; SHANK3: P = 0.016), with no group differences in event-related percent change in power (ER%) (P = 0.49, BF₁₀ = 0.53, and error% = 0.00107; fig. S5C). At 0.1× concentration, these odors produced similar, although weaker (TD: P = 0.028; SHANK3: P = 0.028) and nonsignificant, time-frequency responses in both groups (TD versus SHANK3, P = 0.50, BF₁₀ = 0.52, and error% = 0.00105; fig. S5C). In other words, children with SHANK3 exhibited EEG responses, indicating sensitivity to odors and odor gradients comparable to those of TD children.

Together, these observations highlight hyporeactivity to odor hedonics—but not to odors per se—in children with SHANK3 mutations, suggesting a potential link between SHANK3 deficiency and olfactory anhedonia.

Given the observation of olfactory valence processing deficits in children with SHANK3 mutations, we used Shank3B^−/−mice—harboring a comparable loss-of-function mutation—to investigate whether this phenotype is conserved in rodent models. Using a refined performance index (PI) to quantify odor-evoked valence responses (19), we observed pronounced differences between wild-type (WT) and Shank3B^−/− mice (KO). WT mice displayed strong aversion to negatively valenced odorants such as 2,5-dihydro-2,4,5-trimethylthiazoline (TMT) and pentanal and robust attraction to positively valenced odorants including isoamyl acetate (IAA) and 2-phenylethanol (2-PE) (Fig. 2A). In contrast, Shank3B^−/− mice showed markedly blunted aversion to TMT and pentanal, as well as reduced—or in some cases reversed—attraction to appetitive odorants strongly preferred by WT mice. These alterations indicate that Shank3 deficiency disrupts the normal coding of both attractive and aversive odor valence. On the basis of these results, we selected 2-PE and TMT as representative appetitive and aversive odorants, respectively, for subsequent mechanistic experiments (32).

TMT application elicited robust defensive responses in WT mice, characterized by rapid escape (reflected by increased maximum velocity) followed by prolonged freezing. In contrast, Shank3B^−/− mice failed to show either flight or freezing in response to TMT (Fig. 2, D and E, WT versus KO mice, in D: P = 0.359 in water, P = 0.907 in 2-PE, and P = 0.008 in TMT; in E: P = 0.562 in water, P = 0.884 in 2-PE, and P < 0.001 in TMT) and instead spent significantly more time investigating the TMT source (Fig. 2C). Shank3B^−/− mice also exhibited a marked reduction in attraction to 2-PE (Fig. 2C, WT versus KO mice, P = 0.479 in water, P = 0.001 in 2-PE, and P = 0.002 in TMT). Apart from the TMT-evoked changes in maximum velocity and immobility, the total distance traveled did not differed between groups (Fig. 2F), indicating that the behavioral alterations in Shank3B^−/− mice reflect selective impairments in odor-evoked valence processing rather than generalized motor deficits. Therefore, consistent with children carrying the SHANK3 mutations, Shank3B^−/− mice failed to engage in normal responses toward odors with distinct valence.

To investigate whether the abnormal response observed during the presentation of attractive and aversive odorants in Shank3B^−/− mice was a result of the inability to differentiate between odors, we trained mice to associate a set of odors with differential outcomes (fig. S6A). Each training trial began with a conditioned stimulus (CS) of an odor, followed by a 2-s delay (the response window) and then an unconditioned stimulus (US; water, nothing, or air puff) (fig. S6, B and C). As training progressed, both WT and Shank3B ^−/− mice exhibited anticipatory licks only during the delay before water trials, indicating the establishment of an odor-based CS-US association (fig. S6, D and F). In this associative learning paradigm, there was no significant difference between WT and Shank3B^−/− mice (fig. S6E), suggesting that Shank3B^−/− mice are capable of perceiving and discriminating between different odorants, as evidenced by their ability to form odor-specific conditioned responses in a classical associative learning task.

Next, we sought to identify the neural circuit substrate that was affected by the loss of Shank3 and that played key roles in the observed valence coding defects in Shank3B^−/− mice. Because a previous study indicated that the CoA is essential for behavioral responses toward appetitive and aversive odors (19), we asked whether the activity of the CoA was abnormal in Shank3 mutant mice. Immunohistochemical analysis of the immediate early gene product c-Fos revealed a robust elevation of the number of c-Fos⁺ neurons in the OB, the CoA, and the PIR after 2-PE and TMT stimulation in WT mice when contrasted with the Shank3 group. However, c-Fos induction was significantly lower in the CoA of Shank3B^−/− mice compared to WT mice, indicating attenuation of c-Fos in the Shank3 group (P = 0.523 in water, P = 0.006 in 2-PE, and P < 0.001 in TMT). In contrast, there were no significant between-group differences in the number of c-Fos⁺ neurons in the OB or PIR (Fig. 3, A and B).

