C57BL/6 mice obtained from The Jackson Laboratory, Bar Harbor, ME, USA were housed in cages with water, bedding, and chow (Prolab, RMH 3000) under standard conditions of a 12-hour light/dark cycle (lights on at 8:00 AM) and a temperature of 22°C at the animal facility of the Facultad de Ciencias de la Vida, Universidad Andrés Bello (Santiago, Chile). All experimental procedures and care were performed according to current regulations and guidelines of the Animal Welfare Committee of the Facultad de Ciencias de la Vida (Bioethics approval certificate number 012/2021) and Agencia Nacional de Investigación y Desarrollo (ANID). Daily supervision of mice was performed by a veterinarian. Mice were euthanized by inhaling 5% isoflurane/O₂ according to AVMA guidelines (51).

Gestational HTX was induced in pregnant mice according to previous reports (52–55). Six- to eight-week-old C57BL/6 mice were mated in a proportion of 2 females and 1 male. The next day, a vaginal smear was collected and analyzed to search for spermatozoa. Mice with a positive smear were considered pregnant, and that day was referred to as embryonic day 1 (E1). Each dam was individually placed in a cage. On E10, pregnant mice were randomly distributed into three groups (N = 7 pregnant mice per group). The first group received regular tap water during the entire pregnancy (untreated dams); therefore, gestation was conducted under euthyroid conditions. The second group received 0.025% w/v of 2-mercapto-1-methylimidazole (methimazole, MMI) (M8506, Sigma-Aldrich, USA) in the drinking water from E10 to E14 (MMI-treated dams). To reverse the MMI-induced phenotypes, the third group was treated with 0.025% w/v MMI along with a daily subcutaneous injection of 25 μg/kg of T₄ dissolved in PBS from E10 to E14 (MMI+T₄-treated dams) (52–55). This treatment aimed to demonstrate that the observed effects induced by MMI are due to maternal T₄ reduction and not a side effect of MMI. Untreated and MMI-treated dams did not receive a vehicle intraperitoneal injection to avoid the related stress of a daily injection. Both the MMI and MMI+T₄ treatments were administered daily from E10 to E14 at 9:00 AM. The water for untreated dams was also renewed during E10 to E14. The bottles containing MMI solution were prepared daily with fresh tap water and protected from light. Consequently, the experimental groups of this work were the respective progenies named after the mother’s treatment: Control-offspring, HTX-offspring, and HTX+T₄-offspring.

To verify the correct induction of gestational HTX, blood samples were collected from the facial vein of each pregnant mouse on E14 and total (t)THs and TSH were quantified. Additionally, a blood sample was also obtained from each adult mouse on postnatal day 50 (P50) to evaluate thyroid function. Serum was separated from the blood samples by centrifugation at 1,000 xg for 15 min at 4°C. Levels of tT₃ and TSH were quantified by sandwich ELISA using mouse CUSABIO® ELISA kits (catalog N°E05086m and N°E05116m, respectively) following the manufacturer’s instructions. Concurrently, tT₄ serum levels were determined by chemiluminescence in an external certified veterinary laboratory (LQCE, Santiago de Chile). After quantification, the tT₄ range for MMI-treated dams was established at [0.1 – 1.9] (μg/dL), whereas the range of tT₄ in the untreated dams was [2.2 – 4.5] (μg/dL). Since the administration of tT₄ during MMI treatment is expected to restore their levels and reverse the induced phenotype, the accepted tT₄ range in the MMI+T₄-treated dams was set equal to that of the untreated group ([2.2 – 4.5] (μg/dL)) (52–55).

Control, HTX, and HTX+T₄ offspring were weaned on P28. To mitigate the impact of the “litter effect”, experimental groups were formed by randomly selecting 2-3 mice of each sex from each litter. Consequently, each group comprised 30 mice, conformed by 15 females and 15 males. Mice of the same sex were housed together in cages (4-5 mice per cage) from weaning until behavioral assessment. Behavioral testing was conducted during adulthood (from P55 to P64), specifically during the dark phase of the circadian cycle (between 9.00 AM and 2.00 PM), consistent with a previous report of behavioral assessment in mice (56). The sequence of behavioral testing was arranged as follows: marble burying (MB) on P55, elevated plus maze (EPM) on P56, tube dominance test from P57 to P63 (considering training and confrontation steps), and three-chamber social preference test on P64. The behavioral performance of each mouse in the EPM, tube dominance, and three-chamber tests was conducted under 60 decibels (dB) of white noise and recorded using an FHD 1080p/30 fps camera. Data were manually scored and analyzed by three trained investigators blinded to each experimental group.

