Early life epilepsy is associated with increased risk of intellectual disability and autism 
and epilepsy occurs at a significantly higher rate in autistic patients relative to the general population 
. The interaction between epilepsy and autism is likely to be multifactorial, but the present study supports a significant role for perturbations in mTORC1 signaling pathway in linking the pathogenesis of these two disorders, both of which have been attributed at least in part to synaptic dysfunction. Early life is characterized by enhanced synaptogenesis and synaptic plasticity, and here we show that there is a commensurate increase in activation of mTORC1-dependent downstream signaling in the immature brain. While mTORC1 has been shown to regulate synaptic function and plasticity in vitro
, we now demonstrate that seizures sequentially activate up- and downstream components of plasticity related mTORC1 signaling, including upstream activators PI3K/Akt, MAPK/ERK and BDNF and downstream effectors 4E-BP1, p70S6K and ribosomal protein S6. The mTORC1 pathway is therapeutically targetable, and a major finding of this study is that pharmacologic suppression of mTORC1 activity with acute rapamycin treatment disrupts subsequent development of epilepsy as well as autistic-like behavior, suggesting that the mTORC1 pathway may be a common, and reversible, mediator for the interaction between early life epilepsy and autism.
This is the first report to show that markers of mTORC1 activity are transiently increased at baseline during the first three postnatal weeks in rodents, consistent with heightened synaptic plasticity. The expression of phosphorylated p70S6K (Thr389), one of the best-characterized downstream targets of mTORC1 
, was highest during the first postnatal weeks in both hippocampus and neocortex. The increased mTORC1 signaling appears to be occurring coincident with the developmental upregulation of known upstream activators of this pathway, including glutamate receptors, BDNF and Rheb 
, which supports its critical involvement in synaptic and network development 
. Interestingly, we found a developmental time lag between the increase in phospho-p70S6K and phospho-S6. To date, the timing of activation of these pathways has only been studied in cell cultures or acute slices, and our in vivo
results are especially intriguing given the reported rapid kinetics of these phosphorylation events in vitro
. The activity of mTORC1 is highly regulated and balanced: for example, activation of p70S6K blocks further activation of this pathway by inhibiting PI3K/Akt signaling 
. During development, this negative feedback loop might be initially stronger then the ability of p70S6K to activate S6, at least to a level detectable by western blot in brain tissue. Interestingly, the delay in S6 activation was more pronounced in the neocortex, where phospho-p70S6K levels were highest. Another interesting finding was the apparent disconnect between the steady increase in total mTOR protein and the downregulation of mTORC1 signaling with increasing age, which would suggest that in the adult brain, the activity of mTORC2 predominates over mTORC1. This hypothesis is supported by the developmental regulation of phospho-mTOR (Ser2448) showing a progressive increase with maturation, at least in the hippocampus. Although the functional relevance of this phosphorylation site in neurons remains to be determined, it has been previously reported that phospho-mTOR (Ser2448) can be associated with both mTORC1 and mTORC2 complexes 
. Future analyses, beyond the scope of this study, will need to examine the developmental regulation of specific mTORC2 targets, including phospho-Akt (Ser 473) 
This study also reveals upregulation of multiple mTORC1 pathway components following neonatal seizure induction in both the hippocampus and neocortex. Specifically, we found that mTORC1 downstream signaling is transiently activated, likely through induction of the putative upstream activators PI3K/Akt, MAPK/ERK and BDNF pathways (). This is consistent with synaptic plasticity models 
, and status epilepticus in adult rats 
. We show a sequential transient increase in BDNF, phospho-Akt and phospho-ERK between 1 and 3 h after seizure induction, followed by upregulation of the mTORC1 downstream effectors phospho- 4E-BP1, phospho-p70S6K and phospho-S6 at 12–24 h post-seizure ().
Potential mTORC1-dependent mechanisms involved in epileptogenesis and altered social behavior following neonatal seizures.
Seizures can rapidly elevate both glutamate and BDNF levels in adult and immature animal models of epilepsy 
, which is in agreement with our findings. Interestingly, in addition to the early increases of BDNF in hippocampus and neocortex, we found a later upregulation at 12 h in hippocampus, which is consistent with the fact that multiple waves of activity-induced BDNF-dependent protein synthesis are required for hippocampal memory consolidation, beyond the first few hours post-training 
. BDNF, in addition to its role in neuronal proliferation, differentiation and survival, represents a crucial regulator of synaptic function and plasticity, in particular, it can activate the PI3K/Akt and MAPK/ERK pathways promoting dendritic growth, dendritic branching, and spine formation 
Phospho-Akt and phospho-ERK may be important in promoting epileptogenesis, beyond their function of activating mTORC1. Akt has an important role in cell proliferation and cell survival 
and may have a neuroprotective role in this model. ERK signaling can enhance protein synthesis by directly phosphorylating S6 at Ser235/236 
as well as p70S6K at multiple sites 
. In addition, ERK can modulate the transcription of multiple plasticity genes by activating the transcriptional regulator cAMP response element-binding CREB 
. Interestingly, we found that phospho-ERK responses were more robust in the hippocampus, consistent with its critical role in hippocampal synaptic plasticity, learning and memory 
Notably, the activation of downstream mTORC1 signaling in our mild neonatal seizures model occurs somewhat later than the reported phospho-S6 upregulation following induction of status epilepticus in adult rats (1–6 h post) 
. In addition to more severe seizures, the adult models also cause neuronal death, whereas our neonatal model exhibits no cell death and shows ongoing subclinical excitability over the first 24 to 48 h after seizure onset 
. Furthermore, hypoxia was used to induce neonatal seizures in this study, and it is known that hypoxia itself can suppress mTORC1 signaling 
. Although the duration of hypoxia used for seizure induction was only 15 min, which is much less then needed to induce a marked inhibition of phospho-p70S6K 
, we cannot exclude that this may have prolonged the time lag between the seizure-induced activation of mTORC1 upstream and downstream components. Nevertheless, the lag in activation may present a window of opportunity for intervention.
