Insulin-Like Growth Factor-1 Promotes Synaptogenesis Signaling, a Major Dysregulated Pathway in Malformation of Cortical Development, in a Rat Model

Malformation of cortical development (MCD) is one of the main causes of intractable epilepsy in childhood. We explored a treatment based on molecular changes using an infant rat model of methylazoxymethanol (MAM)-induced MCD established by injecting MAM at gestational day 15. The offspring were sacrificed on postnatal day (P) 15 for proteomic analysis, which revealed significant downregulation in the synaptogenesis signaling pathway in the cortex of MCD rats. Recombinant human insulin-growth factor-1 (rhIGF-1) was injected from P12 to P14 twice daily and the effect of IGF1 on N-methyl-d-aspartate (NMDA)-induced spasms (15 mg/kg of NMDA, i.p.) was tested; the onset of P15 single spasm was significantly delayed (p = 0.002) and the number of spasms decreased (p < 0.001) in rhIGF1-pretreated rats (n = 17) compared to those in VEH-treated rats (n = 18). Electroencephalographic monitoring during spasms showed significantly reduced spectral entropy and event-related spectral dynamics of fast oscillation in rhIGF-1 treated rats. Magnetic resonance spectroscopy of the retrosplenial cortex showed decreased glutathione (GSH) (p = 0.039) and significant developmental changes in GSH, phosphocreatine (PCr), and total creatine (tCr) (p = 0.023, 0.042, 0.015, respectively) after rhIGF1 pretreatment. rhIGF1 pretreatment significantly upregulated expression of cortical synaptic proteins such as PSD95, AMPAR1, AMPAR4, NMDAR1, and NMDAR2A (p < 0.05). Thus, early rhIGF-1 treatment could promote synaptic protein expression, which was significantly downregulated by prenatal MAM exposure, and effectively suppress NMDA-induced spasms. Early IGF1 treatment should be further investigated as a therapeutic strategy in infants with MCD-related epilepsy.


Introduction
The development of the cerebral cortex is accomplished through well-orchestrated neurogenesis, neuronal migration, cell proliferation, and organization involving various proteins and transcription factors [1,2]. Any dysregulation results in a broad-spectrum disorder, termed as malformation of cortical development (MCD). Many genetic or environmental factors [3,4] are associated with the pathogenesis of MCD. Patients with MCD suffer from developmental problems, neurological deficits, and epilepsy [1,5,6]. In particular, MCD is the most common cause of intractable epilepsy in pediatric populations [2,5,7,8]. The earlier the epilepsy begins, the more frequent and severe cognitive impairment occurs [2]. MCD is also linked to development of epileptic spasms, and a recent study with infantile spasms (IS) indicates that MCD constitutes more than 10% of etiology in children with IS [9]. Despite many clinical and translational research [10,11], epileptogenesis of the malformed brain is still unclear. The afterbirth diagnosis of MCD makes these structural alterations in fetal brain development permanent, ultimately leaving physicians with the choice of only symptomatic treatment.
The offspring from methylazoxymethanol (MAM)-treated rats have developmental brain anomalies, including migration failure, ventricular enlargement, and disorganization of neocortical and hippocampal structures [10,12], which resemble the pathologic findings observed in patients with MCD [5,12]. Our previous studies have reported cognitive impairment and increased seizure susceptibility with increased fast oscillation (FO) as well as decreased dendritic arborization or cortical neurons during infancy in this MAM-induced MCD rat model [5,13,14]. These MCD rat also demonstrated increased seizure susceptibility to N-methyl-d-aspartate (NMDA) in their postnatal period, as in a previous rat model of infantile spasms with prenatal betamethasone exposure and postnatal NMDA injections [15]. Using this MAM-induced MCD model [4,16], we aimed to identify the most severely deteriorated canonical pathway in the early postnatal period.
Insulin-like growth factor-1 (IGF-1), produced by all central nervous system (CNS) cell types, plays important roles in brain development and neuroplasticity. IGF-1 is involved in cell organization, neural circuitry formation, and maturation of synaptic efficacy in the early brain, CNS development, and neuronal cell growth and proliferation [17][18][19]. In the developing brain, IGF-1 modulates the axonal development and synapse formation through the phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT) pathway [17,18]. Several studies [20][21][22][23] have reported that the application of IGF-1 could improve outcomes after brain injury and other studies [24,25] showed significantly low CSF IGF-1 in patients with symptomatic infantile spasms; however, the role of IGF-1 in epilepsy or early brain development is still controversial [19].
We hypothesized that IGF-1, which plays extremely critical role in brain growth, might benefit children with early brain injury and that its beneficial effect can also mitigate the brain damage. In this study, we investigated whether recombinant human IGF-1 (rhIGF-1) pretreatment at early postnatal period inhibits NMDA-induced spasms in a rat MCD model [4,16] and if IGF-1 pretreatment reverses the key pathway involved in pathologic MCD brain.

