This study used the L7-Pcp2 Cre/Tsc1 floxed animal model of autism in mice (Mus musculus) with a C57BL/J6 genetic background (Jackson Laboratories). After weaning, mice were housed with littermates of the same sex at Miami University’s Laboratory of Animal Resources facility, with controlled temperature and humidity and a 12:12 light cycle starting at 7:00 am. Mice were provided with water and a standard rodent diet ad libitum (Lab Diet 5,001, Cincinnati Lab Supply Inc., Cincinnati, OH, USA). All animal experimentation was approved by the Institutional Animal Care and Use Committee of Miami University (protocol #1044). Cre-negative Tsc1^flox/flox mice were crossed with Cre-positive Tsc1^flox/wt mice to generate Cre-positive Tsc1^flox/wt, referred to here as Tsc1^mut/wt, and Cre-positive Tsc1^flox/flox mice, referred to here as Tsc1^mut/mut mice for experiments. These pairings also generated Cre-negative Tsc1^flox/flox mice, which were used as littermate controls in behavioral studies. L7-Pcp2 Cre-positive animals (lacking loxP sites) and Cre-negative Tsc1^flox/flox have previously been shown to not differ from wild type mice in behavioral assays (Barski et al., 2000; Ehninger et al., 2008; Tsai et al., 2012, 2018). Hemizygous Cre-positive animals were also crossed with the Ai14 Cre reporter tool strain (Jackson Laboratory, strain # 007914).

Parasagittal cerebellar slices (25 μm) of fixed and cryoprotected brains were used for imaging the tdTomato fluorescent signal in Cre-positive Purkinje neurons. Images were acquired using a Zeiss LSM-710 confocal microscope using either a 63X oil emersion objective or 20X objective. Z-series stack images with 1 μm intervals or single plane snapshots were used to assess tdTomato expression in Cre-reporter mice.

Spontaneous and evoked action potentials were recorded from Purkinje neurons in parasagittal brain sections acutely isolated from 5 to 8 week-old animals. Brain sections were perfused with warmed (33–34°C) artificial cerebral spinal fluid (ACSF) that was continuously bubbled with 95% O2/5% CO₂. Borosilicate glass recording pipettes were filled with an internal solution containing (in mM): 0.2 EGTA, 3 MgCl₂, 10 HEPES, 8 NaCl, 4 Mg-ATP, and 0.5 Na-GTP. Recording pipette resistances were 2–4 MΩ. Electrophysiology records were acquired using a dPatch amplifier and SutterPatch acquisition software (Sutter Instrument). Evoked action potential recordings were taken following gap-free spontaneous firing recordings. For evoked firing protocols, cells were initially injected with a −0.5 nA hyperpolarizing current for 100 ms before stepping current injections to 0, 0.1, 0.2, and 0.3 nA for 700 ms. Input resistance and capacitance measurements were obtained from a whole-cell voltage-clamp recording in which membrane voltage was stepped from a −80 mV holding potential to −90 mV (for 100 ms) and to −70 mV (for 100 ms) in a subsequent sweep. Membrane capacitance was calculated by integrating the area of fast capacitive transient current during voltage steps and dividing by the change in membrane voltage (10 mV). Electrophysiology recordings were analyzed using SutterPatch (Sutter Instrument), Microsoft Excel (Microsoft), and Prism (GraphPad) software.

Behavioral tests were conducted on male and female mice in cohorts of animals that were 9–11 weeks of age and repeated on animals in the Tsc1^mut/mut and control cohorts at 16–24 weeks of age. A single experimenter performed all behavioral tests for each age group. Age ranges reflect the amount of time necessary for the individual experimenter to complete these behavioral tests. For each age group, elevated balance beam tests were completed before social interaction tests. All tests were performed during the light cycle between 9:00 and 17:00.

Statistical analyses were completed using GraphPad Prism 8 software (GraphPad Software Inc., La Jolla, CA, USA). Data from whole cell electrophysiology recording measurements were compared across sexes and genotypes using unpaired Student’s t-tests. Data from elevated balance beam experiments were compared across sexes and genotypes using repeated measures (RM) two-way ANOVA tests. Social interaction experiment results were assessed using two-way ANOVA tests; comparisons of the percent of time investigating the unfamiliar animal used sex and genotype as between-subject factors, and comparisons of the number of direct contacts used sex and chamber (containing stranger animal or containing empty cage) as between-subject factors within each genotype. Šídák’s corrections were applied in multiple comparisons tests. p < 0.05 was considered to be statistically significant. p-values for all statistical tests comparing behavioral performance measures across sex are reported in Supplementary Table 1 for the 9–11 week animal cohorts, and in Supplementary Table 2 for the 16–24 week-old animal cohorts.