We then recorded real-time calcium signals of CoA and PIR neurons in head-fixed mice during 2-PE and TMT stimulations after intracranial delivery of AAV9-CaMKIIα-GCaMP6s into these brain regions (Fig. 3, C to E), which enables selective labeling of excitatory neurons (33). We observed a strong transient increase in GCaMP fluorescence in WT mice elicited by 2-PE and TMT. In contrast, there was only a mild increase in genetically encoded calcium indicator GCaMP (comprising GFP, calmodulin, and the M13 peptide) fluorescence in Shank3B^−/− mice (Fig. 3F). The normalized GCaMP fluorescence change (ΔF/F) during 2-PE and TMT stimulations was significantly lower in Shank3B^−/− mice compared to that in WT mice (Fig. 3H, P = 0.009 in 2-PE and P = 0.029 in TMT). Unlike in the CoA, we did not observe a significant difference in calcium responses in the PIR (Fig. 3, G and I). These data suggest that reduced activation of CoA excitatory neurons may contribute to the olfactory valence deficit in Shank3B^−/− mice.

Diminished activation of CoA excitatory neurons may underlie the olfactory valence deficit observed in Shank3B^−/− mice. To investigate the potential impact of Shank3 loss on synaptic functions in the CoA, we conducted whole-cell patch-clamp recording on pyramidal neurons in Layer I of the CoA, where axons from the OB projection neurons terminate. We recorded the miniature excitatory postsynaptic currents (mEPSCs) after blocking sodium channels with QX314 in the pipette solution and action potentials and inhibitory synaptic transmission with tetrodotoxin (TTX) and picrotoxin in the perfusion solution. We found that both mEPSC frequency and amplitude were significantly reduced in Shank3B^−/− mice compared with those littermate control mice (Fig. 4, A to C), suggesting a weaker excitatory synaptic strength in Shank3B^−/− mice. To directly measure the synaptic strength of OB-CoA synapses, we selectively stimulated OB-CoA synapses by positioning the stimulating electrodes in CoA Layer I and recorded the evoked AMPA receptor (AMPAR)–mediated EPSCs from CoA pyramidal neurons in response to stimuli of different intensities (Fig. 4D). We found that the amplitude of EPSCs was significantly lower in Shank3B^−/− mice (Fig. 4, E and F). Consistent with the smaller AMPAR-mediated EPSCs, a lower AMPA-to-NMDA (N-methyl-d-aspartate) ratio was observed in Shank3B^−/− mice (Fig. 4, G and H). To assess whether this synaptic alteration was region specific, mEPSC recordings from pyramidal neurons in the PIR showed no significant differences in frequency, amplitude, rise time, or decay time between WT and Shank3B^−/− mice (fig. S7, A to C). Together with these findings, the lower frequency of mEPSCs observed in the CoA of Shank3B^−/− mice implies a potential contribution of presynaptic deficit. To exclude this possibility, we measured the presynaptic release probability by assessing the paired-pulse ratio (PPR) of EPSCs, and we found no differences in the PPR between genotypes (fig. S7, D and E). These results demonstrate that excitatory synaptic transmission of OB-CoA synapses was impaired in Shank3B^−/− mice and that this deficit is unlikely to be caused by altered presynaptic release probability.

Given that Shank3 plays a critical role in both synapse formation and function (3), we sought to determine whether the Shank3 deletion would induce structural changes in the pyramidal neurons of the CoA. To examine the dendritic processes and spines of CoA pyramidal neurons, we made local injections of a mixture of AAV9-CaMKIIα-Cre (1:10,000) and AAV9-EF1α-DIO-EGFP to sparsely label neurons with green fluorescent protein (GFP) in WT and Shank3B^−/− mice. We then analyzed the dendritic processes and spines of pyramidal neurons in Layer I of the CoA where they receive olfactory inputs from the OB. The Sholl analysis indicated that the dendritic complexity of CoA pyramidal neurons in Shank3B^−/− mice was lower compared to that of WT mice (Fig. 4, I and J). We also found a decrease in the number of intersections between 30 and 120 μm away from the soma in Shank3B^−/− mice. Significantly lower total dendritic spine density was also observed in Shank3B^−/− mice. We further classified different types of spines. Among the different spine types, the ratios of mushroom and stubby were reduced, whereas long thin spines increased, and filopodia spines remained unaltered in Layer I of the CoA in Shank3B^−/− mice (Fig. 4, K and L). These observations demonstrated dendritic and spine morphology abnormalities consistent with synaptic function defects caused by the loss of Shank3.

