AMPA Receptor antagonist NBQX attenuates later-life epileptic seizures and autistic-like social deficits following neonatal seizures


  • Jocelyn J. Lippman-Bell,

    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    2. Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A
    Search for more papers by this author
  • Sanjay N. Rakhade,

    Corresponding author
    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    • Address correspondence to Frances E. Jensen, Department of Neurology, Perelman School of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104-4283, U.S.A. E-mail: or Sanjay Rakhade, Clinical Sciences, Genzyme, a Sanofi Company, U.S.A. E-mail:

    Search for more papers by this author
  • Peter M. Klein,

    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Makram Obeid,

    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Michele C. Jackson,

    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Annelise Joseph,

    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    Search for more papers by this author
  • Frances E. Jensen

    Corresponding author
    1. Department of Neurology, Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, U.S.A
    2. Perelman School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A
    3. Program in Neuroscience, Harvard Medical School, Boston, Massachusetts, U.S.A
    • Address correspondence to Frances E. Jensen, Department of Neurology, Perelman School of Medicine, University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104-4283, U.S.A. E-mail: or Sanjay Rakhade, Clinical Sciences, Genzyme, a Sanofi Company, U.S.A. E-mail:

    Search for more papers by this author

  • JJL, SNR, and PMK contributed equally to this manuscript.



To determine whether AMPA receptor (AMPAR) antagonist NBQX can prevent early mammalian target of rapamycin (mTOR) pathway activation and long-term sequelae following neonatal seizures in rats, including later-life spontaneous recurrent seizures, CA3 mossy fiber sprouting, and autistic-like social deficits.


Long-Evans rats experienced hypoxia-induced neonatal seizures (HS) at postnatal day (P)10. NBQX (20 mg/kg) was administered immediately following HS (every 12 h × 4 doses). Twelve hours post-HS, we assessed mTOR activation marker phosphorylated p70-S6 kinase (p-p70S6K) in hippocampus and cortex of vehicle (HS + V) or NBQX-treated post-HS rats (HS + N) versus littermate controls (C + V). Spontaneous seizure activity was compared between groups by epidural cortical electroencephalography (EEG) at P70–100. Aberrant mossy fiber sprouting was measured using Timm staining. Finally, we assessed behavior between P30 and P38.

Key Findings

Postseizure NBQX treatment significantly attenuated seizure-induced increases in p-p70S6K in the hippocampus (p < 0.01) and cortex (p < 0.001). Although spontaneous recurrent seizures increased in adulthood in HS + V rats compared to controls (3.22 ± 1 seizures/h; p = 0.03), NBQX significantly attenuated later-life seizures (0.14 ± 0.1 seizures/h; p = 0.046). HS + N rats showed less aberrant mossy fiber sprouting (115 ± 8.0%) than vehicle-treated post-HS rats (174 ± 10%, p = 0.004), compared to controls (normalized to 100%). Finally, NBQX treatment prevented alterations in later-life social behavior; post-HS rats showed significantly decreased preference for a novel over a familiar rat (71.0 ± 12 s) compared to controls (99.0 ± 15.6 s; p < 0.01), whereas HS + N rats showed social novelty preference similar to controls (114.3 ± 14.1 s).


Brief NBQX administration during the 48 h postseizure in P10 Long-Evans rats suppresses transient mTOR pathway activation and attenuates spontaneous recurrent seizures, social preference deficits, and mossy fiber sprouting observed in vehicle-treated adult rats after early life seizures. These results suggest that acute AMPAR antagonist treatment during the latent period immediately following neonatal HS can modify seizure-induced activation of mTOR, reduce the frequency of later-life seizures, and protect against CA3 mossy fiber sprouting and autistic-like social deficits.

Symptomatic seizures in infancy are associated with later-life epilepsy and cognitive deficits. Acute seizure control can be problematic with conventional anticonvulsant drugs, and the later-life consequences are poorly understood and not addressed in therapeutic trials. Hypoxia is one of the most common causes of neonatal seizures, occurring in 1–2/1,000 live births (Tekgul et al., 2006; Ronen et al., 2007), and is associated with development of epilepsy, neurobehavioral deficits, and intellectual disability (Ronen et al., 1999; Tekgul et al., 2006). Epilepsy and autistic-like behavior have been described to occur concurrently in 30% of patients with either disorder (Tuchman & Cuccaro, 2011); the rate of co-occurrence is even higher in subjects with both autism and intellectual disability (Amiet et al., 2008; Jeste, 2011; Maski et al., 2011; Tuchman & Cuccaro, 2011; Kohane et al., 2012). There is no known causal relationship between epilepsy and autism, including in acquired models of epilepsy. However, the high association of these disorders suggests that they may share anatomic and molecular mechanisms, especially during brain development.