A Cre-LoxP recombination strategy was used to selectively delete Tsc1 from mouse cerebellar Purkinje neurons. Tsc1 mutant mice with loxP sites flanking exons 17 and 18 of the Tsc1 gene (Kwiatkowski et al., 2002; Jackson laboratory, strain # 005680) were crossed with a transgenic mouse strain that expresses Cre recombinase under the control of the mouse Purkinje cell-specific L7 promoter (Jackson laboratory, strain # 010536; Figure 1A). Progeny from this cross included hemizygous Cre-positive animals with either heterozygous or homozygous Tsc1 floxed alleles, used as test group animals, and Cre-negative animals. Cre-negative animals with homozygous Tsc1 floxed alleles were used as control group animals. To verify selective Cre expression in cerebellar Purkinje neurons, hemizygous Cre-positive animals were also crossed with the Ai14 Cre reporter tool strain (Jackson Laboratory, strain # 007914; Figure 1A), which express tdTomato fluorescence following Cre-mediated recombination (Madisen et al., 2010). Parasagittal cerebellar sections from the Cre-positive progeny of this cross revealed robust tdTomato expression that was selective to Purkinje neurons in the cerebellum (Figure 1B) of both male and female animals (see Supplementary Figure 3). Cre-expressing Purkinje neurons were identified in both adult (Figure 1B) as well as neonatal (P16) animals. In these sections, no tdTomato fluorescence was identified in cells outside the Purkinje neuron layer.

Targeted deletion of Tsc1 from male (mouse) cerebellar Purkinje neurons has previously been shown to result in attenuated repetitive firing in adult Purkinje neurons (Tsai et al., 2012). To determine if Cre-positive Tsc1^flox/flox progeny (described above), referred to here as Tsc1^mut/mut, have similar deficits in repetitive firing, and to test if changes in membrane excitability is conserved across Tsc1^mut/mut Purkinje neurons from male and female animals, we performed whole cell patch current-clamp recordings from adult (5–8 week-old) Tsc1^mut/mut and wild type Purkinje neurons in acutely isolated cerebellar sections (see Methods). These recordings were performed at near physiological temperatures (33–34°C) and revealed the spontaneous repetitive firing frequency in Tsc1^mut/mut Purkinje neurons is significantly (p < 0.001, Student’s t-test) reduced compared to age-matched wild type cells (see Figures 2A,B). Reduced repetitive firing frequencies measured in Tsc1^mut/mut Purkinje neurons were not, however, significantly different between cells isolated from male and female mice (Figure 2C). Depolarizing current injections after a common hyperpolarizing (−0.5 nA) current step revealed similar impaired evoked firing in Tsc1^mut/mut Purkinje neurons, and no difference between male and female Tsc1^mut/mut Purkinje neurons (Figures 2D,E). Measures of membrane capacitance were also not significantly different across male and female Tsc1^mut/mut Purkinje neurons, however, membrane input resistance was found to be significantly (p = 0.02, Student’s t-test) lower in male Tsc1^mut/mut Purkinje neurons compared to cells from Tsc1^mut/mut females (Figure 2F). No sex-specific differences in capacitance and input resistance were measured in wild type controls cells (data not shown).