To establish a causal link between the loss of Shank3 in CoA neurons and defective odor valence–driven behaviors, we used a targeted strategy to selectively knock out Shank3 expression in the CoA of adult C57BL/6J mice using CRISPR-Cas9 technology. We bilaterally injected AAV9-pCALM1-SpCas9-miniU6-sgRNA^Shank3 into the CoA to generate an acute Shank3 knockout model (Fig. 5, A and B, and fig. S8). Consistent with observations in Shank3B^−/− mice, acute ablation of Shank3 in CoA neurons led to a decrease in the frequency of mEPSCs (Fig. 5, C to E), whereas the presynaptic release probability, as assessed by PPR, remained intact (fig. S9, A and B). Using the AAV-SpCas9-sgRNA system, we achieved a highly specific knockdown of Shank3 in the CoA, with Shank3 mRNA levels reduced to 41.5% ± 6.3% of control levels (fig. S8B) and a corresponding loss of Shank3 protein (46.0% ± 4.7%; fig. S8, C and D), with no detectable changes in the adjacent PIR. These results demonstrate that the CRISPR-Cas9 manipulation efficiently and selectively disrupts Shank3 expression within the CoA while sparing surrounding olfactory cortical regions (fig. S8, B to D).

Furthermore, the amplitude of evoked AMPAR-mediated EPSCs in response to stimuli at 0.4 and 0.5 mA was reduced, indicating impaired postsynaptic responsiveness. Assessment of attraction and aversion responses with 2-PE and TMT revealed dysfunctional odor valence assessment in acute CoA Shank3 knockout mice, as evidenced by reduced and increased investigation time toward 2-PE and TMT, respectively (Fig. 5G, sgRNA^Scr versus sgRNA^Shank3 mice, P = 0.459 in water, P = 0.007 in 2-PE, and P < 0.001 in TMT). Consistent with their altered valence response, sgRNA^Shank3 mice showed significantly attenuated the maximum velocity and reduced immobility in response to TMT compared to mice injected with the control AAV expressing a scrambled (Scr) single guide RNA (sgRNA) (Fig. 5, H and I, sgRNA^Scr versus sgRNA^Shank3 mice; in H: P = 0.996 in water, P = 0.728 in 2-PE, and P < 0.001 in TMT; in I: P = 0.285 in water, P = 0.841 in 2-PE, and P < 0.001 in TMT). Furthermore, compared with sgRNA^Scr mice, Shank3 acute ablation mice exhibited lower calcium signals in the CoA during 2-PE and TMT stimulations (Fig. 5, K and L, P = 0.015 in 2-PE and P = 0.002 in TMT). Collectively, these findings demonstrate that acute loss of Shank3 specifically within the CoA was sufficient to damage odor valence perception.

To further determine whether the loss of Shank3 in the CoA is essential to odor valence–induced behavioral abnormalities in Shank3 knockout mice, we selectively restored Shank3 expression in the CoA. We used a Shank3 conditional knock-in (CKI) mouse model where Shank3 expression can be reinstated upon Cre recombinase expression in a Shank3 knockout background (34). In the Shank3 CKI mouse model, exons 13 to 16 encoding the PDZ domain are inverted and transcriptionally silenced using the FLEx switch. Delivery of AAV-Cre into the CoA induces Cre-dependent recombination, restoring the correct orientation of the inverted exon cassette. This recombination reinstates endogenous Shank3 expression specifically in CoA neurons that receive Cre, thereby enabling region-specific rescue of Shank3 function (Fig. 6A). Similar to Shank3B^−/− mice, Shank3 CKI mice displayed defective behavioral responses toward 2-PE and TMT (fig. S10). We then injected AAV8-CaMKIIα-Cre-EYFP into the CoA of Shank3 CKI mice to restore the reexpression of most Shank3 isoforms in the CoA but not in the PIR (Fig. 6, A to C, and fig. S11). In electrophysiological analyses, we found that reduced mEPSC frequency in Shank3 CKI mice was effectively restored after Cre expression (Fig. 6, D to F). The impaired synaptic strength of OB-CoA synapses was also reversed (Fig. 6G), whereas the presynaptic release probability was intact after the restoration of Shank3 expression (fig. S12, A and B).