Potential common mechanistic pathways appear to be that of the mammalian target of rapamycin (mTOR) and of activity-dependent pathways downstream of excitatory glutamatergic synapses. Patients with tuberous sclerosis (TSC), a genetically acquired disorder caused by overactivation of the mTOR pathway, display autistic-like behavior and seizures (Meikle et al., 2007; Greer & Greenberg, 2008; Greer et al., 2010; Waltereit et al., 2011). Activation of the mTOR pathway mediates molecular changes in neurotransmitter receptor and ion transporter function, synaptic reorganization, programmed cell death, and neurogenesis; all also have been implicated in neuronal network formation, epileptogenesis, and autism spectrum disorders (Kwon et al., 2006; Nie et al., 2010; Wong, 2010; Kirschstein, 2012; Russo et al., 2012; Talos et al., 2012; Tsai et al., 2012; Wang et al., 2012; Cambiaghi et al., 2013; Sun et al., 2013). Our rodent model of hypoxia-induced neonatal seizures (HS) (Jensen, 1995) results in mTOR pathway up-regulation, long-term epilepsy, and autistic-like social behavior deficits, and these effects are blocked by early treatment with the mTOR inhibitor rapamycin (Talos et al., 2012).

Additional consequences of early life HS include network hyperexcitability, impaired long-term potentiation, and CA3 mossy fiber sprouting (Sanchez et al., 2001, 2005; Rakhade et al., 2008, 2011; Zhou et al., 2011). The function of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) subtype of ionotropic glutamate receptors (AMPAR), which is enhanced in the immature brain (Silverstein & Jensen, 2007), plays a critical role in susceptibility to seizures, and in early epilepsy-induced dysregulation of synaptic signaling (Rakhade et al., 2008; Zhou et al., 2011). A critical early step in epileptogenesis appears to be rapid seizure-induced posttranslational modification of the AMPAR GluA1 subunit (Rakhade et al., 2012). Notably, blocking this modification with AMPAR antagonists immediately after HS can prevent some of the early increases in AMPAR function (Rakhade et al., 2008), and treatment administered following HS (20 mg/kg, intraperitoneally [i.p.], immediately and q12h for 48 h post-HS) prevents long-term enhanced seizure susceptibility (Koh & Jensen, 2001; Koh et al., 2004; Rakhade et al., 2008; Zhou et al., 2011), suggesting a reversible epileptogenic cascade. We have previously shown a specific effect on AMPARs themselves with enhanced amplitude of excitatory postsynaptic currents (EPSCs) persisting over 48 h in post-HS rats. We hypothesize that this secondary effect of seizures on synaptic and network excitability drives activity-dependent signaling cascades, including mTOR activation, which could lead to or exacerbate the long-term phenotype.

Therefore, here we examined the effects of early postseizure treatment with the AMPAR antagonist 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX) on several long-term sequelae of neonatal HS. Because we have previously implicated seizure-induced activation of AMPARs and the mTOR pathway, we hypothesized that increased AMPAR activity drives several downstream consequences of neonatal HS (Sarbassov et al., 2005; Tavazoie et al., 2005; Sengupta et al., 2010; Bateup et al., 2011). Early treatment with rapamycin can protect against long-term seizures, network hyperexcitability, and the development of autistic-like behavior in later life (Talos et al., 2012). Taken together with our prior findings of immediate post-HS enhancement of AMPAR function, these studies suggest that glutamate receptor activation may interact with the mTOR pathway, with other studies suggesting transcriptional and translational changes at the synapse that could mediate the molecular mechanics of these changes (Gong et al., 2006; Talos et al., 2012; Sun et al., 2013).

We report that administration of NBQX during the initial 48 h following HS in P10 rats prevents the early increase in mTOR signaling pathway activation, and leads to attenuation of later-life spontaneous recurrent seizures, social preference deficits, and aberrant hippocampal mossy fiber sprouting in adult rats. These results suggest that early NBQX treatment is protective against the postseizure development of behavioral, cellular, and molecular changes. This is one of the first demonstrations of attenuation of the long-term sequelae of neonatal seizures by posttreatment with therapeutically targeted antagonists. These results provide proof of concept for the potential to target several long-term comorbidities observed in models of early life seizures, and to translate this into therapeutic strategies for the human population.

Materials and Methods

See supporting information for additional methodologic detail.

Animals, seizure induction, and treatment

Litters of male Long–Evans rats (Charles River Laboratories, Wilmington, MA, U.S.A.) were maintained on a 12 h light/dark cycle. All experiments were approved by the Institutional Animal Care and Use Committee at Boston Children's Hospital (Boston, MA, U.S.A.), in accordance with the National Institutes of Health guidelines. HS were induced in P10 pups, as described previously (Jensen et al., 1991), for 15 min (7% O2 for 8 min, 5% O2 for 6 min, 4% O2 for 1 min). Only rats exhibiting ≥5 tonic–clonic seizures during hypoxia were included in the HS group.