MB was performed to analyze repetitive/persistent behavior (57, 58). Each mouse was placed inside the testing cage (arena size: 24 cm x 17.2 cm, wood chips bedding depth: 5 cm) containing 12 marbles arranged in four columns of three marbles at equidistant distances. After 30 min of exploration, each mouse was returned to its respective cage. The number of buried marbles was counted, considering them buried when 2/3 of their volume was concealed within the wood chips.

The elevated plus maze (EPM) test was employed to analyze anxious-like behavior (59, 60). Each mouse was placed in the center of a suspended cross-shaped maze consisting of two open and two closed arms (25 cm long x 7 cm wide x 24.5 cm high walls of closed arms). After 5 minutes of exploration, each mouse was returned to its respective cage. The number of entries into both the open and closed arms, as well as the time spent in each arm, were recorded. An entrance was only considered valid when all four paws of the mouse crossed the boundary of the maze center into an open or closed arm.

The tube dominance test was used to assess aggressive social interaction (61). A hollowed transparent Plexiglas tube measuring 30 cm in length and 2.5 cm in diameter was used. Two mice from distinct experimental group were introduced to the tube from opposite directions. The critical point occurs when both mice meet in the middle of the tube, and one must retreat to exit, thereby revealing a dominant or subordinate status based on the species hierarchy theory. The training step was conducted over three consecutive days (from P57 to P59) with the goal of teaching each mouse that crossing the tube is the only way to exit without backing up. Subsequently, the testing stage took place over four consecutive days (from P60 to P63). During the confrontation stage, mice from the Control-offspring were paired with those from the HTX-offspring group, the HTX-offspring mice were paired with the HTX+T₄-offspring mice, and the Control-offspring mice were paired with the HTX+T₄-offspring mice in a pairwise fashion. Confrontations were conducted between mice of the same sex, with pairs randomly chosen and confronted three times per day, rotating the initial entry position to minimize bias. After each confrontation, mice were returned to their respective cages to reduce the immediate impact of recent winning or losing. The number of wins for each mouse was recorded, and the total number of wins after four days was used as an indicator of their social status, distinguishing between dominant or subordinate.

The three-chamber social preference test was conducted to assess social interaction abilities (62). Mice were first habituated per 1 h to the three-chamber device (a plexiglass box of 102 cm length x 47 cm width x 45 cm height). The three-chambered apparatus consisted of two lateral compartments, each containing one inverted wire cup, along with one empty central compartment. During the pre-test, each mouse was allowed to acclimate to the setup for 10 minutes. Then, the first phase of the test was initiated. An inanimate object was placed under one cup (named “nonsocial stimulus”), while an unfamiliar mouse, matched in terms of age and sex, was placed under the other cup (named “social stimulus”). Each mouse was then positioned in the central compartment of the three-chamber device and allowed to transit between the nonsocial and social compartments for 10 minutes. After that time, mice were returned to their respective cages. In the second phase of the test, the nonsocial stimulus was replaced with a second unfamiliar, age- and sex-matched mouse (also named “de novo stimulus” or “Stranger 2”), while maintaining the first strange mouse now renamed “Stranger 1”. Each mouse was allowed to transit between the Stranger 1 and Stranger 2 compartments for 10 min. The following parameters were registered: the time that each mouse spent in the social, nonsocial, and de novo compartments, the time of direct interaction with the social, nonsocial, and de novo stimulus, and the number of entrances to each compartment. Direct interaction was defined as the presence of the mouse within a diameter of 5 cm around each cup. Similar to the EPM test, an entry was considered when all four legs of the mouse crossed the center of the chamber toward one of the lateral compartments.

On the day following the last behavioral test (P65), the Control, HTX, and HTX+T₄ offspring were euthanized to obtain the blood, brain, and spleen. Blood was collected via cardiac puncture and allowed to clot at room temperature for 25-30 minutes. Afterward, serum was isolated by centrifugation (1,000 xg, 15 minutes at 4°C), aliquoted, and stored at -80°C until further analysis. Brains were dissected to isolate the prefrontal cortex (PFC) and hippocampus from the telencephalon, following a standardized mouse brain dissection protocol (63). Both tissues were homogenized in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 0.1% Triton X-100) supplemented with phosphatase and protease inhibitors (1 mM Na₃VO₄, 1 mM NaF, 30 mM sodium pyrophosphate, and 1 mM PMSF). After a 30-minute incubation on ice, the homogenates were centrifuged at 15,000 xg for 15 minutes at 4°C. Total protein quantification from the supernatants was performed using the BCA method with a BSA standard curve (PierceTM, BCA™ 23225). Cytokines were quantified in serum samples of 100 μL and in brain tissue portions (0.5 mg of protein) using sandwich ELISA kits for specific cytokines: IL-1β (BioLegend, 432601), IL-6 (BioLegend, 431307), IL-17A (BioLegend, 432501), IL-10 (BioLegend, 555252), and TNF-α (BioLegend, 430901), following the manufacturer’s instructions. Detection was conducted using a microplate spectrophotometer (Epoch™). Serum cytokine concentrations were normalized to a volume of 100 μL, while cytokine levels in brain tissue portions were normalized to the amount of protein used for these measurements. Results are presented as pg/mL for cytokines in serum and as pg/mg for normalized brain-derived cytokines.

