Adult C57 mice and FVB129 mice were bought from the animal center of the Fourth Military Medical University. FMR1^−/− mice were bought from Jax laboratory (stock No.010504). All animal experiments were carried out under protocols approved by the Animal Care and Use Committee of Fourth Military Medical University and according to “Policies on the Use of Animals and Humans in Neuroscience Research,” revised and approved by the Society for Neuroscience.

VPA-treated mice were obtained by breeding the progenies of pregnant mice exposed to VPA at E12.5 (i.p. injection, 500 mg/kg, P4543, Sigma).

For immunohistochemistry, mice were perfused intracardially with 4% paraformaldehyde phosphate buffer. Serial coronal sections were prepared and blocked by PBS containing 3% BSA and 0.3% Triton-X100, and then incubated with primary antibodies overnight at room temperature. The primary antibodies used were as follows: mouse anti-CaMK II (Cell Signaling, 50,049, RRID: AB_2721906, 1: 400), mouse anti-GAD67 (Millipore, MAB5406, RRID: AB_2278725, 1:1000). mouse anti-NeuN (Abcam, ab104224, RRID: AB_10711040, 1:200), and rabbit anti-c-Fos (cell signaling, 2,250, RRID: AB_2247211, 1:1600). After washing with PBS, corresponding secondary antibodies conjugated with donkey anti-rabbit (Alexa Fluor 488, Invitrogen, A-21206, RRID: AB_2535792, 1:800), and donkey anti-mouse (Alexa Fluor 594, Invitrogen, A-21203, RRID: AB_141633, 1:800) were incubated with the sections for 2–4 h at room temperature, protected from light. After washing with PBS, sections were counterstained with Hoechst33342 (1:1000, Sigma) for 20 min.

rAAVhSyn-hChR2-mCherry (serotype 2/9, titer 4 × 10¹² vg/ml) and rAAVhSyn-eNpHR3.0-mCherry (serotype 2/9, titer 4 × 10¹² vg/ml) were purchased from Brain VTA, Wuhan. For virus injection, the animals were anesthetized with isoflurane. The injections were made into the left ACC (0.23 mm left to bregma, 0.93 mm anterior to bregma, and 1.6 mm in depth). Standard injection volumes were set to 200 nL. After the injection, the needle was left in place for 10 min before being withdrawn.

For optogenetic neuronal modulation, the ACC was activated with a 465-nm laser (49 mW, 10 Hz, blue light) for 30 s for hChR2 and inhibited with a 589-nm laser (49 mW, 10 Hz, yellow light) for 30 s for eNpHR 3.0.

USV recordings were performed as described (Hou et al., 2021). Briefly, mice were placed in a clean rectangular cage (60 × 42 × 40 cm) covered with a metal lid. A field microphone (VT UltraMic-384, Virtins Technology) was placed 2 cm below the lid of the cage. The microphone signal was sampled by software (SeaWave), recording frequencies ranging from 20 Hz to 190 kHz. One male experimental mouse and one female WT mouse were put in the USV recording cage. Each group of mice had no interactions prior to the recording and was considered as n = 1 for statistics. USV production was recorded for 10 min. The data were stored and analyzed offline by a researcher blinded to the experimental design. USV was detected using a custom program (deep squeak) with a spectrographic display.

A newly developed TMS device equipped with a 13-layer 3-turn “8”-shape coil, capable of focusing the stimulation region within a minimal size of 0.5 mm³ (Black Dolphin IT-TMS, Solide Company, Xi’an, China), was adopted. Before formal experiments, the brain region specificity was tested by stimulating the hindlimb region of the right motor cortex. Upon single TMS stimulation, the movement of bilateral hindlimbs was videoed. At the same time, electromyographic activity in the bilateral hindlimbs was recorded as described (Saturnino et al., 2020). Briefly, the positive electrode was inserted into the front end of the gastrocnemius muscle. The negative electrode was inserted into the back end of the gastrocnemius muscle, and the grounding electrode was inserted into the root of the mouse tail. The maximum signal sampling rate was set to 32,000 Hz. The single-channel signal sampling rate was set to 8,000 Hz. The signal was filtered through a 50-Hz band-stopping filter, a 20–480-Hz band-passing filter, and a power frequency harmonic filter (Solide Company, Xi’an, China).

rTMS was carried out by placing the probe 1–2 mm above the skull of the left ACC. The parameters of rTMS were as follows: 10 Hz stimulating pulse intensity at 40% of the maximum power of the rTMS device, 20 stimulations per cluster, repeated 40 times. rTMS was conducted for 7 consecutive days, with 800 pulses per day. Starting 3 days prior to the rTMS procedure, the mice were habituated to the coil for 10 min each day. For sham stimulation in the control group, the coil was placed immediately above the skull without magnetic stimulation.