Control (C + V) and HS (HS + N) rats were treated with NBQX 20 mg/kg, i.p., dissolved in 0.9% NaCl immediately prior to use; Sigma-Aldrich, St. Louis, MO, U.S.A.) or vehicle (C + V, and HS + V) immediately following hypoxia and every 12 h for three additional injections (Koh et al., 2004).

Western blot analysis

Rats were euthanized at 12 h post-HS (n = 20–27/group). Western blots were performed as previously described for hippocampal and cortical tissue (Talos et al., 2012; Fig. S1A). Primary antibodies to phospho-p70S6 kinase (p70S6K) (Thr389; 1:500) and total p70S6K (1:500; Cell Signaling Technology, Beverly, MA, U.S.A.) were used. Normalized values for each protein were expressed as percentage of mean expression level of control tissue on the same immunoblot, then phosphoprotein/total protein ratios were calculated.

Long-term video-EEG recordings with implanted cranial electrodes

Twenty-four hours video–electroencephalography (EEG) recordings were acquired from P70 to P100 rats (n = 10–12/group; Fig. S1B), as described previously (Rakhade et al., 2011). Four epidural electrodes were stereotaxically placed in contact with the cortical surface (AP 0.0/−6.5 mm, ML 4.0/−4.0 mm from bregma) at P65 (Plastics One, Roanoke, VA, U.S.A.). Video-EEG recording epochs ranged from 5.42–47.0 h/rat/week, over 1–4 weeks/rat (details provided in Table S1), with average seizure frequency calculated per hour of video-EEG recording (as described in Rakhade et al., 2011).

Timm staining

Timm staining for hippocampal mossy fibers was performed in tissue from P100 to P105 rat pups (Fig. S1B) as described previously (Holmes et al., 1999; Huang et al., 1999; Rakhade et al., 2011). Timm staining was analyzed using both a semiquantitative scale and densitometric measurements in stratum pyramidale and stratum oriens of CA3.

Behavioral testing

In a separate cohort of rats from those used for EEG and Timm staining, control and HS rats treated at P10–12 with either vehicle or NBQX were tested at P30–36 for open field locomotion, three-chamber social choice, and olfactory habituation/dishabituation (Fig. S1C) as described previously (Moy et al., 2004; Silverman et al., 2011; Talos et al., 2012).


NBQX treatment attenuates seizure-induced increases in p70S6 kinase

We have previously demonstrated that seizures in P10 rats increases activation of the mTORC1 signaling pathway, including phosphorylation of p70S6K (Thr 389), and that the mTOR inhibitor rapamycin prevents later development of epileptogenesis and accompanying behavioral deficits (Talos et al., 2012). We thus hypothesized that blocking upstream AMPARs with NBQX should attenuate p70S6K hyperphosphorylation. First, we confirmed our previous report of increased phospho-p70S6K 12 h post-HS in the neocortex and hippocampus of HS + V rats (p < 0.0001 by analysis of variance [ANOVA] for all groups, p < 0.01 HS + V, 122 ± 4.6% normalized to 100% controls, vs. C + V, 100 ± 3.7%, posttest in neocortex, n = 26–27/group; p = 0.002 by ANOVA for all groups, p < 0.05 HS + V [129.1 ± 8.4%] vs. C + V [100 ± 5.1%] posttest in hippocampus, n = 20–22/group) relative to vehicle-treated controls (Fig. 1). NBQX treatment immediately following hypoxia decreased phospho-p70S6K levels in HS + N rats back to levels similar to controls in the neocortex (92.7 ± 6.0%, posttest p > 0.05) and hippocampus (97.5 ± 6.4% posttest p > 0.05). Total levels of p70S6K were similar between groups (p > 0.05, Kruskal-Wallis test). Therefore, the phospho-p70S6K/p70S6K ratio was significantly increased in HS + V rats relative to both vehicle-treated controls (HS + V = 123.5 ± 3.6%, C + V = 100 ± 3.2%, p < 0.001 in neocortex, HS + V = 122 ± 6%, C + V = 100 ± 6.2%, p < 0.05 in hippocampus; Kruskal-Wallis test) and NBQX-treated HS rats (HS + N = 93.5 ± 5.4%, n = 26, p < 0.001 in neocortex; 96.3 ± 8.0%, n = 22, p < 0.01 in hippocampus), indicating that postseizure hyperphosphorylation of p70S6K could be rescued by NBQX treatment.

Figure 1.

NBQX treatment attenuates seizure-induced increases in the mTORC1 signaling pathway. Representative Western blots of 12 h post-HS neocortical and hippocampal tissue probed with phospho-p70S6K (Thr 389) and p70S6K. The ratio of p-p70S6K:p70S6K in the neocortex was significantly higher in HS + V rats (123.5 ± 3.6%, n = 27) than in either C + V (100.0 ± 3.2%, n = 26, p = 0.0003) or HS + N rats (93.5 ± 5.4%, n = 26, p < 0.0001). The p-p70S6K:p70S6K ratio was also significantly increased in the hippocampus of HS + V rats (122.8 ± 5.9%, n = 20) relative to that in C + V (100.0 ± 6.2%, n = 20, p = 0.042) and HS + N rats (96.3 ± 8.1%, n = 22, p = 0.003). *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate SEM.