Animal Experiments
Animal experiments were approved by the Institutional Animal Care and Use Committee and conformed to the Revised Guide for the Care and Use of Laboratory Animals (8th Edition, 2011). Timed-pregnant Sprague-Dawley rats were purchased (Orient Bio Inc., Seoul, Korea) at gestational day 14 (G14) and housed under a 12 h light/dark cycle with free access to food and water. On G15, two doses of MAM (15 mg/kg intraperitoneally, MRIGlobal, Missouri) or normal saline were injected into pregnant rats at 8:00 A.M. and 6:00 P.M. Delivery occurred consistently on gestational day 21 for all the rats, which was considered postnatal day (P) 0 for the offspring.

Proteomics Analysis
Prenatally MAM-exposed rats (n = 4) and control rats (n = 4) were sacrificed on P15 and their cortex were separated for proteomic analysis.

Sample Preparation and Nano-liquid Chromatography Electrospray Ionization Tandem Mass Spectrometry (LC-ESI-MS/MS) Analysis
Brain tissues were carefully washed in phosphate-buffered saline (PBS) on ice to remove blood. The cortices of prenatally MAM-exposed rats and controls were individually cryopulverized using a Cryoprep device (CP02, Covaris) as previously described (PMID: 24678027). Peptide separation was performed using Dionex UltiMate 3000 RSLCnano system (Thermo-Fisher Scientific). Mass spectra were acquired in a data-dependent mode with an automatic switch between a full scan with 20 data-dependent MS/MS scans.

Database Searching and Label-Free Quantitation (LFQ) and Functional Enrichment and Gene Ontology Analysis
The acquired MS/MS spectra were searched using the SequestHT on Proteome discoverer (version 2.2, Thermo Fisher Scientific) against the SwissProt database (July 2019). False discovery rates (FDRs) were set for 1% for each analysis. For the differential analysis of the relative abundance of proteins between samples, Perseus (version 1.6.13.0) was used. Proteins with a q-value of < 0.05 and log2 fold change ± 1 were considered differentially regulated proteins.
To generate multiple spasms, prenatally MAM-exposed rats were injected with NMDA at P12 and then randomly assigned into rhIGF-1 or VEH group. Rats were treated with rhIGF-1 or VEH from P12 at 6:00 P.M. to P15 at 8:00 A.M. All rats received additional spasm triggers on P13 and P15. The rats were monitored on P15 to determine the effects of rhIGF-1 pretreatment on spasms.

Cortical Electroencephalography (EEG) Recording and Analysis
For intracranial EEG recording, two cortical electrodes were surgically implanted into bilateral somatosensory cortices in each of five rats treated with rhIGF-1 or VEH under ketamine/xylazine sedation (50/7 mg/kg in 10 mL/ kg saline i.p.) at P12. At P15, spasms were triggered by a single dose of NMDA (15 mg/kg i.p.; Sigma), and EEGs of the two groups of rats were recorded with simultaneous videos using the Twin EEG system (Grass Technologies) for 90 min before NMDA injection (pre-ictal period) and 120 min after injection or until the end of spasms. The sampling rate was 400 Hz with a 0.1 Hz high-pass filter, and 5 min of artifact-free data of unipolar recordings were collected for each. Before analysis, the data were preprocessed using the EEGLAB toolbox of MATLAB 2015b. From each rat, 100 epochs (1 s duration) of pre-ictal and spasms periods per rat were extracted from the EEG data.
For quantitative estimation, averaged spectral entropy (SE) and event-related spectral dynamics (ERSP) of FO , power spectral density (PSD) from each epoch were calculated using the EEGLAB toolbox of MATLAB 2017b.

Statistical Analysis
Statistical analysis was performed using IBM SPSS (ver. 22.0; IBM Corp., Armonk, NY, USA). Level of significance was preset to p < 0.05. Two-group comparisons of the concentrations of neuro-metabolites, cortical protein expression, behavioral assessments, and spasms data were performed using the Mann-Whitney U test. Repeated measure-analysis of variance (RM-ANOVA) with Bonferroni correction was used to test the difference between two groups on the time course data of freezing behaviors and developmental changes of neuro-metabolites. Intracranial EEG recording data was analyzed using a linear mixed model.