To examine potential sex-specific differences in motor coordination and balance resulting from the targeted deletion of Tsc1 in Purkinje neurons, the elevated balance beam behavioral assay (Carter et al., 2001) was performed on 9–11 week-old and 16–24 week-old mice. For each age-group, the balance beam test was performed across four consecutive testing days. The amount of time it took the animal to traverse a 10 mm wide beam (see Figure 3A), and the number of hindlimb foot-slips while traversing the beam (see Methods) were recorded. The mean (±SEM) time to cross and hindlimb foot-slips are plotted in Figure 3 for each of the four testing days. Results for the testing of 9–11 week-old animals are plotted in Figures 3B,C, and again for the 16–24 week-old age group in Figures 3D,E. These data revealed that at both ages (9–11 week-old and 16–24 week-old) Tsc1^mut/mut animals have impaired motor performance compared to control (Tsc1^flox/flox; Cre -) mice, taking significantly (Figures 3B,D; p < 0.0001, RM two-way ANOVA) longer to cross the 10 mm balance beam and having significantly (Figures 3C,E; p < 0.0001, RM two-way ANOVA) more hindlimb foot-slips while traversing the 10 mm beam. Animals with targeted deletion of a single Tsc1 allele (Tsc1^mut/wt) performed similarly to control animals (p = 0.9866, RM two-way ANOVA, Figure 3B). For all genotypes tested on the 10 mm balance beam, no significant differences were recorded across sex (within a genotype). These data suggest there are no significant differences between sexes in ASD-associated deficits in fine motor coordination and balance. This testing paradigm was also performed using a more challenging, 8 mm balance beam, and similar to the 10 mm balance beam, revealed no significant differences across sex within a given genotype (see Supplementary Figures 1, 2).

A three-chamber apparatus was used to measure differences between sexes in impairments for social approach preference and social novelty preference in Tsc1^mut/mut mice. The three-chamber apparatus consisted of an empty center chamber and two exterior chambers with wire cups large enough to comfortably hold an adult mouse. In this assay, preference for social approach is reflected as an increase in the investigation of an unfamiliar animal (in one of the wire cups) compared to the alternate empty wire cup (Figure 4A). In Figure 4B, the mean (±SEM) percentage values for the time spent investigating the unfamiliar mouse, as a percentage of the total time spent investigating the unfamiliar mouse and the empty wire cup, are plotted across sex and genotype for 9–11 week-old cohorts. Results from these tests reveal Tsc1^mut/mut mice spend a significantly (p < 0.05, two-way ANOVA) lesser percentage of time investigating the unfamiliar animal compared to the control (Tsc1^flox/flox/Cre-negative) cohort. In this analysis, animals with targeted deletion of one Tsc1 allele (Tsc1^mut/wt) were not significantly different from Tsc1^mut/mut or control animal cohorts. Similar to the results from the elevated balance beam assay, within each tested genotype (Tsc1^mut/mut, Tsc1^mut/wt, and control), we found no significant difference in these values based on sex (p = 0.2681, two-way ANOVA, Supplementary Table 1).

As another method to quantify the performance on the social approach behavioral test, the number of the test mouse’s direct contacts with the cage holding the unfamiliar animal and the number of direct contacts with the empty cage are plotted in Figures 4C–E, across sexes, for Tsc1^mut/mut, Tsc1^mut/wt, and control genotypes, respectively. These data reveal control mice of both sexes exhibit social approach preference (p < 0.01, two-way ANOVA, Figure 4C), having a higher number of direct contacts with the unfamiliar animal compared to the empty cage. Interestingly, while male Tsc1^mut/mut and Tsc1^mut/wt mice did not exhibit significant preference for the unfamiliar animal over the empty cage (Tsc1^mut/mut: p = 0.1627, Tsc1^mut/wt: p = 0.1065, two-way ANOVA, Figures 4D,E), indicating impaired social approach behavior in these genotypes, female Tsc1^mut/wt mice did exhibit significant preference for the unfamiliar animal (p < 0.01, two-way ANOVA, Figure 4D) similar to control animals; a phenotype not shared by Tsc1^mut/wt males.

In older (16–24 week-old) animals, however, the social approach testing paradigm revealed a clear difference between sexes in Tsc1^mut/mut mice (Figures 5A,B). Male Tsc1^mut/mut mice spent a significantly smaller proportion of time than female test animals investigating the unfamiliar mouse (mean difference: -18.43 ± 5.1, p < 0.01, N = 8 males and N = 9 females, two-way ANOVA, Figure 5B). No difference between sexes was measured in control mice (p = 0.99, N = 8 males and N = 6 females, two-way ANOVA, Figure 5B). In Figures 5C,D, the number of the test mouse’s direct contacts with the cage holding the unfamiliar animal versus the empty cage are plotted between sexes for control and Tsc1^mut/mut animals, respectively. Consistent with the results presented in Figure 5B, control test mice from both sexes exhibited a significant preference for contacts with the unfamiliar animal compared to the empty cage (p = <0.0001, two-way ANOVA, Figure 5C). In control mice, we measured no significant difference in preference between males and females (p = 0.1123, two-way ANOVA, Figure 5C). Conversely, male Tsc1^mut/mut mice exhibited a significantly (p = 0.0117, two-way ANOVA, Figure 5D) lesser preference for contacting the unfamiliar animal compared to female Tsc1^mut/mut mice. Differences in the number of contacts (with the empty cage or with the unfamiliar mouse cage) may be driven by sex-specific differences in the total time spent in each chamber. We explored this possibility by analyzing the total time spent in each chamber for male and female animals across genotypes during the social approach and social novelty tasks (see Supplementary Figures 4, 5). These data reveal that in older (16–24 week-old) control and Tsc1^mut/mut cohorts, male and female mice did not significantly differ in the amounts of time in each chamber.