NBQX treatment following neonatal seizures blocks the development of later-life seizures

Because post-HS NBQX treatment prevented the increase in mTOR activation that we have previously shown to be associated with long-term epilepsy and autistic-like behavior, we next sought to test whether NBQX could block these long-term post-HS outcomes, beginning with the development of spontaneous recurrent seizures. As per our prior study (Rakhade et al., 2011), we used intracranial cortical electrodes to record video-EEG recordings from post-HS rats, obtained weekly in 24 h epochs from P70 to P100 (average of 21.7 h epoch/week, recorded once a week for an average of 3.2 weeks; Table S1). Electrographic seizures consisted of runs of spikes and sharp waves (Fig. 2A) accompanied by sudden behavioral arrest, staring episodes, head jerking, and facial automatisms (Rakhade et al., 2011; Talos et al., 2012). Consistent with prior reports, post-HS vehicle-treated rats displayed significantly more spontaneous seizures (3.23 ± 0.99 seizures/h, median = 0.20 seizures/h [interquartile range; IQR 0.0–1.66], n = 11) than vehicle treated controls (0.36 ± 0.18, median = 0.0 seizures/h [IQR 0.0–0.11], n = 10, p = 0.004, Fig. 2B). Although some control rats exhibited spontaneous seizures, seizures in wild-type Long-Evans rats have been reported previously (Shaw, 2007; Shaw et al., 2009; Rakhade et al., 2011), and were significantly less frequent than in post-HS rats. In contrast, post-HS rats treated with NBQX had significantly fewer seizures than vehicle-treated rats post-HS (0.14 ± 0.06 seizures/h, median = 0.0 seizures/h [IQR 0.0–0.08], n = 12, p = 0.0004, ANOVA with multiple comparison post hoc test). Taken together, our findings indicate that AMPAR activation may be upstream of post-HS changes that could account for the development of epilepsy.

Figure 2.

Increased frequency of seizure events in post-HS rats is attenuated by treatment with NBQX. (A) Representative spontaneous electrographic seizure recorded from an intracranial cortical electrode in an adult rat post neonatal HS. The cortical EEG recordings show baseline activity prior to seizure onset of seizures (arrow). Insets show the EEG trace with an expanded time scale demonstrating seizure initiation with spikes, progression to the tonic phase with high frequency, large amplitude spikes, and the termination phase of the seizure with smaller amplitude spikes and postictal slowing. Representative cortical EEG recordings from C + V (B) and HS + N (C) rats show predominantly unaltered baseline activity. Scale bars = 1 s, 250 μV. (D) The frequency of seizures observed per hour of video-EEG recording analyzed significantly increased in HS + V rats (3.23 ± 0.99 seizures/h, median = 0.20 [IQR 0.0–1.66] seizures/h, n = 11), compared with C + V rats (0.36 ± 0.18 seizures/h, median = 0.0 [IQR 0.0–0.11] seizures/h, n = 10, p = 0.004). Rats that were treated with NBQX following HS were found to have significantly fewer frequent seizures (0.14 ± 0.06 seizures/h, median = 0.0 [IQR 0.0–0.08] seizures/h, n = 12, p = 0.0004) than HS + V rats. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate standard error of the mean (SEM).

NBQX blocks increased mossy fiber sprouting in area CA3 of the hippocampus following neonatal seizures

Hippocampal mossy fiber sprouting and the aberrant projection of these axonal terminals increase in animal models of epileptogenesis and in human cases of epilepsy (Mathern et al., 1996; Holmes et al., 1999). Because we and others have previously reported mossy fiber sprouting in CA3 following neonatal seizures (Anderson et al., 1999; Holmes et al., 1999; Sogawa et al., 2001; Rakhade et al., 2011), we evaluated mossy fiber sprouting into stratum pyramidale and stratum oriens of CA3 in P100–105 rats following neonatal HS as an additional outcome measure to assess the disease-modifying properties of NBQX in this model. Distribution of Timm granules was significantly increased (Timm score 2.683 ± 0.15, n = 6) in the HS + V group compared to C + V rats (1.635 ± 0.14, n = 6, p < 0.0001, ANOVA with post hoc test, Fig. 3A,B), most prominently in the septal regions of the hippocampus. Consistently, densitometric analysis of the relative intensities of Timm staining showed significantly stronger staining intensity in vehicle-treated post-HS rats (174 ± 11%, n = 6) compared to littermate controls (100 ± 8.7%, n = 6, p < 0.0001, Fig. 3C). Compared to vehicle, NBQX treatment significantly attenuated the average Timm scores (2.14 ± 0.15, n = 7, p = 0.036) and Timm staining density (115 ± 8%, n = 7, p = 0.0005) in post-HS rats. Therefore, NBQX treatment in rats that had experienced hypoxic seizures led to an attenuation of the increased mossy fiber sprouting observed in the CA3 region of the hippocampus.