The relative expression of the glutamatergic proteins NLGN3 (94 kDa) and HOMER1 (40 kDa) in the PFC and hippocampus was evaluated by western blot. Initially, 20 µg of protein from both tissues was denatured at 95°C for 5 min and then separated on a 12.5% SDS-PAGE gel at 100 V for 120 min. The PageRuler Plus Prestained Protein Ladders of 10-250 kDa (Thermo Scientific™ 26619) and 10-180 kDa (Thermo Scientific™ 26616) were used to estimate the molecular weight of NLGN3 and HOMER1, respectively, and to monitor the progress of electrophoretic separation and electrotransference. Proteins were transferred onto nitrocellulose membranes (Thermo Scientific) for 1 h at 100 V on ice. Thereafter, membranes were blocked with 5% nonfat milk dissolved in Tris buffered saline with 0.1% Tween-20 (TBS-T, pH 7.6) for 1h at R.T with constant shaking. Membranes were separately probed with the primary antibodies anti-NLGN3 (2 μg/mL) (ab192880) and anti-HOMER1 (0,75 μg/mL) (ab211415), both prepared in 5% nonfat milk in TBS-T, overnight at 4°C and constant rotation. Moreover, α-tubulin (1 μg/mL) (ab6046) was used as a loading control. On the next day, membranes were washed (3 times, 5 min each) using TBS-T and then incubated with the secondary antibody anti-rabbit HRP-conjugated (1 μg/mL) (Bethyl A120-101P) in 5% nonfat milk in TBS-T for 1h at R.T and constant rotation. Membranes were washed (3 times, 5 min each) using TBS-T and incubated with ECL detection substrate (Thermo ScientificTM 34076) for 5 min. Bands were visualized and images were recorded using the Alliance Q9 Advanced chemiluminescence and spectral fluorescence imaging system (UVITEC Cambridge). Band intensities were analyzed using ImageJ software (NIH, Bethesda, MD, USA) and the relative expression of NLGN3 and HOMER1 was normalized to α-tubulin.

Spleens were separately placed in fresh RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) and then homogenized through a 70 µm cell strainer. Cell suspensions were centrifuged at 460 xg for 5 min at 4°C and then treated for 5 min at R.T with ammonium-chloride-potassium buffer (ACK, pH 7.2) to lyse erythrocytes. After two washes with fresh 1X PBS (5 mL each), total splenic leukocytes (also called splenocytes) were resuspended in 1 mL of PEB buffer (PBS 1X, 0.5% BSA, 2 mM EDTA), and counted using the Countess 3 Automated Cell Counters equipment (Thermo Fisher Scientific) using the trypan blue exclusion test of cell viability analysis (64).