To test for a preference for social novelty across sexes, the three-chamber apparatus was used to measure the preference of test animals for investigating an unfamiliar animal (in one chamber) compared to a familiar animal (in the alternative chamber). An image and cartoon depiction of this testing paradigm is presented in Figure 6A. Social novelty preference is measured as the preference for investigating the unfamiliar animal compared to the familiar animal. These tests were performed directly after social approach testing (see Methods). Also similar to our previous behavioral tests, in 9–11 week-old animals, a cohort of Tsc1^mut/wt animals was also tested. Interestingly, in these social novelty preference tests, we measured no significant difference in preference for investigating the unfamiliar animal across (9–11 week-old) control, Tsc1^mut/wt, and Tsc1^mut/mut genotypes (p = 0.2972, two-way ANOVA, Figure 6B). This is in contrast to previous findings in which Tsc1 mutant mice were found to exhibit no preference for investigating the socially novel animal (Tsai et al., 2012). Interestingly, within each genotype, control and Tsc1^mut/wt genotypes were found to have significant differences in social novelty preference based on sex. For instance, in control mice, female test animals investigated the unfamiliar animal a significantly lower percentage of time compared to male test animals (mean difference: 18.9 ± 5.8, p = 0.0055, two-way ANOVA, Figure 6B). These differences in preference based on sex were not, however, measured in the Tsc1^mut/mut cohort (p = 0.9986, two-way ANOVA, Figure 6B).

Interestingly, when number of cage contacts was used to assess social novelty preference in the 9–11 week-old animals, we measured no sex-specific differences in control animals (p = 0.9222, two-way ANOVA, Figure 6C), however, in Tsc1^mut/wt (Figure 6D) and Tsc1^mut/mut (Figure 6E) cohorts, there were significant differences based on sex, with females in both Tsc1^mut/wt and Tsc1^mut/mut cohorts contacting the unfamiliar animal cage less than males of the respective cohort (Figures 6D,E). Together, these data suggest female test animals, from all genotypes tested, exhibit a lower preference for social novelty compared to male test animals; importantly, this sex-specific difference was measured in control animals, which makes assessing sex-specific effects of Tsc1 deletion difficult. This difference between sexes was not found to be present in the Tsc1^mut/mut group. A previous study which compared social behavioral phenotypes of prairie voles to C57BL/6 mice also identified sex-specific differences in wild type C57BL/6 mice social novelty preference (Wu et al., 2022). Similar to the results presented here, Wu et al. (2022) determined male mice exhibit significantly greater social novelty preference (Figure 6B) compared to female mice.

Social novelty performance was assessed again in 16–24 week old control and Tsc1^mut/mut cohorts, with results presented in Figures 6F–H. 16–24 week-old Tsc1^mut/mut mice spent a significantly smaller percentage of time investigating the unfamiliar animal than control mice (p = 0.0438, two-way ANOVA, Figure 6F), revealing a deficit in typical (control) social novelty preference. No significant differences between male and female test animals of either genotype were measured (control-p = 0.5644, Tsc1^mut/mut-p = 0.0917, two-way ANOVA, Figures 6G,H). Male and female control mice exhibited a significant preference for contacting the socially novel animal (p < 0.0001, two-way ANOVA, Figure 6G), and, in contrast to the results from the 9–11 week-old cohorts, Tsc1^mut/mut mice did not exhibit preference for the unfamiliar mouse based on contact number (males: p = 0.1889, females: 0.0942, two-way ANOVA, Figure 6H). These data suggest older (16–24 week-old) Tsc1^mut/mut develop an impairment in social novelty preference that is not significant at younger (9–11 week-old) ages.