Figure 3.

NBQX blocks increased mossy fiber sprouting in hippocampal CA3 hippocampus following neonatal seizures. (A) Photomicrographs of Timm staining in hippocampi of adult, post-HS + V rats show a marked increase in aberrant mossy fiber sprouting into the stratum pyramidale and stratum oriens layers of CA3 as compared to C + V rats. NBQX treatment immediately following seizure induction in HS + N rats decreased Timm staining. Insets show higher magnification of Timm granule distribution in stratum pyramidale and stratum oriens (scale bar = 250 μm). (B) Average Timm scores, calculated using a semiquantitative scoring scale, from HS + V rats were significantly higher (2.683 ± 0.15, n = 6) than from either C + V (1.635 ± 0.14, n = 6, p < 0.0001) or HS + N rats (2.14 ± 0.15, n = 7, p = 0.036). (C) Mean density measurements of Timm staining measured in the stratum pyramidale-oriens confirmed increased aberrant mossy fiber sprouting in HS + V rats (174 ± 11%, n = 6) over C + V levels (100 ± 8.7%, n = 6, p < 0.0001) and the attenuation of this increase in Timm granules following treatment with NBQX in HS + N rats (115 ± 8%, n = 7, p = 0.0005). *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate SEM.

Autistic-like deficits in social behavior following neonatal seizures were attenuated by early AMPA receptor blockade

We hypothesized that epilepsy and cognitive dysfunction, which often develop after neonatal seizures, may occur through similar pathways. We previously demonstrated that following neonatal HS, rats display autistic-like social deficits (Talos et al., 2012). Given the attenuation of p70S6K phosphorylation and seizures following NBQX administration, we next examined whether NBQX treatment would also attenuate social deficits (n = 12–16 rats/group for all behavioral tests). Using a three-chamber social choice test, we found that all four groups showed a significant preference for sociability, spending more time with a novel social stimulus than with a nonsocial object (all p < 0.0001 by paired t-test of time spent with social vs. nonsocial stimulus for each rat, Fig. 4A). However, when we replaced the nonsocial object with a novel rat, HS + V rats had no preference for interacting with the novel (71.0 ± 12 s) over the familiar rat (65.2 ± 14.5 s, p = 0.77, paired t-test, Fig. 4B). This differed from the significant preference for social novelty in both C + V (99.0 ± 15.6 s with novel vs. 42.0 ± 7.6 s with familiar rat, p = 0.006, paired t-test) and C + N rats (101.3 ± 20.1 s with novel vs. 33.5 ± 6.9 s with familiar rat, p = 0.02, paired t-test). Preference for social novelty following neonatal HS was restored to control levels after NBQX treatment (114.3 ± 14.1 s with novel vs. 45.6 ± 7.6 s with familiar rat, p = 0.003, paired t-test).

Figure 4.

NBQX rescues behavioral post-HS social deficits. (A) All rats showed a preference for a novel rat over a novel object in a social preference test. (B) When a novel rat replace the novel object, HS + V rats failed to shift their attention from the familiar to the novel rat, showing a deficit in preference for social novelty by spending the same amount of time with the novel rat (71.0 ± 12 s) as the familiar rat (65.2 ± 14.5 s; p > 0.05), as compared to C + V rats that spent more time with the novel (99.0 ± 15.6 s) vs. familiar rat (42.0 ± 7.6 s; p < 0.01). NBQX treatment rescued this deficit, as post-HS NBQX-treated rats spent more time with the novel (114.3 ± 14.1 s) vs. familiar rat (45.6 ± 7.6 s; p < 0.01). (C) Rat performance in the open field assay was comparable between groups, measured by time in center and (D) distance traveled, although NBQX treatment in post-HS rats caused a slight, but not significant, increase in overall distance traveled. (E) Each group demonstrated normal olfaction, including social olfaction (cages 1 and 2), with normal dishabituation to each odor (trials 1, 2, and 3 for each). C, control; H, post-HS; V, vehicle; N, NBQX. *p < 0.05, **p < 0.01, ***p < 0.001. Error bars indicate SEM.

The social deficit was not accompanied by alterations in other behaviors we measured, suggesting that other behavioral abnormalities were not a factor. Baseline tests of open field activity were similar between all groups, in total distance traveled (C + V = 82.9 ± 13.6, HS + V = 88.1 ± 8.2, C + N = 89.5 ± 9.9, and HS + N = 111.7 ± 14.9 m, p = 0.44, Kruskal–Wallis test, Fig. 4C) and time spent in the center (C + V = 54.1 ± 11.3, HS + V = 44.5 ± 10.9, C + N = 51.8 ± 1.6, and HS + N = 61.8 ± 13.2 s, p = 0.63, Kruskal-Wallis test, Fig. 4D). Olfactory habituation/dishabituation was also similar between groups (p = 0.88 by Kruskal-Wallis test, Fig. 4E), indicating no deficits in the rats' ability to differentiate between nonsocial or social odors.