For electric field simulation, the open-access electromagnetic simulation software SimNIBS was used as described (Saturnino et al., 2019, 2020). The coil inductance was approximately 17 μH. The capacitance was 140 μF. The reference voltage was 1,600 V, and the reference frequency was 3,250 Hz. The maximum current change rate of the coil was approximately 9.4e7 A/s. According to the TMS stimulation intensity, the current change rate was set to approximately 40% of the maximum current change rate.

All behavior analysis and statistics were performed by an investigator who was blinded to the experimental design. No sample calculation was performed. For morphological analysis, three mice were included in each group, and at least five brain sections were quantified for each mouse. For behavior analysis, eight mice were included in each group. No mice were excluded in terms of data analysis. Mice of the same genotype were randomly assigned to the same treatment group. Each behavioral test was conducted using distinct groups of animals. Data are presented as the mean ± standard error. Normal distribution is assessed using the Shapiro–Wilk test. Statistical comparisons were made using the Student’s t-test or one-way ANOVA. A p-value of less than 0.05 was considered statistically significant.

To examine if the ACC neurons respond to social-induced USV, we put a male mouse and a female mouse together for 10 min to induce USV and conducted double-immunostaining of c-Fos with pan-neuronal marker NeuN, excitatory neuronal marker CaMKII, and inhibitory neuronal marker GAD67 in the bilateral ACC following USV emission. The results showed a significantly larger number of c-Fos/NeuN-positive cells in the left ACC of the USV-emitting mice compared to the right ACC of the USV-emitting mice and that of the bilateral ACC of the resting mice (one-way ANOVA). P_{Rest-L vs. USV-L} = 0.0002. P_{Rest-R vs. USV-L} = 0.0007. P_{USV-L vs. USV-R} = 0.0004 (Figures 1A,B). Similarly, a significant increase in the c-Fos/CaMKII-positive cells was observed in the left ACC of USV-emitting mice, compared to the right ACC of the USV-emitting mice and the bilateral ACC of the resting mice (one-way ANOVA). P_{Rest-L vs. USV-L} = 0.011. P_{Rest-R vs. USV-L} = 0.0002. P_{USV-L vs. USV-R} = 0.001 (Figures 1C,D). No significant difference in the c-Fos/NeuN- and c-Fos/CaMKII-positive cells was found between the bilateral ACC of the resting mice and the right ACC of USV-emitting mice (Figures 1B,D). Co-expression of c-Fos with GAD67 was rarely detected in the bilateral ACC of both USV-emitting mice and resting mice (Supplementary Figure S1). These data demonstrated that the left ACC was highly responsive to social-induced USV.

To examine whether the left ACC was required for social-induced USV, we carried out optogenetic modulation of the ACC neurons. AAV expressing light-sensitive eNpHR3.0 was injected into the left ACC of WT mice. The ACC injection and virus infection were verified by mCherry expression (Supplementary Figure S2). Optogenetic inhibition of the ACC neurons was achieved by laser irradiation at 589 nm for 30 s, and USV was evaluated immediately within the following 10 min. The results showed that the optogenetic inhibition of the left ACC significantly reduced the numbers (Student’s t-test. p = 0.0078, Figure 2A), as well as the sinuosity of USV emitted (Student’s t-test. p = 0.029, Figure 2B). Although no change in average call length or mean power of USV was found before or after laser irradiation, suppressing neuronal activity seemed to shift calls toward longer duration and higher power (Figures 2C,D). These data indicated that the left ACC neurons may play a role in social communication.

We next tested whether the left ACC was involved in the social communication deficits in FMR1^−/− mice and VPA-pretreated mice. Considering that FMR1^−/− mice were derived from the FVB129 background, we used FVB129 mice as a WT control for FMR1^−/− mice. USV induced a large number of c-Fos/NeuN- and c-Fos/CamKII-positive cells in the left ACC of FVB129 mice. Significant reductions of c-Fos/NeuN- and c-Fos/CamKII-positive cells were observed in the left ACC of FMR1^−/− mice, compared to those in FVB129 control mice (one-way ANOVA, P_{FVB-USV-L vs. FMR1-KO-USV-L} = 0.002 for c-Fos/NeuN-positive cells). P_{FVB-USV-L vs. FMR1-KO-USV-L} = 0.02 for c-Fos/CamKII-positive cells (Figures 3A,B). USV did not induce a c-Fos response in the right ACC of both FMR1^−/− mice and FVB129 control (Figures 3A,B).