Numerous ASD-linked behavioral deficits were originally identified in (male) Tsc1^mut/mut mice and these phenotypes were linked to two types of cerebellar impairment. In older (>10 week-old) Tsc1^mut/mut animals, Tsai et al. (2012) measured reduced numbers of Purkinje neurons in the cerebellar vermis, as well as a significant increase in the proportion of Purkinje neurons expressing apoptotic factors TUNEL and Caspase 9. No changes in Purkinje neuron number were measured in younger (6 week-old) animals, however, there were marked deficits in Purkinje neuron repetitive firing properties (Tsai et al., 2012), similar to the results reported here (Figure 2). Because impaired behaviors were also recorded in younger (~6 week-old animals) it’s likely that Tsc1 deletion causes cerebellar circuit deficits through both impaired Purkinje neuron excitability, and, as the animals become older, by way of Purkinje neuron apoptosis. The stand-out result from the current investigation is that in older Tsc1^mut/mut animals, impaired social interaction behavior (social approach preference) is more severe in Tsc1^mut/mut males compared to females, and importantly, this sex-specific difference is not present in behaviors related to balance and motor coordination. This result indicates the sex-specific impairments vary across distinct cerebellar circuits. However, the progression of Purkinje neuron apoptosis in Tsc1^mut/mut animals has not been measured and directly compared across males and females.

Previous work has also suggested the selective deletion of Tsc1 from Purkinje neurons differentially affects distinct cerebellar circuits. Rapamycin treatment, which has been shown to rescue deficits associated with Tsc1 deletion (Meikle et al., 2008; Zhou et al., 2009), was administered at various ages of Tsc1^mut/mut mice to determine at which ages the treatment was effective in rescuing behavioral impairments. Interestingly, the ages at which rapamycin was effective at alleviating behavioral impairments were dependent on the type of behavior being rescued. For instance, rapamycin treatment in 10 week-old animals rescued deficits in motor function (measured via the rotarod test), but deficits in social interaction behaviors were not improved (Tsai et al., 2018). A clear hypothesis for this difference in rescue is that rapamycin-mediated rescue is no longer possible if Purkinje neurons in the relevant cerebellar circuit have undergone apoptosis. Extended to the work presented here, it may be that in male Tsc1^mut/mut animals, there is greater apoptosis and Purkinje neuron loss in circuits contributing to social interaction behaviors compared to Tsc1^mut/mut females, and that this sex-specific difference does not occur in cerebellar circuits contributing to balance and motor coordination. It will be interesting to test these ideas directly. Purkinje neurons functional in circuits dedicated to balance and motor coordination have been shown to localize to spinocerebellar (vermis and paravermis) regions (Unverdi and Alsayouri, 2024); while Purkinje neurons functional in higher order behaviors, such as social interactions have been associated with Purkinje neurons localized in hemispheric Crus I and Crus II cerebellar subregions (Matas et al., 2021; Stoodley et al., 2017; Van Overwalle et al., 2020).

In addition, or perhaps in place of sex-specific differences in Purkinje neuron apoptosis, there may be sex-specific and circuit-specific differences in the deficits in Purkinje neuron membrane excitability of Tsc1^mut/mut animals. It’s of note that the sex-specific difference in social approach behaviors of Tsc1^mut/mut animals was not measured in younger (9–11 week-old) animals. This result is in line with the progressive nature of behavioral deficits in Tsc1^mut/mut animals (Tsai et al., 2018; 2012), which also may result from progressive Purkinje neuron apoptosis that occurs differentially across distinct cerebellar circuits. Beyond identifying the cellular and circuit-level substrates responsible for the sexual dimorphism in this ASD model, it will be interesting to also investigate the physiological drivers of such differences. For instance, there is some evidence that female sex hormones may provide a neuroprotective function in neurodegenerative disorders (Sumien et al., 2021; Túnez et al., 2006; Henderson et al., 1994; Garcia-Segura et al., 2001), resulting in sex-specific outcomes, although differences in the expression properties of sex chromosomes and/or sex hormone receptor expression (Ng and Hazrati, 2022) may also contribute to these differences. In future work aimed at effectively linking Tsc1^mut/mut behavioral impairments with impairments in cerebellar functioning, it will be necessary to categorize experiments based on animal sex and age, and in cellular-based experiments, careful localization and targeting of Purkinje neurons should be used to link the functioning of discrete cerebellar circuits with the relevant behavioral phenotype.