Adaptative and myeloid immune cell populations from the splenocytes were analyzed by flow cytometry. A portion of 1.0×10⁷ splenocytes were stimulated in vitro with phorbol myristate acetate (PMA; 5 ng/mL), ionomycin (500 ng/mL), and brefeldin-A (BFA, 10 μg/mL) for 5 h at 37°C and 5% CO₂. Cells were harvested by centrifugation (600 xg, 5 min, 4°C) and then incubated with the following antibodies for 45 min at 4°C to assess Treg and Th17 populations: anti-CD4-BUV395 (BD 563790), anti-CD25-PeCy7 (BioLegend 102016), and anti-IL-17A-APCCy7 (BD 560821). Cells were washed and fixed with 2% formaldehyde in PBS for 15 min at R.T. After washing with PEB buffer, the cells were permeabilized using buffer IV (BD Phosflow 560746) during 15 minutes at 4°C. Cells were then washed and incubated with anti-FOXP3-Pe (Invitrogen 12-5773-82) and anti-RORγt-BV421 (BD 562894) during 45 min at 4°C. Antibodies were used at 0.2 mg/mL. Cells were washed and fluorescent beads (Thermo Fisher C36950) were added to each sample at a concentration of 667 beads/mL to determine the absolute number of events of interest. Absolute numbers of Treg and Th17 cells were also used to determine the Th17/Treg ratio as previously reported (65). Finally, 300,000 events per sample were recorded using the BD LSRFFortessa X-20 cytometer from the Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile. FlowJo software (version 10.8) was used to analyze the immune cell populations. Concurrently, to analyze myeloid immune cell populations, a portion of 1.0×10⁷ splenocytes were directly stained with the following antibodies to analyze myeloid immune cell populations: anti-CD45-FITC (BioLegend 109806), anti-CD3-PercpCy5.5 (BioLegend 100218), anti-CD11b-PeCy7 (Invitrogen 25-0112-82), anti-CD49b-APC (BioLegend 103515), and anti-NK1.1-BV605 (BioLegend 108739) for Natural killers (NKs); anti-CD45-FITC (BioLegend 109806), anti-F4/80-BUV395 (BD 565614), anti-CD80-BV605 (BioLegend 104729), and anti-CD68-PeCy7 (BioLegend 137015) for M1-like macrophages; and anti-CD45-FITC (BioLegend 109806), anti-F4/80-BUV395 (BD 565614), anti-CD163-APC (BioLegend 156705), and anti-CD206-BV421 (BioLegend 141717) for M2-like macrophages, during 45 min at 4°C and darkness. All antibodies were used at 0.2 mg/mL. Cells were washed and analyzed by flow cytometry as previously described.

Sample sizes were determined using G*Power 3.1.9.7 software, based on a previous report (66). For pregnant mice groups, one-way ANOVA in F tests was conducted, with an effect size set to 0.80, α set to 0.05, and a test power set to 0.80, and considering 3 groups. This calculation yielded a total sample size of 21, with 7 mice allocated to each group. For experimental groups (Control-offspring, HTX-offspring, and HTX+T₄-offspring), sample size was determined using a mixed-effects model in F test ANOVA, with an effect size set to 0.5 (effect size specification according to Cohen, 1988 (67)), α set to 0.05, a test power set to 0.99, number of measurements set to 41 (comprising all independent measurements obtained from each evaluated parameter in this study) and considering 3 groups. This resulted in a total sample size of 90, distributing 30 mice per experimental group, and separating 15 mice per sex. This allocation of mice for behavioral assessment and subsequent molecular analysis on the same mice aligns with the reduction and refinement principles of the 3 Rs in bioethics (68). Data analysis and statistical tests were performed using Prism 9.0.2 software (GraphPad, Inc). Thyroid hormone levels were compared among pregnant mice groups using one-way ANOVA followed by Tukey’s post-hoc test. To conduct multiple comparisons between experimental groups and both sexes for each parameter evaluated in this study, while mitigating the impact of the “litter effect”, a mixed-effects model was employed followed by Tukey’s post-hoc test. This approach enhances the reliability and accuracy of the results, ensuring robust statistical analysis (69). Pearson correlation analysis was applied to associate maternal T₄ levels with all outcomes evaluated in the offspring. The results are presented as the mean ± standard error of the mean (SEM). Significance was established at p<0.05. p values in figures and text are presented as follows: *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001.

Gestational HTX was induced in pregnant mice by administering 0.025% w/v of methimazole (MMI) through the tap drinking water from embryonic day 10 (E10) to E14 (refer to Materials and methods). The E10 to E14 timeframe for MMI treatment was chosen because it is comparable to the first trimester of gestation in human brain development (70). MMI acts as an allosteric inhibitor of the thyroid peroxidase enzyme, hindering the coupling of molecular iodine to the tyrosine residues in the thyroglobulin molecule, thereby inhibiting the TH synthesis (52–55). Several reports have shown that short-term administration of this drug at this concentration induces gestational HTX rather than gestational hypothyroidism (52–55). To confirm the successful induction of gestational HTX in dams, the levels of tT₄, tT₃, and TSH were quantified from serum samples obtained on E14. Results are shown in Figure 1 . The levels of tT₄ in the MMI-treated dams were reduced compared to untreated dams, however, this decrease was reversed by the addition of T₄ during the MMI administration ( Figure 1A ). Additionally, the levels of tT₃ ( Figure 1B ) and TSH ( Figure 1C ) remained similar among untreated, MMI-treated, and MMI+T₄-treated dams. The exclusive reduction in tT₄ confirmed the successful induction of HTX. Additionally, tT₄, tT₃, and TSH were quantified in the Control, HTX, and HTX+T₄ offspring of both sexes on P50. The results indicate that the levels of tT₄ ( Figure 1D ) tT₃ ( Figure 1E ) and TSH ( Figure 1F ) were unchanged across the groups and sexes, indicating that the MMI treatment did not impact the offspring’s thyroid function in adulthood.