We then injected a mixture of AAV9-CaMKIIα-FLP (1:10,000) and AAV9-EF1α-fDIO- mCherry into Shank3 CKI mice to sparsely label pyramidal neurons in Layer I of the CoA and analyzed the anatomical features of dendritic processes after Shank3 restoration. We found elevated number of intersections and spines in the Layer I portion of CoA pyramidal neurons after reinstating Shank3 expression in Shank3 CKI mice (Fig. 6, H to L). Intriguingly, reexpression of Shank3 in the CoA alone was sufficient to rescue attraction and avoidance elicited by 2-PE and TMT in Shank3 CKI mice, respectively [Fig. 6, M to O; one-way analysis of variance (ANOVA) followed by Dunnett’s test; in M: CKI + GFP versus WT + Cre, P = 0.721; CKI + Cre versus WT + Cre, P = 0.916, in water; CKI + GFP versus WT + Cre, P = 0.002; CKI + Cre versus WT + Cre, P = 0.222, in the 2-PE group; CKI + GFP versus WT + Cre, P < 0.001; CKI + Cre versus WT + Cre, P = 0.206, in TMT; in N: CKI + GFP versus WT + Cre, P = 0.957; CKI + Cre versus WT + Cre, P = 0.858, in water; CKI + GFP versus WT + Cre, P = 0.435; CKI + Cre versus WT + Cre, P = 0.997, in the 2-PE group; CKI + GFP versus WT + Cre, P = 0.002; CKI + Cre versus WT + Cre, P = 0.102, in TMT; in O: CKI + GFP versus WT + Cre, P = 0.114; CKI + Cre versus WT + Cre, P = 0.161, in water; CKI + GFP versus WT + Cre, P = 0.852; CKI + Cre versus WT + Cre, P = 0.097, in the 2-PE group; CKI + GFP versus WT + Cre, P = 0.040; CKI + Cre versus WT + Cre, P = 0.867, in TMT]. These data strongly argue that Shank3 expression in the CoA is essential to the intact olfactory valence coding in mice.

A total of 99 children, accompanied by their caregivers, participated in the human testing experiment. Of these, 22 participants (8 SHANK3 and 14 TD children) were excluded due to data quality issues. Consequently, the data of 77 participants were included in the final analysis. For the nasal airflow assessment, the sample comprised 21 children with SHANK3 deficiency (9 girls and 12 boys, 6.32 ± 2.09 years) and 42 age- and sex-matched TD children (18 girls and 24 boys, 6.36 ± 1.98 years). For the scalp EEG assessment, seven children with SHANK3 deficiency (three girls and four boys, 6.52 ± 1.98 years) and seven age- and sex-matched TD children (three girls and four boys, 6.64 ± 2.06 years) were included. SHANK3 and TD groups did not differ in sex or age (nasal airflow: χ² = 0, P = 1, t₆₁ = −0.07, and P = 0.94; EEG: χ² = 0, P = 1, t₁₂ = −0.11, and P = 0.91). SHANK3 mutations were identified through genetic testing, and gene reports were provided by caregivers. Demographic information and mutation details for the SHANK3 group are summarized in table S1. None of the participants had respiratory allergies or upper respiratory infections at the time of testing. Written informed consent was obtained from the primary caregiver of each child to participate in the procedure, which was approved by the Institutional Review Boards at the Institute of Psychology (H22108) and the Shenzhen Institute of Advanced Technology (SIAT-IRB-220215-1-1-0587) of the Chinese Academy of Sciences.

All animal experiments were performed in accordance with the Shenzhen Institute of Advanced Technology (SIAT), Chinese Academy of Sciences guidelines for animal care and were approved by SIAT’s Institutional Animal Care and Use Committee (IACUC, SIAT-IRB-180129-NS-HY-A0398). Animals were maintained on a 12-hour:12-hour light:dark cycle (lights on at 08:00) with food and water provided ad libitum. Genotyping of Shank3B mice (the Jackson Laboratory, 017688) and Shank3 CKI mice (a gift from W. Wang) was performed following standard protocols. We also used C57BL/6J mice (aged 6 to 8 weeks; Beijing Vital River Laboratory Animal Technology Co. Ltd.) in selective deletion of Shank3 experiments. Our reported sample sizes in each experiment are comparable to those found in previous publications. Animals were randomly allocated to either a sgRNA^Shank3 or a sgRNA^Scr group. Experimenters were blind to the experimental groups, and the order of testing was counterbalanced during behavioral experiments. No data were excluded from the analyses.