Proteome Changes in the Cortices of Prenatally MAM-Exposed Rats at P15
We found a total of 3943 proteins, of which 3736 proteins had quantitative information. IPA analysis identified the top 30 enriched canonical pathways in MAM-exposed rat cortex (Table S1). Among them, top ten enriched pathways are indicated (Fig. 1A) and the most affected genes in each pathway were also listed (Table S1). Synaptogenesis signaling was the most significantly downregulated pathway in rats with MAM-induced MCD compared to that in normal controls (-Log 10 [P value] = 21, z-score = − 1.455). The significantly altered proteins of synaptogenesis signaling pathway are presented in supplementary data (Table S1). MAM-exposed rats showed significant activation of IGF-1 signaling pathway affecting neurogenesis (PMID: 26879907) (-Log 10 [P value] = 8.1, z-score = 1.789).
The IPA of total dataset showed general decrement of neurotransmission in prenatally MAM-exposed rat cortices at their infancy (Fig. 1B). HOMER1, GRIN2A, and CAMK2A levels were significantly downregulated with the predicted inhibition of the upstream regulator, NMDAR, in MAM-exposed rat cortices compared with those in controls (Fig. 1C).
We performed quantitative measurement of FO from EEGs after single-dose NMDA-induced spasms (Fig. 5C). At baseline before spasm, the fast oscillation-spectral entropy (FO-SE) was significantly lower in rhIGF-1 pretreatment group than in the VEH group (n = 5, 0. 39