Retrospective studies in humans have associated gestational HTX with autistic behavioral traits in the offspring (17, 18). However, comprehensive causal studies, along with an investigation into the cellular and molecular mechanisms underlying these ASD-like manifestations in the offspring, have yet to be conducted. Therefore, using a mouse model, the primary aim of this study was to validate the notion that gestational HTX leads to the development of ASD-like behavioral alterations in the offspring. Furthermore, considering potential sex-specific effects, both females and males were included in the exploration of these outcomes. For this purpose, four behavioral tests were employed to evaluate repetitive behavior, anxious-like behavior, and social interaction abilities in the HTX-gestated offspring of both sexes and compare their performance with that of Control and HTX+T₄ offspring of both sexes. Behavioral testing was performed during adulthood given it is known that the brain reaches maturity at synaptic level and behavior-relevant areas, enabling a more precise investigation of ASD-like associated cognitive functions (70). Data are provided in Supplementary Table 1 .

First, repetitive behavior was assessed using the marble burying test, where increased number of buried marbles compared to the control group indicates ASD-like repetitive behavior (57, 58). Female and male Control, HTX, and HTX+T₄ offspring were individually placed in the testing cage containing twelve glass marbles for 30 min to evaluate their burying behavior. After that time, mice were returned to their respective cages and the number of buried marbles was counted (see Materials and methods). The percentages of buried marbles are shown in Supplementary Table 1.1 . The representative images illustrate the buried marbles of both the female Control ( Figure 2A ), HTX ( Figure 2B ), and HTX+T₄ ( Figure 2C ) offspring, as well as the male Control ( Figure 2D ), HTX ( Figure 2E ), and HTX+T₄ ( Figure 2F ) offspring. Quantified data indicate that both female and male HTX-offspring significantly buried fewer marbles than the Control and HTX+T₄-offspring ( Figure 2G , Supplementary Table 1-1.1 ), suggesting the HTX-gestated offspring do not manifest a repetitive ASD-like behavior.

Since anxiety is a widespread comorbid condition in ASD (71), the second evaluated aspect was the anxious-like behavior through the elevated plus maze (EPM) test (59, 60). Mice were individually placed in the cross-shaped maze and the time spent in both the open and closed arms and the number of entrances were registered (see Materials and methods). Results are shown in Figure 3 and Supplementary Table 1-1.2 . The findings indicate that both female and male HTX-offspring spent considerably less time exploring the open arms ( Figure 3A , Supplementary Table 1-1.2a ), which was consistent with a lower number of entries into the open arms ( Figure 3B , Supplementary Table 1-1.2b ), compared to the Control and HTX+T₄ offspring. Conversely, both female and male HTX-offspring spent significantly more time hiding in the closed arms, compared to the Control and HTX+T₄ offspring ( Figure 3C , Supplementary Table 1-1.2c ). No differences in the number of entrances to the closed arms were detected across experimental groups and sexes ( Figure 3D , Supplementary Table 1-1.2d ). Considering that when mice spending more time hiding in the closed arms of the maze and less exposure to open spaces points to anxious-like behavior, these results suggest that the HTX-offspring manifests a behavior indicative of anxiety.

Subsequently, two specific tests were employed to evaluate two different dimensions of social behavior. The first test was tube dominance, design to assess socially aggressive behavior, aiding in the differentiation between subordinate and dominant statuses based on the principles of the hierarchy species theory (61). The test involves positioning two mice from distinct experimental groups at opposite entrances of a transparent hollow tube and allowing them to enter. Once they meet in the middle, a confrontation occurs, with one of the two compelling its opponent to retreat to exit. To evaluate the aggressive behavior of the HTX-gestated offspring, the following confrontation design was employed: Control-offspring vs. HTX-offspring, HTX+T₄-offspring vs. HTX-offspring, and Control-offspring vs. HTX+T₄-offspring (details in the Materials and Methods section). The percentage of wins for each experimental group was determined and provided in Figure 4 and Supplementary Table 1-1.3 . The findings revealed that both female and male HTX-offspring displayed lower winning percentages when confronted with their respective female and male Control-offspring ( Figure 4A , Supplementary Table 1-1.3a ), and HTX+T₄-offspring ( Figure 4B , Supplementary Table 1-1.3b ), suggesting that the HTX-gestated offspring have a subordinate status in the context of confrontation. Additionally, when Control-offspring was confronted with HTX+T₄-offspring, regardless the sex, no significant differences in the winning percentage were detected ( Figure 4C , Supplementary Table 1-1.3c ).