For human nasal airflow assessment, olfactory stimuli consisted of two pleasant fragrant oils—pineapple [50% (v/v) in propylene glycol] and chocolate [33% (v/v) in light mineral oil]—and two unpleasant odor compounds—trimethylamine [1% (v/v) in propylene glycol] and isovaleric acid [2% (v/v) in propylene glycol]. Trimethylamine has a rotten fishlike odor, whereas isovaleric acid has a sour stinky footlike odor. Each odor was presented using a felt-tip pen filled with 3 ml of the corresponding odor solution. For presentation, the cap was removed and the pen tip was positioned ~2 cm in front of both nostrils. For scalp EEG recordings, pineapple [5 and 0.5% (v/v) in propylene glycol] and trimethylamine [0.1 and 0.01% (v/v) in propylene glycol] at two concentration levels (1× and 0.1×) served as the olfactory stimuli, with purified air included as a control condition. Odors were delivered via a computer-controlled olfactometer providing a constant airflow of 2.5 liter/min.

The olfactory stimuli used for the Olfactory Valence Test comprised a panel of 11 odors that span diverse chemical categories and innate valence properties, including TMT, pentanal, furfural, 2-hexanone, propylene glycol (vehicle control), agmatine sulfate (AST), benzaldehyde, 2,3-butanedione, hexanal, IAA, and 2-PE. All odorants were prepared at a final concentration of 85 μmol/ml and dissolved in propylene glycol. For each trial, 50 μl of the odor solution was applied to a piece of filter paper and placed in a designated corner of the arena. The olfactory stimuli were presented for subsequent mechanistic behavioral experiments; we focused on two representative odorants—2-PE (appetitive) and TMT (aversive) [2-PE at 1% (v/v) in water, and the aversive odor, TMT, at a concentration of 300 μmol/ml in water]. During the testing phase, a 50-μl aliquot of each odor solution was applied to a piece of filter paper and placed in a designated corner of the experimental arena (19).

Mice (6 to 8 weeks of age) were anesthetized using chloral hydrate (1%, w/v) and positioned in a stereotaxic injection frame (RWD Life Science). A digital mouse brain atlas was linked to the injection frame to guide the identification and targeting of the CoA or the PIR. Injections were performed at a rate of 20 to 30 nl/min with a Hamilton syringe (10 μl, model 1701) using a Nanoliter 2000 microinjection system (WPI). Stereotaxic coordinates were used to direct injections into either the CoA [anteroposterior (AP), −1.40 mm; mediolateral (ML), −2.70 mm; and dorsoventral (DV), −5.80 mm) or the PIR (AP, −1.10 mm; ML, −3.70 mm; and DV, −4.70 mm]. To prevent backflow of fluid, the pipette was left in the brain for at least 5 min after completion of the injection. Once the injection pipette was removed, the mouse was removed from the stereotaxic apparatus, and the incision was closed with wound clips. During all surgical procedures, mice were kept on a heating pad and brought back to their home cages after recovery of movement.

We followed a published protocol for fiber photometry (55). Briefly, male mice were unilaterally injected with 300 nl of AAV9-CaMKIIα-GCaMP6s (2.71 × 10¹² vector genomes/ml), followed by the insertion of a ceramic cannula housing an optical fiber [1.25 mm in outer diameter (OD); 0.37 numerical aperture (NA)] toward the CoA or PIR. The ceramic cannula was then fixed with dental acrylic. Mice were individually housed for 2 weeks to allow GCaMP expression.

For sparse labeling (56), mice were injected with 100-nl mixtures of AAV9-CaMKIIα-Cre (1:10,000) and AAV9-EF1α-DIO-EGFP for Shank3B mice or AAV9-CaMKIIα-FLP (1:10,000) and AAV9-EF1α-fDIO-mCherry for Shank3 CKI mice with methods mentioned above. Mice were brought back to their home cages for 3 weeks to allow reporter expression.