In VPA-induced ASD mice, a smaller number of c-Fos/NeuN- and c-Fos/CamKII-positive cells were found in the left ACC following USV compared to the left ACC of the control mice (one-way ANOVA, P_{C57-USV-L vs. VPA-USV-L} = 0.0003 for c-Fos/NeuN-positive cells). P_{C57-USV-L vs. VPA-USV-L} = 0.001 for c-Fos/CamKII-positive cells (Figures 3C,D). Few c-Fos-positive cells were detected in the right ACC of both C57 mice and VPA-pretreated mice (Figures 3C,D). These data demonstrated a weak or absent response of the left ACC in ASD mice to USV.

We then evaluated the effects of increasing neuronal activity in the left ACC on social communication in ASD mice. AAV expressing hChR2 was injected into the left ACC of these two mice. Two weeks later, optogenetic activation of the ACC neurons was achieved by laser irradiation at 453 nm for 30 s, and USV was evaluated immediately within the following 10 min. Laser irradiation significantly increased the number of USV calls in both FMR1^−/− and VPA-pretreated mice (Student’s t-test. p = 0.016 for FMR1^−/− mice. p = 0.046 for VPA-pretreated mice Figures 4A–D). No changes in USV duration and frequency were detected in both mice before and after optogenetic activation (Supplementary Figure S3).

It is thought that the syllable composition of USV contains biological meanings (Von Merten et al., 2014; Chen et al., 2021). Next, we classified the syllables into seven types as described with modification (Gao et al., 2019): (1) short, a duration <10 ms; (2) flat, a consistent duration >10 ms; (3) split, an emission with a very short split; (4) downward; (5) upward; (6) chevron, a continuous increase followed by a continuous decrease in frequency; (7) complex, long duration with variation of frequency. The results showed that FMR1^−/− mice exhibited five types of syllables. Activating the left ACC neurons resulted in six types of syllables and increased the percentage of type 1 syllables. VPA-pretreated mice showed all seven types of syllables under normal conditions. Optogenetic activation of the left ACC neurons did not generate more types of syllables in the VPA-pretreated mice but increased the percentage of type 1 syllables. These data indicate that activating the left ACC neurons could rescue the USV defects of FMR1^−/− and VPA-pretreated mice.

To modulate the activity of the left ACC neurons in a clinically appliable way, we developed a precise target TMS, which could confine magnetic stimulation to a size of approximately 0.5–1.5mm³. We first tested the region-specificity of this precise target TMS by placing the coil on the surface of the right motor cortex (Figure 5A). The left hindlimb movement was triggered immediately following TMS stimulation without affecting the movement of the right hindlimb (Supplementary Video S1). Electromyographic recording showed typical movement-evoked potential in the left hindlimb (but not in the right hindlimb) upon TMS stimulation (Figure 5B), illustrating the brain region specificity of this precise target rTMS.

Then, we stimulated the left ACC of WT mice using a 10-Hz protocol, which presumably activated the neuronal activity for 7 consecutive days and assessed USV emission (Figure 5C). Within 10 min of recording, significantly more USV calls were detected in the rTMS-treated mice (Student’s t-test. p = 0.048. Figures 5D,E). As for the syllable subgroups, the rTMS treatment did not elicit new types of USV syllables. Interestingly, type 1 syllables were more frequently observed in the rTMS-treated mice (Figures 5F,G). These data indicate that modulating the left ACC by rTMS is able to promote USV emissions.

We next applied the same rTMS protocol to the left ACC of FMR1^−/− and VPA-pretreated mice for 7 days. A significant increase in call numbers was recorded in rTMS-treated mice, compared to sham treatment control mice (Student’s t-test. p = 0.002 for FMR1^−/− mice. p = 0.012 for VPA-pretreated mice Figures 6A–D). To validate whether the effect was left ACC-specific, we evaluated the USV emission following rTMS stimulation of the right ACC. In this case, no significant changes in call numbers were observed (Supplementary Figure S4).

Regarding syllable composition, the rTMS treatment of the left ACC resulted in one more subtype of USV syllables in FMR1^−/− mice, as did optogenetic activation. In addition, rTMS increased the percentage of type 1 syllables in both FMR1^−/− and VPA-pretreated mice (Figures 6E,F), indicating that modulating the left ACC by rTMS was beneficial for social communication in ASD.