Autism spectrum disorder (ASD) is a highly prevalent neurodevelopmental disorder that is characterized by impaired social interaction and communication, repetitive behaviors, and restricted interests (1, 2). Early behavioral intervention is recommended for children diagnosed with ASD. Currently, its core symptoms cannot be cured and there is a need to develop pharmacological treatments. In order to develop effective pharmacological treatments that can be started during the early developmental stage, the pathogenesis of ASD needs to be understood. As extensively demonstrated in both humans (3) and rodent models (4), MeCP2 is one of the few genes known to play a causal role in ASD (5). Using genetically manipulated rodent models, researchers have demonstrated that both loss and gain of MeCP2 function can alter synaptic transmission and disrupt the overall excitation/inhibition balance in neural circuits (6). Consistent with this, dendritic spines in cortical neurons, which are postsynaptic to excitatory synapses and a proxy for synaptic densities, are affected in people with ASD (7). In fact, multiple ASD-related genes are involved in synaptic function (8).

Evidence for the possibility that loss of MeCP2 function results in developmental dysregulation of N-methyl-D-aspartate receptor (NMDAR) expression has attracted interest to NMDAR as a therapeutic target for rett syndrome (RTT), a neurodevelopmental disease caused by disruption of the MeCP2 gene (9, 10). One class of molecules that have shown promise in preclinical models of RTT are channel-blocking NMDAR antagonists (10). Kron et al. (11) demonstrated that treatment of heterozygous female MeCP2 mutant mice with a subanesthetic dose of ketamine (8 mg/kg) is highly effective in acutely reversing disease phenotypes, including abnormal patterns of neuronal activation in cortical and subcortical structures as well as sensorimotor dysfunction. Although MeCP2 gain-of-function and loss-of-function in genetically engineered rodents recapitulates typical pheno-types of patients with autism, it is still not known where ketamine affects the rodent brain, and whether/how it relates to autism pathology due to MeCP2 mutation. Therefore, the present study investigated the induction of ASD behaviors by knocking down MeCP2 in the hippocampus, and the effects of ketamine on the behaviors and synaptic molecules in the hippocampus of rats infused with lenti-shMeCP2.

Behavioral validation of the neonatal VPA-induced animal models of autistic-like behaviors

VPA rats were generated by intraperitoneally injecting pregnant SD rats on embryonic day 12.5 (E12.5) with a single dose of VPA (500 mg/kg) (Fig. 1A). Behavioral assays showed that the VPA rats did not display any difference in locomotor activity compared to saline rats (Fig. 1B: t₅₄ = 1.27, P = 0.2095). However, the VPA rats spent less time in the middle zone in the OFT (Fig. 1B: t₅₄ = 2.139, P < 0.05), indicating that they had anxiety-like behaviors. Autistic-like behavior is characterized by repetitive behaviors (12). To test if VPA rats displayed this type of behavior, we performed the self-grooming test. The VPA rats spent significantly longer self-grooming than the saline rats (Fig. 1C: t₅₄ = 2.247, P < 0.05). Aberrant reciprocal social interaction is a core symptom of autistic-like behaviors (13, 14). We evaluated social interaction with the social approach. In this assay, the VPA rats spent less time sniffing at the social stimulus than the saline rats (Fig. 1D: t₁₈ = 3.59, P < 0.01), indicating that they displayed impaired social interaction.

Ketamine ameliorates autistic-like social deficit features in VPA rats

Next, we used a three-chamber apparatus to assess sociability and social novelty preference for social interaction, which may be relevant to autistic-like behaviors (15, 16). In the three-chamber sociability test, if the test animal spends more time with the empty wire cage (E) than with the wire cage containing the stranger (S), this points to a deficit of sociability (Fig. 1E). In the first session, we assessed sociability by measuring staying time in the compartment with a stranger rat in the wire cage versus in the one with the empty wire cage. We found that VPA rats treated with saline demonstrated significantly reduced social interaction compared with saline rats, as indicated by the relative amount of time spent investigating the stranger cage compared with the empty cage (Fig. 1F). This measurement yields the sociability index (Fig. 1G). To determine whether ketamine ameliorates autistic-like social deficits in the VPA rats, we tested the effect of ketamine treatment on social inter-action in the three-chamber test. Administration of ketamine (20 mg/kg) to VPA rats 1 h before the three-chamber test significantly attenuated the reduction in social interaction, restoring this behavior to levels comparable to those in saline rats. In contrast, ketamine treatment of saline rats had no effect on investigation time (Fig. 1F) and sociability index (Fig. 1G). In addition, VPA rats spent significantly less time sniffing around a caged stranger rat when analyzed by sniffing time (Fig. 1H) and sniffing index (Fig. 1I), and both of these measures were recovered by ketamine treatment.