For selective deletion, 500 nl of AAV9-pCALM1-SpCas9-miniU6-sgRNA^Shank3 or AAV9-pCALM1-SpCas9-miniU6-sgRNA^Scr (5 × 10¹² vg/ml) was bilaterally injected into the CoA of C57BL/6J. Mice were brought back to their home cages for 4 weeks to ensure enough time for Shank3 deletion.

For conditional knockout and rescue experiments, 500 nl of AAV8-CaMKIIα-Cre-EYFP or AAV8-CaMKIIα-EYFP (5 × 10¹² vg/ml) was bilaterally injected into the CoA. Mice were brought back to their home cages for 3 weeks to allow Shank3 reexpression.

Replication was confirmed across different batches of animals and viral preparations.

A fiber photometry system (Thinker Tech) was used to record GCaMP signals in neurons. To record fluorescence signals, a 488-nm laser beam was reflected by a dichroic mirror, focused by a 10× objective lens, and then coupled to an optical commutator. An optical fiber (OD, 125 μm; NA = 0.37, 5.5/5 m long) guided the light between the commutator and the implanted optical fiber. The laser power was adjusted at the tip of the optical fiber to a low level of 0.01 to 0.02 mW to minimize bleaching. The GCaMP fluorescence was band-pass filtered (MF525-39, Thorlabs) and collected by a photomultiplier tube (R3896, Hamamatsu). An amplifier (C7319; Hamamatsu) was used to convert the photomultiplier tube current output to voltage signals, which was further filtered through a low-pass filter (40-Hz cutoff, Brownlee 440, AutoMate Scientific). The analog voltage signals were digitalized at 500 Hz and recorded by a Power 1401 digitizer and Spike2 software (CED) (55).

Head-fixed mice with optical fibers were connected to the fiber photometry system for recording. Responses were recorded following stimulation of 2-PE, TMT, or air presented for 3 s for each trial. Fiber photometry data were analyzed using MATLAB (MathWorks). The fluorescence change (ΔF/F) was calculated as (F − F₀)/F₀, where F₀ was the mean fluorescence signal during a 1-s baseline period in each trial. The area under the curve (AUC) was calculated as mean ΔF/F during a 10-s period (0 to 10 s from onset).

The olfactory preference test was conducted with modifications from a previously established protocol (19, 57). Before each behavioral session, mice were transferred to a dedicated testing room and allowed to acclimate for at least 30 min. Adult male mice were tested individually in a 40 cm–by–40 cm–by–45 cm (width × length × height) open-field arena. At the beginning of each trial, a clean piece of filter paper (1.5 cm by 2 cm) was placed in the designated target quadrant and remained there throughout the entire 15-min session.

Mice were first allowed to freely explore the arena for 5 min without odor stimulation. This odor-free period was used to quantify each mouse’s baseline preference for the target quadrant. Immediately after the baseline phase, 50 μl of the odorant solution (85 μmol/ml in propylene glycol) was gently applied to the filter paper without disturbing the animal, and the mouse continued exploring the arena for an additional 10 min. All sessions were video recorded, and the time spent in the target quadrant was analyzed using EthoVision XT.

For each mouse, the PI obtained under the solvent-alone condition (propylene glycol) was used as the individualized baseline. PI values obtained during odor stimulation were then normalized to this solvent PI, such that positive values reflected attraction and negative values reflected aversion relative to the solvent control.

Behavioral assays were performed in a 40 cm–by–40 cm–by–45 cm open-field chamber. A piece of filter paper (1.5 cm by 2 cm) was fixed in the target quadrant for the entire 15-min session. During the first 5 min (baseline period), no odorant was applied; during the subsequent 10 min (test period), 50 μl of the odorant solution (odorant diluted in water) was gently added to the filter paper without disturbing the animal. Water alone served as the vehicle control, and all performance indices were normalized to the vehicle baseline (58). All odorant exposures were performed in thoroughly cleaned chambers, with 75% ethanol used between trials to eliminate residual olfactory cues.

Behavioral videos were recorded and analyzed using EthoVision XT (Noldus). The investigation time was defined as the cumulative duration during which the mouse’s nose entered the predefined filter paper region of interest (ROI). Maximum velocity was calculated as the peak instantaneous speed within the target quadrant during odor exposure. Immobility time was quantified using EthoVision’s standard mobility thresholds, defined as episodes in which whole-body movement fell within 0.05 to 60% of the activity detection threshold for at least 2 s. Total distance traveled was extracted as an index of general locomotor activity and used to rule out nonspecific motor deficits.