Discussion
Development of the cerebral cortex is very intricate and mediated by various factors and processes [1]. Any dysregulation in these processes or factors causes malformation in the brain development [3,4,27]. Abnormal brain development, especially MCD, is the main cause of intractable epilepsy in pediatric [27,28] patients, leading to sequelae such as cognitive impairment affecting whole life [2,5,29]. However, the precise pathogenetic mechanism underlying epileptogenesis in MCD brain is yet unknown [10,11]. Hence, treatment is usually focused 0.68 ± 0.14, p = 0.037), and PSD95 (0.98 ± 0.14 vs. 0.75 ± 0.15, p = 0.004) is significantly increased in infant rats with MCD after rhIGF-1 pretreatment. CaMKII expression is significantly reduced after rhIGF-1 pretreatment (1.43 ± 0.08 vs. 1.53 ± 0.09, p = 0.030). D Predictive schematic diagram of synaptic protein expression in infant rats with MCD after rhIGF-1 pretreatment. Based on western blot results, alterations in the cortical protein expression, including increase in PSD-95, NMDAR, and AMPAR subunits and decrease in CaMKII, are observed after rhIGF-1 pretreatment in the malformed brain only on symptom relief or surgical resection of the dysplastic cortex. To identify the epileptogenetic mechanism of MCD, we conducted studies on a rat model using MAM and found clinical phenotypes similar to those in patients with MCD [5,13]. Using this model, we aimed to determine the major disrupted pathway in the MCD cortex.
Quantitative proteome analysis revealed that synaptogenesis signaling was the most significantly downregulated  1, p = 0.002). B Although the onset of tailing and spasms was not different between rhIGF-1 pretreatment group and VEH controls (n = 6), the number of spasms is fewer in rhIGF-1 pretreatment group (n = 6, 17.3 ± 9.5 vs. 38.2 ± 13.3, p = 0.016, RM-ANOVA, F(1, 10) = 5.374, p = 0.043). C The fast oscillation-event-related spectral dynamics (FO-ERSP) is significantly decreased (4.43 ± 0.35 vs. 5.04 ± 3.29, p < 0.001, linear mixed model analysis) in rhIGF-1 pretreatment group (n = 5) compared to that in VEH group during spasms (n = 5). In addition, fast oscillationspectral entropy (FO-SE) is reduced in rhIGF-1 pretreatment group at inter-ictal and ictal periods compared to that in VEH (inter-ictal, 0.39 ± 0.04 vs. n = 5, 0.42 ± 0.02, p < 0.001; ictal, 0.38 ± 0.02 vs. 0.44 ± 0.02, p < 0.001, linear mixed model analysis). Although there is no difference in rhIGF-1 pretreatment group, the FO-SE and FO-ERSP increased in ictal periods than in inter-ictal periods in VEH group (FO-SE, 0.42 ± 0.02 vs. 0.44 ± 0.02, p < 0.001; FO-ERSP, 4.42 ± 0.28 vs. 5.04 ± 3.29, p < 0.001, linear mixed model analysis) canonical pathway in the malformed cortex of prenatal rats exposed to MAM at P15 (Fig. 1, Table S1), consistent with the decrease in CaMKIIA along with an increase in the susceptibility to NMDA-induced spasms [5]. Our previous work demonstrated the poor dendritic spine development and reduced neuronal population in the RSC of prenatally MAM-exposed rats [13]. Altered early synaptogenesis is reported in neurodevelopmental disorders, including epilepsy, intellectual disability, and autism spectrum disorders [30]. The neuronal connections of the CNS comprise both inhibitory and excitatory synapses where GABA and glutamate are involved in the major inhibitory and excitatory actions. In this experimental model, the second trimester of gestation when the fetus is exposed to MAM is the time of the formation of the excitatory synapses [31] and overexpression of the genes associated with neurodevelopmental disorder [32].
To confirm the changes in the excitatory/inhibitory synaptic development in rats with malformed cortex, the profiles of glutamate receptors, CaMKII, and PSD95 were examined. We observed a significant decrease in the expression of AMPAR2 and CaMKII in the MAM-exposed rat cortex at P15 (Fig. 2A, C). The expression level of CaMKII even decreases with rhIGF-1 treatment (Fig. 4C). CaMKII is abundantly found in the brain and plays a crucial role in synaptic plasticity and function, including synaptic spine formation [33]. Alterations in CaMKII activity and expression have been confirmed in various neuropsychiatric diseases [33]. In particular, the reduced activity of CaMKII is known to be related to epilepsy [33,34]. This decreased expression of cortical CaMKII in malformed brain may be related to the previously reported dendritic arborization [13] and increased spasm susceptibility [5]. However, this is inconsistent with our findings of reduction of CaMKII after rhIGF-1 treatment, and there is no clear consensus on the precise role that CaMKII plays in epilepsy [35]. Given the complex interaction between postsynaptic density proteins and other synaptic and channel proteins, downregulated CaMKII is not the only factor in excitability, but other changes in the synaptic proteins (such as PSD95, AMPAR1, AMPAR4, NMDAR1, and NMDAR2A) and their downstream signaling may be more important factors that modulate the excitability [36].
IGF-1, a member of the insulin-like peptides (ILPs) family, is a polypeptide that plays an essential role in early brain development. IGF-1 is produced in all cell types in the CNS and is involved in neuronal growth, polarity, maturation, and neuroplasticity [17,37]. However, the function of IGF-1 is ambivalent in relation to neurological diseases [19] and understudied. In a preclinical study, IGF-1 ameliorates some behavioral abnormalities in girls with Rett syndrome [38]. Some studies suggest that IGF-1 and IGF-1 signaling pose a risk of epilepsy with increasing seizure activity [19,39]. Furthermore, various effects of IGF-1 on synapses were reported [37,40] that IGF-1 application increases the AMPAR-mediated synaptic transmission and increases excitatory postsynaptic potentials (EPSP) [20] or increases the expression of NMDAR2A and NMDAR2B in addition to increasing the complexity of synapses [40][41][42]. However, most results were limited to aged rats with disorders other than epilepsy. In this study, we tried to the effect of rhIGF-1 