These results were comparable to our previous report (Talos et al., 2012) and add the finding that over-activation of AMPARs is likely to be involved in the development of social deficits. It is important to note that unlike our earlier studies using rapamycin (Talos et al., 2012), NBQX did not affect behavior of the control rats, reinforcing the potential therapeutic value of AMPAR antagonists.

Taken together, our findings indicate that AMPAR activation may begin a potential common pathway through which neonatal seizures could lead to molecular, cellular, and behavioral changes that could account for the development of epilepsy and autistic-like features later in life in animals experiencing HS.


Neonatal seizures represent a significant insult to the developing brain. The immature brain has lower thresholds to seizures in part due to a normal developmental period of enhanced excitation and reduced inhibition, caused by maturational expression patterns of key neurotransmitter receptors, ion transporters, and synaptic proliferation (Rakhade & Jensen, 2009). Similar to clinical observations, early life seizures in rodents can result in both later-life social behavioral deficits and epilepsy. Our prior results indicate that these may be in part mediated by seizure-induced activation and posttranslational modifications of AMPARs, plus activation of the mTOR pathway (Talos et al., 2012). However, no information existed as to whether these events were related and were common mechanisms for both the long-term epileptogenic changes and the autistic-like behavioral deficits. Therefore, we tested whether inhibition of AMPARs with NBQX through the first 48 h after induction of neonatal seizures, which we have previously identified as a specific window of therapeutic intervention (Koh et al., 2004), could modulate the mTOR pathway and attenuate long-term sequelae of neonatal seizures. The principal findings of this study are that post-HS NBQX treatment in rats significantly attenuates over activation of p70S6K, reduces seizure frequency, diminishes the cellular changes in mossy fiber sprouting in the CA3 region, and restores the preference for social novelty. Taken together, these data provide a proof of concept for attenuating the major long-term effects of neonatal seizures, including chronic spontaneous seizures and social behavior deficits.

Coactivation of AMPAR and mTOR pathway following early life seizures

Here, we studied phosphorylation of p70S6K which is directly downstream of mTORC1 and a useful marker of mTOR pathway activity. We and others have previously shown the widespread activation of mTOR signaling acutely and chronically following epileptic seizures (Zhou et al., 2009; McDaniel & Wong, 2011; Wong, 2011; Talos et al., 2012; Zhang & Wong, 2012). We have previously shown using immunohistochemistry that phospho-S6, a downstream marker of mTOR pathway activation that is phosphorylated by p70S6K, is up-regulated 24 h post-HS at P10, 12 h after increased p70S6K phosphorylation (Talos et al., 2012). The increase was specific to dendrites and somas of CA1 hippocampal neurons and layer V cortical neurons, and was not seen in astrocytes or in interneurons. We therefore did not repeat the localization measurements here, instead using p70S6K Western blots to test only whether AMPAR could act upstream of the increased mTOR pathway activation we have reported previously.

Multiple potential mechanisms contributing to synaptic dysplasticity in epileptogenesis and social behavior deficits(Rakhade & Jensen, 2009), such as changes in neurotransmitter receptor and ion transporter function, synaptic reorganization, programmed cell death, and neurogenesis, are dependent on protein synthesis and biochemical signaling pathways regulated by mTOR pathway members (Kwon et al., 2006; Nie et al., 2010; Wong, 2010; Kirschstein, 2012; Russo et al., 2012; Talos et al., 2012; Tsai et al., 2012; Wang et al., 2012; Cambiaghi et al., 2013; Sun et al., 2013). The glutamate released during seizures can activate the mTOR pathway directly, or indirectly, through Ca2+-mediated signaling cascades (Lenz & Avruch, 2005; Gong et al., 2006). In addition, mTOR has been linked to the onset of epilepsy; hyperactivation of the mTOR pathway in hippocampal dentate gyrus cells developed spontaneous seizures within 4 weeks, showing a causal link between excessive mTOR activation and epilepsy (Nadif Kasri et al., 2009).