Mice were water deprived for 48 hours and habituated to a head-fixed setup. Odor delivery, water reward, and lick detection were controlled by an Arduino-based system interfaced with MATLAB. During Pavlovian conditioning, a 1-s odor stimulus (CS) preceded water delivery (US) by 2 s. Anticipatory licking was quantified during a 2-s response window before reward delivery. The intertrial interval was randomized between 6 and 8 s. Mice received ~40 trials per session per day, and behavior was analyzed offline in MATLAB. Well-trained mice (three to six sessions) exhibited robust anticipatory licking (≥80% of maximal lick rate) following odor presentation.

Whole-cell patch-clamp recording was performed as described previously (59, 60). Briefly, the mice were anesthetized with isoflurane and brains were removed and placed in ice-cold dissection solution saturated with 95% O₂ and 5% CO₂ containing the following: 115 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 0.1 mM CaCl₂, 3 mM MgCl₂, 25 mM NaHCO₃, 25 mM glucose, 0.4 mM l-ascorbic acid, 3 mM myo-inositol, and 2 mM sodium pyruvate. Coronal slices (300 μm) containing the CoA region were prepared using a vibratome (VT1200S, Leica). Slices were then transferred to a recovery chamber containing the following: 115 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 2 mM CaCl₂, 1 mM MgCl₂, 25 mM NaHCO₃, 25 mM glucose, 0.4 mM l-ascorbic acid, 3 mM myo-inositol, and 2 mM sodium pyruvate (saturated with 95% O₂ and 5% CO₂, pH 7.4), for 30 min at 32°C and >30 min at room temperature before recording. Recordings were performed with internal solutions for recording excitatory synaptic responses, containing the following: 125 mM Cs-Meth, 20 mM CsCl, 5 mM Na₂Phosphocreatine, 4 mM MgATP, 0.3 mM NaGTP, 0.5 mM EGTA, and 2 mM QX-314 (pH 7.3). Patch pipettes were pulled from borosilicate glass capillary tubes using a PC-100 pipette puller (Narishige). The resistance of pipettes filled with intracellular solution varied between 4 and 5 MΩ.

The recordings were obtained using a Multiclamp 700B amplifier (Molecular Devices). For all whole-cell recordings, membrane resistance was monitored before and after each run. Cells were rejected if Ra was >30 MΩ and leaky current >500 pA throughout the experiment. Recordings were digitized and filtered with a Digidata 1550B. Clampex 10.7 was used for acquisition and analysis. Experimenters were unaware of the genotype of all samples throughout data collection and analysis.

For the recording of PPR and AMPAR-mediated input-output curve, the cells were voltage clamped at −70 mV in the presence of 50 μM D-AP5 to block NMDA receptors (NMDARs) and 50 μM picrotoxin to block inhibitory synaptic transmission. The evoked AMPAR-mediated synaptic responses were triggered by a bipolar electrode placed in Layer I (100 to 150 μm from the soma of neurons recorded). The paired stimuli were delivered at an interstimulus interval of 50 ms. The PPR is calculated by the ratio of the peak response of the second EPSC to that of the first EPSC. For the AMPA-to-NMDA ratio, the AMPAR- and NMDAR-mediated EPSCs were recorded in the presence of 50 μM picrotoxin at holding membrane potentials of −70 and +40 mV, respectively. The AMPA-to-NMDA ratio was calculated as the ratio of the average AMPA-EPSC peak amplitude to the NMDA-EPSC amplitude. For the recording of mEPSCs, the cells were voltage clamped at −70 mV in the presence of 50 μM picrotoxin to block GABA_A (γ-aminobutyric acid type A) receptors and 1 μM TTX to block action potentials. The mEPSCs were digitized at 20 kHz, filtered at 2 kHz, and analyzed with Mini Analysis (Synaptosoft Inc.).