The second social aspect evaluated was the social interaction preferences through the three-chamber social preference test (62). Mice from the three experimental groups were exposed to interact with different stimuli placed in opposite compartments of a three-chamber device to assess social preferences (details provided in Materials and methods). The percentage of time spent in each compartment and the number of entrances are illustrated in Supplementary Table 1-1.4 , Supplementary Figures 1 , 2 . The percentage of time spent for direct interaction with each stimulus is shown in Figure 5 and Supplementary Table 1-1.4 . During the first step of the test, female HTX-offspring spent more time in the nonsocial compartment only compared to the HTX+T₄ offspring, while male HTX-offspring spent more time in the nonsocial compartment compared to both the Control and HTX+T₄ offspring ( Supplementary Table 1-1.4a , Supplementary Figure 1A ). Likewise, male HTX-offspring spent more time in the nonsocial compartment even compared to female HTX-offspring ( Supplementary Table 1-1.4a , Supplementary Figure 1A ). Additionally, only male HTX-offspring made a higher number of entrances to the nonsocial compartment ( Supplementary Table 1-1.4a , Supplementary Figure 1B ) and spent more time directly interacting with the nonsocial stimulus compared to both the Control and HTX+T₄ offspring and even compared to female HTX-offspring ( Figure 5A , Supplementary Table 1.4c ). Conversely, only male HTX-offspring spent less time in the social compartment ( Supplementary Table 1-1.4d , Supplementary Figure 1C ), however, the number of entrances into the social compartment remained similar between experimental groups and sexes ( Supplementary Table 1-1.4e , Supplementary Figure 1E ). Finally, both female and male HTX-offspring spent less time directly interacting with the social stimulus compared to the Control and HTX+T₄ offspring ( Figure 5B , Supplementary Table 1-1.4f ).

In the second phase of the test, the nonsocial stimulus was replaced with a new stranger mouse (referred to as the ‘de novo stimulus’ or ‘Stranger 2’), while maintaining the first stranger mouse, now renamed as ‘Stranger 1’. Despite no differences were identified in the time spent in the Stranger 1 compartment ( Supplementary Table 1-1.4g , Supplementary Figure 2A ), only male HTX-offspring made more entries to that compartment compared to Control and HTX+T₄ offspring ( Supplementary Table 1-1.4h , Supplementary Figure 2B ). Additionally, both female and male HTX-offspring spent more time directly interacting with the Stranger 1 stimulus compared to Control and HTX+T₄ offspring ( Figure 5C , Supplementary Table 1-1.4i ). Conversely, HTX-offspring (both sexes) spent less time in the Stranger 2 compartment ( Supplementary Table 1-1.4j , Supplementary Figure 3C ), but only females displayed a lower number of entrances to the Stranger 2 compartment ( Supplementary Table 1-1.4k , Supplementary Figure 3D ). Finally, both female and male HTX-offspring spent less time directly interacting with the de novo stimulus compared to Control and HTX+T₄ offspring ( Figure 5D , Supplementary Table 1-1.4l ). Complementing these results with the first step, the HTX-gestated offspring show a reduced preference for interacting with the social stimuli but exhibit a stronger preference for interacting with the same social stimuli when the object is replaced with a second unfamiliar one in the second step of the test. These findings reinforce the notion of impaired interaction abilities resembling ASD-like behavior in this offspring.

Beyond behavioral alterations, immune dysregulation has also been associated with ASD, mainly by the determination of proinflammatory cytokines in blood samples of patients and the examination of neuroinflammatory traits and immune cell populations in secondary lymphoid organs in mouse models (45, 46). Hence, we investigated whether gestational HTX leads to ASD-like inflammatory status in offspring.