A major consequence of mTOR activation is increased protein synthesis in dendrites, which is important for induction of synaptic plasticity both in vitro and in vivo, including for the late phase of long-term potentiation and memory consolidation (Stoica et al., 2011). mTOR controls protein synthesis presumably through phosphorylation of its substrates, which include eukaryotic initiation factor 4E-binding protein and p70S6 kinase (S6K). mTOR can regulate both translational initiation and elongation, the latter by promoting the inactivation of elongation factor 2 kinase. We have previously observed increased phosphorylation of p70S6 kinase following seizures in this rodent model (Talos et al., 2012; Sun et al., 2013), and here aimed to determine whether increased mTOR activation post-HS could be linked to the upstream AMPAR overactivation we reported previously (Koh et al., 2004; Rakhade et al., 2008; Zhou et al., 2011). Loss of Tsc1 protein in mice results in enhanced AMPA EPSCs (Bateup et al., 2011), and our lab has shown increased AMPAR subunit GluA1 in neurons of TSC patients (Talos et al., 2008). Taken together, this supports the use of AMPAR antagonists in models that include seizures and autistic characteristics. We demonstrate here that increased mTOR activation can indeed be prevented with acute NBQX treatment, which may serve as a predictive marker for the attenuation of some of the long-term effects of neonatal seizures.

Multiple studies show that the mTOR pathway can regulate AMPARs (Wang et al., 2006; Slipczuk et al., 2009); here we demonstrate that the reverse may be true as well. Based on our results showing that NBQX blocks post-HS overactivation of mTOR, we hypothesize that AMPAR activation drives activity-dependent up-regulation of the mTOR pathway. Taken together with prior studies of mTOR regulation of AMPARs (Wang et al., 2006; Talos et al., 2008; Slipczuk et al., 2009; Bateup et al., 2011), we conclude that the interaction between AMPARs and mTOR is likely bidirectional.

Acute NBQX treatment suppresses development of chronic spontaneous seizures and aberrant mossy fiber sprouting

A major objective of current research is to design a treatment to prevent the development of epilepsy and cognitive deficits following neonatal seizures. Conceptually, seizures in the developing brain initiate an epileptogenic process that culminates in unprovoked seizures after a period of weeks, months, or years, with a latent period between the initial seizures and the appearance of epileptic seizures (Baulac & Pitkanen, 2008; Rakhade & Jensen, 2009). The rodent model of epilepsy used here also exhibits a latent period, prior to appearance of epileptic seizures. This latent period represents a window of opportunity in which an appropriate treatment can be targeted with the aim to stop or modify the disease process.

In this study, we show that treatment with NBQX, a potent AMPAR antagonist, administered following hypoxic seizures in rat pups, reduces the frequency and number of seizures during adulthood. We used a 3 sec or longer duration of rhythmic spike activity associated with behavioral change as the minimal criteria for seizure activity to increase our sensitivity for therapeutic effect. We have previously analyzed the frequency of seizures using longer thresholds for the duration of events (Rakhade et al., 2011) and using thresholds ranging from 3-10 seconds, we observed a statistically significant difference between animals experiencing HS in early life and normoxic control animals. A subset of animals from a smaller cohort in the current study had results consistent with prior results. Using this sensitive criteria in controls, the posthypoxic rats had an almost 800% increase in seizure events 3 sec or longer, and these events were blocked by NBQX treatment. The ability to reduce the potential development of a seizure disorder is particularly valuable with a postinsult paradigm, since it provides proof-of-concept that the development of seizures can be modified following the acute insult. Therefore, these results may be translated into potential therapeutic intervention, especially during the latent phase following the initial insult, to attenuate the development of epileptic seizures.

NBQX posttreatment also reduced mossy fiber sprouting, a common consequence in CA3 following early life seizure models (Anderson et al., 1999; Holmes et al., 1999; Sogawa et al., 2001; Osterweil et al., 2010; Rakhade et al., 2011). Mossy fiber sprouting in the dentate gyrus or hilar regions of immature animals postseizure is more limited than in CA3 and less correlated with behavioral outcomes (de Rogalski Landrot et al., 2001). Because NBQX posttreatment also reduced chronic spontaneous seizures, this suggests that mossy fiber sprouting may be involved in the process of epileptogenesis. However, our study was not designed to determine relationships between mossy fiber sprouting and either spontaneous recurrent seizures or autistic behavior: these three outcomes were measured independently to test the effects of treatment with NBQX. The results here, though supportive, do not clarify this relationship. In addition, the significance of mossy fiber sprouting in autism has never been studied. It will be important for future studies to more specifically identify the necessary downstream targets mediating epileptogenesis and autistic behavior that are reversed by NBQX treatment.