In the second session of the three-chamber test we assessed social interaction by measuring the preference for the stranger1 cage (S1) versus the familiar cage (F) (Fig. 1E). A new rat was added to the previously empty compartment and social preference between the familiar rat and novel rat was assessed by measuring the time spent close to the cage with the familiar versus the stranger rat. VPA rats treated with saline displayed a similar social novelty preference to that of saline rats treated with (Fig. 1J, K). Ketamine (20 mg/kg) did not significantly alter the investigation time (Fig. 1J) and preference index (Fig. 1K) both in saline and VPA rats. Similar results were obtained using preference sniffing time as measured by sniffing time (Fig. 1L) and preference sniffing time index (Fig. 1M). Together, these results indicate that prenatal VPA induces autism-like social deficits in social interaction, similar to previous findings (17) and ketamine alleviates the observed social deficits (Fig. 1J-M).

Ketamine recovers PTEN expression in VPA rats

We first examined whether endogenous MeCP2 levels were decreased in VPA rats and found that MeCP2 mRNA (Supplementary Fig. 1A: t₁₂ = 3.386, P < 0.01) and protein (Supplementary Fig. 1B: t₆ = 2.528, P < 0.05) levels were indeed significantly lower in the VPA rats on postnatal day 28 (P28). Given this result, we next examined the effect of ketamine on MeCP2 expression by investigating the changes in expression of various regulators related to synaptic function and MeCP2 target genes. Quantitative real-time PCR analyses indicated that the level of Pten mRNA was significantly lower in hippocampal lysates from the VPA rats, while the levels of Psd95, Glur1, Synapsin1, Rab3d and Vamp3 mRNAs were largely unchanged (Supplementary Fig. 1C, D). There are reports that abnormalities in PTEN lead to neurological disorders such as autism, seizures and schizophrenia (18-21). Ketamine treatment significantly increased Pten mRNA, while the Psd95, Glur1, Synapsin1, Rab3d and Vamp3 mRNAs were not altered (Supplementary Fig. 1C, D).

Hippocampal MeCP2 knockdown produces autistic-like behaviors in rats

To test whether hippocampal MeCP2 knockdown generates autistic-like behaviors, we infused lentivirus expressing shRNA targeted against rat MeCP2 (lenti-shMeCP2) into the dentate gyrus (DG), 4 weeks before ketamine injection (Fig. 2A). A re-covery period of 4 weeks was chosen because effective knock-down was achieved by infusion of lentivirus-mediated shMeCP2 into rats after 21 days (22). Lenti-shMeCP2 infusions lowered MeCP2 mRNA (Fig. 2B: t₁₄ = 6.169, P < 0.001) and protein (Fig. 2C: t₈ = 3.072, P < 0.05) levels in the hippocampal DG of rats compared to control lenti-shNC rats. We investigated whether the deficiency in MeCP2 led to autistic-like behaviors, including anxiety and repetitive behaviors. The locomotor activity of the MeCP2 knockdown rats was unchanged (Fig. 2D: t₅₁ = 0.6303, P = 0.5313), but they spent less time in the middle zone than the lenti-shNC rats, indicating that they ex-perienced increased anxiety (Fig. 2D: t₅₁ = 2.129, P < 0.05). They also spent significantly more time self-grooming than the lenti-shNC rats (Fig. 2E: t₅₁ = 2.431, P < 0.05), and less time sniffing in the direct social approach (Fig. 2F: t₄₁ = 4.675, P < 0.001). Taken together, these results indicate that reducing hippo-campal MeCP2 levels leads to pronounced autistic-like beha-viors such as anxiety, repetitive behavior, and social deficit.

Ketamine ameliorates autistic-like social deficits features in the hippocampal MeCP2 knockdown model

We examined the impact of hippocampal MeCP2 knockdown on social deficits in young (4 weeks) male rats in the three-chamber test. Lenti-shMeCP2 rats displayed significantly reduced social interaction compared with lenti-shNC rats, as indicated by the relative amount of investigation time in a stranger cage compared with an empty cage (Fig. 3A), and the sociability index (Fig. 3B). When a single injection of ketamine at a dose of 20 mg/kg was applied 1 h prior to behavioral tests it significantly elevated the sociability index of the lenti-shMeCP2 rats, suggesting that ketamine alleviates the observed social deficits in the lenti-shMeCP2 rats (Fig. 3B). Similar results were obtained for sociability sniffing time and sniffing index (Fig. 3C, D) as well as social novelty preference, investigation time and sniffing time (Fig. 3E-H). Taken together, these results indicate that MeCP2 knockdown rats display pronounced deficits in social interaction, and ketamine administration can rescue the autistic-like social deficits of these rats.

Ketamine ameliorates the levels of synaptic molecules in MeCP2 knockdown rats

We then sought to determine the molecular mechanisms that may underlie the amelioration of social deficits by ketamine. Glutamate receptors are a potential key target of ketamine since diminished synaptic signals at glutamatergic synapses are strongly linked to autistic-like phenotypes, including social deficits and repetitive behaviors (23, 24). We examined synaptic plasticity-related gene expression in rats infused with lenti-shMeCP2 into the hippocampal DG. Quantitative real-time PCR analyses indicated that the level of Glur1 mRNA was significantly lower in hippocampal DG lysates from lenti-shMeCP2 rats, while the levels of Psd95, Synapsin1, Pten, Rab3d and Vamp3 mRNAs were largely unchanged (Fig. 4). Although the level of Psd95 mRNA was unaffected in lenti-shMeCP2 rats, its level, together with that of Glur1 was significantly elevated upon ketamine treatment, while Synapsin1, Pten, Rab3d and Vamp3 mRNAs were unchanged (Fig. 4). Consistently, ketamine administration increased levels of the postsynaptic proteins GluR1 in lenti-shMeCP2 rats but not in lenti-shNC rats (Supplementary Fig. 2A). However, ketamine only induced a tendency to increase the expression of PSD95 in lenti-shMeCP2 rats (Supplementary Fig. 2B).