on a wide range of aberrant signaling molecules, particularly synaptogenesis signaling which was the most severely affected canonical pathways in these young rats with prenatal MAM-exposures. In these rats with malformed cortices, early postnatal systemic rhIGF-1 treatment increased the expression of some subunits of AMPARs (AMPAR1 and AMPAR4) and NMDARs (NMDAR1 and NMDAR2A) as well as PSD-95 (Fig. 4B, C). The neurotransmission through these glutamate receptors takes an important role in synaptic connections of early postnatal life and the receptor composition of these receptors are highly regulated during development and the low AMPAR signaling during early period are associated with exclusive NMDAR2B expression and strong incorporation of AMPAR4 [43]. Therefore, the increased expression of these glutamate receptors and the reduction of NMDA-induced spasms in MCD rats with early postnatal rhIGF-1 treatment suggest that rhIGF-1 may promote the development of these malformed brain to an extent closer to normal rather than hyperexcitation (Fig. 5). In addition to AMPAR and NMDAR, increase in the expression of PSD-95, a post-synaptic density protein that promotes synaptic maturation [37], suggests that IGF-1 treatment during early developmental period can contribute to the modulation in synaptogenesis. These positive effects of IGF-1 supplementation contrasted with the increased IGF-1 signaling pathway in MCD-rats compared to controls at P15 (Table S1). However, a closer look at the most affected proteins in the IGF-1 signaling reveals that such as AKT, PIK3C3, PIK3CB, PIK3R1, PIK3R4, RAF1, RAP1A, RARA, and RRAS are also key molecules of the synaptogenesis signaling pathway and some of the were downregulated (Tables S1, S2). Considering diverse action of IGF-1, IGF-1 may affect neuronal excitability indirectly through the synaptogenesis signaling pathway, rather than directly acting on the IGF-1 signaling.
Moreover, we demonstrated the increase in NeuN expression in early rhIGF-1-treated rats (Fig. 4A), which is consistent with previous study in transgenic mice with IGF-1 overexpression [42] or in neuronal cell culture study observing neuronal growth and migration [44] or study of IGF-1 on hippocampal neurogenesis in old rats [45]. Despite these neuronal changes, there is no behavioral improvement, including short-term and long-term memories, in rhIGFtreated adolescent rats with MCD (Fig. S2) and increased neurogenesis or activation of IGF-1 signaling could not be shown in this study. In future research, the increased neurogenesis after IGF-1 treatment in these young animals should be evaluated.
To demonstrate the in vivo anti-seizure efficacy of rhIGF-1 pretreatment, we tested NMDA-induced spasm susceptibility after rhIGF-1 pretreatment or randomized treatment protocols using this infant rat model [5,13,14]. Both pretreatment or randomized treatment with rhIGF-1 could effectively reduce the number or delay the onset of spasms (Fig. 5A, B). EEG also supported the neuronal changes after rhIGF-1 pretreatment, as evident from the reduced FO in MCD infant rats (Fig. 5C). The FO plays a crucial role in the integration of neuronal networks and is related to the synchronized activation of interconnected excitatory pyramidal neurons and inhibitory interneurons [14,46]. In our previous study, we reported increased FO-ERSP, a time-related shift of the FO band frequency [14,47], consistent with an increase in seizure susceptibility in these MCD rats at P15 [5,14]. In the present study, rhIGF-1 pretreatment significantly reduced the ictal FO-ERSP that suggested attenuation of neuronal network dysregulation. Furthermore, the SE-FO significantly reduced in rats with rhIGF-1 pretreatment compared to that in rats subjected to VEH treatment during interictal and ictal periods. SE is a measure of the irregularity of neuronal network signals, and higher SE was reported in patients with drug-resistant epilepsy than in healthy controls [48]. Previous reports have shown that IGF-1 reduces excitatory post-synaptic currents and partially rescues immature synaptic functions in MeCP2 mutant mice [49] and tripeptide (1-3) IGF-1 abolishes spasms and EEG abnormality in tetrodotoxin model of infantile spasms [50]. Similarly, rhIGF-1 pretreatment could suppress the overwhelming pathologic FO in rats with MCD in this study.
After rhIGF-1 pretreatment in MAM-induced MCD rats, GSH level significantly decreased after rhIGF-1 pretreatment, and there were significant developmental changes in Cr and GSH concentration after rhIGF-1 pretreatment (Fig. 3B, C). GSH, a tripeptide composed of glutamate, glycine, and cysteine, is an antioxidant that protects cells from the damage caused by reactive oxygen species (ROS) [51][52][53]. Changes in GSH levels are known to be related to neurological disorders; in particular, reductions in GSH levels are closely associated with an increase in oxidative stress and are related to epilepsy [51,53]. Although there are the studies of GSH in epilepsy patients or patients with focal cortical dysplasia [53,54], the role of GSH in epileptic brain is unclear. Creatine (Cr) is a marker for energy metabolism [55,56], and recent studies have reported the elevation of Cr levels in malformed cortices of patients with epilepsy and suggests Cr as hypometabolic marker during inter-ictal period [55,56]. Thus, stabilization of cortical Cr and GSH after rhIGF-1 treatment may add evidence of GSH/Cr as a marker of neuronal stabilization in the MCD cortex. Epileptogenesis in MCD is intricately intertwined with the timing of insult, etiology, extent of disease, and patient's age [2]. In this model of MCD, which experiences a midgestation insult, synaptogenesis signaling was markedly disrupted. Early rhIGF-1 pretreatment could attenuate the spasms susceptibility induced by NMDA at P15, along with alterations of in synaptic protein expression. These results suggest that rhIGF-1 can potentially serve as a therapeutic agent in patients with MCD-associated epilepsy and may modulate early synapse formation, one of the main target pathways of epilepsy.