Modification of chronic behavioral deficits suggests acquired autistic-like social behavior can be modulated following NBQX treatment

The rate of autism spectrum disorder (ASD) in young children with epilepsy has been reported to be as high as 37% (Irwin et al., 2000), although the overlap varies based on study criteria, and is higher in children also presenting with intellectual disability (Lindemann et al., 2011; Tuchman & Cuccaro, 2011; van Eeghen et al., 2013). The rat neonatal HS model in part recapitulates this comorbidity, exhibiting later epilepsy and at least one component of an autistic phenotype, social deficits (Talos et al., 2012). In humans, the preference to pay more attention to novel faces develops as early as 1–2 years of age (Carver et al., 2003; Burden et al., 2007). Of interest, it has been reported that children with ASD do not show as much difference in attention to a familiar versus a novel face (Dawson et al., 2002). We show here, as in our previous results (Talos et al., 2012), that like human children, adolescent rats show a preference for novel over familiar social stimuli, and that following neonatal HS, rats lack this preference, perhaps similarly to children with ASD. The prevention of later social deficits with NBQX treatment that we show here is similar to what we reported with brief rapamycin treatment (Talos et al., 2012), but unlike rapamycin, NBQX has no effect on the social behaviors of control rats, and hence may represent a treatment strategy with a better safety profile. We have not yet tested the other two core deficits of autism in humans: stereotyped behaviors and communication deficits. We plan to test these in the future to determine whether HS is associated with all components of ASD.

Patients with childhood onset epilepsy are highly likely (53%) to fall into the range for ASD on the Social Responsiveness Scale, a test commonly given to diagnose ASD (van Eeghen et al., 2013). In addition, autistic behaviors are higher after infantile spasms than in the general population (Koo et al., 1993). Supporting a causal role for seizures in cognitive deficits, blocking infantile spasms can improve cognitive outcome (Jambaque et al., 2000). Taken together with our data here and that previously published, we conclude that seizures during this early, critical period of synaptic development (Rakhade & Jensen, 2009) can cause dysfunctional maturation/development of circuits and pathways necessary for proper cognitive functioning later in life.

AMPAR activation can set in motion many mechanisms that could potentially affect cognitive behavior acquired following early life seizures. Previous reports show several significant acute deficits in synaptic potentiation and development of silent synapses after neonatal seizures in rodents that were blocked by NBQX (Zhou et al., 2011). As described here, NBQX can also block long-term cognitive changes, leading us to hypothesize that the early postseizure molecular and cellular changes that we have shown here and previously can be blocked by AMPAR antagonists (Koh et al., 2004; Zhou et al., 2011), likely alter the development of normal behavior. Two disorders on the autism spectrum that also show seizures, fragile X syndrome (FXS) and TSC, also show alterations in AMPARs. Fmr1 KO mice show increased basal transcription of AMPAR subunit GluA1 (Muddashetty et al., 2007), although not necessarily increased function of AMPARs, whereas decreasing Tsc1 (as in TSC) enhances AMPAR expression and/or function in humans (Talos et al., 2008) and mice (Bateup et al., 2011). Taken together with our results, we conclude that AMPAR modulation may present a useful therapy for preventing autistic-like social behavior that can develop following early life seizures.

Effects of hypoxia

We have previously studied the effect of hypoxia alone without seizure induction in Long-Evans rat pups (Zhou et al., 2011). In slices removed from these rats (<5% of total of animals), we failed to observe enhanced amplitude or frequency of AMPA receptor–mediated spontaneous excitatory postsynaptic currents (sEPSCs). Slices removed at 48–72 h after hypoxia in these rats previously also showed no alterations in synaptic function, or in the number of silent synapses (Zhou et al., 2011). These data suggest that seizure activity, not hypoxia alone, is required in the short-term to mediate alterations in synaptic function; we hypothesize that a similar lack of long-term alterations may be observed in rats exposed to hypoxia alone.

In conclusion, this study demonstrates the feasibility of targeting long-term consequences of early life seizures. Our data suggest that modulating glutamatergic neurotransmission in the appropriate setting can improve social deficits in animal models, providing opportunities to target specific comorbidities, a significant clinical problem with inadequate therapies, and to attenuate long-term development of epilepsy. Future studies will evaluate the specific aspects of symptomatic improvement and determine the exact molecular mechanisms involved in improvement of social behavior deficits. However, the current study suggests that AMPAR antagonists, such as NBQX, GYKI compounds, talampanel, and perampanel may have an expanded role in attenuating long-term comorbidities of early life seizures and epileptogenesis.


This work was supported by National Institutes of Health Grants NS 031718, DP1 OD003347 (FEJ; from the Office of the Director), and F32 NS068161 (JJL), and Intellectual Developmental Disabilities Research Center Grant P30 HD18655 (National Institute of Child Health and Human Development). We would like to thank Dr. Jacqueline Crawley for helpful discussion regarding rodent behavioral studies.

Author Contributions

JJL and SNR designed, performed experiments, analyzed data, and wrote the manuscript. PMK and MO designed, performed experiments, and analyzed data. MCJ and AJ performed experiments and analyzed data. FEJ obtained funding, designed experiments, and wrote the manuscript.


SNR is currently an employee of Genzyme, a Sanofi Company; no competing financial interest declared during the conduct and publication of this study. FEJ is currently funded by an investigator-initiated grant from Eisai, Inc., that did not support the current study. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.


  • Image of creator

    Jocelyn J. Lippman-Bell

  • Image of creator

    Sanjay N. Rakhade

  • Image of creator

    Peter M. Klein