Enhancing the significance of seizure-induced activation of mTORC1 in neonatal seizures, we found that in vivo
rapamycin treatment attenuated a number of epileptogenic consequences previously reported in this model 
. Effective inhibition of mTORC1 signaling at P10 during HS protected against enhanced susceptibility to KA-induced seizures at P13, but also reversed the long term effects of HS in inducing later life spontaneous seizures. Importantly, these effects were not due to suppression of acute HS, as their severity and duration were unaffected by rapamycin treatment. This is consistent with previous studies in adult rats that show little or no immediate effects of rapamycin on acute status epilepticus in vivo
, and in vitro
studies showing no direct effect of rapamycin on excitatory synaptic transmission or neuronal excitability 
Rapamycin treatment not only blocked the increases in early mTORC1 activity markers and later seizure susceptibility, but also prevented seizure-induced changes in synaptic function. We have previously shown that epileptogenesis in this model is selectively mediated by enhanced activation and modification of AMPA receptor subtype of ionotropic glutamate receptors, as evidenced by HS-induced increases in AMPA receptor mEPSCs and sEPSCs 
. Importantly, treatment with AMPA receptor antagonists in the subacute period following HS prevents long term increased network excitability and impaired synaptic transmission 
. Similarly, we found that rapamycin significantly attenuated the increased AMPA receptor mEPSCs and sEPSCs amplitude in HS rats, without affecting the basal synaptic transmission. Notably, components of the mTORC1 pathway co-localize with post-synaptic scaffolding protein post-synaptic density 95 (PSD95), facilitating local long-lasting changes in synaptic function 
. mTORC1 regulates translation of immediate early genes, and of genes encoding ion channels, protein kinases or cytoskeletal proteins critical to spine and synapse function 
. As the mRNAs encoding glutamate receptor subunits are localized in dendrites and can undergo local translation in an activity-dependent manner 
, activation of mTORC1 could serve to increase local translation of AMPA receptor subunits themselves, or of related proteins involved in their trafficking and insertion 
. This is consistent with our data showing the co-occurrence of increased phospho-S6 in dendrites of pyramidal neurons at 24 h post-HS and subacute increases in AMPA receptor-mediated sEPSC and mEPSCs amplitude.
This study provides the first evidence for development of autistic-like social behavior deficits following neonatal seizures in wild type animals, manifested as a lack of preference for social novelty. Similar social novelty deficit have been shown in several autism mouse models 
, including genetic models that involve dysregulation of mTORC1 pathway components 
. Although currently there is no known casual relationship between epilepsy and autism 
, the high association of these two disorders suggests that they may share some anatomical and molecular mechanisms 
. The same in vivo
rapamycin treatment that prevented subacute changes in seizure susceptibility and long term epilepsy also ameliorated these autistic-like behavioral deficits. Thus the present study implicates the mTORC1 pathway not only in epileptogenesis, but also in the behavioral consequences of early life seizures. These results raise the possibility that modifying epileptogenesis in the immature brain may also prevent other manifestations of network dysfunction involved in later life neurobehavioral co-morbidities. While these results suggest a disease modifying effect of rapamycin for both epilepsy and secondary autism, caution must be placed on potential clinical translation. While rapamycin showed a beneficial effect in post-seizure animals, there was a mild effect on later social preferences in control rats that had been treated at P10. Hence the protection here is predicated on pathological increases in mTORC1 activity by early life seizures, and in the absence of this pathological upregulation, suppression of the mTORC1 pathway may have actually caused subtle developmental abnormalities in the naïve controls 
. Safety studies of mTORC1 inhibition must address these issues.
In conclusion, these results show a time-dependent activation of up and downstream components of mTORC1 pathway, beginning within an hour after seizure onset and lasting up to 24 h in the immature rat brain (). Activation of the mTORC1 pathway, and subsequent increased AMPA receptor function, may have a critical role in epilepsy as well as autistic-like behavior as a consequence of early life seizures and may be one of probably multiple convergence points underlying an interaction between autism and neonatal seizures (). These results suggest that this interaction is not simply limited to TSC patient population, and that mTORC1 inhibition may have more widespread application as a preventative, disease modifying therapeutic strategy following refractory early life seizures with impact on subsequent brain function.