Mice were anesthetized with chloral hydrate (1%, w/v) and perfused transcardially with cold phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA) to remove blood. Thick coronal sections (40 μm) were collected throughout the whole brain. Following PBS washes, samples were blocked in BlockAid (B10710, Invitrogen) containing 0.3% (w/v) Triton X-100 for 1 hour at room temperature. Sections were overlaid with anti-c-Fos antibody (1:400, 9F6, Cell Signaling Technology), anti-HA antibody (1:400, C29F4, CST), or anti-GFP antibody (1:500, 600-301-215, Rockland) and incubated for 48 hours at 4°C. After washing off the primary antibody, samples were overlaid with a 1:500 dilution of Alexa Fluor 488 Goat-anti-Rabbit IgG (111-545-144, Jackson ImmunoResearch) or Alexa Fluor 488 Goat-anti-Mouse IgG (115-545-003, Jackson ImmunoResearch) and incubated at room temperature for 2 hours. After washing, sections were coverslipped with Fluoroshield mounting medium with 4′,6-diamidino-2-phenylindole (DAPI; ab104139, Abcam).

c-Fos–positive cells were quantified from anatomically matched coronal sections identified according to the Allen brain atlas. For each animal, two sections were collected from the OB (bregma: +4.2 to +3.8 mm), three from the CoA and the PIR (bregma: −1.2 to −2.4 mm), using evenly spaced 120-μm intervals to cover posteromedial (CoApm) and posterolateral (CoApl) subdivisions. For each mouse, the reported value represents the average c-Fos count across all sections for each region. Each group included three mice, and all individual animal data points are shown.

Images were acquired with an Axio Imager Z2 microscope (Zeiss) or an LSM 900 Confocal microscope (Zeiss). ROIs were defined using atlas-guided anatomical boundaries and verified by a blinded experimenter. c-Fos⁺ nuclei were counted using consistent thresholding parameters in Imaris (Oxford Instruments). For each mouse, the reported value represents the average c-Fos count across the two/three sections for each region.

Total RNA was extracted using RNAiso Plus (9109, Takara) according to the manufacturer’s instructions. cDNA was synthesized from total RNA using a reverse transcription kit (11155ES60, Yeasen). Quantitative polymerase chain reaction (qPCR) was subsequently performed using the SYBR Green Master Mix (311185ES03, Yeasen). All primers were synthesized by Tsingke Biotechnology (Shenzhen, China). The primer sequences were as follows: Shank3: forward 5′-CCTATGACAGCCTCACTTCACACAG-3′, reverse 5′-GACCCACCACTTGCTTGTGTCCAAC-3′; β-actin (internal control): forward 5′-AGTGTGACGTTGACATCCGT-3′, reverse 5′-TGCTAGGAGCCAGAGCAGTA-3′.

For qPCR analysis, each biological sample (tissue from an individual mouse) was assayed in three technical replicates (triplicate wells). The mean of the three technical replicates was used as the value for each biological replicate for subsequent statistical analysis.

CoA and PIR tissues were microdissected in ice-cold oxygenated ACSF (artificial cerebrospinal fluid) and lysed in RIPA (radioimmunoprecipitation assay) buffer (20115ES60, Yeasen) containing protease inhibitors. After centrifugation, protein concentrations were determined using a BCA assay (23235, Invitrogen). Equal amounts of protein were resolved by SDS–polyacrylamide gel electrophoresis and transferred to PVDF (polyvinylidene difluoride) membranes. Membranes were blocked with 5% nonfat milk and incubated overnight at 4°C with a primary antibody against Shank3 (1:1000, 64555S, CST) and Tubulin (1:5000, 66031, Proteintech), followed by HRP (horseradish peroxidase)–conjugated secondary antibody (1:1000, 7074P2, CST). Signals were detected using enhanced chemiluminescence and quantified in ImageJ. Shank3 expression was normalized to Tubulin for each lane.

For Western blot analyses, biological replicates consisted of protein lysates prepared from three independent mice per group, with each mouse contributing one sample to the analysis.

All experiments were conducted blinded, including the electrophysiology and behavioral analyses. The number of replicates (n) is indicated in the figure legends and refers to the number of experimental subjects independently treated in each experimental condition. Statistical analyses were performed in SPSS (version 22, IBM), JASP (2023, version 0.17), and GraphPad (version 9, Prism). Sniff response differences between means in the pleasant and unpleasant odorant conditions were compared against zero using one-sample Student’s t tests. The modulation differences between means across TD and SHANK3 groups were tested for statistical significance using the two-tailed unpaired t test. Additional Bayesian independent samples t tests were conducted to compare the sniff response differences between subgroups of children with SHANK3. The sex ratio between groups was assessed by the chi-square test. Normally distributed data were tested by one-way ANOVA followed by Dunnett’s test or Tukey’s test for multiple comparisons and unpaired two-sample Student’s t test for two-group comparisons. Probabilities of P < 0.05 were considered to be statistically significant.