After conducting the behavioral tests, mice were euthanized, and specific cytokines were measured from serum samples (refer to Materials and Methods) ( Figure 6 , Supplementary Table 2-2.1 ). The results indicate that both female and male HTX-offspring had increased concentrations of the proinflammatory cytokines TNF-α ( Figure 6A , Supplementary Table 2-2.1a ), IL-6 ( Figure 6B , Supplementary Table 2-4.1b ), IL-1β ( Figure 6C , Supplementary Table 2-2.1c ), and IL-17A ( Figure 6D , Supplementary Table 2-2.1d ), in comparison to Control and HTX+T₄ offspring. Additionally, IL-1β was even higher in the male HTX-offspring than in females ( Figure 6C , Supplementary Table 2-2.1c ). The anti-inflammatory cytokine IL-10 was significantly reduced only in male HTX-offspring compared to male Control and HTX+T₄ offspring ( Figure 6E , Supplementary Table 2-2.1e ). The increased serum concentration of proinflammatory cytokines in the HTX-offspring is in concordance with ASD-like molecular features. Considering the neuroinflammatory component of ASD (47), IL-17A and its counterpart IL-10 were quantified in brain regions related to social cognition (prefrontal cortex (PFC) and hippocampus) ( Figure 7 , Supplementary Table 2-2.2 ). In the PFC, both female and male HTX-offspring had an increased concentration of IL-17A ( Figure 7A , Supplementary Table 2-2.2a ), while a reduction of IL-10 in the PFC was only observed in male HTX-offspring, compared to Control and HTX+T₄ offspring ( Figure 7B , Supplementary Table 2-2.2b ). Interestingly, the male Control-offspring had a higher content of IL-10 in the PFC compared to female Control-offspring ( Figure 7B , Supplementary Table 2-2.2b ). Moreover, IL-17A in the hippocampus was increased in HTX-offspring of both sexes ( Figure 7C , Supplementary Table 2-2.2c ), whereas IL-10 was reduced only in the hippocampus of male HTX-offspring compared to Control and HTX+T₄ offspring ( Figure 7D , Supplementary Table 2-2.2d ). These results suggest that the HTX-offspring exhibits a sex-dependent neuroinflammatory-like component which can also be indicative of an ASD-like phenotype.

Finally, to broaden our comprehension of the immunological aspect of the HTX-offspring regarding the proportion of immune cell populations (48, 49), total splenocytes were isolated, and myeloid and specific T cells were assessed by flow cytometry. Total splenocytes were subjected to an in vitro stimulation with PMA, ionomycin, and BFA. Thereafter, harvested cells were labeled with appropriate antibodies to identify Tregs and Th17 cells and analyzed by flow cytometry (see Materials and methods). The results are presented in Figure 8 and the employed gating strategy for Treg and Th17 cell selection is shown in Supplementary Figures 3A and B , respectively. The findings revealed that both female and male HTX-offspring had a lower absolute number of CD4⁺CD25^highFOXP3⁺ cells (Tregs), compared to Control and HTX+T₄ offspring ( Figures 8A, B ). Conversely, an increased absolute number of CD4⁺IL-17A⁺RORγt⁺ cells (Th17) was observed in HTX-offspring of both sexes, also compared to Control and HTX+T₄ offspring ( Figures 8C, D ). Consistent with these findings, both female and male HTX-offspring had a higher Th17/Treg ratio ( Figure 8E ). Myeloid cells, particularly M1/M2-like macrophages and NK cells were also evaluated in splenocytes by flow cytometry. The results are depicted in Figure 9 and the employed gating strategy is shown in Supplementary Figures 4A-C . Similarly, both female and male HTX-offspring had a higher absolute number of CD45⁺CD3^-F4/80⁺CD80⁺CD68⁺ cells (M1-like macrophages), than the Control and HTX+T₄ offspring ( Figures 9A, B ). Nonetheless, no differences were observed in the absolute numbers of CD45⁺CD3^-F4/80⁺CD206⁺CD163⁺ cells (M2-like macrophages) ( Figures 9C, D ) and CD45⁺CD3^-CD11b⁺NK1.1⁺CD49b⁺ cells (NKs) ( Figures 9E, F ). These findings, in conjunction with the previous ones, indicate that the HTX-offspring also exhibits an imbalanced proportion of immune cells, overall pointing to an exacerbated proinflammatory status that aligns with an ASD-like phenotype.

The effects of gestational HTX on offspring behavior and immune status encourage exploration into whether this condition also affects the expression of glutamatergic proteins, such as NLGN3 and HOMER1, whose altered levels are associated with ASD-like behaviors in mice (72). For this purpose, total proteins from both the PFC and hippocampus were isolated, and the expression of NLGN3 and HOMER1 was determined by using western blot (refer to Material and methods for details). The results of this analysis are shown in Figure 10 .