MeCP2 loss-of-function in genetically-engineered animals including rodents produces typical features of autism, yet where MeCP2 loss affects the rodent brain and whether/how this relates to autism pathology remain unknown. Here we report marked and reproducible effects of knockdown of hippocam-pal MeCP2 on repetitive and social behaviors. These behavioral abnormalities can be reversed by treatment with a sub-psychotomimetic dose of ketamine, which also rescues synaptic molecules.

Studies of signaling and metabolisms have revealed the complexity of ASD and its characteristics (25, 26). Simple yet reproducible animal models of human ASD are needed for the understanding of therapeutic mechanisms and development of novel treatments. Differences in the molecular networks between humans and rodents may limit the utility of rodent models for human diseases. However, the rat model is well suited for studies on neurodevelopmental diseases, and previous studies have described similarities in neuronal structure and synaptic development to humans (27, 28). Further advantages are that rats are easy to handle, mature rapidly, and are reproductively efficient. The present study revealed simultaneous changes in synaptic and behavioral phenotypes in VPA and MeCP2 knockdown models, although the synaptic molecules that underwent changes were slightly different. While the VPA-induced model has time to adapt to chemical damage at the developmental periods, the MeCP2 rats infused with lenti-shMeCP2 are presumed to display autistic behavioral patterns clearly as synaptic function is compromised by the strong gene suppression. This may lead to subtle difference in the expression of synaptic molecules between the two models, which may in turn produce slightly different behavioral responses.

The autistic-like behaviors induced by neonatal VPA exposure and hippocampal inhibition of MeCP2 were both rescued by treatment with ketamine. This is in line with previous results showing that autistic-like phenotypes are induced by functional deficits in NMDA receptor function (29). It will be interesting to examine whether ketamine works in other ASD models such as those induced by local inhibition of MeCP2 or TSC1 in the dorsal striatum (30).

For the first time, we have demonstrated that hippocampal MeCP2 knockdown leads to behavioral abnormalities linked to autism-like traits in rats, and that ketamine prevents these effects. These findings provide a novel strategy for testing the effects of ASD treatments. Although the animal model injected with lenti-shMeCP2 does not mimic all the changes that occur at the system level in human autism, it would be useful as a quick and simple experimental model for testing the effects of potential therapeutic agents.

The detailed methods are described in the “Supplementary Materials and Methods”.

Intraperitoneal (i.p.) injection of VPA to pregnant rats

Pregnant SD female rats were administered a single i.p. injection of sodium valproate (Sigma, ST. Louis, MO) in 0.9% saline (500 mg/kg), or 0.9% saline alone (VPA-untreated controls), at embryonic day 12.5 (E12.5). All behavioral tests were performed on SD rats 4 weeks of age. Only male rats were used for be-havioral experiments.

Three-chamber test

The tests for sociability and social preference were performed in a three-chamber apparatus as previously described (31). The animals used here were all age- and sex-matched littermates; SD rats were used as the stranger rats. The three-chamber apparatus was a 120 cm × 40 cm × 58 cm black plastic box. The first session was a 5 min habituation period. The test animal was introduced and allowed to stay in the central area. After habituation, a stranger animal was introduced into the wire cage of the right compartment (stranger zone) for the sociability test. The test animal in the central area was allowed to explore the three-chamber apparatus after removal of the gate blocking the central area. Time spent in the stranger zone and around the cage was measured for 10 min. The social preference test was conducted for 10 min directly after termination of the sociability test. While the subject animal was confined in the central area, a novel animal (stranger 1; S1) was intro-duced into the wire cage of the left compartment (new stranger zone) followed by measurement of the time spent in each compartment as in the previous session. The preference index was calculated as (S − E) / (S + E) for sociability, and (S1 − F) / (S1 + F) for social novelty preference.

This work was supported by a National Research Foundation of Korea (NRF) Grant (No.2019R1A2C2003616 to H.S.), and a Medical Research Center Grant (No. 2017R1A5A2015395 to H.S.); it was also supported by Basic Science Research Program NRF Grants (No. 2017R1D1AB03032858 and No. 2020 R1I1A1A01060863 to M.C., and No. 2021R1I1A1A01054879 to S.Y.K.) funded by the Ministry of Science and Technology, Republic of Korea; and an Institute of Information & Communications Technology Planning & Evaluation (IITP) grant funded by the Korea Government (MSIT) (No.2020-0-01373, Artificial Intelligence Graduate School Program (Hanyang University)) and the research fund of Hanyang University (HY-202000000700013).

The authors have no conflicting interests.