The expression levels of NLGN3 in the PFC were comparable among the HTX, Control, and HTX+T₄ offspring, regardless the sex ( Figures 10A, B ), however, only male HTX-offspring showed a significant increase in the expression levels of hippocampal NLGN3 compared to male Control and HTX+T₄ offspring ( Figures 10C, D ). At the same time, the expression levels of hippocampal NLGN3 in male HTX-offspring was significantly higher than in female HTX-offspring ( Figures 10C, D ). On the other hand, the expression levels of HOMER1 in the PFC was similar when comparing the HTX, Control, and HTX+T₄ offspring of the same sex ( Figures 10E, F ), however, it is worth noting that the expression levels of HOMER1 in the PFC of males in the Control, HTX, and HTX+T₄ offspring was simultaneously higher than in females of the Control, HTX, and HTX+T₄ offspring ( Figures 10E, F ). Finally, only male HTX-offspring showed higher expression levels of hippocampal HOMER1 compared to male Control, HTX+T₄, and even female HTX-offspring ( Figures 10G, H ). These results indicate that the effects of gestational HTX on the offspring go beyond behavior and immune response, but also induce sex-dependent alterations in the expression of glutamatergic proteins, which collectively can be interpreted as ASD-like outcomes.

Additionally, the results of this work were also collectively analyzed (without separation by sex) and are presented from Supplementary Figures 5 – 14 .

In this study, gestational HTX was defined as a reduction of maternal tT₄ within a range of [0.1 – 1.9] (μg/dL), whereas the Control and HTX+T₄ ranges were established at [2.2 – 4.5] (μg/dL). Nonetheless, it is relevant to note that these ranges were defined within the context of natural variability. For this reason, to investigate whether maternal T₄ levels could influence the development of ASD-like phenotypes in the offspring, a Pearson correlation analysis was conducted (73) to measure the association between maternal T₄ levels (considering all HTX, Control, and HTX+T₄ dams’ levels) and all the observed outcomes ( Table 1 , Supplementary Figure 15 ).

The results revealed significant positive correlations between maternal T₄ levels and the percentage of buried marbles (r=0.578, p<0.00001), the number of entries into open arms during the EPM test (r=0.591, p<0.00001), the percentage of time directly interacting with social stimuli in the Three-chamber test (r=0.559, p<0.00001), and IL-10 concentration in serum (r=0.578, p<0.00001). Conversely, significant inverse associations were noted between maternal T₄ levels and IL-1β (r=-0.641, p<0.00001) and IL-17A (r=-0.721, p<0.00001) levels in the blood ( Table 1 ). These associations suggests that maternal thyroid status influence the development of ASD-like phenotypes in the offspring.

Additional significant correlations between behavioral parameters and proinflammatory outcomes are shown in Supplementary Figures 17A, B . Interestingly, the IL-1β levels, regardless of sex, were negatively correlated with specific behavioral parameters, including the percentage of buried marbles in the MB test (r=-0.580, p<0.00001), the number of entries into the open arms in the EPM test (r=-0.600, p<0.00001), the time of direct interaction with the social stimuli in the three-chamber test (r=-0.586, p<0.00001), and positively correlated with the time of direct interaction with the Stranger 1 stimuli in the three-chamber test (r=0.653, p<0.00001) ( Supplementary Figure 15 ). At the same time, the IL-1β levels were positively correlated with other proinflammatory outcomes, including the blood levels of IL-17A (r=0.632, p<0.00001) and TNF-α (r=0.649, p<0.00001), and the number of M1-like macrophages in the spleen (r=0.588, p<0.00001) ( Supplementary Figure 15 ). Similar trends are evident in the associations between the blood levels of IL-6, IL-17A, and TNF-α with behavioral parameters ( Supplementary Figure 15 ). Moreover, the levels of IL-17A in the hippocampus were negatively correlated with the percentage of buried marbles (r=-0.640, p<0.00001), while the IL-17A levels in the PFC were positively correlated with the levels of TNF-α in blood (r=0.634, p<0.00001) ( Supplementary Figure 15 ). No significant correlations were identified regarding the levels of IL-10 in either the PFC or hippocampus ( Supplementary Figure 15 ). At the cellular level, the Th17/Tregs ratio was negatively correlated with the MB test performance (r=-0.603, p<0.00001), while the number of M1-like macrophages showed a negative correlation with the number of entries to the exposed arms in the EPM test (r=-0.557, p<0.00001), and a positive correlation with the direct interaction time with the Stranger 1 stimuli (r=0.666, p<0.00001), and the blood levels of IL-1β (r=0.588, p<0.00001), IL-6 (r=0.582, p<0.00001), IL-17 (r=0.636, p<0.00001), and TNF-α (r=0.630, p<0.00001) ( Supplementary Figure 15 ).